Citation
Renal Dysfunction after Surgery for Urologic Cancer and Other Diseases

Material Information

Title:
Renal Dysfunction after Surgery for Urologic Cancer and Other Diseases
Creator:
Bozorgmehri, Shahab
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Epidemiology
Committee Chair:
CANALES,MUNA THALJI
Committee Co-Chair:
BEYTH,REBECCA J
Committee Members:
COOK,ROBERT L
LU,XIAOMIN
GILBERT,SCOTT MICHAEL

Subjects

Subjects / Keywords:
bladder
cancer
cystectomy
kidney
urology

Notes

General Note:
Kidney disease is a costly and growing public health problem with several negative health consequences. Bladder cancer is common among men in the United States. Radical cystectomy (RC) and urinary diversion (UD), a surgical treatment for bladder cancer, may increase risk of overt renal dysfunction. However, studies to date have been uncontrolled, conflicting and did not identify key predictors of renal function decline after RC plus UD. Additionally, acute kidney injury (AKI) is common among hospitalized patients and is associated with substantial morbidity and mortality. However, the incidence, risk factors and renal outcomes of AKI are unclear in patients undergoing urologic procedures. In this retrospective study, we evaluated the effect of RC&UD for bladder cancer treatment on renal function, as measured by estimated glomerular filtration rate (eGFR) and identified the predictive factors for renal function decline among 384 bladder cancer patients who sought care in a tertiary health care center between 2000 and 2014. Additionally, we retrospectively examined 1,557 patients who underwent urologic procedures between 2000 and 2010 to identify the predictive factors for AKI and partial or no renal recovery after AKI. We found that bladder cancer patients who underwent RC&UD experienced a faster decline in renal function over time, as measured by eGFR slope, as compared to patients who did not undergo the procedure, independent of patient characteristics, co-morbid conditions, and medications use. Patients with RC&UD had higher rate of renal deterioration and odds of rapid decline in renal function in unadjusted analysis, but not after adjusting for confounding variables. Clinical cancer stage, race, chemotherapy, and certain medications were independently associated with faster renal function decline in patients with RC&UD. With respect to AKI after urologic surgery, we found that AKI occurred in 39% of patients following urologic procedures. Type of procedure, worse baseline eGFR, weekend admission, co-morbid conditions, and certain medications were independently associated with higher odds of AKI. Additionally, 23% had partial or non-recovery of renal function after AKI. Type of surgery, severity of AKI, increasing age and baseline eGFR were independently associated with higher odds of partial or non-recovery of renal function after AKI.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2018

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 RENAL DYSFUNCTION AFTER SURGERY FOR UROLOGIC CANCER AND OTHER DISEASES By SHAHAB BOZORGMEHRI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE D EGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016

PAGE 2

2 2016 Shahab Bozorgmehri

PAGE 3

3 To my mother and the memory of my father

PAGE 4

4 ACKNOWLEDGMENTS I thank my mother and si ster. I thank my disser tation comm ittee chair, Dr. Muna Canales, for her mentorship, supervision and guidance. I am sure it was equally exciting and challenging for her to help me finish this dissertation as it was for me. I also thank my disse rtation committee members, Drs Robert Cook S cott Gilbert Xiaomin Lu and Rebecca Beyth for their encouragement, valuable recommendations, and guidance during this enduring journey. I thank Dr. Azra Bihorac for her support Finally I thank all my friends and peers who supported me during this proce ss

PAGE 5

5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ ........ 11 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 18 Background and Significance ................................ ................................ ................. 18 Chronic Kidney Disease: Definition ................................ ................................ .. 18 Chronic Kidney Disease is Common, Costly, and Deadly ................................ 18 Risk Factors of Chronic Kidney Disease ................................ .......................... 19 Bladder Cancer Epidemiology ................................ ................................ .......... 20 Radical Cystectomy plus Ur inary Diversion for Bladder Cancer Treatment ...... 21 Acute Kidney Injury is Common, Costly, and Deadly ................................ ....... 22 Literature Review and Gaps in the Literature ................................ .......................... 23 Impact of Radical Cystectomy plus Urinary Diversion on Renal Function ........ 23 Predictors of Renal Function Decli ne after Radical Cystectomy plus Urinary Diversion ................................ ................................ ................................ ....... 26 Acute Kidney Injury after Urologic Surgery ................................ ....................... 27 Public Health Importance ................................ ................................ ........................ 28 Conceptual Framework ................................ ................................ ........................... 30 Source of Data ................................ ................................ ................................ ........ 30 2 EFFECT OF RADICAL CY STECTOMY PLUS URINARY DIVERSION FOR BLADDER CANCER TREATMENT ON RENAL FUNCTION ................................ 40 Background ................................ ................................ ................................ ............. 40 Methods ................................ ................................ ................................ .................. 41 Study Participants ................................ ................................ ............................ 41 Radical Cystectomy plus Urinary Diversion ................................ ...................... 41 Measurement of Ren al Function ................................ ................................ ...... 42 Covariates ................................ ................................ ................................ ........ 43 Statistical Analysis ................................ ................................ ............................ 44 Results ................................ ................................ ................................ .................... 47 Participants Characteristics ................................ ................................ .............. 47 Renal Function ................................ ................................ ................................ 48 Discussion ................................ ................................ ................................ .............. 51

PAGE 6

6 3 PREDICTORS OF RENAL FUNCTION DECLINE AFTER RADICAL CYSTECTOMY PLUS URINARY DIVERSION AMONG PATIENTS WITH BLADDER CANCER ................................ ................................ ............................... 77 Bac kground ................................ ................................ ................................ ............. 77 Methods ................................ ................................ ................................ .................. 78 Study Participants ................................ ................................ ............................ 78 Radical Cystectomy p lus Urinary Diversion ................................ ...................... 79 Measurement of Renal Function ................................ ................................ ...... 79 Covariates ................................ ................................ ................................ ........ 81 Statistical Analysis ................................ ................................ ............................ 82 Results ................................ ................................ ................................ .................... 84 Characteristics of Study Participants ................................ ................................ 84 Predictors of Renal Function Decline ................................ ............................... 86 Discussion ................................ ................................ ................................ .............. 91 4 ACUTE KIDNEY INJURY AFTER UROLOGIC SURGERIES ............................... 137 Background ................................ ................................ ................................ ........... 137 Methods ................................ ................................ ................................ ................ 138 Study Participants ................................ ................................ .......................... 138 Acute Kidney Injury (AKI) ................................ ................................ ............... 138 Renal Outcome after AKI ................................ ................................ ................ 139 Predictors ................................ ................................ ................................ ....... 139 Statistical Analysis ................................ ................................ .......................... 141 Results ................................ ................................ ................................ .................. 142 Characteristics of Patients ................................ ................................ .............. 142 Distribution of AKI ................................ ................................ ........................... 144 Predictors of AKI ................................ ................................ ............................ 144 Predictors of Partial or Non Recovery of Renal Function after AKI ................ 147 Discussion ................................ ................................ ................................ ............ 149 5 CONCLUSIONS ................................ ................................ ................................ ... 177 Summary of Findings ................................ ................................ ............................ 177 Discussion ................................ ................................ ................................ ............ 180 Future Directions ................................ ................................ ................................ .. 185 APPENDIX A ICD 9 CM DIAGNOSIS CODES FOR BLADDER CANCER ................................ 187 B PROCEDURE CODES IN MANAGEMENT OF BLADDER CANCER .................. 188 C ASSESSMENT OF PROPORTIONAL HAZARDS ASSUMPTION ....................... 189 D ICD 9 CM PROCEDURE CODES ................................ ................................ ........ 192

PAGE 7

7 LIST OF REFERENCES ................................ ................................ ............................. 193 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 205

PAGE 8

8 LIST OF TABLES Table page 1 1 GFR Categories in Chronic Kidney Disease ................................ ..................... 35 1 2 The American Joint Committee on Cancer (AJCC) TNM classificat ion of urinary bladder cancer ................................ ................................ ........................ 36 1 3 Treatment recommendatio n s for bladder cancer ................................ .............. 37 1 4 Risk, Injury, Failure, Loss, and End stage Kidne y (RIFLE) classification .......... 39 2 1 Sociodemographi c and Clinical Characteristics of Patients with Bladder Cancer between 2000 and 2014 ................................ ................................ ......... 59 2 2 Co morbid Conditions of Patients with Bladder Cancer by Radical Cystectomy plus Urinary Diversion Group between 2000 and 2014 ................... 60 2 3 Clinical stage of bladder cancer, the AJCC TNM classification, by radical cystectomy plus urinary diversion group between 2000 and 2014 ..................... 61 2 4 Medications Use in Patients with Bladder Cancer by Radical Cystectomy plus Urinary Diversion Group ................................ ................................ .............. 63 2 5 Association between radical cystectomy plus urinary diversion with chang e in MDRD eGFR over time ................................ ................................ .................. 64 2 6 Association between radical cystectomy plus urinary diversion and time to per 30% ................................ ............................. 66 2 7 Association between radical cystectomy plus urinary diversion and rapid decline in renal function (MDRD eGFR >3 ml/min/1 73m 2 /year) ........................ 68 3 1 Sociodemographic and clinical characteristics of bladder cancer patients with radical cystectomy by type of urinary diversion between 2000 and 2014 .... 98 3 2 Co morbid cond itions of patients with radical cystectomy according to the type of urinary diversion between 2000 and 2014 ................................ .............. 99 3 3 Clinical stage of bladder cancer, the AJCC TNM classification, of patients with radical cystectomy according to the type of urinary diversion between 2000 and 2014 ................................ ................................ ................................ .. 100 3 4 Medications use in patients with radical cystectomy according to the type of urinary diversio n between 2000 and 2014 ................................ ........................ 102 3 5 Association between type of urinary diversion and change in MDRD eGFR with time using ................................ ................................ ................................ 104

PAGE 9

9 3 6 Association between clinical variables and change in MDRD eGFR with time in patients with radical cystectomy plus urinary diversio n ................................ 105 3 7 Multivariable linear mixed regression model pr edicting change in MDRD eGFR with time in patients with radical cystectomy plus urinary diversion ....... 114 3 8 Association between clinical variables and time to percent change in MDRD 30% in patients with radical cystectomy plus uri nary diversion ............ 118 3 9 Association between clinical variables and time to percent change in MDRD 30% in patients with radical cys tectomy plu s urinary diversion ............ 121 3 10 Association between clinical variables and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) in patients with radical cystectomy plu s urinary div ersion ................................ ................................ ....................... 122 3 11 Association between clinical variables and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) in patients with radical cystectomy plus urinary diversion ................................ ................................ ....................... 125 4 1 Risk, Injury, Failure, Loss, and End stage Kidney (RIFLE ) classification of AKI ................................ ................................ ................................ ................... 154 4 2 Sociodemographic and clinical c haracteristics of patients according to urologic procedure group between 2000 and 2010 ................................ .......... 155 4 3 Co morbid conditions of patients according to urologic procedure group between 2000 and 2010 ................................ ................................ ................... 157 4 4 Medication use in patients according to urologic procedure group between 2000 and 2010 ................................ ................................ ................................ .. 159 4 5 Frequency and distribu tion of AKI in patients according to urologic procedure group between 2000 and 2010 ................................ ................................ ......... 161 4 6 Association between predictive factors and AKI after urologic procedures ..... 162 4 7 Association between predictive factors and AKI after ur ologic procedures ..... 164 4 8 ROC analysis of models to predict AKI after urologic procedu res ................... 165 4 9 Frequency and distribution of renal outcome after AKI in patients with AKI according to urologic procedure groups ................................ ........................... 166 4 10 Association between predictive factors and partial recovery or non recovery of renal function following AKI in patients with urologic proce dures ................. 167 4 11 Association between predi ctive factors and partial recovery or non recovery of renal function following AKI in patients with ur ologic procedures ................. 169

PAGE 10

10 4 12 ROC analysis of models to predict partial recovery or non rec overy of renal function following AKI in patients with urologic procedures .............................. 170

PAGE 11

11 LIST OF FIGURES Figure page 1 1 The isotope dilution mass spectro metry (IDMS) traceable the Modification of Diet in Renal Disease (MDRD) study equation ................................ ................... 32 1 2 Types of ur inary diversion ................................ ................................ ................. 33 1 3 Conceptual framework ................................ ................................ ...................... 34 2 1 Plot of predicted mean of MDRD eGFR over time by radical cystectomy plus urinary diversion group ................................ ................................ ....................... 70 2 2 Plot of predicted mean of MDRD eGFR over time by radical cystectomy plus urinary diversion group ................................ ................................ ....................... 71 2 3 Kaplan Meier survival curves of renal deterioration free ( 30%) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion ................................ ................................ .......................... 72 2 4 Kaplan Meier survival curves of renal deterioration free ( 2 0%) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion ................................ ................................ .......................... 73 2 5 Kaplan Meier survival curves of renal deterioration free ( 25%) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion ................................ ................................ .......................... 74 2 6 Kaplan Meier survival curves of renal deterioration free ( 40%) survival of bladder ca ncer patients with and without radical cystectomy plus urinary diversion ................................ ................................ .......................... 75 2 7 Kaplan Meier survival curves of renal deterioration free ( 57%) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion ................................ ................................ .......................... 76 3 1 Plot of predicted mean of MDRD eGFR over time in bladder cancer patients according to the type of urinary diversion ................................ ......................... 126 3 2 Plot of predicted mean of MDRD eGFR over time in bladder cancer patients according to the type of urinary diversion ................................ ......................... 127 3 3 Plots of predicted mean of MDRD eGFR over time in bladder cancer patients with radical cystectomy plus urinary diversion according to (A) gender, (B) smoking status, (C) diabetes mellitus, and (D) chemotherapy ....... 128

PAGE 12

12 3 4 Plots of predicted mean of MDRD eGFR over time in bladder cancer patients with radical cystectomy plus urinary diversion according to (A) angiotenesin II receptor blockers, (B) diuretics, (C) platinum based antineoplastic, and (D) statins ................................ ................................ .......... 129 3 5 Kaplan Meier survival curves of renal deterioration 30) survival of bladder cancer patients according to the type of urinary diversion .. 130 3 6 Kaplan Meier survival curves of renal deterioration 20) survival of bladder cancer patients according to the type of urinary diversion .. 131 3 7 Kaplan Meier survival curves of renal deterioration 25) survival of bladder cancer patients according to the type of urinary diversion .. 132 3 8 Kaplan Meier survival curves of renal deterioration 40) survival of bladder cancer patients according to the type of urinary diversion .. 133 3 9 Kaplan Meier survival cu rves of renal deterioration 57) survival of bladder cancer patients according to the type of urinary diversion .. 134 3 10 Cumulative hazards curves of renal deterioration (MD 30) in patients with radical cystectomy plus urinary diversion according to predictive factors ................................ ................................ ................................ ............... 135 3 11 ROC analysis of the fitted logistic regression model to predict rapid decline in renal function (MDRD eGFR > 3 ml/min/1.73m 2 /year) in patients with radical cystectomy plus urinary diversion ................................ ........................ 136 4 1 Performance of the fitted models to predict AKI after after urologic procedures using ROC analysis ................................ ................................ ....... 171 4 2 Cross validation of the fitted model (Model 4) to predict AKI after urologic procedures using ROC analysis ................................ ................................ ....... 173 4 3 Performance of the fitted models to predict partial recovery or non recovery of renal function following AKI in patients with urologic procedures using ROC analysis ................................ ................................ ................................ .... 174 4 4 Cross validation of the fitted model (Model 4) to predict partial recovery or non recovery of renal function following AKI in patients with urologic procedures using ROC analysis ................................ ................................ ....... 176

PAGE 13

13 LIST OF ABBREVIATION S ACE Angiotensin Converting Enzyme AKI Acute kidney injury ANOVA One Way Analysis of Variance ARBs Angiotensin II Receptor Blockers ARF Acute Renal Failure ASA Acetylsalicylic Acid AUC Area Under the Curve BMI Body Mass Index CHF Congestive He art Failure CI Confidence Interval CKD Chronic Kidney Disease CKD EPI Chronic Kidney Disease Epidemiology Collaboration CPT Current Procedural Terminology CTSI Clinical and Translational Science Institute DM Diabetes Mellitus eGFR Estimated Glom erular Filtration Rate ESRD End Stage Renal Disease FCS Conditional Specification FORDS Facility Oncology Registry Data Standards HTN Hypertension ICD 9 CM International Classification of Disease, Ninth Revision, Clinical Modification

PAGE 14

14 IDMS Isoto pe Dilution Mass Spectrometry IDR Integrated Data Repository IVs Instrumental Variables M Metastasis MAR Missing at Random MDRD Modification of Diet in Renal Disease MI Myocardial Infarction MIBC Muscle Invasive Bladder Cancer N Node NCI Nati onal Cancer Institute NMIBC Non Muscle Invasive Bladder Cancer OR Odds Ratio PVD Peripheral Vascular Disease QOL Quality of Life RC Radical cystectomy RCT Randomized Controlled Clinical Trial RIFLE R renal risk, I injury, F failure, L loss of ki dney function, and E end stage renal disease RRT Renal Replacement Therapy SCr Serum Creatinine SES Socioeconomic status T Tumor TMP SMX Trimethoprim Sulfamethoxazol

PAGE 15

15 TURBT Transurethral Resection of The Bladder Tumor UD Urinary Diversion UTI Urin ary Tract Infection

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RENAL DYSFUNCTION AFTER SURGERY FOR UROLOGIC CANCER AND OTHER DISEASES By Shahab Bozorgmehri December 2016 Chair: Muna T. Canales Major: Epidemiology Kidney disease is a costly and growing public health problem with several negative health consequences. Bladder cancer is common among men in the U nited States. Radical cystectomy (RC) and urinary diversion (UD), a surgical treatment for bladder cancer, may increase risk of overt renal dysfunction. However, studies to date have been uncontrolled, conflicting and did not identify key predictors of ren al function decline after RC plus UD. Additionally, acute kidney injury (AKI) is common among hospitalized patients and is associated with substantial morbidity and mortality. However, the incidence, risk factors and renal outcomes of AKI are unclear in pa tients undergoing urologic procedures. In this retrospective study, we evaluated the effect of RC&UD for bladder cancer treatment on renal function, as measured by estimated glomerular filtration rate (eGFR) and identified the predictive factors for renal function decline among 384 bladder cancer patients who sought care in a tertiary health care center between 2000 and 2014. Additionally, we retrospectively examined 1,557 patients who underwent urologic procedures between 2000 and 2010 to identify the pre dictive factors for AKI and partial or no renal recovery after AKI.

PAGE 17

17 We found that bladder cancer patients who underwent RC&UD experienced a faster decline in renal function over time, as measured by eGFR slope, as compared to patients who did not undergo the procedure, independent of patient characteristics, co morbid conditions, and medications use. Patients with RC&UD had higher rate of renal deterioration and odds of rapid decline in renal function in unadjusted analysis, but not after adjusting for co nfounding variables. C linical cancer stage, race, chemotherapy, and certain medications were independently associated with faster renal function decline in patients with RC&UD. With respect to AKI after urologic surgery, we found that AKI occurred in 39% o f patients following urologic procedures. Type of procedure, worse baseline eGFR, weekend admission, co morbid conditions, and certain medications were independently associated with higher odds of AKI. Additionally, 23% had partial or non recovery of renal function after AKI. Typ e of surgery, severity of AKI, in creasing age and baseline eGFR were independently associated with higher odds of partial or non recovery of renal function after AKI.

PAGE 18

18 CHAPTER 1 INTRODUCTION Background and Significance Chronic Kid ney Disease: Definition Kidney or renal function is defined by glomerular filtration rate (GFR). The gold standard method for measurement of GFR is the use of exogenous filtration markers such as inulin, 99mTc DTPA, 51Cr EDTA, iohexol, or 125I iothalamate (Endre, Pickering, & Walker, 2011; Stevens, Coresh, Greene, & Levey, 2006) However, since accurate measurement of GFR using these methodologies is cumbersome and costly, in clinical practice or research surrogate m easures such as serum creatinine based estimation equations are commonly used (Levey & Coresh, 2012) Therefore, GFR estimation equations have been developed to easily assess renal function with use of a single blood test, the serum creatinine. The most widely used formula for estimating GFR is the Modification of Diet in Renal Dis ease (MDRD) study equation that incorporates serum creatinine (SCr), age, gender, a nd race (Figure 1 1 ) (Levey et al., 1999; Stevens et al., 2006) (Levey et al., 2006) Chronic Kidney Disease (CKD) is a general term for a diverse group of disorders affecting kidney function and structure (Levey & Coresh, 2012) CKD is defined based on decreased renal function for more than three months or/and the presence of kidney damage (Levey & Coresh, 2012) On the basis of eGFR CKD is classified into five stages regardles s of its underlying cause ( Table 1 2 ) (Levey, Coresh, Balk, Kausz, Levin, & Steffes, 2003) Chronic Kidney Disea se is Common, Costly, and Deadly CKD is a worldwide public health problem with an increasing incidence and prevalence (Levey, Coresh, Balk, Kausz, Levin, & Steffes, 2003) CKD, as defined by

PAGE 19

19 MDRD eGFR < 60 ml/min/1.73m 2 affects over 16 million individuals in the United States (Coresh et al., 2007) CKD is associated with numerous complications including anemia, bone disease, and hypertension (Eckardt et al., 2013) Furthermore, CKD negatively impacts quality of life (QOL) and independently increases the risk of cardiovascular disease, hospitalization, and death (Go, Chertow, Fan, McCulloch, & Hsu, 2004) For example, a meta analysis of fourteen studies of general popula tion cohorts showed that increasing severity of CKD was independently associated with increasing risk of all cause mortality: adjusted hazard ratio of death for eGFR at 60, 45, and 15 ml/min/1.73m 2 when compared to eGFR at 95 ml/min/1.73m 2 was 1.18 (95%C I: 1.05 1.32), 1.57 (95%CI: 1.39 1.78), and 3.14 (95%CI: 2.39 4.13), respectively (Matsushita et al., 2010) Several studies indicated impairment of QOL in patients with CKD. For example, a cross sectional study of 9 67 outpatients found that patients with eGFR<60 ml/min/1.73m 2 was associated with worse overall health (OR:1.65; 95%CI: 1.21 2.24), despite multivariate adjustment (Odden, Whooley, & Shlip ak, 2006) Given the numerous negative health consequences of CKD, the cost of caring for CKD patients in the United States is staggering; CKD patients incur per person per year costs of $22,348 before dialysis and $87,945 on dialysis (i.e. end stage rena l disease) ( National Institutes of H ealth 2013) Risk Factors of Chronic Kidney Disease It is important to recognize that several factors might increase the risk of CKD For example, in a community based longitudinal cohort study of 2,585 participants, the Framingham Offspring Study, i t was shown that age, baseline GFR, body mass index (BMI), smoking, diabetes, and hypertension were associated with and increased risk of new onset kidney disease (Fox et al., 2004) The other established risk factor s for CKD

PAGE 20

20 include African American (Freedman, Soucie, & McClellan, 1997) Nat ive America n (Pettitt, Saad, Bennett, Nelson, & Knowler, 1990) or Mexican American race (Tareen et al ., 2005) family history of kidney disease (Freedman et al., 1997; Jurkovitz, Franch, Shoham, Bellenger, & McClellan, 2002) heavy consumption of alcohol (Shankar, Klein, & Kle in, 2006) metabolic syndrome (Tanner, Brown, & Muntner, 2012) low socioeconomic status (McClellan et al., 2010) acute kidney injury ( Chawla, Eggers, Star, & Kimmel, 2014; Heung & Chawla, 2012) history of kidney stones (Vupputuri, Soucie, McClellan, & Sandler, 2004) low birth weight (Vikse, Irgens, Leivesta d, Hallan, & Iversen, 2008) and cancer chemotherapy (Ries & Klastersky, 1986) Bladder Cancer Epidemiology Ca ncer is the second leading cause of death in the United States (Yoon, Bastian, Anderson, Collins, & Jaffe, 2014) Bladder cancer is the fourth most common cancer in men in the United States (Siegel, Ma, Zou, & Jemal, 2014) It is estimated that approximately 2.7 million people have a histo ry of bladder cancer worldwide with the incidence rates and subsequent burde n of the disease being highest in developed countries (Ploeg, Aben, & Kiemeney, 2009) Bladder cancer is three to four times more common in men than in women (Kirkali et al., 2005) In 2014 alone, it is estimated that 74,690 new cases of bladder cancer were diagnosed in the United States, with 56,390 and 18,300 in men and women, respectively (Siegel et al., 2014) According to National Cancer Institute (NCI), there has been a statistically significant increasing trend in bladder cancer incidence rates from 1975 t o 2010 (Howlader, 2013) .Despite remarkable advancement in diagnosis and treatment of bladder cancer in recent decades, the morbidity and mortality from bladder cancer remain substantial. It is estimated that

PAGE 21

21 11,170 m en and 4,410 women will die due to bl adder cancer in the United States in 2014 (Siegel et al., 2014) The relative 5 year survival rate of bladder cancer ranges from 15% to 88% depending upon stage at diagnosis (ACS, 2014) Radical Cystectomy plus Urinary Diversion for Bladder Cancer Treatment The tumor, node, metastasis (TNM) classi fication system is used for treatmen t recommendations (Table 1 2 ). There are several treatment modalities available chemotherapy, radiation therapy, immunotherapy, trans urethral resection of the bladder tumor (TURBT), partial cystectomy and radical cystectomy (RC) plus uri nary diversion (UD) ( Table 1 3 ; Figure 1 2 ). The RC plus UD has long been gold standard treatment for muscle invasive bladder cancer (MIBC) and for tre atment failure in non muscle invasive bladder cancer (NMIBC) (Babjuk et al., 2013; Stein et al., 2001; Witjes et al., 2014) A radical cystectomy is an operation to remove the entire bladder, as well as the surroun ding lymph nodes. RC typically includes removal of the prostate and seminal vesicles (in males) or removal of the uterus, ovaries and part of the vagina (in females). A urinary diversion is any one of the several surgical procedures to reroute urine flow f rom its normal pathway. The common types of UD include ileal conduit, Indiana pouch reservoir, and orthotopic neobladder ( Figure 1 2 ). An ileal conduit urinary diversion is a surgical procedure where a segment of the intestine is used to create a tube to d irect urine through a stoma to an external collecting bag (urostomy bag). An Indiana pouch reservoir is a surgically created urinary diversion where a small reservoir is made out of portions of intestine in order to store urine until it is drained via a ca theter inserted through the stoma. An orthotopic neobaldder is a surgically created urinary diversion

PAGE 22

22 where a bladder like reservoir is made out of portions of intestine and connected to urethra, which allows patient to urinate. Acute Kidney Injury is Com mon, Costly, and Deadly Acute kidney injury (AKI) is a common clinical condition, frequently encountered in hospitalized patients, and is associated with a significantly increased risk of morbidity and mortality (Uchino et a l., 2005) AKI during hospitalization, even with small changes in SCr level, is independently associated with long term risk of death (Bihorac et al., 2009) The incidence of AKI has been increasing, presumably as a result of the expansion of invasiv e medical and surgical procedures (Xue et al., 2006) The incidence of postoperative AKI ranges from 16.7% to 30% in surgical patients (Case, Khan, Khalid, & Khan, 2013) The RIFLE criteria (R renal risk, I injury, F failure, L loss o f kidney function, and E end stage renal disease (Bellomo et al., 2004) Table 1 4 ), a cens us classification for AKI, represent a widely accepted tool for detection and classification of AKI, and independently correlate with clinical outcomes (Bagshaw, George, Dinu, & Bellomo, 2008) AKI related mortality in hos pitalized patients ranges from 15% to 41%, depending upon the RIFLE category, with odds ratios of 2.5, 4.5, and 5.4 for risk, injury, and failure, respectively, when compared to no AKI (Uchino, Bellomo, Goldsmi th, Bates, & Ronco, 2006) Very few studies of post operative AKI focused on urologic surgeries were limite d by reliance on SCr alone, did not address the predictive factors, and lack ed follow up to see the full outcome of AKI (i.e. complete recovery, par tial recovery, and non recovery) (Campbell, Novick, Streem, Klein, & Licht, 1994; Gratzke et al., 2009) The latter is important because the outcomes of AKI include three recognized courses: increased short and lon g term mortality, survival with full recovery from AKI, and survival with progressive CKD (Ishani et al., 2009; Schmitt et al., 2008)

PAGE 23

23 Literature Review and Gaps in the Literature Impact of Radical Cystectomy plus Urinary Diversion on Renal Function RC plus UD offers a reasonable possibility of disease control at 5 years, with a disease specific survival of 74% (Manoharan, Ayyathurai, & Soloway, 2009) However, RC plus UD is associated with early and late complications in 44% and 51% of patients respectively (Nieuwenhuijzen et al., 2008) Patients with urinary diversion are at risk of several complications affecting urinary tract and kidney such as reservoir uretero reflux, hydronephrosis, kidney sto ne formation, and pyelonephritis (Hautmann, Volkmer, Schumacher, Gschwend, & Studer, 2006; Madersbacher et al., 2003; Pycha et al., 2008) Most importantly, it has been suggested that urinary diversion increases the risk of developing overt renal dysfunction (Kristjansson & Mansson, 2004) Thus far, a few studies have investigated the impact of UD for bladder cancer and non cancer disease states on renal function. However, these studies have yielde d conflicting results and were limited by at least one of the followings: inadequate ascertainment of the renal outcome, small sample size, heterogeneous population (i.e. cancer and non cancer patients), variable follow up time, lack of repeated measures o f renal function, loss to follow up, and none had a control group Also, some studies examined only one type of UD. For example, Samuel et al. (Samuel, Bhatt, Montague, Clarke, & Ramani, 2006) assessed long term follow up of renal function measured by isotopic GFR, after ileal conduit UD for bladder cancer in 174 patients and reported 29% of patients had a deteriorating renal function defined by a persistent trend lower with a d ecrease in the GFR of 5% or greater from the baseline postoperative GFR, with a mean follow up of 8.2 years. Another study reported renal failure requiring dialysis as a long term (>90 days after surgery) complication in 6 out of 130 patients with inconti nent

PAGE 24

24 UD for bladder cancer (Pycha et al., 2008) These two studies (Pycha et al., 2008; Samuel et al., 2006) examined only incontinent UD and did not examine continent cutaneous reservoir or orthotopic neobladder. In a retrospective study of 31 patients with blad der cancer, Lantz el al. (Lantz, Saltel, & Cagiannos, 2010) assessed renal function following cystectomy and neobladder reconstruction and reported statistically significa nt increase in SCr levels by 12 months af ter surgery. However, this study was limited by use of SCr to estimate renal function. SCr alone is not an accurate estimate of renal function due to various limitations (Leve y et al., 2006) In a retrospective study of 50 patients with ileal conduit diversion and 111 patients with orthotopic neobladder substitution, Jin et al. (Jin et al., 2012) reported deterioration of renal function after 10 years in 36% and 21% of patients who underwent ileal conduit urinary diversion and orthotopic neobladder, respectively. However, th e authors investigated a heterogeneous population and included only those who lived with a UD excluding those who died within 10 years after the UD from the analysis Another retrospective study of 70 patients with RC plus UD, reported that mean eGFR decreased from 74.6 ml/min/1.73m 2 before surgery to 63.9 ml/min/1.73m 2 and 34% of patients showed reduced renal function, defined by a greater than 25% decrease in eGFR from the baseline, with a median follow up period of 34.5 months (Osawa et al., 2013) A recent study of a 5% Medicare sample (n=1,565), consisting of 66% and 34% patients with and without bladder can cer, respectively, reported that at 5 years after UD 16% of the patients experienced re nal failure or impairment, defined according to ICD 9 diagnosis and procedure codes as well as HCPCS service codes for ESRD, acute kidney failure, dialysis, and kidney transplantation (Gilbert, Lai, Saigal, Gore, & Project, 2013) In a

PAGE 25

25 retrospective stud y of a heterogeneous group of 126 patients with continent UD, Jonsson et al. (Jo nsson, Olofsson, Lindholm, & Tornqvist, 2001) examined long term results after UD and reported a significant decline in renal function; however, the authors suggested that decrease in renal function could be due to the aging process. While aforementioned studies (Gilbert et al., 2013; Lantz et al., 2010; Osawa et al., 2013; Pycha et al., 2008; Samuel et al., 2006) reported renal function deterioration after RC plus UD, a few others reported no change in renal functio n (Perimenis, Burkhard, Kessler, Gramann, & Studer, 2004; Thoeny, Sonnenschein, Madersbacher, Vock, & Studer, 2002) For example, Perimenis et al. (Perimenis et al., 2004) conducted a study of 129 men who survived mor e than 5 years after RC plus UD and reported that the mean creatinine levels remained unchanged. In another study, Thoeny et al. (Thoeny et al., 2002) reported preoperatively mean (SD) SCr level was 98(19) mol/l and 10 years thereafter it was 83(27) mol/l in 76 pa tients with orthotopic neobladder with a median follow up of 84 months. A nother study (Canter et al., 2011) reported 11% and 14 % increase in eGFR and creatinine clearance, resp ectively, at 3 months after surgery in 194 bladder cancer patients Given the context of present literature, the impact of RC plus US for bladder cancer on renal function over time is unclear. This study ad d s to t he literature in that it is the first to examine the impact of RC plus UD on renal function over time using all of the following: 1) large sample size; 2) control group; 3) extended follow up; 4) homogeneous population; 5) repeated measures of renal function over time ; and 6) use of three different definitions of renal function decline.

PAGE 26

26 Predictors of Renal Function Decline after Radical Cystectomy plus Urinary Diversion To date only few studies have investigated the predictive factors of renal func tion decline after RC plus UD for bladder cancer; however, these studies were limited b y at least one of the following : inadequate ascertainment of the renal outcome, small sample size, heterogeneous population, variable follow up time, lack of repeated me asures of renal function, and loss to follow up. For example, in a retrospective study of 174 cancer patients, Samuel et al. indicated an initial postoperative GFR<50, recurrent sepsis, and hypertension were predictors of deteriorating renal function defi ned by a persistent trend lower with a decrease in the GFR of 5% or greater from the baseline postoperative GFR, with a mean follow up of 8.2 years (Samuel et al., 2006) However, only patients with ileal conduit ur inary diversion were examined (Samuel et al., 2006) In another retrospective study, Jin et al. examined long term renal function outcomes after urinary diversion in a heterogeneous population (i.e. bladder cancer, shrunken bladder, hemorrhagic cystitis, and neurogenic bladder) and identified hypertension, diabetes, and obstruction as risk factors for renal function deterioration (Jin et al., 2012) In a retrospective study of 70 patients with bladder cancer, Osawa et al. indicated that postoperative acute pyelonephritis and chemotherapy were associated with renal deterioration defined by a decrease in the eGFR of 25% or greater, after RC plus UD (Osawa et al., 2013) In the 2nd study of this dissertation, we identify predictors of renal function decline among bladder ca ncer patients undergoing RC plus UD. For this project, we leverage a rich clinical database with extensive longitudina l follow up and all of th e following: assessment of different types of urinary diversion, large sample size group,

PAGE 27

27 extended follow up, homogeneous population, and repeated measures of renal function over time to address this gap Acute Kidney Injury after Urologic Surgery Little is known with respect to frequency, risk factors, and renal outcome of AKI after urologic surgeries. The majority of studies that have investigated incidence and risk factors of postoperative AKI examined populations other than ur ologic surgery (e.g. cardiothoracic surgery or ge neral surgery). For example, a retrospective study of 11,080 adult surgical patients (Bihorac et al., 2009) reported that the frequency of RIFLE defined AKI was 10%, 20%, 28%, and 42% in neurosurgery, vascular surgery, general surgery, and cardiothoracic surgery, respectively. Few studies examined AKI in patients with urologic cancers unde rgoing surgery. For example, a prospective study of 117 patients with RCC (G ratzke et al., 2009) reported SCr values increased 1 week after surgery in 3.8, 4.6, and 8.1% of patients after open radical nephrectomy, retroperitoneoscopic radical nephrectomy, and partial nephrectomy, respectively. In a study of 259 patients who had u ndergone partial nephrectomy for renal tumors, Campbell et al. (Campbell et al., 1994) reported tha t 33 patients experienced acute renal failure (ARF) of whom 9 required dialysis. The authors added that predisposing factors for ARF were t umor size>7cm, parenchymal excision >50% and ischemia time >60 min A review of the existing literature indicates that few studies investigated the incidence and predictive factor of AKI in urologic surgery, particularly for patients with urologic cancers. Given the fact that AKI is associated with high mortality, increased costs of care, prolonged hospitalization, and in creased likelihood of CKD, it is important to examine incidence and predictive factors of AKI in patients with urologic surgery.

PAGE 28

28 In the 3rd study of this dissertation, we investigate the incidence and predictive factors, and renal outcome of acute kidney injury within 30 days after urologic surgeries. The novel key points of our proposed research include: 1) assessing frequency and predictive factors for acute kidney injury in urologic surgery, 2) assessing renal outcomes after acute kidney injury episode, and 3) large sample size. Public Health Importance It is well recognized that the current trend of an aging population corresponds to increasing burden of chronic diseases and cancer. Patients with bladder cancer may benefit from advancement in variety of treatment modal ities and improved overall survival. However, the possibility of developing renal impairment following treatment can severely complicate the course of disease and limit survival. CKD is a pressin g public health problem with increasing in cidence and prevalence rates (Levey, Coresh, Balk, Kausz, Levin, Steffes, et al., 2003) CKD is associated with numerous complications, negatively impacts QOL and increases t he risk of cardiovascular disease, hospitalization, and death (Go et al., 2004) CKD is costly an d imposes significant burdens on the healthcare system. In fact, the combined CKD and ESRD population are associated with twenty four percent of the Medicare budget ( National Institutes of Health 2013) In addition, CKD limits patient s chemotherapy (Canter et al., 2011) The first study of this dissertation will determine whether RC plus UD for bladder cancer leads to renal function decline w hen compared to controls with bladder cancer who do not undergo RC plus UD. The second study will go further to determine predictive factor s for renal function decline among those who undergo RC plus UD for bladder cancer. This insight will increase patien

PAGE 29

29 regarding risk of renal impairment after RC plus UD for bladder cancer and key factor s that underlie this risk. Positive results from this study will be to highlight the importance of careful monitoring of renal func tion after RC plus UD for bladder cancer. With increased awareness of renal risk associated with RC plus UD and underlying risk factors, additional care may be taken to minimize renal risk and related complications. In addition, this work will inform decis ion making process and identify patients in need of closer monitoring for renal impairment after surgery. For example, using information from this study, modifiable risk factors for renal impairment (e.g. obesity) can be addressed before and after the oper ation or throughout the course of treatment, limited exposure to nephrotoxic drugs or drug adjustments can be implemented Furthermore, counseling of the patient can be performed regarding the risks and benefits of the procedure, and patients at higher ris k of developing renal dysfunction can be monitored closely using SCr and urine tests. Ultimately, prevention of CKD after RC plus UD for bladder cancer will reduce cost of care and negative health consequences related to CKD. Identification of predictive factors of acute kidney injury and renal outcome among patients who undergo urologic surgery is of utmost importance owing to the fact that acute kidney injury is independen tly associated with short term and long term mortality, a 3.5 day increase in hospi tal length of stay, and approximately $7,500 in excess hospital costs (Chertow, Burdick, Honour, Bonventre, & Bates, 2005) The results from this study will advance the field by understanding what factors increase the risk of acute renal injury among patients who undergo urologic surgery. This insight can guide the decision making pr ocess with respect to the management of this patient

PAGE 30

30 population in clinical settings. In addition, this study will inform patients and providers about risk and predictive factors of AKI and partial or non recovery following AKI in this population. Thus, mo difiable pre, intra or post operative factors may be addressed to help prevent development of AKI and related negative health outcomes. For example, hemodynamic optimization, avoidance of nephrotoxic drugs, or appropriate drug adjustment can be implement ed, particularly in patients who are at higher risk for developing post operative AKI. Conceptual Framework It order to understand the relationship between predictive factors and renal dysfunction in bladd er cancer patients who undergo urinary diversion an d account for potential confounders, we have developed a conceptual framework (Figure 1 3 ). This c onceptual model indicates various factors that may have an impact on outcome of RC plus UD chemoth erapy) may affect renal function. Urinary diversion may also increase the risk of urinary tract infections and urinary stones which consequently increase the risk of renal dysfunction. enal function. in a retrospective study design, but nonetheless they are important and needed to be considered in a conceptual framework. Source of Data For the first and second studies, examining renal function decline after UC plus UD, w e utilize d both UF Health Shands Hospital integrated data repository (IDR) and UF Oncology Data Center as our data source s The UF Health IDR, supported by UF Clinical and Transl ational Science Institute (CTSI) and Shands Healthcare, is a large

PAGE 31

31 and research enterprises. The UF Health IDR contains information on various inpatients and outpatient clinical encounter data, diagnoses, demographics, procedures, lab results, co mo r bi dity measures, and medications. Some essential data elements, such as cancer staging data, are obtained using the UF Oncology Data Center. The UF Oncology Data Center colle cts data on all cancer cases diagnosed at UF Health Shands Hospital. The process of compi ling the data for this study was as follows: after obtaining the UF IRB approv al, the UF Health IDR team acted as an honest broker and provide d us with requested data set. In the next s tep, cancer staging data were added by linking to medical record numbers by the UF Oncology Data Center. For the 3 rd study of predictors of AKI after urologic surgeries the data source is an existing comprehensive database created by in tegrating the administrative, clinical, pharmacy, laboratory, and surgery data at UF Health Shands Hospital. First, adult surgical patients were identified using administrative data. The administrative data phics admission type and service, discharge disposition, and costs. Additionally, using the administrative data, for each admission, information was obtained on up to fifty medical conditions and fifty procedures according to the International Classificat ion of Disease, Ninth Revision, Clinica l Modification (ICD 9 CM) codes. The administrative data were linked to laboratory data and pharmacy data using the medical record number. The final comprehensive database allow ed us to examine the relevant lab result s such as SCr measurements.

PAGE 32

32 Figure 1 1. The isotope dilution mass spectrometry (IDMS) traceable the Modification of Diet in Renal Disease (MDRD) study equation

PAGE 33

33 Figure 1 2 Types of u rinary d iversion (Cleveland, 2013) Reprint ed with permission, Cleveland Clinic Center for Medical Art & Photography 2016. All Rights Reserved.

PAGE 34

34 Figure 1 3 Conceptual framework

PAGE 35

35 Table 1 1 GFR Categories in Chronic Kidney Disease (Inker et al., 20 14) GFR Category GFR, mL/min per 1.73 2 Terms G1 Normal or high G2 60 89 Mildly deceased (relative to young adult level) G3a 45 59 Mildly to moderately decreased G3b 30 44 Moderately to severely decreased G4 15 29 Severely decreased G5 <15 (or dialysis) Kidney failure

PAGE 36

36 T able 1 2. The American Join t Committee on Cancer (AJCC) TNM classification of urinary bladder cancer (Greene FL, 2002) T Primary tumour TX Primary tumour cannot be assessed T0 No evidence of primary tumour Ta Non invasive papillary carcinoma Tis Carcinoma in situ T1 Tumour invades subepithelial connective tiss ue T2 Tumour invades muscle T2a Tumour invades superficial muscle (inner half) T2b Tumour invades deep muscle (outer half) T3 Tumour invades perivesical tissue T3a Microscopically T3b Macroscopically (extravesical mass) T4 Tumour invades any of the following: prostate, uterus, vagina, pelvic wall, abdominal wall T4a Tumour invades prostate, uterus or vagina T4b Tumour invades pelvic wall or abdominal wall N Lymph nodes NX Regional lymph nodes cannot be assessed N0 No regional lymph node metast asis N1 Metastasis in a single lymph node in the true pelvis (hypogastric, obturator, external iliac, or presacral) N2 Metastasis in multiple lymph nodes in the true pelvis (hypogastric, obturator, external iliac, or presacral) N3 Metastasis in common iliac lymph node(s) M Distant metastasis MX Distant metastasis cannot be assessed M0 No distant metastasis M1 Distant metastasis

PAGE 37

37 Table 1 3. Treatment recommendations for bladder cancer (Babjuk et al., 2013; Wi tjes et al., 2014) Treatment recommendations in non muscle invasive (Ta, T1 tumors and CIS) bladder cancer according to risk stratification. The standard initial therapy for Ta and T1 papillary bladder tumors is complete macroscopic transurethral rese ction (TURB), including a part of the underlying muscle. CIS cannot be eradicated by TURB and further treatment is mandatory. It is recommended to stratify patients according to prognostic factors into three risk groups that will facilitate treatment recom mendations. Risk Category Definition Treatment Recommendation Low risk tumors Primary, solitary, Ta, LG/G1, < 3 cm, no CIS One immediate instillation of chemotherapy Intermediate risk tumors All cases between categories of low and high risk O ne immediate instillation of chemotherapy followed by further instillations, either chemotherapy for a maximum of 1 year or 1 year full dose BCG High risk tumors Any of the following: T1 tumors; HG/G3 tumors; CIS; Multiple and recurrent and la rge (> 3 cm) Ta G1G2 tumors (all these conditions must be presented) Intravesical full dose BCG instillations for 1 3 years or cystectomy (in highest risk tumors) Subgroup of highest risk tumors T1G3 associated with concurrent bladder CIS, multiple and/or large T1G3 and/or recurrent T1G3, T1G3 with CIS in prostatic urethra, micropapillary variant of urothelial carcinoma, LVI Radical cystectomy should be considered BCG failures Radical cystectomy is recommended

PAGE 38

38 Table 1 3. Continued Treatmen t recommendation for muscle invasive bladder cancer (MIBC) For MIBC radical cystectomy and urinary diversion is the curative treatment of choice In T2N0M0 selected patients Multimodality bladder sparing therapy can be considered for T2 tumors (i.e. alter native, not the standard option) Radical cystectomy is recommended in T2 T4a, N0M0, and high risk non MIBC. Neoadjuvant chemotherapy is recommended for T2 T4a, cN0M0 bladder cancer and should always be cisplatin based combination therapy. Surgically non c urable tumors Primary radical cystectomy in T4b bladder cancer is not a curative option. Radical cystectomy may be a therapeutic/palliative option. Intestinal or nonintestinal forms of urinary diversion can be used, with or without, palliative cystectomy. Recommendation for metastatic disease First line treatment for fit patients Use cisplatin containing combination chemotherapy with GC, PCG, MVAC, preferably with G CSF, or HD MVAC with G CSF. First line treatment in patients ineligible (unfit) for cis platin For cisplatin ineligible (unfit) patients, with PS2 or impaired renal function, as well as those with 0 or 1 poor Bajorin prognostic factors and impaired renal function, treatment with carboplatin containing combination chemotherapy, preferably with gemcitabine/carboplatin is indicated.

PAGE 39

39 Table 1 4 Risk, Injury, Failure, Loss, and End stage Kidney (RIFLE) classification (Bagshaw, George, Bellomo, & Comm, 2008) Class Serum creatinine/GFR criteria Urine output criteria Risk Serum creatinine 1.5 or decrease in GFR < 0.5 ml/kg/hour 6 hours Injury Serum creatinine 2 or decrease in GFR < 0.5 ml/kg/hour 12 hours Failure Serum creatinine 3, or decrease in GFR acute rise > 0.5 mg/dl < 0.3 ml/kg/hour 24 hours, or anuria 12 hours Loss Complete lo ss of kidney function > 4 weeks End stage kidney disease Complete loss of kidney function > 3 months

PAGE 40

40 CHAPTER 2 EFFECT OF RADICAL CYSTECTOMY PLUS URINARY DIVERSION FOR BLADDER CANCER TREATMENT ON RENAL FUNCTION Background In the United States bladder cancer is the fourth most common cancer among men, and in 2014 alone there were 74,690 new cases and 15,580 related deaths (Siegel et al., 2014) Radical cystectomy (RC) plus urinary diversion (UD) has long been the reference standard treatment for muscle invasive bladder cancer with a disease specific 5 year survival of 74% (Manoharan et al., 2009) However, p atients with RC plus UD are at particularly increased risk of overt renal dysfunction (Kristjansson & Mansson, 2004) This is concerning given that chronic kidney disease (CKD) is a costly and growing public health problem that negatively impacts quality of life (QOL) and independently increases the risk of cardiovascular disease, hospitalization, and death (Go et al., 2004; Levey, Coresh, Balk, Kausz, Levin, Steffes, et al., 2003) Thus far, only a few studies have investigated the impact of RC plus UD for bl adder cancer on renal function and have been limited by small sample size, lack of control groups, heterogeneous population and inadequat e ascertainment of the outcome (Gilbert et al., 2013; Jin et al., 2012; Lantz et al., 2010; Pycha et al., 2008; Samuel et al., 2006) Given the context of present literature, the impact of RC plus US for bladder cancer on renal function over time is unclear. Therefore, we conducted a retrospective study of patients wit h bladder cancer who sought care in a tertiary health care center between the years 2000 to 2014 to d etermine the effect of RC plus UD for bladder cancer treatment on renal function, as measured by estimated glomerular filtration rate ( GFR ) using Modification of Diet in Renal Disease (MDRD) equation. We hypothesized that p atients who undergo RC plus UD for bladder cancer have more rapid decline in

PAGE 41

41 renal function as compared to bladder cancer patien ts who do not undergo RC plus UD. We bel ie ve results from this study help inform patients and providers about the risk of renal impairment after RC plus UD for bladder cancer. This insight can guide the decision making process with respect to the management of patient s with bladder cancer in clinical settings. Methods Study Participants W e identified 517 men and women aged 18 yea rs and older diag nosed with bladder cancer according to the International Classification of Disease, Ninth Revision, C linical Modification (ICD 9 CM) (Appendix A) We included bladder cancer patients with a minimum of two measurements of serum creatinine (SCr) because havi ng at least two measurements of SCr was required to calculate the change in renal function over time. We excluded those who had CKD stage G5, defined as an MDRD eGFR< 15ml/min/1.73m 2 or ESRD at baseline and those with <30 days between their firs t and last SCr measurements. Three hundred eighty four out of 517 adult patients with bladder cancer met our in clusion and exclusion criteria, and form ed our analytic cohort We used previously collected inpatient and outpatient information in the University of Flor ida integrated data repository and cancer registry of our health care center. The study was approved by the University of Florida Institutional Review Board. Radical Cystectomy plus Urinary Diversion The predictor of interest was RC plus UD compared to no RC plus UD The surgical procedure information for each patient was obtaine d according to Current Procedural Terminology (CPT) codes and Facility Oncology Registry Data Standards (FORDS) site specific surgery codes (Appendix B).

PAGE 42

42 Bladder cancer patients were assigned to the surgery group (RC plus UD group) if they had undergone RC plus UD according to either CPT or FORDS codes. We used the following CPT codes for ascertainment of RC plus UD : 51570, 51575, 51590, 51595, 51596, and 51597 (Appendix B).We use d the following FORDS site specific surgery codes for ascertainment of RC plus UD : 50, 60, 61, 62, 70, 71, and 72 (Appendix B). Bladder cancer patients who sought care at our academic health center from January 2000 to December 2014 and had not undergone RC plus UD were assigned to the control group (no RC plus UD group). Measurement of Renal Function We used the Modification of Diet in Renal Disease (MDRD) (Levey et al., 2006) study equation to estimate glomerular filtration rate (GFR) based on the SCr, age, gender, and race (Levey et al., 1999; Stevens et al., 2006) The isotope dilution mass spectrometry (IDMS) traceable MDRD stu dy equation is expressed as follows: eGFR (ml/min/1.73m 2 ) = 175 x (Scr) 1.154 x (Age) 0.203 x (0.742 if female) x (1.212 if African American) (conventional units) (Levey et al ., 2006) For patients with RC plus UD we included a baseline SCr measurement within four weeks before the date of surgery and all the following SCr measurements after 30 days from the surgery. For patients who did not undergo RC plus UD (controls) we i ncluded a baseline SCr (first encounter SCr) and all recorded SCr measurements >30 days after the baseline. Our outcome of interest was renal function decline that was defined by the following methods: (a) change in (MDRD) (Levey et al., 2006) eGFR over tim e (eGFR slope); (b) time to 30% (Coresh et al., 2014) ; and (c) rapid decline in renal function, defined by a decrease in the MDRD eGFR > 3 ml/min/1.73m 2 /year (de

PAGE 43

43 Boer et al., 2009; Rifkin et al., 2008) The MDRD eGFR slope was determined using a linear mixed model regression of change in MDRD eGFR with time using all estimated GFR measurements from the baseline and follow up period. Percentage change in MDRD eGF R was calculated as (last eGFR baseline eGFR)/(baseline eGFR)*100% (Coresh et al., 2014) 30% was considered an event. Also, in a sensitivity analysis, cutoff points of percentage change 20% were assessed. For estimat ion of rapid d ecline in renal function, in 201 patients with a follow up time of one year or longer, a subject specific MDRD eGFR slope was determined as an annual change estimated from a least square regression model using all estimated GFR measurements f rom the baseline and follow up period. Rapid decline in renal function was defined by a decrease in the MDRD eGFR >3 ml/min/1.73m 2 /year based upon previously published literature (de Boer et al., 2009; Rifkin et al., 2008) and represents a three fold increase compared to annual change in eGFR due to normal aging (Hemmelgarn et al., 2006; Lindeman, Tobin, & Shock, 1985) Covariates The conceptual framework and the previous literature guided the selection of covariates. We obtained information on treatment modalities including chemotherapy, radiotherapy, and intravesical instillation of anticarcinogenic agents according to CPT cs included age, gender, race, marital status, primary insurance, body mass index (BMI), and smoking status. Comorbid conditions at the baseline included congestive heart failure (CHF), peripheral vascular disease (PVD), myocardial infarction (MI), cerebrova scular disease, chronic pulmonary

PAGE 44

44 disease, connective tissue disease rheumatic disease, diabetes mellitus (DM), dementia, HIV/AID, alcohol abuse, hypertension (HTN), urinary tract infection s (UTI s ), nephrolithiasis, urinary obstructions, and hydronephrosis We obtained information on the stage of the bladder cancer according to the American Joint Committee on Cancer (AJCC) tumor, node, and metastasis (TNM) staging classification. We obtained both clinical and pathologic staging data, but due to incompletene ss of the pathologic data, we included only clinical staging data in our analysis. We used all available medication data from the baseline and the follow up period. We imputed the missing data using multiple imputation (PROC MI). The proportions (%) of mis sing data were as follows: co morbid conditions (28%), BMI (48%), smoking status (29%), clinical TNM stage (3%), and medications (2%). Statistical Analysis The baseline demographics and clinical characteristics were examined in bladder cancer with and without radically cystectomy plus urinary diversion, and statistical differences were assessed using one way analysis of variance (ANOVA) and chi squared tests for continuous and categorical variables, respectively. Propensity score approach was used to address the confounding by indication, whe re the clinical conditions (e.g. cancer stage or patients characteristics) that determine the choice of the treatment given (e.g. RC plus UD) are independently associated to the outcome of interest (e.g. rena l i mpairment). Propensity score was estimated using multivariable logistic regression model with RC plus UD as dependent variable and with age, gender, race, BMI, baseline MDRD eGFR, primary insurance, marital status, smoking, co morbid conditions, clinical T NM stage, clinical AJCC stage group, and general cancer stage as independent variables. These variables were

PAGE 45

45 selected based upon a priori identification of key confounding variab les in the literature and upon clinical judgment. The area under the receiver operating characteristics curve (AUC) for the propensity mode was 0.88 (95%Cl 0.85 0.92). The estimated the propensity score was added to the outcome regression models as a continuous variable. The association between RC plus UD and change in MDRD eGFR (e GFR slope) was assessed using multivariable linear mixed models regression of change in MDRD eGFR with time using all estimated GFR measurements from the baseline and follow up period. We constructed four linear mixed models (PROC MIXED) with residual max imum likelihood estimation (REML), random intercept, and unstructured covariance. The four constructed linear mixed models were as follows: model 1: RC plus UD, time, and interaction of RC plus UD with time; model 2: RC plus UD, time, interaction of RC p lus UD with time, and age; model 3: RC plus UD, time, interaction of RC plus UD with time, age, and propensity score; and model 4: fully adjusted model. Candidate variables included in th e full model were age, propensity score, and baseline eGFR a priori plus other clinically relevant confounding variables that were associated with both predictor (RC plus UD) and outcome (eGFR slope) at statistical significance of p<0.1. The association between RC plus UD and time to percentage change in MDRD 30% (last eGFR baseline eGFR)/(baseline eGFR)*100% (Coresh et al., 2014) ) was assessed using multivariable cox proportional hazards models (PROC PHREG). We plotted surviv al probability of not having a renal deterioration event, defined by a decrease in the MDRD eGFR of 30% or greater, in patients with and without RC plus UD using Kaplan Meier curves. We compared the renal deterioration free survival

PAGE 46

46 distributions between p atients with and without RC plus UD, using log rank test. In the sensitivity analysis ,w e repeated these analyses for time to percentage change in MDRD eGFR 57% (corresponding to doubling of 20% (Hemmelgarn et al., 2006) Four cox proportional hazards models were constructed as follows: model 1 : RC plus UD; model2: RC plus UD and age; model 3: RC plus UD, age and propensity score; and model 4: fully adjusted model. The fully adjusted cox proportional hazards model (model 4) was fitted using a backward elimination procedure. Candidate variables i ncluded in the backward elimination procedure were age, propensity score, and baseline eGFR a priori plus other clinically relevant confounding variables that were associated with both predictor (RC plus UD) and outcome (time to percentage change in MDRD 30%) at statis tical significance of p<0.1. The adequacy of the Cox proportional hazards model was examined and the predictors in the model satisfied the proportional hazards assumption (Lin, Wei, & Ying, 1993) (proportionality test p value=0.189; Appendix C ). The association between RC plus UD and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) was assessed using multivariable logistic regression models (PROC LOGISTIC) up time (n=201). Four logistic regression models were constructed as follows: model 1: RC plus UD; model2: RC plus UD and age; model 3: RC plus UD, age and propensity score; and model 4: fully adjusted model. The fully adjusted logistic regression model (model 4) was fitted using a backward elimination procedure. Candidate variables included in the backward elimination procedure were age, propensity score, and baseline eGFR a priori plus other clinically relevant confounding variables that were

PAGE 47

47 associated with bot h predictor (RC plus UD) and outcome (rapid decline in renal function; MDRD eGFR >3 ml/min/1.73m 2 /year) at statistical significance of p<0.1. W e assumed a missing at random (MAR) mechanism for the missing data on co morbid conditions BMI, smoking s tatus, and cance r stage Multiple imputation with fully conditional specification (FCS) method was used to create 5 complete data sets using PROC MI. We used linear regression model and discriminant function method for imputation of continuous and categori cal variables, respectively. Each of the 5 complete data sets was analyzed using the appropriate statistical test (e.g. PROC MIXED, PROC PHREG, and PROC LOGISTIC). The parameter estimates (e.g. coefficients and standard errors) obtained from each analyzed data set were then combined to generate statistical inference using PROC MIANALYZE. We did not have any missing data on age, gender, race, surgical procedures, and SCr; therefore, we did not impute any data with respect to the main predictor and primary ou tcome variables. The statistical significance level was set at two sided alpha=0.05. Statistical Analysis Software (SAS) version 9.4 (SAS Institute Inc, Cary, North Carolina) was used to perform the statistical analyses. Results Participants Characte ristics The mean age of the 384 men and women who were included in the study was 6812 years. The m ajority of patients were males (76.0%), Caucasian (86.2%), married (62.5%), an d former smoker (63.4%). Table 2 1 compares the demographics and baseline char acteristics of bladder cancer p atients who underwent RC plus UD to those who did not undergo the procedure. Out of 384 patients, 172 (44.8%) patients underwent RC plus UD Compared to patients who did not undergo RC plus UD those

PAGE 48

48 who underwent the procedu re were more likely to be married (55.2% vs.71.5%, p value= 0.008). The baseline MDRD eGFR values were 73.828.3 and 71.624.9 for patients with and without RC plus UD respec tively (p value= 0.420). Table 2 2 compares co morbid conditions between bladde r cancer pati ents with and without RC plus UD The most prevalent co morbid conditions in these patients were hypertension (61%), followed by diabetes mellitus (23%) and chronic pulmonary disease (22%). Compared to patients without radically cystectomy plu s urinary diversion, those with the procedure were less likely to have peripheral vascular disease (10.0% vs. 3.4%, p value= 0.026) and hypertension, although the latter was not statistically significant (66.9% vs. 55.8%, p value= 0.058 ). Table 2 3 presen ts distribution of clinical stage of bladder cancer according to the American Joint Committee on Cancer (AJCC) TNM classification by RC plus UD group. Compared to patients without RC plus UD, those with RC plus UD, were more likely to have primary tumor ( T2), and less likely to have primary tumor (Ta) an d (Tis), and distant metastasis (M1) (Table 2 3) Table 2 4 presents medications use in patients according to RC plus UD group. Compared to patients without RC plus UD those with the procedure were more l ikely to use analgesics, aminoglycosides, antineoplastics, beta blockers, beta lactam antibiotics, H2 blockers, nonsteroidal anti inflammatory drugs (NSAIDS), platinum based antineoplastics, and vancomycin (Table 2 4) Renal Function The linear mixed m odel showed that bladder cancer patients who underwent RC plus UD had a steeper decline in renal function compared to those who did not undergo the procedure (parameter estimates of intera ction between RC plus UD and time =

PAGE 49

49 3.153; 95% CI: 4.649, 1.657; p value < 0.001)(Table 2 5). In other words, patients with RC plus UD had a mean MDRD eGFR decline of 3.85 ml/min/1.73m2/year (95% CI: 5.000, 2.721 ml/min/1.73m2/year), as compar ed to a mean MDRD eGFR decline of 0.698 ml/min/1.73m2/year (95% CI: 1.621, 0.232 ml/min/1.73m2/year) in patients without the procedure (Figures 2 1 and 2 2). The association between RC plus UD and a steeper decline in mean MDRD eGFR remained statically significant after adjusting for age (parameter estimates of intera ction between RC plus UD and time= 3.143; 95% CI: 4.636, 1.649; p value < 0.001) and then propensity score (parameter estimates of interac tion between RC plus UD and time= 3.145; 95% CI: 4.638, 1.652; p value < 0.001) (Table 2 5). In t he fully adjusted li near mixed model, after accounting for age, propensity score, radiotherapy, chemotherapy, intravesical anticarcinogenic, angiotensin converting enzyme inhibitor s (ACEIs) /angiotensin II receptor blocker s (ARBs) anticonvulsant, antineoplastic, acetylsalicy lic acid ( ASA ) beta lactam antibiotics, calcineurin inhibitor, H2 blocker, penicillin, platinum based antineoplastic, vancomycin, vasodilator, and xanthine oxidase inhibitor, we found that RC plus UD is strongly associated with faster decline in mean MDRD eGFR (parameter estimates of interaction between RC plus UD and time = 3.145; 95% CI: 4.638, 1.652; p value < 0.001) (Table 2 5). The Kaplan Meier survival method was used to estimate the survival probability of not having a renal deterioration event, d efined by a decrease in the MDRD eGFR of 30% or greater in patients with and without RC plus UD (Figure 2 3). This analysis was repeated for time to percentage change in MDRD eGFR with cutoff values of MDRD 57% (correspondi ng to doubling of SCr

PAGE 50

50 concent ration) (Figures 2 4, 2 5, 2 6, and 2 7). The median of follow up time was 13 (5 29) months Time to event analysis indicated that out of 384 patients with bladder cancer, 141 (77 and 64 among patients with and without RC plus UD respectively) experienced a renal deterioration event, defined by a decrease in the MDRD eGFR of 30% or greater throughout study period. Furthermore, the estimated survival function for patients with RC plus UD lay below that of patients without the p rocedure (log rank test p value= 0.0002) (Figure 2 3). Repetition of this analysis with different cutoff v alues 57% showed similar results (log rank test 25%,and 40%, respectively), except 57% (log rank test p value= 0.1718) (F igures 2 4, 2 5, 2 6, and 2 7). The unadjusted cox proportional hazards 30% among patients with RC plus UD was 88% larger than among patients without the procedure throughout the study period (HR=1.88, 95% CL: 1.3 5 2.63; p value <0.001) (Table 2 6). After adjusting for age, the 30% among patients with RC plus UD was 87% larger than among patients without the procedure (HR=1.87, 95% CL: 1.34 2.62; p value <0.001). After further adjustmen t for propensity score, the hazards of incident 30% among patients with RC plus UD was 30% larger than among patients without the procedure, but this difference was not statically significant (HR=1.30, 95% CL: 0.81 2.08; p value= 0.273). The ful ly adjusted cox proportional 30% among patients with RC plus UD was similar to patients without the procedure, after accounting for age, propensity score, diabetes mellitus, hydronephrosis, cl inical T, clinical N,

PAGE 51

51 antiviral s beta lactam antibiotics, and vancomycin (HR=1.01, 95% CL: 0.62 1.63; p value= 0.976) (Table 2 6). Out of 201 bladder cancer patients with one year or longer follow up time, 73 (41 and 32 amo ng patients with and without RC plus UD respectively) had rapid decline in renal function, defined as MDRD eGFR >3 ml/min/1.73m 2 /year In the unadjusted logistic regression model, RC plus UD was strongly associated with increased odds of r apid decline in renal function, defined as MDRD eGFR >3 ml/min/1.73m 2 /year (OR=2.07; 95% CI: 1.15 3.70; p value= 0 .015) (Table 2 7). After adjusting for age, this positive association persisted (OR=2.01; 95% CI: 1.11 3.63; p value= 0.021). However, after further adjustme nt for propensity score this association was attenuated and no longer statically significant (OR=1.34; 95% CI: 0.60 3.16; p value = 0.473) (Table 2 7). In the final regression model using stepwise backward elimination, after adjusting for age, propensity s core baseline eGFR, DM ACEIs / ARBs H2 blocker, and beta lactam antibiotics RC plus UD was not associated with r apid decline in renal function, defined as MDRD eGFR >3 ml/min/1.73m 2 /year (OR=1.30; 95% Cl: 0.5 4 2.87; p value= 0.556) (Table 2 7). Discuss ion To the best of our knowledge, this is the first study to examine the impact of RC plus UD for treatment of bladder cancer on renal function over time using bladder cancer patients without the procedure as control gro up. In this retrospective study of 3 84 adult patients with bladder cancer, we found that patients who underwent RC plus UD experienced a faster decline in renal function over time, as measured by MDRD eGFR slope, compared to patients who did not undergo the procedure, despite adjustment for age, propens ity score, and other confounding variables Furthermore, we

PAGE 52

52 found that bladder cancer patients who underwent RC plus UD had a higher risk of 30% and greater odds of rapid decline in eGFR compared to controls; this persisted despit e adjustment for age but was no longer statistically significant after adjustment for propens ity score, and other confounding variables Although these characteristics, co morbid conditions, stage of bladder cancer, and medications use, the strength of statically significant association between RC plus UD and mean MDRD eGFR slope in the fully adjusted linear mixed model suggests that the attenuation of the observed associations in the adjusted time to event and logistic regression analyses might be due to insufficient statisti cal power resulting from use of dichotomous outcomes. Our findings that bladder cancer patients with RC plus UD experience steep decline in renal function are consistent with the majority of the prior studies that have examined this relationship without t he inclusion of a control group For example, in a retrospective study of 222 patients with ileal conduit diversion with a median follow up of 91 months, Rouanne et al. found the median MDRD eGFR decreased from 66 ml/min/1.73m 2 to 59 ml/min/1.73m 2 (Rouanne et al., 2015) .The authors added that the decline in renal function was steeper in the first and second year after the UD ( 9 ml/min/1.73m 2 and 4 ml/min/1.73m 2 in the 1st and 2nd year, respectively) (Rouanne et al., 2015) In another retrospective study of 1,631 patients with RC, the authors found 2 decrease in CKD EPI eGFR, was 71% and 74% for continent and incontinent diversion, respectively, by 10 yea rs of follow up (Eisenberg et al., 2014) A retrospective study of 169 patients with

PAGE 53

53 RC plus UD reported renal deterioration, defined by a decrease in the eGFR of 20% or greater, in 46% of patients with a mean follo w up of 104 months (Nishikawa, Miyake, Yamashita, Inoue, & Fujisawa, 2014) While the majority of prior studies (Eisenberg et al., 2014; Gilbert et al., 2013; Lantz et al., 20 10; Nishikawa et al., 2014; Osawa et al., 2013; Pycha et al., 2008; Rouanne et al., 2015; Samuel et al., 2006) reported renal function deterioration after RC plus UD, a few others reported no change in renal function (Perimenis et al., 2004; Thoeny et al., 2002) For example, Perimenis et al. (Perimenis et al., 2004) conducted a study of 129 men who survived more than 5 years after RC plus UD and reported that the mean SCr levels remained unchanged. The main differen ces between our study and prior studies are the inclusion of bladder cancer patients without RC plus UD as a control group, use of three widely acceptable definitions of renal function decline, and appropriate statistical adjustment for potentially confoun ding variables using propensity score and multivariable regression models. In our study, not only we found that renal function decline occurred at a high rate among bladder cancer patients with RC plus UD, but also we showed that bladder cancer patients wi th RC plus UD experience a faster decline in renal function compared to bladder cancer patients without RC plus UD. The primary outcome of this study, renal function decline, was defined as change in MDRD eGFR over time (eGFR slope) (Coresh et al., 2014) In addition to eGFR slop, 30% (Coresh et al., 2014) and rapid decline (i.e. MDRD eGFR >3 ml/min/1.73m 2 /year) (Li ndeman et al., 1985) The MDRD eGFR slop was determined using a linear mixed model using all estimated GFR measurements from the baseline

PAGE 54

54 and follow up period. This approach was based on the assumption that GFR decline or CKD progression follows a steady linear trajectory over time (Hunsicker et al., 1997) However, it has been suggested that kidney disease progression may follow a nonlinear GFR trajectory consists of periods of acceleration or non progression (Li et al., 2012) Li et al. (Li et al., 2012) suggested using a Bayesian approach (Crainiceanu, Ruppert, & Wand, 2005) jectory. In our study, GFR decline was assumed to be linear because linearity assumption of kidney disease progression has been commonly used in research (Coresh et al., 2014; Hunsicker et al., 1997; Sakaguchi et al. 2013) conceptually convenient, and made the statistical approach of view. A nonlinear GFR decline assumption requires sophisticated statistical analysis and ident ification of time dependent variables (Li et al., 2012) A percentage change in M 30% was chosen as an alternative outcome becaus it is strongly and consistently associated with the risk of end stage renal disease (ESRD) and mortality (Coresh et al., 2014) This cutoff value is smaller than a doubling of SCr concentration, 57%, which is the traditional endpoint of CKD progression (Coresh et al., 2014; Lewis, Hunsicker, Bain, & Rohde, 1993) However, 57% is a late event and occurs less frequently, requiring a large sample size and long period of follow up (Coresh et al., 2014) 20% were investigated as well. For rapid decline in renal function, an MDRD eGFR >3 ml/min/1.73m 2 /year was chosen because it represents a three fold increase compared to annual change in eGFR due to normal aging (Hemmelgarn et al., 2006; Lindeman et

PAGE 55

55 al., 1985) and it has been commonly used in research (de Boer et al., 2009; Rifkin et al., 2008) Confounding by indication is a major source of concern in this study. Confounding by indication, or sometimes called channeling bias, is a common problem in observational studies when the exposure of interest is the assigned medication or surgical procedure (Blais, Ernst, & Sui ssa, 1996; Greenland & Neutra, 1980; Walker, 1996) In observational studies, the methods available to address or minimize confounding by indication can be categorized into two groups: study design methods or statistical analysis methods. In this study, w e used the propensity score method and regression adjustment to address confounding by indication. The rationale for the use of propensity score is based on the fact that use of propensity score balances the compared groups with respect to the observed cov ariates, and the propensity score estimates are much more robust compared to conventional multivariate regression models (Lunceford & Davidian, 2004) Also, propensity score method simplifies the final regression model (i.e. the final model will consists of the treatment varia ble, estimated propensity score, and a few important confounding factors) because all observed pretreatment variables and their potential interactions are included in the propensity score model without being preoccupied with the model fit. Another source of concern in this retrospective study was incomplete or missing data. Potential reasons for missingness are lost to follow up, competing risks (e.g. death), non complaint, and non response. Also, in our study, co morbid conditions data for some patients w ere not available in the database. Missing data leads to loss of information, loss of statistical power, less reliable results, and bias (Steyerberg, 2009) In

PAGE 56

56 this study, the data obtained from the integrated data repository was merged with the data obtained from the cancer registry. In order to reduce the proportion of missing data, we limited our study subjects to patients who had records in cancer registry. In this study, we assumed that the mechanism of missing data was missing at random (MAR), the probability that an observation is missing depends only on observed data (Steyerberg, 2009) .The method commonly used to address MAR is multiple imputation (Schafer & Graham, 2002) The multiple imputation method replaces each missing value with a set of plausible values, rep resenting the uncertainty about the correct value. The multiple imputation method was used for imputation of missing data on co morbid conditions TNM staging, BMI, and smoking status. Our study has several strengths. First, to the best of our knowledge t his is the first study to investigate the effect of RC plus UD on renal function over time having bladder cancer patients without RC plus UD as control group. Second, we used three well established definitions of renal function decline which included the M DRD eGFR slope, time to 30% from the baseline, and rapid decline in renal function, defined as MDRD eGFR >3 ml/min/1.73m 2 /year. Furthermore, we addressed the confounding by indication using propensity score with regression adj ustment, and we accounted for clinically relevant and statically significant confounding factors. Our study also has several limitations. First, this study is limited by its retrospective nature, and by the fact that data were collected from various clini cal and administrative information systems. Second, our academic health care center is a tertiary referral center; therefore, the patient populations are more likely to have co

PAGE 57

57 morbid conditions that may contribute to renal function decline. We tried to ad dress this by appropriate statistical adjustment for clinically relevant and confounding factors. Although in the final multivariable regression models, we included age, propensity score, and clinically relevant factors with statistically significant assoc iation with both RC plus UD and renal function decline, there is potential for colinearity between variables and overadjustment (Schisterman, Cole, & Platt, 2009) and it is possible that the point estimates might have been undermined and our results might have been biased toward the null hypothesis. Furthermore, we used the MDRD study equation (Levey et al., 1999) for estimation of GFR. The MDRD equation is commonly used in clinical practice and researc h, and it is reasonably accurate for the assessment of renal function in non hospitalized patients with kidney disease (i.e. the MDRD equation was developed in patients with kidney disease ) (Stevens et al., 2006) T he MDRD equation has a sensitivity and specificity of 95% and 82% to detect a measured GFR<60 ml/min/1.73 m 2 using iothalamate as gold standard (Levey et al., 2009) However, the MDRD equation has some limitations. The GFR estimated by the MDRD equation does not accurately reflect renal function in the healthy individuals and at the early stages of renal impairment (Stevens et al., 2006) In this population, the MDRD equation tends to underestimate the GFR when directly measured GFR (i.e. by use of an exogenous filtration marker such as inulin or 125I iothalamate) is less than 90 ml/ min/1.73 m 2 leading to a false positive diagnosis of kidney disease (Stevens et al., 2006) We chose to use MDRD eGFR; however, because it is the most widely used method and it is assumed to be reasonably accurate for the assessment of renal function in population under study (i.e. older individuals with bladd er cancer). As mentioned before, the use of

PAGE 58

58 MDRD equation may lead to misclassification of renal impairment. However, this misclassification of renal impairment can be considered non differential since it occurs with equal likelihood among bladder cancer p atients with and without RC plus UD. The problem with a non differential misclassification is that it biases the results toward the null (Aschengrau & Seage III, 2008) Finally, the generalizability of our findings is limited because the study subjects were mostly older C aucasian males. In summary, we found that bladder cancer patients who underwent RC plus UD experienced a faster decline in renal function over time, as measured by MDRD eGFR slope, as compared to patients who did not under go the procedure. P atients with RC plus UD had a mean MDRD eGFR decline of 3.84 ml/min/1.73m 2 /year, whereas those without RC plus UD had a mean MDRD eGFR decline of 0.69 ml/min/1.73m 2 /year. characteristics, co m orbid conditions and medications use. Additionally, we found that 30 and increased odds of rapid decline in renal function. However, these associations may be partially explained eristics, co morbid conditions and medications use. It is possible that colinearity between variables, overadjustment, and insufficient statistical power for categorical outcomes might have biased these estimates toward the null. We believe future prospe ctive observational studies or controlled trials are needed to elucidate these relationships. In the mean time, the results from this study may help inform patients and providers about the risk of renal impairment after RC plus UD for bladder cancer. This insight may guide the decision making process with respect to the management of patients with bladder cancer in clinical settings.

PAGE 59

59 Table 2 1 Sociodemographic and Clinical Characteristics of Patients with Bladder Cancer between 2000 and 2014 No Radica l Cystectomy plus Urinary Diversion (n=212) Radical Cystectomy plus Urinary Diversion (n=172) All (n=384) p value Sociodemographic Age ,mean (SD) years 67(13) 68(10) 68(12) 0.813 Male, n (%) 155(73.11) 137(79.65) 292(76.04) 0.136 Race/ethnicity n (%) White 181(85.38) 150(87.21) 331(86.20) 0.553 African American 11(5.19) 6(3.49) 17(4.43) Other 20(9.43) 16(9.30) 36( 9.38 ) Marital Status (%) Married 117(55.19) 123(71.51) 240(62.50) 0.008 Primary Insurance, n (%) Medicar e 134(63.21) 114(66.28) 248(64.58) 0.153 Medicaid 15(7.08) 14(8.14) 29(7.55) Private 54(25.47) 34(19.77) 88(22.92) No insurance 9(4.25) 6(3.49) 15(3.92) VA 0(0.00) 4(2.33) 4(1.04) Clinical Characteristics BMI mean (SD) 28.9(5.7) 28. 0(5.8) 28.4(5.8) 0.254 Smoking Status + n (%) Current Smoker 27(21.26) 24(16.44) 51(18.68) 0.560 Former Smoker 79(62.20) 94(64.38) 173(63.37) Never Smoker 21(16.54) 28(19.18) 49(17.95) Baseline MDRD eGFR, mean(SD) 71.6(24.9) 73.8(28.3) 72.6(26. 5) 0.420 BMI data were available for 199 out of 384 patients with n =110 and n=89 in patients with and without RC plus UD, respectively. + Smoking Status data were available for 273 out of 384 patients with n=146 and n=127 in patients with and without RC plus UD, respectively.

PAGE 60

60 Table 2 2 Co morbid Conditions of Patients with Bladder Cancer by Radical Cystectomy plus Urinary Diversion Group between 2000 and 2014 No Radical Cystectom y plus Urinary Diversion (n=212 ) Radical Cystectom y plus Urinary Diversion (n=172 ) All (n=384 ) p value Co morbid Conditions n (%) Congestive Heart Failure 9(6.92) 5(3.40) 14(5.05) 0.182 Myocardial Infarction 17(13.08) 12(8.16) 29(10.47) 0.183 Peripheral Vascular Disease 13(10.00) 5(3.40) 18(6.50) 0.026 Cerebrovascular Disease 8(6.15) 5(3.40) 13(4.69) 0.28 Dementia 2(1.54) 1(0.68) 3(1.08) 0.491 Chronic Pulmonary Disease 35(26.92) 27(18.37) 62(22.38) 0.088 Connective Tissue Disease/Rheumatic Disease 7(5.38) 2(1.36) 9(3.25) 0.059 Diabetes 36( 27.69) 29(19.73) 65(23.47) 0.119 Hypertension 87(66.92) 82(55.78) 169(61.01) 0.058 AIDS/HIV 1(0.77) 0(0.00) 1(0.36) 0.287 UTI 25(19.23) 24(16.33) 49(17.69) 0.527 Nephrolithiasis 3(2.31) 5(3.40) 8(2.89) 0.588 Urinary Obstruction and Hydroneph rosis 21(16.15) 25(17.01) 46(16.61) 0.849 Alcohol Abuse 8(6.15) 4(2.72) 12(4.33) 0.161 Co morbid conditions data were available for 277 out of 384 patients with n =147 and n=130 in patients with and without RC plus UD, respectively

PAGE 61

61 Table 2 3. C linical stage of bladder c ancer, the AJCC TNM classification, by r adical cystectomy plus urinary diversion g roup between 2000 and 2014 All (n=384) No Radical Cystectomy plus Urinary Diversion (n=212) Radical Cystectomy plus Urinary Diversion (n=172) p va lue Primary tumor (T) n (%) Ta 69(18.35) 66(32.04) 3(1.76) <0.001 Tis 16(4.26)) 8(3.88) 8(4.71) T1 62(16.49) 42(20.39) 20(11.76) T2 98(26.06) 28(13.59) 70(41.18) T3 32(8.51) 12(5.83) 20(11.76) T4 14(3.72) 10(4.85) 4(2.35) TX 85(22.61) 40 (19.42) 45(26.47) Regional lymph nodes (N) N0 256(68.09) 149(72.33) 107(62.94) 0.167 N1 9(2.39) 4(1.94) 5(2.94) N2 12(3.19) 8(3.88) 4(2.35) N3 4(1.06) 1(0.49) 3(1.76) NX 95(25.27) 44(21.36) 51(30.00) Distant metastasis (M) M0 292(78. 28) 160(78.43) 132(78.11) 0.011 M1 13(3.49) 12(5.88) 1(0.59) MX 68(18.23) 32(15.69) 36(21.30)

PAGE 62

62 Table 2 3. Continued All (n=384) No Radical Cystectomy plus Urinary Diversion (n=212) Radical Cystectomy plus Urinary Diversion (n=172) p va lue Clinical Stage Group 0a 69(18.35) 65(31.55) 4(2.35) <0.001 0is 14(3.72) 8(3.88) 6(3.53) I 63(16.76) 42(20.39) 21(12.35) II 80(21.28) 21(10.19) 59(34.71) III 25(6.65) 8(3.88) 17(10.00) IV 35(9.31) 22(10.68) 13(7.65) Unknown 90(23.94) 4 0(19.42) 50(29.41) TNM classification data were available for 376 out of 384 patients with n =170 and n=206 in patients with and without RC plus UD, respectively AJCC, American Joint Committee on Cancer

PAGE 63

63 Table 2 4 Medications Us e in Patients with Bladder Cancer by Radical Cystectomy plus Urinary Diversion Group No Radical Cystectomy plus Urinary Diversion (n=212) Radical Cystectomy plus Urinary Diversion (n=172) All (n=384) p value Medications, n (%) ACE inhibitor/a ngiotensin II receptor blocker 87(41.84) 73(43.71) 160(42.67) 0.714 Analgesics 165(79.33) 148(88.62) 313(83.47) 0.016 Aminoglycoside 54(25.96) 74(44.31) 128(34.13) <.001 Anticonvulsant 3(1.44) 2(1.20) 5(1.33) 0.837 Antineoplastic 110(52.88) 58(34.73) 168(44.80) <.001 Antiviral 9(4.33) 4(2.40) 13(3.47) 0.310 ASA 76(36.54) 61(36.53) 137(36.53) 0.998 Beta blocker 110(52.88) 129(77.25) 239(63.73) <.001 Beta lactam antibiotics 133(63.94) 153(91.62) 286(76.27) <.001 Calcineurin inhibitor 8(3.85) 5(2.99) 13(3.47) 0.654 Calcium channel blocker 54(25.96) 58(34.73) 112(29.87) 0.065 Diuretic 78(37.50) 78(46.71) 156(41.60) 0.072 Fluoroquinolone 157(75.48) 117(70.06) 274(73.07) 0.240 H2 blocker 49(23.56) 74(44.31) 123(32.80) <.001 NSAID 48(23.08) 72(43.11) 120(32.00) <.001 Platinum based antineoplastic 39(18.75) 46(27.54) 85(22.67) 0.043 Statin 97(46.63) 74(44.31) 171(45.60) 0.654 Vancomycin 68(32.69) 99(59.28) 167(44.53) <.001 Vasodilator 29(13.94) 36(21.56) 65(17.33) 0.053 Xanthine oxidase inhibitor 6(2.88) 10(5.99) 16(4.270 0.139 Treatment, n (%) Intravesical anticarcinogenic 96(45.28) 15(8.72) 111(28.91) <.001 Chemotherapy 133(62.74) 99(57.56) 232(60.42) 0.302 Radiotherapy 18(8.49) 8(4.65) 26(6.77) 0.137

PAGE 64

64 Tabl e 2 5. Association between radical cystectomy plus urinary diversion with change in MDRD eGFR over time using multivariable linear mixed models Model 1 Model 2 Referent Estimates 95%CL p value Estimates 95%CL p value Intercept 71.078 (67.217, 74.940) <0 .001 126.03 (110.17, 141.88) <0.001 RC plus UD No RC plus UD 0.503 ( 6.241, 5.235) 0.864 0.354 ( 5.786, 5.079) 0.899 Time 0.698 ( 1.585, 0.189) 0.123 0.703 ( 1.588, 0.182) 0.12 RC plus UD*Time No RC plus UD*Time 3.153 ( 4.649, 1.657) <0.001 3 .143 ( 4.636, 1.649) <0.001 Age 0.815 ( 1.043, 0.587) <0.001 Propensity Score

PAGE 65

65 Table 2 5 Continued Model 3 Model 4 Referent Estimates 95%CL p value Estimates 95%CL p value Intercept 124.77 (1 07.879, 141.660) <0.001 119.426 (84.35, 154.50) <0.001 RC plus UD No RC plus UD 0.414 ( 6.110, 6.937) 0.901 1.672 ( 5.350, 8.695) 0.641 Time 0.706 (0.452, 1.591) 0.118 0.693 ( 1.579, 0.193) 0.125 RC plus UD*Time No RC plus UD*Time 3.145 ( 4.638, 1.652) <0.001 3.144 ( 4.640, 1.649) <0.001 Age 0.815 ( 1.044, 0.587) <0.001 0.83 ( 1.063, 0.596) <0.001 Propensity Score 1.774 ( 6.572, 10.119) 0.677 2.442 ( 5.991, 10.876) 0.57 RC plus UD, radical cystectomy plus urinary diversion Model 1 un adjusted Model 2 adjusted for age Model 3 adjusted for age and propensity score Model 4 adjusted for age, propensity score, radiotherapy, chemotherapy, intravesical anticarcinogenic, ACE inhibitor/angiote nsin II receptor blocker anticonvulsant, antin eoplastic, ASA, beta lactam antibiotics, calcineurin inhibitor, H2 blocker, penicillin, platinum based antineoplastic, vancomycin, vasodilator, and xanthine oxidase inhibitor

PAGE 66

66 Table 2 6 Association between radical cystectomy plus urinary diversion and time to percent change in MDRD 30% using unadjusted and multivariable cox proportional hazards models Model 1 Model 2 Referent Hazard Ratio 95%CL p value Hazard Ratio 95%CL p value RC plus UD No RC plus UD 1.88 (1.35 2.63) <0 .001 1.87 (1.34 2.62) <0.001 Age 1.02 (1.00 1.03) 0.032 Propensity Score

PAGE 67

67 Table 2 6. Continued Model 3 Model 4 Referent Hazard Ratio 95%CL p value Hazard Ratio 95%CL p value RC plus UD No RC plus UD 1. 3 (0.81 2.08) 0.273 1.01 (0.62 1.63) 0.976 Age 1.02 (1.00 1.03) 0.025 1.02 (1.01 1.04) 0.009 Propensity Score 2.3 (1.09 4.84) 0.029 2.08 (0.76 5.67) 0.153 RC plus UD, radical cystectomy plus urinary diversion Model 1 unadjusted Model 2 adjuste d for age Model 3 adjusted for age and propensity score Model 4 adjusted for age, propensity score, diabetes mellitus, hydronephrosis, clinical T, clinical N, antiviral, beta lactam antibiotics, and vancomycin

PAGE 68

68 Ta ble 2 7. Association between radical cystectomy plus urinary diversion and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) using unadjusted and multivariable logistic regression models Model 1 Model 2 Referent OR 95%CL p value OR 95%CL p value RC plus UD No RC plus UD 2.07 (1.15 3.70) 0.015 2.01 (1.11 3.63) 0.021 Age 1.03 (1.00 1.06) 0.024 Propensity Score

PAGE 69

69 Table 2 7. Continued Model 3 Model 4 Referent OR 95%CL p value OR 95%CL p value RC plus UD No RC plus UD 1.34 (0.60 3.16) 0.473 1.3 0 (0.54 2.87) 0.556 Age 1.03 (1.00 1.06) 0.025 1.02 (0.99 1.06) 0.157 Propensity Score 1.13 (0.18 7.12) 0.892 1.61 (0.19 13.76) 0.657 RC plus UD, radical cystectomy plus urinar y diversion ; OR, Odds Ratio Model 1 unadjusted Model 2 adjusted for age Model 3 adjusted for age and propensity score Model 4 adjusted for age, propensity score, baseline eGFR, diabetes mellitus, ACE inhibitor/ angiotensin II receptor blocker, H2 block er, and beta lactam antibiotics

PAGE 70

70 Figure 2 1. Plot of predicted mean of MDRD eGFR over time by radical cystectomy plus urinary diversion group

PAGE 71

71 Figure 2 2. Plot of predicted mean of MDRD eGFR over time by radical cystectomy plus urinary d iversion group

PAGE 72

72 Figure 2 3. Kaplan Meier survival curves of renal deterioration free ( 30% ) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion

PAGE 73

73 Figure 2 4. Kaplan Meier survival curves of renal deter ioration free ( 2 0% ) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion

PAGE 74

74 Figure 2 5. Kaplan Meier survival curves of renal deterioration free ( 25 % ) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion

PAGE 75

75 Figure 2 6. Kaplan Meier survival curves of renal deterioration free ( 40 % ) survival of bladder cancer patients with and without radical c ystectomy plus urinary diversion

PAGE 76

76 Figure 2 7. Kap lan Meier survival curves of renal deterioration free ( 57 % ) survival of bladder cancer patients with and without radical cystectomy plus urinary diversion

PAGE 77

77 CHAPTER 3 P REDICTORS OF RENAL FUNCTION DECLINE AFTER RADICAL CYSTECTOMY PLUS URINARY DIV ERSION AMONG PATIENTS WITH BLADDER CANCER Background Radical cystectomy (RC) plus urinary diversion (UD) has long been the gold standard treatment for muscle invasive bladder cancer (MIBC) and for treatment failure in non muscle invasive bladder cancer ( NMIBC) (Babjuk et al., 2013; Stein et al., 2001; Witjes et al., 2014) RC plus UD for bladder cancer is associated with significant risks of mortality and morbidity, particularly renal function decline and chronic kidney disease (CKD) (Kristjansson & Mansson, 2004; Manoharan et al., 2009) There are several factors that might contribute to an increased risk of renal function decline after RC plus UD for bladder cancer. These include older age, diabetes mellitus (DM) hypertension (HTN) body mass index (BMI), smoking, low baseline glomerular filtration rate (GFR) (Fox et al., 2004) family history of kidney disease (Freedman et al., 1997; Jurkovitz et al., 2002) heavy consumption of alcohol (Shankar et al., 2006) metabolic syndrome (Tanner et al., 2012) low socioec onomic status (McClellan et al., 2010) history of kidney stones (Vupputuri et al., 2004) cancer chemotherapy (Ries & Klastersky, 1986) and African American (Freedman et al., 1997) Native America (Pettitt et al., 1990) or Mexican American race (Tareen et al ., 2005) Given the association between CKD and several negative health outcomes and related high cost (Eckardt et al., 2013; Go et al., 2004; Matsushita et al., 2010; National Institutes of Health 2013) it is e ssential to identify most significant risk factors for developing renal function decline after RC plus UD for treatment of bladder cancer. Therefore, we conducted a retrospective study of patients with bladder cancer who underwent RC plus UD and sought car e at a tertiary health care center between the years 2000 to 2014 to to assess

PAGE 78

78 whether the type of urinary diversion affects the renal function and to identify the predictive factors for renal function declin e, as measured by estimated g lomerular filtratio n rate (e GFR ) after RC plus UD for bladder cancer. We hypothesized that the odds of renal impairment or rapid decline in renal function after RC plus UD will significantly differ across different types of urinary diversion, and patient characteristics and co morbid conditions We believe that the results from this study will identify predictive factors that lead to increased risk of renal impairment after RC plus UD for bladder cancer, thus leading to targeted risk factor reduction prior to RC plus UD. Ult imately, prevention of CKD after RC plus UD for bladder cancer would reduce negative health and cost of care consequences related to CKD. Methods Study Participants In this single center retrospective study, we identified 517 men and women aged 18 years an d older diagnosed as having bladder cancer according to the International Classification of Disease, Ninth Revision, Clinical Modification (ICD 9 CM) who sought care at our academic health care center from January 2000 to December 2014. We included bladder cancer patients who underwent RC plus UD according to Current Procedural Terminology (CPT) codes or Facility Oncology Registry Data Standards (FORDS) site spec ific surgery codes (Appendix B) and who had a minimum of two measure ments of serum creatinine (S Cr). We excluded those who had CKD stage G5, defined as an MDRD eGFR< 15ml/min/1.73m 2 or dialysis at the baseline and those with <30 days between their f irst and last SCr measurements. O f 517 adult patients with bladder cancer 172 patients met our inclus ion and exclusio n criteria and formed our analytic cohort. We used already collected inpatient and outpatient information using the

PAGE 79

79 University of Florida integrated data repository and cancer registry of our health care center. The study was approved by th e In stitutional Revi ew Board at t he University of Florida. Radical Cystectomy plus Urinary Diversion The surgical procedure information for each pati ent was obtained according to Current Procedural Terminology (CPT) codes and Facility Oncology Registry Data Standards (FORDS) site specific surgery codes (Appendix B).We used the following CPT codes for ascertainment of radical cystectomy: 51570, 51575, 51590, 51595, 51596, and 51597 (Appendix B).We used the following FORDS site specific surgery codes for a scertainment of radical cystectomy: 50, 60, 61, 62, 70, 71, and 72 (Appendix B). We categorized type of urinary diversion into thr ee categories: ileal conduit, cu taneous or orthotopic continent diversion, and other type of reconstruction. We identified pat ients with ileal conduit diversion according to CPT codes of 51590 or 51595 or FORDS site specific surgery code of 61 concomitant with a CPT or FORDS site specific surgery codes for radical cystectomy. We identified patients with cutaneous or orthotopic co ntinent diversion according to CPT code of 51596 or FORDS site specific surgery code of 62 concomitant with a CPT or FORDS site specific surgery codes for radical cystectomy. Those patients with a with a CPT or FORDS site specific surgery codes for radical cystectomy, but with no CPT or FORDS site specific surgery code for UD were categorized as other type of reconstruction. Measurement of Renal Function We used the Modification of Diet in Renal Disease (MDRD) (Levey et al., 2006) study equation to estimate g lomerular filtration rate based on the SCr, age, gender, and race (Levey et al., 1999; Stevens et al., 2006) The isot ope dilution mass spectrometry

PAGE 80

80 (IDMS) traceable MDRD study equation is expressed as follows: eGFR (ml/min/1.73m 2 ) = 175 x (Scr) 1.154 x (Age) 0.203 x (0.742 if female) x (1.212 if African American) (conventional units) (Levey et al., 2006) For all patients we included a baseline SCr measurement within four weeks before the date of surgery and all the subsequent SCr measurements after 30 days from the surgery. Our outcome of interest was renal function decline which was defined by the following methods: (a) change in (MDRD) (Levey et al., 2006) eGFR over time (eGFR slope) ; (b) time to percentage 30% (Coresh et al., 2014) ; and (c) rapid decline in renal function, defined by a decrease in the MDRD eGFR > 3 ml/min/1.73m 2 /year (de Boer et al., 2009; Rif kin et al., 2008) The MDRD eGFR slope was determined using a linear mixed model regression of change in MDRD eGFR with time using all estimated GFR measurements from the baseline and follow up period. Percentage change in MDRD eGFR was calculated as (la st eGFR baseline eGFR)/(baseline eGFR)*100% (Coresh et al., 2014) 30% was considered an event. We additionally performed sensitivity analyses examining cutoff points of percentage 40%, 20% F or estimation of rapid decline in renal function, in 90 patients with a follow up time of one year or longer, a subject specific MDRD eGFR slope was determined as an annual change estimated from a least square regression model using all estimated GFR measu rements from the baseline and follow up period. Rapid decline in renal function was defined by a decrease in the MDRD eGFR >3 ml/min/1.73m 2 / year based upon previously published literature (de Boer et al., 2009;

PAGE 81

81 Rifki n et al., 2008) The cutpoint represents a three fold increase compared to annual change in eGFR due to normal aging (Hemmelgarn et al., 2006; Lindeman et al., 1985) Covariates The conceptual framework and the pr evious literature guided the selection of covariates. We obtained information on treatment modalities including chemotherapy, radiotherapy, and intravesical instillation of anticarcinogenic agents according to CPT d characteristics included information on age, gender, race, marital status, primary insurance, body mass index (BMI), and smoking status. Co morbid conditions at the baseline included congestive heart failure (CHF), peripheral vascular disease (PVD), myo cardial infarction (MI), cerebrovascular disease, chronic pulmonary disease, connective tissue disease rheu matic (DM dementia, HIV/AI D, alcohol abuse, hypertension ( HTN ) urinary tract infection (UTI), nephrolithiasis, urinary obstructions, and hydronephro sis. We obtained information on the stage of the bladder cancer according to the American Joint Committee on Cancer (AJCC) tumor, node, and metastasis (TNM) staging classification. We obtained both clinical and pathologic staging data, but due to incomplet eness of the pathologic data, we included only clinical staging data in our analysis. We used all available medication data from the baseline and the follow up period. We imputed the missing data using multiple imputation (PROC MI). The proportions (%) of missing data were as follows: co morbid conditions ( 15 %), BMI ( 36 %), smoking status ( 15 %), clinical TNM stage ( 2 %), and medications (3 %).

PAGE 82

82 Statistical Analysi s The baseline demographics and clinical characteristics compared according to the type of UD usi ng one way analysis of variance (ANOVA) and chi squared or tests for continuous and categorical variables, respectively. We identified predictive factors such as type of UD patient characteristics, co morbid conditions, treatment modalitie s, and medications for (a) change in (MDRD) (Levey et al., 2006) eGFR over time (eGFR slope), (b) time to percentage 30% and (c) odds of rapid decline in renal function, defined by a decrease in the MDRD eGFR > 3 ml/min/1.73m 2 /year For the outcome of eGFR slope, we used linear mixed models regression of change in MDRD eGFR w ith time using all estimated GFR measurements from the baseline and follow up period. We constructed a multivariable linear mixed model (PROC MIXED) with residual maximum likelihood estimation (REML), random intercept, and unstructured covariance. For the outcome of time to percentage change in MDRD 30%, we used cox proportional hazards models (PROC PHREG). We plotted survival probability of not having a renal deterioration event, defined by a decrease in the MDRD eGFR of 30% or greater, according to the type of UD using Kaplan Meier curves. We compared the renal deterioration free survival distributions across different types of UD, using log rank test. We repeated these analyses for time to percentage change in MDRD eGFR with cutoff values of MDRD e 25%, and 20% (Hemmelgarn, et al., 2006). Finally, for the outcome of odds of rapid decline, we used logistic regression models (PROC LOGISTIC). For each outcome, candidate variables included in the full model were type of UD, age, gender, baseline eGFR, DM, and HTN a priori plus other clinically relevant variables that were associated with

PAGE 83

83 outcome at statistical significance of p<0.1. For the outcomes of time to eGFR % 30% and odds of rapid eGFR decline, we constructed mul tivariable cox proportional hazard and multivariable logistic regression model s respectively, using a backward elimination procedure. The goodness of fit of the fitted logistic regression model was assessed using Hosmer Lemeshow test (Hosmer & Lemeshow, 2000) The predictive performance of the fitted logistic regression model to de tect a rapid dec line in renal fu nction was assessed using area under the receiver operating characteristics curve (AUC). We assumed a missing at random (MAR) mechanism for the missing data on co morbid conditions BMI, smoking status, and cancer stag e Multiple i mputation with fully conditional specification (FCS) method was used to create 5 complete data sets using PROC MI. We used linear regression model and discriminant function method for imputation of continuous and categorical variables, respectively. Each o f the 5 complete data sets was analyzed using the appropriate statistical test (e.g. PROC MIXED, PROC PHREG, PROC LOGISTIC). The parameter estimates (e.g. coefficients and standard errors) obtained from each analyzed data set were then combined to generate statistical inference using PROC MIANALYZE. We did not have any missing data on age, gender, race, surgical procedures, and SCr; therefore, we did not impute any data with respect to the main predictor and primary outcome variables. The statistical s ignificance level was set at two sided alpha=0.05. Statistical Analysis Software (SAS) version 9.4 (SAS Institute Inc, Cary, North Carolina) was used to perform the statistical analyses.

PAGE 84

84 Results Characteristics of Study Participants The mean age of the 17 2 men and women who were included in the study was 6810 years. Majority of patients were males (79.7%), Caucasian (87.2%), married (71.5%), and former smoker (64.4%). Table 3 1 summarizes the demographics and baseline characteristics of bladder cancer pat ients who underwent radical cysctectomy according to the type of UD Out of 172 patients, 133 (77%), 25 (15%), and 14(8%) patients had ileal conduit, cutaneous or orthotopic continent diversion, and other types of reconstruction, respectively. Compared to patients who had ileal conduit or other types of reconstruction, those who had cutaneous or orthotopic continent diversion were younger (589 vs. 699 and 699 years; p value <0.001), less likely to be married (56% vs. 74% and 71%; p value= 0.038), and mo re likely to have private insurance as their primary payer (60% vs. 14% and 7%; p value <0.001). Compared to patients with ileal conduit and cutaneous or orthotopic continent diversion, those with other type of reconstruction were less likely to be male (1 4% vs. 84% and 92%; p value <0.001). Compared to patients with ileal conduit diversion, those with cutaneous or orthotopic continent diversion, and other types of reconstruction had higher baseline MDRD eGFR, although it was not statically significant (MDR D eGFR of 72.029.3, 79.124.0 and 81.024.8 ml/min/1.73m 2 for patients with ileal conduit, cutaneous or orthotopic continent diversion, and other types of reconstruction, respectively; p value= 0.420) (Table 3 1). Table 3 2 presents distributions of co mo rbid conditions at the baseline according to the type of UD. Approximately, 56% of patients had hypertension, 20% had DM 18% had chronic pulmonary disease, 17% had urinary tract obstruction or hydronephrosis,

PAGE 85

85 and 16% had UTI s Distributions of co morbid c onditions were relatively similar across three different types of UD. Only notable differences were distributions of hypertensio n and urinary obstruction/ hydronephrosis. Compared to patients who had ileal conduit or cutaneous or orthotopic continent divers ion those with other types of reconstruction were more likely to have hypertension, even though this difference was not statistically significant (85% vs. 52.6% and 52.6%; p value=0.060). None of the patients who underwent cutaneous or orthotopic continen t diversion had urinary obstruction or hydronephrosis at the baseline, whereas 20% and 14% of patients who underwent ileal conduit and other types of reconstruction had urinary obstruction or hydronephrosis (p value= 0.092) (Table 2 2). Table 3 3 presents distribution of clinical stage of bladder cancer according to the American Joint Committee on Cancer (AJCC) TNM classification by the type of UD. Approximately, 41% of patients had clinical T2, 63% had clinical N0, and 78% had clinical M0. Accordingly, 34 % of patients were categorized as clinical stage II. The relative distributions of clinical TNM groups were similar across the three different types of UD (Table 3 3). Table 3 4 compares medications use in patients according to the type of UD. In this cohort of bladder cancer patients, commonly used medications were beta lactam antibiotics (91.6%), analgesics (88.6%), beta blockers (77.3%), and fluoroquinolones (70.1%). Approximately, 58% of the patients received chemotherapy, 5% received radiotherapy, and 9% had intravesical instillation of anticarcinogenic agents (e.g. BCG). The relative proportions of medications use were similar across different types of UD with the exception of H2 blockers, nonsteroidal anti inflammatory drugs (NSAIDs), and

PAGE 86

86 statins. Compared to patients with ileal conduit or other types of reconstruction, patients with cutaneous or orthotopic continent diversion were more likely to use NSAIDs and less likely to use statins (Table 3 4). Predictors of Renal Function Decline T able 3 5 presents the association between type of UD and change in eGFR over time (eGFR slope) using a linear mixed model. Patients with cutaneous or orthotopic continent diversion had a steeper decline in renal function as compared to those with ileal con duit diversion (coefficient of interaction between cutaneous or orthotopic continent diversion and time: 3.614; 95% CI: 6.402, 0.826; p value= 0.011). In other words, patients with cutaneous or orthotopic continent diversion UD had mean eGFR decline of 6.775 ml/min/1.73m2/year (95% CI: 9.570, 4.533 ml/min/1.73m2/year) as compared to patients with ileal conduit diversion who had mean eGFR decline of 3.161 ml/min/1.73m2/year (95% CI: 4.516, 1.806 ml/min/1.73m2/year) (Figures 3 1 and 3 2). Patients wit h other type of reconstruction had mean eGFR decline of 1.931 ml/min/1.73m2/year (95% CI: 5.064, 1.092 ml/min/1.73m2/year) although this was not statically significant as compared to those with ileal conduit diversion (coefficient of interaction between other type of reconstruction and time: 1.230; 95% CI: 3.141, 5.600; p value= 0.581) (Table 3 5). Table 3 5 shows unadjusted associations between several parameters and change in eGFR over time using linear mixed models. Results from unadjusted linear mixe d models indicated that males experienced faster decline in eGFR over time as compared to females ( coefficient of interaction between male and time : 4.606 ; 95% CI: 7.390, 1.821 ; p val ue= 0.001). In other words, the mean eGFR decreased by 4.848 ml/min/1 .73m 2 /year and 0.242 ml/min/1.73m 2 /year in males and females, respectively

PAGE 87

87 (Figure 3 3). Compared to never smoker, current smokers and former smokers had faster decline of mean eGFR over time (coefficient of interaction between current smoker and former s moker and time: 7.016, 95% CI: 11.026, 3 007 ; p val ue < 0.001 and 5.034, 95% CI: 8.213 1.855 ; p val ue = 0.002, respectively). The mean eGFR in patients with and without diabetes decreased by 6.829 ml/min/1.73m 2 /year and 3.060 ml/min/1.73m 2 /year, r espectively (coefficient of interaction between diabetes and time: 3.769, 95% CI: 6.579, 0 .960 ; p val ue= 0.009) (Table 3 6 and Figure 3 3). Non platinum based and platinum based antineoplastic use was associated with faster decline in mean MDRD eGFR (co efficient of interaction between non platinum based and platinum based antineoplastic and time: 3.186, 95% CI: 5.496, 0 877 ; p val ue= 0.007 and 3.784, 95% CI: 6.090, 1 479 ; p val ue = 0.001, respectively). Patients who used diuretics or statins exper ienced faster decline in mean MDRD eGFR over time (coefficient of interaction between diuretic and statin and time: 3.093, 95% CI: 5.377, 0 809 ; p val ue= 0.008 and 3.878, 95% CI: 6.165, 1 591 ; p val ue < 0.001, respectively). Other variables associate d with faster decline in mean MDRD eGFR over time were primary tumor (clinical T3 as compared to Ta or Tis), regional lymph nodes (clinical N1 and N2, as compared to N0), AJCC stage group (stage IV as compared to 0 or 0is), primary payer (Medicare, Medicai d or VA, as compared to private insurance), higher baseline eGFR, and use of xanthine oxidase inhibitors (Table 3 6). Patients who received radiotherapy had increased mean MDRD eGFR over time. Angiotensin II receptor blocker s use slowed the decline in mea n MDRD eGFR over time, but this association was not statically significant (coefficient of interaction between a ngiotensin II receptor blocker s and time: 4.585, 95% CI: 9.846, 0.675 ; p val ue= 0.088).

PAGE 88

88 Table 3 7 presents adjusted association between progn ostic factors and change in eGFR over time using a multivariable linear mixed model. In the fully adjusted model, patients with cutaneous or orthotopic continent diversion had faster decline in mean eGFR over time as compared to patients with ileal conduit diversion; however, this difference was no longer statistically significant (coefficient of interaction between cutaneous or orthotopic continent diversion and time: 2.573, 95% CI: 0.246, 5.392 ; p val ue= 0.074). Other prognostic factors associated with c hange in mean eGFR over time were regional lymph nodes (clinical N1, N2, and NX, as compared to N0), distant metastasis (MX, as compared to M0), radiotherapy, beta blocker and xanthine oxidase inhibitor use (Table 3 7). The Kaplan Meier survival met hod was used to estimate the survival probability 30% from the baseline, in patients according to the type of UD (Figure 3 5). This analysis was repeated for time to percen tage change in eGFR with cutoff values of 57% (Figures 3 6 to 3 9). The median of fol low up time was 13 (5 24) months Throughout the study period, renal deterioration, defined by a decrease in the eGFR of 30% or greater, wa s observed in 77(45%) of the 172 patients with RC, including 60(45%), 10(40%), and 7(50%) patients in the ileal conduit diversion, cutaneous or orthotopic continent diversion, and other type of recons truction group respectively. The time to 30%) analysis indicated that the estimated survival function for patients with RC did not differ according to the type of UD (log rank test p value= 0.9605) (Figure 3 3). Repetition of this analysis with different cutoff values 40%, 57% confirmed that the estimated survival

PAGE 89

89 function for patients with radical cystectomy did not differ according to the type of UD 57%, respectively) (Figur es 3 6 and 3 9). Table 3 8 shows unadjusted associations between several parameters and time to renal deterioration, defined by a decrease in the eGFR of 30% or greater, using Cox proportional hazards models. There was no difference in rates of renal dete rioration throughout the study period across different types of UD (unadjusted HR= 0.97 and 1.11; 95%Cl: 0.50 1.90 and 0.51 2.43; p value=0.930 and 0.797 for cutaneous or orthotopic continent diversion and other types of reconstruction, respectively, as co mpared to ileal conduit). The rate of renal deterioration throughout the study period among patients who received chemotherapy was 88% larger than among those who did not receive chemotherapy (unadjusted HR=1.88, 95% CL: 1.17 3.05; p value= 0.010). Other v ariables significantly associated with higher rates of renal deterioration throughout the study period were regional lymph nodes (clinical N2, as compared to N0), race (other, as compared to white), and beta blocker, f luoroquinolone and v ancomycin use (Ta ble 3 8). Table 3 9 presents adjusted association between several parameters and time to renal deterioration, defined by a decrease in the eGFR of 30% or greater, using a multivariable Cox proportional hazards model. Patients with cutaneous or orthotopic continent diversion and other types of reconstruction had higher rates of renal deterioration, as compared to patients with ileal conduit diversion, but the difference was not statistically significant (adjusted HR=1.45 and 2.16; p value =0.332 and 0.152 for cutaneous or orthotopic continent diversion and other types of reconstruction

PAGE 90

90 respectively). In the adjusted model, variables significantly associated with higher rates renal deterioration throughout the study period were race (other, as compared to white) regional lymph nodes, (clinical NX, as compared to N0), chemotherapy, and vancomycin use (Table 3 9). Figure 3 10 illustrates cumulative hazards curves of renal 30%) according to the predictive factors using multivariable Cox p roportional hazards model. Among 90 patients with a follow up period of one year or longer, rapid decline in renal function, defined by eGFR >3 ml/min/1.73m 2 /year, was observed in 45(50%) patients including 32(48%), 7(58%), and 6(55%) patients in the ile al conduit diversion, cutaneous or orthotopic continent diversion, and other type of recons truction groups, respectively. Table 3 10 shows unadjusted associations between several parameters and r apid decline in renal function, defined by MDRD eGFR >3 ml/mi n/1.73m 2 /year, among 90 patients with a follow up period of one year or longer, using logistic regression models. Patients with cutaneous or orthotopic continent diversion and other types of reconstruction had higher odds of developing r apid decline in ren al function, as compared to patients with ileal conduit diversion, but these associations were not statistically significant (unadjusted OR=1.53 and 1.31; 95%Cl: 0.44 5.31 and 0.37 4.72; p value =0.332 and 0.152 for cutaneous or orthotopic continent divers ion and other types of reconstruction respectively). We did not find any statistically significant association between selected variables and odds of rapid decline in renal function in the univariate analysis (Table 3 10). Table 3 11 presents adjusted ass ociation between pre selected clinically relevant variables and rapid decline in renal function, defined by eGFR >3 ml/min/1.73m 2 /year,

PAGE 91

91 among 90 patients with a follow up period of one year or longer, using a multivariable logistic regression model. None o f the selected variables were associated with odds of developing rapid decline in renal function (Table 3 10). We assessed the predictive performance of the fitted multivariable logistic regression model to detect rapid decline in renal function, defined by eGFR >3 ml/min/1.73m 2 /year, using receiver operating characteristics curve (ROC) analysis. The ROC curve analysis indicated that t his fitted model poorly predicted developing rapid decline in renal function among 90 patients with a follow up period of o ne year or longer (AUC= 0.596; 95%Cl: 0.478 0.714) (Figure 3 11). Discussion In this retrospective study of 172 patients with RC plus UD, we found that patients who underwent cutaneous or orthotopic continent diversion experienced a faster decline in ren al function over time as measured by eGFR slope, as compared to those with ileal conduit diversion. We found that current and former smoking status, presence of diabetes at basel ine, clinical T3, N1, N2, Medicare, Medicaid or VA as primary payer, and high er baseline eGFR, and use of non platinum based and platinum based antineoplastics, statins, and xanthine oxidase inhibitor were associated with faster decline in mean eGFR, using unadjusted linear mixed models. But, in the adjusted linear mixed model, cut aneous or orthotopic continent diversion was no longer associated with faster decline in eGFR. In the adjusted model, clinical N1, N2, NX and use of xanthine oxidase inhibitor were associated with faster decline in mean eGFR. Using time to event analysis, we did not find any statically significant differences in rates of renal deterioration, defined by a decrease in the eGFR of 30% or greater, across different types of UD Results from the unadjusted Cox proportional hazards

PAGE 92

92 models suggested that chemothera py, clinical N2, race (other, as compared to white), and use of beta blocker, fluoroquinolone, and vancomycin were associated with higher rates of renal deterioration. In the adjusted Cox proportional hazards model, race (other, as compared to white), clin ical NX, chemotherapy, and vancomycin use remained associated with higher rates of renal deterioration throughout the study period. Our finding suggests that half of the patients with follow up period of one year or longer developed rapid decline in renal function, defined by eGFR >3 ml/min/1.73m 2 /year. Although the rate of rapid decline was higher in those with cutaneous or orthotopic continent diversion, as compared to ileal conduit diversion, the difference was not statistically significant. Unadjusted a nd adjusted logistic regression models failed to identify any predictive factors for rapid decli ne in renal function. To date a few studies have investigated the predictive factors of renal function decline after RC plus UD for bladder cancer. For example, in a retrospective study of 174 cancer patients, Samuel et al. indicated an initial postoperative GFR<50, recurrent sepsis, and hypertension were predictors of decreased GFR. However, only patients with ileal conduit UD were examined (Samuel et al., 2006) In another retrospective study, Jin et al. examined long term renal function outcomes after UD in a heterogeneous population (i.e. bladder cancer, shrunken bladder, hemorrhagic cystitis, and neurogenic bladder) an d identified hypertension, diabetes, and obstruction as risk factors for renal function deterioration (Jin et al., 2012) In a retrospective study of 70 patients with bladder cancer, Osawa et al. indicated that post operative acute pyelonephritis and chemotherapy were associated wi th renal deterioration after RC plus UD (Osawa et al., 2013) Nishikawa et al. found that baseline hypertension and acute

PAGE 93

93 pyelonephritis were indepen dently associated with renal deterioration, defined by a decrease in the eGFR of 25% or more, in a retrospective study of 169 patients with RC plus UD (Nishikawa et al., 2014) Nishikawa et al. reported that the typ e of UD was not associated with renal deterioration (Nishikawa et al., 2014) It should be noted that these studies were limited by at least one of the followings: inadequate ascertainment of the renal outcome, smal l sample size, heterogeneous population, inadequate adjustment for potentially confounding factors, and lack of repeated measures of renal function. Osawa et al. and Nishikawa et al. used the standard Japanese formula for eGFR (Nishikawa et al., 2014; Osawa et al., 2013) .Our study was different for the reason that we used three different widely acceptable methods to evaluate renal impairment in patients after RC plus UD. In our study, we found that patients who under went cutaneous or orthotopic continent diversion had a faster decline in renal function over time as measured by MDRD eGFR slope, as compared to those who underwent ileal conduit diversion. However, we did not find this association to be statically signif icant when we accounted for confounding variables. Our finding suggests that patients who underwent cutaneous or orthotopic continent diversion were 50% more likely to have rapid decline in renal function ( MDRD eGFR >3 ml/min/1.73m 2 /year ) than those with ileal conduit diversion, but this association was not statically significant. Furthermor e, we did not find any significant association between 30%) using time to event analysis. Prior studies did not report any significant association between type of UD and renal impairment (Gershman et al., 2015; Gilbert et al., 2013; Nishikawa et al., 2014; Zabell, Adejoro, Konety, & Weight, 2015) We found that chemotherapy was associated with

PAGE 94

94 renal deterioration as it was shown in prior studies (Osawa et al., 2013) Our results also sugges ted that clinical N1, N2, and NX and use of xanthine oxidase inhibitor and v ancomycin were independently associated with renal impairment. One of the strengt h of our study, in comparison with pr ior studies, was the evaluation of impact of medication use on renal impairment in patients with RC plus UD. Unadjusted analyses showed associations between non platinum based and platinum based antineoplastics, statins, xanthine oxidase inhibitor, beta bl ocker, fluoroquinolone, and vancomycin and renal impairment. Although in adjusted analyses only xanthine oxidase inhibitor and vancomycin found to be independently associated with renal impairment, it is important to note the effects of these medications o n renal function. Vancomycin induced renal toxicity has been reported in 10% to 40% of patients depending on the dosage (Elyasi, Khalili, Dashti Khavidaki, & Mohammadpour, 2012) In our study, we showed that among t his group of patients, the rate of renal deterioration throughout the study period in those with vancomycin use was two times larger than in those without vancomycin use. Increased risk of renal impairment with use of certain medications in this group of p atients suggests that more emphasis should be put on the avoidance of nephrotoxic drugs or appropriate adjustment of dosage of such medications. Our study had several limitations. First, this study is limited due to the retrospective nature of the analy sis. Second, we did not have data on death among this group of older patients. Therefore, it is possible that some patients with concomitant co morbid conditions might have died before they developed renal deterioration. In this situation, death would be c onsidered as competing risk. In other words, a competing

PAGE 95

95 risk is an event, which either precludes the observation of the outcome of interest or modifies its probability of occurrence (Gooley, Leisenring, Crowley, & S torer, 1999; Pintilie, 2007) In conventional methods for time to event analysis, such as the Kaplan Meier method or conventional Cox proportional hazards model, which we used in this study, the censoring mechanism is assumed to be noninformative, meaning that the survival time is assumed to be independent of the cause of censorship; the censoring occurs at random; and censored subjects are likely to have the same probability of having the outcome of interest as non censored subjects (Lau, Cole, & Gange, 2009; Noordzij et al., 2013; Satagopan et al., 2004) In the presence of competing risks, the subjects who have experience the competing event (e.g. death) should not be treated as censored at random since the competi ng event (e.g. death) hinders the probability that subjects experience the outcome of interest. Moreover, noninformative censoring requires that prognosis does not influence the censorship; which is not true in the presence of a competing event such as dea th. Therefore, the conventional methods for time to event analysis, such as the Kaplan Meier method or conventional Cox model, would be unsuitable since they ignore competing risks (Lau et al., 2009; Lin, 1997; Noord zij et al., 2013; Pepe & Mori, 1993) The proper approach to address this competing risk issue would be use of the cause specific proportional hazards model (Lunn & McNeil, 1995) the subdistribution proportional hazards model (Fine & Gray, 1999) or parametric mixture model (Lau, Cole, & Gange, 2011) However, all these approaches require the data on the competing risk (i.e. death). Since the data on mortality was not available, we were limited to use conventional methods for time to event analysis. That being said, it is important to re cognize that the risk is overestimated

PAGE 96

96 by standard Kaplan Meier analysis or conventional Cox proportional hazards model when competing events (i.e. death) are more common (Grams et al., 2012) Third, in this study we did not examine collinearity between variables. Therefore, potential multicollinearity between variables and overadjustment (Schisterman et al., 2009) might have been biased our results toward the null hypothesis. Additionally, i t is possible that small number of patients in cutaneous or orthotopic continent diversion and other type of reconstruction groups might have lead to our study being underpowered for analyzing categorical outcomes Finally, the results from this study have limited generalizability. In summary, we did not find any independent association between type of UD and renal function decline. We showed the chemotherapy was associated with higher risk of renal deterioration, defined by a decrease in the MDRD eGFR of 30% or greater, throughout the study period in both unadjusted and adjusted models. We also found certain medications such as vancomycin are associated with higher risk of renal deterioration throughout the study period. Furthermore, cons istent with prior studies, we showed that, following RC plus UD, patients experienced high rate of renal deterioration irrespective of the type of UD. This finding highlights the importance of close monitoring of renal function among these patients, and po tential need for nephrology consultation for all patients undergoing RC plus UD. It is likely that appropriate management of co morbid conditions such as diabetes and hypertension that are traditionally associated with renal impairment and avoidance of nep hrotoxic drugs such as vancomycin may prevent steep decline in renal function among this group of patients. We think more research needs to be conducted to further investigate the association between type of UD and renal impairment. The future prospective observational studies or clinical

PAGE 97

97 controlled trials should focus on not only recognition of other predictive factors of renal function decline, but also development of a risk assessment tool for identification of patients at increased risk of renal impairm ent following RC plus UD.

PAGE 98

98 Table 3 1 Sociodemographic and clinical characteristics of bladder cancer patients with radical cystectomy by type of urinary d iversion between 2000 and 2014 Type of Urinary Diversion All (n=172) Cutaneous or Orthoto pic Continent Diversion (n=25) Ileal Conduit (n=133) Other Types of Reconstruction (n=14) p value Age ,mean (SD) years 68(10) 58(9) 69(9) 69(9) < 0 .001 Male n, (%) 137(79.65) 23(92.00) 112(84.21) 2(14.29) < 0 .001 Race/ethnicity, n (%) White 1 50(87.21) 24(96.00) 114(85.71) 12(85.71) 0.147 African American 6(3.49) 0(0.00) 4(3.01) 2(14.29) Other 16(9.30) 1(4.00) 15(11.28) 0(0.00) Marital Status (%) Married 123(71.51) 14(56.01) 99(74.44) 10(71.43) 0.038 Primary Insurance, n (%) Medicare 114(66.28) 6(24.00) 98(73.68) 10(71.43) < 0 .001 Medicaid 14(8.14) 2(8.00) 10(7.52) 2(14.29) Private 34(19.77) 15(60.00) 18(13.53) 1(7.14) No insurance 6(3.49) 2(8.00) 3(2.26) 1(7.14) VA 4(2.33) 0(0.00) 4(3.01) 0(0.00) BMI + me an (SD) 28.0(5.8) 28.1(7.0) 27.9(5.7) 28.1(4.5) 0.987 Baseline MDRD eGFR, mean(SD) 73.8(28.3) 79.1(24.0) 72.0(29.3) 81.0(24.8) 0.319 Smoking Status n (%) Current Smoker 24(16.44) 5(26.32) 16(14.16) 3(21.43) 0.610 Former Smoker 94(64.38) 12(63.16 ) 74(65.49) 8(57.14) Never Smoker 28(19.18) 2(10.53) 23(20.35) 3(21.43) + BMI data were available for 110 out of 172 patients, with n =17 n= 86 and n= 7 in patients with continent diversion ieal conduit, and other types of reconstruction, respectively. *Smoking status data were available for 146 out of 172 patients, with n =19 n= 113 and n= 14 in patients with continent diversion, ieal conduit, and other types of reconstruction, respectively.

PAGE 99

99 Table 3 2. Co morbid conditions of patients with radical c ystectomy according to the type of urinary d iversion between 2000 and 2014 Type of Urinary Diversion All (n=172 ) Cutaneous or Orth otopic Continent Diversion (n=25 ) Il eal Conduit (n=1 33 ) Other Types of Reconstruction (n=14) p value Co morbi d Conditions n (%) Congestive Heart Failure 5(3.40) 0(0.00) 5(4.39) 0(0.00) 0.473 Myocardial Infarction 12(8.16) 0(0.00) 12(10.53) 0(0.00) 0.151 Peripheral Vascular Disease 5(3.40) 0(0.00) 5(4.39) 0(0.00) 0.473 Cerebrovascular Disease 5( 3.40) 1(5.26) 4(3.51) 0(0.00) 0.706 Dementia 1(0.68) 0(0.00) 1(0.88) 0(0.00) 0.864 Chronic Pulmonary Disease 27(18.37) 6(31.58) 17(14.91) 4(28.57) 0.129 Connective Tissue Disease/Rheumatic Disease 2(1.36) 0(0.00) 2(1.75) 0(0.00) 0.746 Diabetes 29(19.73) 2(10.53) 24(21.05) 3(21.43) 0.558 Hypertension 82(55.78) 10(52.63) 60(52.63) 12(85.71) 0.060 UTI 24(16.33) 3(15.79) 17(14.91) 4(28.57) 0.426 Nephrolithiasis 5(3.40) 2(10.53) 2(1.75) 1(7.14) 0.107 Urinary Obstruction or Hydronephrosis 25(17.01) 0(0.00) 23(20.18) 2(14.29) 0.092 Alcohol Abuse 4(2.72) 1(5.26) 3(2.63) 0(0.00) 0.651 Co morbid condition data were available for 147 out of 172 patients with n =19, n=114, and n=14 in patients with continent diversion, ieal conduit, and ot her types of reconstruction, respectively.

PAGE 100

100 Table 3 3 Clinical stage of bladder cancer, the AJCC TNM classification, of patients with radical cystectomy according to the type of urinary diversion between 2000 and 2014 Type of Urinary Diversi on All (n=172) Cutaneous or Orthotopic Continent Diversion (n=25) Ileal Conduit (n=133) Other Types of Reconstruction(n=14) p value Primary tumor (T) n (%) Ta 3(1.76) 0(0.00) 2(1.52) 1(7.69) 0.776 Tis 8(4.71) 1(4.00) 7(5.30) 0(0.00) T1 20(11.76) 2(8.00) 17(12.88) 1(7.69) T2 70(41.38) 11(44.00) 51(38.64) 8(61.54) T3 20(11.76) 2(8.00) 17(12.88) 1(7.69) T4 4(2.35) 1(4.00) 3(2.27) 0(0.00) TX 45(26.47) 8(32.00) 35(26.52) 2(15.38) Regional lymph nodes (N) N0 107(62.94) 13(5 2.00) 85( 64.39 ) 9( 69.23 ) 0.244 N1 5(2.94) 0(0.00) 5(3.79) 0(0.00) N2 4(2.35) 2(8.00) 2(1.52) 0(0.00) N3 3(1.76) 0(0.00) 2(1.52) 1(7.69) NX 51(30.00) 10(40.00) 38(28.79) 3(23.08) Distant metastasis (M) M0 132(78.11) 19(76.00) 102(77.86) 11( 8 4.62 ) 0.954 M1 1(0.59) 0(0.00) 1(0.76) 0(0.00) MX 36(21.30) 6(24.00) 28(21.37) 2(15.38)

PAGE 101

101 Table 3 3 Continued Type of Urinary Diversion All Cutaneous or Orthotopic Continent Diversion Ileal Conduit Other Types of Reconst ruction p value Clinical Stage Group 0a 4(2.35) 1(4.00) 2(1.52) 1( 7.69 ) 0.495 0is 6(3.35) 0(0.00) 6(4.55) 0(0.00) I 21(12.35) 2(8.00) 19(14.39) 0(0.00) II 59(34.71) 8(32.00) 43(32.58) 8(61.54) III 17(10.00) 2(8.00) 14(10.61) 1(7.69) IV 13 (7.65) 2(8.00) 10(7.58) 1(7.69) Unknown 50(29.41) 10(40.00) 38(28.79) 2(15.38) AJCC, American Joint Committee on Cancer TNM classification data were available for 170 out of 172 patients with n =25, n=132, and n=13 in patients with continent diversi on, ieal conduit, and other types of reconstruction, respectively.

PAGE 102

102 Table 3 4. Medications use in patients with radical cystectomy according to the type of urinary d iversion between 2000 and 2014 Type of Urinary Diversion All (n=1 72 ) Cutaneous or Orthotopic Continent Diversion (n=25 ) Ileal Conduit (n=133 ) Other Types of Reconstruction (n=14) p value Medications n (%) ACE inhibitor 50(29.94) 6(26.09) 38(29.23) 6(42.86) 0.520 Analgesics 148(88.62) 22(95.65) 113(86.92) 13(92.86) 0.417 Aminoglycoside 74(44.31) 9(39.13) 60(46.15) 5(35.71) 0.654 Angiotensin II receptor blocker 30(17.96) 3(13.04) 23(17.69) 4(28.57) 0.484 Anticonvulsant 2(1.20) 1(4.35) 1(0.77) 0(0.00) 0.317 Antineoplastic 58(34.73) 7(30.43) 47(36.15) 4(2 8.57) 0.764 Antiviral 4(2.40) 0(0.00) 4(3.08) 0(0.00) 0.558 ASA 61(36.53) 7(30.43) 48(36.92) 6(42.86) 0.734 Beta blocker 129(77.25) 15(65.22) 105(80.77) 9(64.29) 0.126 Beta lactam antibiotics 153(91.62) 21(91.30) 120(92.31) 12(85.71) 0.698 Calcineurin inhibitor 5(2.99) 1(4.35) 3(2.31) 1(7.14) 0.553 Calcium channel blocker 58(34.73) 11(47.83) 42(32.31) 5(35.71) 0.353 Diuretic 78(46.71) 7(30.43) 65(50.00) 6(42.86) 0.213 Fluoroquinolone 117(70.06) 16(69.57) 91(70.00) 10(71.43) 0.992 H2 blocker 74(44.3 1) 8(34.78) 64(49.23) 2(14.29) 0.027 NSAID 72(43.11) 17(73.91) 50(38.46) 5(35.71) 0.006 Platinum based antineoplastic 46(27.54) 6(26.09) 36(27.69) 4(28.57) 0.984 Statin 74(44.31) 5(21.74) 59(45.38) 10(71.43) 0.011 Vancomycin 99(59.28) 15(65.22) 76(58. 46) 8(57.14) 0.819 Vasodilator 36(21.56) 4(17.39) 31(23.85) 1(7.14) 0.307 Xanthine oxidase inhibitor 10(5.99) 0(0.00) 10(7.69) 0(0.00) 0.220

PAGE 103

103 Table 3 4 Continued Type of Urinary Diversion All (n=172 ) Cutaneous or Orth otopic Continent Dive rsion (n=25 ) Ileal Con duit (n=133 ) Other Types of Reconstruction (n=14) p value Treatment, n (%) Intravesical anticarcinogenic 15(8.72) 3(12.00) 10(7.52) 2(14.29) 0.570 Chemotherapy 99(57.56) 12(48.00) 77(57.89) 10(71.43) 0.360 Radio therapy 8(4.65) 2(8.00) 5(3.76) 1(7.14) 0.587 *Medications data were availa ble for 167 out of 172 patients, with n =23, n=130, and n=14 in patients with continent diversion, ieal conduit, and other types of reconstruction, respectively.

PAGE 104

104 Table 3 5 Association between type of urinary diversion and change in MDRD eGFR with time using a linear mixed model Referent Estimate 95%CL p value p value 1 Intercept 68.73 0 (63.833, 73.627) <0.001 Type of urinary diversion Cutaneous or orth otopic continent diversion Ileal conduit 7.369 ( 4.771, 19.510) 0.234 0.366 Other types of reconstruction Ileal conduit 7.323 ( 8.395, 23.042) 0.361 Time 3.161 ( 4.516, 1.806) <0.001 <0.001 Time*Type of urinary diversion Cutaneous or orthotopi c continent diversion Time*Ileal conduit 3.614 ( 6.402, 0.826) 0.011 0.025 Other types of reconstruction Time*Ileal conduit 1.23 0 ( 3.141, 5.600) 0.581 1 Type 3 tests of fixed e ffects

PAGE 105

105 Table 3 6 Association between clinical variable s and change in MDRD eGFR with time in patients with radical cystectomy plus urinary diversion using linear mixed regression models Estimate 95%CL p value UD (ileal conduit) Referent UD (cutaneous or orthotopic continent diversion) 7.369 ( 4.771, 19 .510) 0.234 UD (other types of reconstruction) 7.323 ( 8.395, 23.042) 0.361 Time (years) 3.161 ( 4.516, 1.806) <0.001 UD (ileal conduit)*Time Referent UD (cutaneous or orthotopic continent diversion)*Time 3.614 ( 6.402, 0.826) 0.011 UD (oth er types of reconstruction)*Time 1.230 ( 3.141, 5.600) 0.581 Age (years) 0.795 ( 1.220, 0.369) <0.001 Time (years) 2.114 ( 11.022, 6.795) 0.642 Age*Time 0.026 ( 0.156, 0.104) 0.698 Gender (male) 2.556 ( 13.186, 8.074) 0.637 Time (years) 0.2 42 ( 2.714, 2.230) 0.848 Gender (male)*Time 4.606 ( 7.390, 1.821) 0.001 Race (white) Referent Race (black) 12.494 ( 35.279, 10.291) 0.282 Race (other) 18.922 ( 33.354, 4.491) 0.010 Time (years) 3.868 ( 5.065, 2.670) <0.001 Race (white)* Time Referent Race (black)*Time 0.764 ( 4.697, 6.224) 0.784 Race (other)*Time 0.972 ( 6.283, 4.340) 0.720 Marital status (married) 2.582 ( 12.048, 6.884) 0.593 Time (years) 3.997 ( 5.864, 2.130) <0.001 Marital status (married)*Time 0.218 ( 2 .139, 2.576) 0.856

PAGE 106

106 Table 3 6 Continued Estimate 95%CL p value Primary insurance (Medicare) 4.619 ( 15.414, 6.177) 0.402 Primary insurance (Medicaid) 16.162 ( 1.120, 33.444) 0.067 Primary insurance (none) 16.973 ( 7.605, 41.551) 0.176 Primary insurance (VA) 13.120 ( 42.985, 16.745) 0.389 Primary insurance (private) Referent Time (years) 0.570 ( 3.951, 2.812) 0.741 Primary insurance (Medicare)*Time 3.297 ( 6.934, 0.339) 0.076 Primary insurance (Medicaid)*Time 6.231 ( 10.90 7, 1.555) 0.009 Primary insurance (none)*Time 2.157 ( 9.024, 4.712) 0.538 Primary insurance (VA)*Time 29.752 ( 54.588, 4.915) 0.019 Primary insurance (private)*Time Referent BMI(kg/m 2 ) 0.027 ( 0.381, 0.434) 0.897 Time (years) 1.511 (4.066, 7 .089) 0.594 BMI*Time 0.190 ( 0.383, 0.002) 0.052 Smoking status (never smoker) Referent Smoking status (current smoker) 17.778 (3.300, 32.256) 0.017 Smoking status (former smoker) 5.641 ( 4.830, 16.111) 0.288 Time (years) 0.704 ( 2.150, 3.558) 0.629 Smoking status (never smoker)*Time Referent Smoking status (current smoker)*Time 7.016 ( 11.026, 3.007 ) <0.001 Smoking status (former smoker)*Time 5.034 ( 8.213, 1.855 ) 0.002 Baseline MDRD eGFR (ml/min/1.73m 2 ) 0.762 (0.653, 0.871 ) <0.001 Time (years) 1.135 ( 1.837, 4.106) 0.454 Baseline MDRD eGFR*Time 0.072 ( 0.111, 0.032) <0.001

PAGE 107

107 Table 3 6 Continued Estimate 95%CL p value Congestive Heart Failure 2.364 ( 16.960, 21.688) 0.810 Time (years) 3.896 ( 5.099, 2.694) <0.001 Congestive Heart Failure*Time 0.269 ( 3.647 4.184) 0.893 Myocardial Infarction 1.443 ( 16.497, 13.610) 0.851 Time (years) 3.861 ( 5.041, 2.680) <0.001 Myocardial Infarction*Time 0.046 ( 4.582, 4.490) 0.984 Peripheral Vascular Disease 3.954 ( 1 5.150, 23.059) 0.685 Time (years) 3.805 ( 4.990, 2.620) <0.001 Peripheral Vascular Disease*Time 0.800 ( 5.180, 3.575) 0.720 Cerebrovascular Disease 3.595 ( 13.774, 20.965) 0.684 Time (years) 4.020 ( 5.186, 2.854) <0.001 Cerebrovascular Disea se*Time 3.348 ( 2.126, 8.821) 0.231 Dementia 7.020 ( 76.887, 62.847) 0.841 Time (years) 3.858 ( 4.997, 2.719) <0.001 Dementia*Time 19.695 ( 91.230, 51.839) 0.590 Chronic Pulmonary Disease 2.516 ( 6.690, 11.722) 0.592 Time (years) 3.750 ( 5.0 28, 2.472) <0.001 Chronic Pulmonary Disease*Time 0.542 ( 3.370, 2.286) 0.707 Connective Tissue Disease/Rheumatic Disease 2.058 ( 33.486, 29.370) 0.897 Time (years) 3.833 ( 4.974, 2.692) <.001 Connective Tissue Disease/Rheumatic Disease*Time 1 4.630 ( 63.905, 34.645) 0.515

PAGE 108

108 Table 3 6 Continued Estimate 95%CL p value Diabetes 0.059 ( 8.597, 8.715) 0.989 Time (years) 3.060 ( 4.342, 1.777) <0.001 Diabetes*Time 3.769 ( 6.579, 0.960) 0.009 Hypertension 1.093 ( 8.533, 10.71 8) 0.817 Time (years) 2.805 ( 4.769, 0.840) 0.005 Hypertension*Time 1.603 ( 4.026, 0.820) 0.195 UTI 1.167 ( 12.288, 9.954) 0.835 Time (years) 4.077 ( 5.320, 2.835) <0.001 UTI*Time 1.379 ( 1.760, 4.517) 0.389 Nephrolithiasis 17.354 ( 40.8 75, 6.166) 0.148 Time (years) 3.954 ( 5.115, 2.793) <0.001 Nephrolithiasis*Time 2.244 ( 3.735, 8.223) 0.462 Urinary Obstruction and Hydronephrosis 10.147 ( 20.574, 0.281) 0.056 Time (years) 3.966 ( 5.241, 2.691) <0.001 Urinary Obstruction an d Hydronephrosis*Time 0.582 ( 2.264, 3.428) 0.688 Alcohol Abuse 0.418 ( 20.149, 20.987) 0.968 Time (years) 4.098 ( 5.275, 2.921) <0.001 Alcohol Abuse*Time 3.904 ( 0.956, 8.764) 0.115

PAGE 109

109 Table 3 6 Continued Estimate 95%CL p value Clinical staging of the bladder cancer according to TNM classification Primary tumor (Ta, Tis) Referent Primary tumor (T1) 10.337 ( 10.713, 31.387) 0.336 Primary tumor (T2) 9.994 ( 8.199, 28.188) 0.282 Primary tumor (T3) 21.698 (0.723, 42.673) 0 .043 Primary tumor (T4) 22.017 ( 54.471, 10.438) 0.184 Primary tumor (TX) 13.090 ( 5.682, 31.862) 0.172 Time (years) 2.898 ( 8.943, 3.147) 0.348 Primary tumor (Ta, Tis)*Time Referent Primary tumor (T1)*Time 0.147 ( 6.355, 6.648) 0.965 Primary tumor (T2)*Time 1.055 ( 7.405, 5.295) 0.745 Primary tumor (T3)*Time 8.063 ( 14.961, 1.165) 0.022 Primary tumor (T4)*Time 3.234 ( 4.339, 10.807) 0.403 Primary tumor (TX)*Time 0.594 ( 5.921, 7.108) 0.858 Regional lymph nodes (N0) Referent Regi onal lymph nodes (N1) 41.905 (15.012, 68.798) 0.002 Regional lymph nodes (N2) 15.438 ( 12.494, 43.371) 0.279 Regional lymph nodes (N3) 7.554 ( 39.595, 24.486) 0.644 Regional lymph nodes (NX) 0.642 ( 8.707, 9.991) 0.893 Time (years) 1.858 ( 3.231, 0.485) 0.008 Regional lymph nodes (N0)*Time Referent Regional lymph nodes (N1)*Time 65.161 ( 94.935, 35.388) <0.001 Regional lymph nodes (N2)*Time 22.199 ( 26.360, 18.038) <0.001 Regional lymph nodes (N3)*Time 3.320 ( 3.975, 10.615) 0.372 Regional lymph nodes (NX)*Time 0.805 ( 3.490, 1.880) 0.557

PAGE 110

110 Table 3 6 Continued Estimate 95%CL p value Distant metastasis (M0) Referent Distant metastasis (M1) 16.464 (38.740, 71.668) 0.559 Distant metastasis (MX) 0.467 ( 9.776, 10.710) 0.929 Time (years) 3.462 ( 4.759, 2.164) <0.001 Distant metastasis (M0)*Time Referent Distant metastasis (M1)*Time 22.420 ( 28.955, 15.886) <0.001 Distant metastasis (MX)*Time 1.445 ( 1.345, 4.235) 0.310 AJCC stage group (0 a 0is) Referent AJCC stage group (I) 13.479 ( 8.273, 35.231) 0.225 AJCC stage group (II) 9.421 ( 9.9151, 28.757) 0.340 AJCC stage group (III) 11.567 ( 10.911, 34.044) 0.313 AJCC stage group (IV) 23.353 ( 0.220, 46.927) 0.052 AJCC stage group (unknown) 13.104 ( 6.426, 32.634) 0.189 Time (years) 3.220 ( 10.413, 3.974) 0.380 AJCC stage group (0, 0is)*Time Referent AJCC stage group (I)*Time 0.278 ( 7.278, 7.834) 0.943 AJCC stage group (II)*Time 1.667 ( 5.860, 9.193) 0.664 AJCC stage group (III)*Time 1.5 33 ( 6.436, 9.502) 0.706 AJCC stage group (IV)*Time 13.305 ( 21.203, 5.407) 0.001 AJCC stage group (unknown)*Time 1.086 ( 6.455, 8.628) 0.778

PAGE 111

111 Table 3 6 Continued Estimate 95%CL p value Intravesical anticarcinogenic 1.533 ( 16.753 13.688) 0.844 Time (years) 4.143 ( 5.336, 2.950) <0.001 Intravesical anticarcinogenic *Time 3.224 ( 0.812, 7.260) 0.117 Chemotherapy 6.837 ( 15.296, 1.623) 0.113 Time (years) 2.175 ( 4.189, 0.161) 0.034 Chemotherapy*Time 2.473 ( 4.916, 0 .029) 0.047 Radiotherapy 4.790 ( 15.218, 24.797) 0.639 Time (years) 4.127 ( 5.291, 2.964) <0.001 Radiotherapy*Time 6.400 (0.713, 12.086) 0.027 ACE inhibitor 6.858 ( 15.587, 1.870) 0.124 Time (years) 2.864 ( 4.506, 1.221) 0.001 ACE inhibitor *Time 1.883 ( 4.165, 0.398) 0.106 Analgesics 4.267 ( 17.557, 9.023) 0.529 Time (years) 2.209 ( 8.105, 3.688) 0.463 Analgesics*Time 1.711 ( 7.721, 4.300) 0.577 Aminoglycoside 2.849 ( 11.104, 5.405) 0.499 Time (years) 4.507 ( 6.105, 2.908) <0.001 Aminoglycoside*Time 1.340 ( 0.941, 3.621) 0.250 Angiotensin II receptor blocker 1.302 ( 12.100, 9.497) 0.813 Time (years) 4.391 ( 5.668, 3.114) <0.001 Angiotensin II receptor blocker *Time 2.573 ( 0.246, 5.392) 0.074

PAGE 112

112 Table 3 6 Continued Estimate 95%CL p value Antineoplastic 2.098 ( 6.797, 10.993) 0.644 Time (years) 2.007 ( 3.768, 0.247) 0.025 Antineoplastic*Time 3.186 ( 5.496, 0.877) 0.007 Antiviral 29.286 ( 58.320, 0.253) 0.048 Time (years) 3.870 ( 5.013, 2.728) <0.001 Antiviral*Time 0.194 ( 18.671, 19.060) 0.984 ASA 7.804 ( 16.316, 0.708) 0.072 Time (years) 4.303 ( 5.869, 2.737) <0.001 ASA*Time 0.974 ( 1.309, 3.258) 0.403 Beta blocker 14.415 ( 24.328, 4.5013) 0.004 Time (years) 3.616 ( 6.682, 0.550) 0.021 Beta blocker*Time 0.267 ( 3.568, 3.033) 0.874 Beta lactam antibiotics 11.919 ( 26.561, 2.723) 0.111 Time (years) 2.052 ( 7.641, 3.537) 0.472 Beta lactam antibiotics*Time 1.863 ( 7.575, 3.848) 0.522 Calcineurin inhibitor 19.390 ( 44.084, 5.303) 0.124 Time (years) 4.071 ( 5.249, 2.894) <0.001 Calcineurin inhibitor*Time 3.328 ( 1.308, 7.964) 0.160 Diuretic 1.393 ( 9.598, 6.813) 0.739 Time (years) 2.411 ( 3.968, 0.853) 0.002 Diuretic*Time 3.093 ( 5.377, 0.809 ) 0.008

PAGE 113

113 Table 3 6 Continued Estimate 95%CL p value Fluoroquinolone 4.869 ( 14.118, 4.380) 0.302 Time (years) 2.141 ( 5.335, 1.053) 0.189 Fluoroquinolone*Time 1.957 ( 5.375, 1.462) 0.262 NSAID 4.513 ( 3.708, 12.733) 0.282 Time (years) 4.469 ( 6.060, 2.878) <0.001 NSAID*Time 1.236 ( 1.045, 3.517) 0.288 Platinum based antineoplastic 5.462 ( 3.846, 14.771) 0.250 Time (years) 2.276 ( 3.770, 0.783) 0.003 Platinum based antineoplastic*Time 3.784 ( 6.090, 1.479) 0.001 S tatin 5.340 ( 13.485, 2.807) 0.199 Time (years) 2.102 ( 3.641, 0.564) 0.007 Statin*Time 3.878 ( 6.165, 1.591) <0.001 Vancomycin 1.081 ( 7.404, 9.567) 0.803 Time (years) 3.350 ( 5.240, 1.460) 0.001 Vancomycin *Time 0.805 ( 3.176, 1.566) 0.506 Vasodilator 6.608 ( 16.797, 3.582) 0.204 Time (years) 3.531 ( 4.827, 2.235) <0.001 Vasodilator*Time 1.377 ( 4.091, 1.337) 0.320 Xanthine oxidase inhibitor 14.048 ( 32.586, 4.491) 0.136 Time (years) 2.968 ( 4.166, 1.769) <0.001 Xanth ine oxidase inhibitor*Time 8.601 ( 12.368, 4.834) <0.001 BMI, body mass index ; UD urinary diversion

PAGE 114

114 Table 3 7 Multivariable linear mixed regression model predicting change in MDRD eGFR with time in patients with radical cystectomy plus urin ary diversion Estimate 95%CL p value Intercept 41.495 1.307 81.683 0.043 UD (ileal conduit) Referent UD (cutaneous or orthotopic continent diversion) 0.993 11.506 9.520 0.853 UD (other types of reconstruction) 2.591 10.264 15.447 0.693 Time ( years) 11.071 12.246 34.387 0.352 UD (ileal conduit)*Time Referent UD (cutaneous or orthotopic continent diversion)*Time 4.585 9.846 0.675 0.088 UD (other types of reconstruction)*Time 2.908 8.610 2.795 0.318 Age (years) 0.194 0.731 0.344 0.4 80 Age*Time 0.068 0.353 0.216 0.636 Gender (male) 1.937 7.012 10.887 0.671 Gender (male)*Time 2.863 6.989 1.263 0.174 Primary insurance (Medicare) 1.81897 12.1817 8.5437 0.7308 Primary insurance (Medicaid) 6.004 7.883 19.892 0.397 Primary ins urance (none) 4.287 15.167 23.741 0.666 Primary insurance (VA) 9.450 32.855 13.955 0.429 Primary insurance (private) Referent Primary insurance (Medicare)*Time 4.165 9.795 1.465 0.147 Primary insurance (Medicaid)*Time 1.294 6.204 8.793 0.735 Primary insurance (none)*Time 5.643 15.370 4.085 0.255 Primary insurance (VA)*Time 10.749 39.039 17.541 0.457 Primary insurance (private)*Time Referent BMI(kg/m 2 ) 0.090 0.483 0.303 0.652 BMI*Time 0.103 0.396 0.190 0.491 Baseline MDRD eGFR ( ml/min/1.73m 2 ) 0.684 0.556 0.813 <.0001 Baseline MDRD eGFR*Time 0.029 0.114 0.056 0.505

PAGE 115

115 Table 3 7. Continued Estimate 95%CL p value Smoking status (never smoker) Referent Smoking status (current smoker) 3.595 10.106 17.296 0.596 Smoking stat us (former smoker) 3.763 5.981 13.508 0.438 Smoking status (never smoker)*Time Referent Smoking status (current smoker)*Time 1.511 8.471 5.450 0.669 Smoking status (former smoker)*Time 1.315 5.432 2.802 0.531 Diabetes 3.976 11.988 4.036 0.33 0 Diabetes*Time 2.820 2.004 7.643 0.252 Hypertension 1.348912 10.247 7.5492 0.7562 Hypertension*Time 0.946 2.545 4.438 0.594 Urinary Obstruction and Hydronephrosis 9.987437 18.5456 1.4293 0.0223 Urinary Obstruction and Hydronephrosis*Time 2.85 8 7.295 1.580 0.206 Primary tumor (Ta, Tis) Referent Primary tumor (T1) 1.737 18.522 15.049 0.839 Primary tumor (T2) 3.508 17.799 10.783 0.630 Primary tumor (T3) 4.328 13.107 21.763 0.626 Primary tumor (T4) 6.140 31.344 19.064 0.633 Primar y tumor (TX) 1.799 15.412 19.010 0.838 Primary tumor (Ta, Tis)*Time Referent Primary tumor (T1)*Time 0.053 7.912 7.806 0.990 Primary tumor (T2)*Time 2.332 9.916 5.252 0.547 Primary tumor (T3)*Time 1.061 8.077 10.200 0.820 Primary tumor (T4)*T ime 1.373 14.222 11.476 0.834 Primary tumor (TX)*Time 0.035 8.935 9.004 0.994

PAGE 116

116 Table 3 7. Continued Estimate 95%CL p value Regional lymph nodes (N0) Referent Regional lymph nodes (N1) 15.859 7.007 38.724 0.174 Regional lymph nodes ( N2) 29.544 54.153 4.936 0.019 Regional lymph nodes (N3) 5.740 29.557 18.077 0.637 Regional lymph nodes (NX) 7.522 20.456 5.411 0.254 Regional lymph nodes (N0)*Time Referent Regional lymph nodes (N1)*Time 59.192 93.762 24.622 0.001 Regiona l lymph nodes (N2)*Time 22.407 33.787 11.027 <0.001 Regional lymph nodes (N3)*Time 3.603 14.759 7.554 0.527 Regional lymph nodes (NX)*Time 17.423 28.998 5.848 0.003 Distant metastasis (M0) Referent Distant metastasis (M1) 43.772 3.005 90.55 0 0.067 Distant metastasis (MX) 3.514 10.841 17.868 0.631 Distant metastasis (M0)*Time Referent Distant metastasis (M1)*Time 2.809 14.026 19.645 0.744 Distant metastasis (MX)*Time 18.353 7.156 29.549 0.001 Radiotherapy 0.876 16.397 14.645 0.912 Radiotherapy*Time 12.255 3.129 21.381 0.009 Angiotensin II receptor blocker 4.958 3.553 13.469 0.254 Angiotensin II receptor blocker *Time 0.449 3.417 4.316 0.820 Antineoplastic 1.042 11.826 9.742 0.850 Antineoplastic*Time 2.837 7.721 2.047 0. 255 Antiviral 5.618 28.279 17.043 0.627 Antiviral*Time 4.695 13.939 23.329 0.620 ASA 0.616 7.558 6.326 0.862 ASA*Time 1.084 4.590 2.421 0.544

PAGE 117

117 Table 3 7 Continued Estimate 95%CL p value Beta blocker 3.448 11.717 4.822 0.414 Beta blocker*Time 5.614 1.206 10.021 0.013 Diuretic 4.165 10.703 2.374 0.212 Diuretic*Time 0.452 2.858 3.763 0.789 Platinum based antineoplastic 3.260 8.409 14.929 0.584 Platinum based antineoplastic*Time 4.681 0.938 10.301 0.103 Statin 0.902 7.843 6.040 0.799 Statin*Time 0.047 3.861 3.767 0.981 Xanthine oxidase inhibitor 0.864 15.292 13.565 0.907 Xanthine oxidase inhibitor*Time 9.913 17.045 2.781 0.007 BMI, body mass index ; UD urinary diversion

PAGE 118

118 Table 3 8. Association b 30% in patients with radical cystectomy plus urinary diversion using univariable Cox proportional hazards models Hazard Ratio 95%CL p value UD (ileal conduit) Referent UD (cutaneous or orthotopic continent diversion) 0.97 0.50 1.90 0.930 UD (other types of reconstruction) 1.11 0.51 2.43 0.797 Age (years) 1.01 0.99 1.03 0.401 Gender (male) 1.69 0.91 3.13 0.096 Race (white) Referent Race (black) 1.44 0.52 3.97 0.485 Race (othe r) 2.51 1.31 4.79 0.005 Marital status (married) 0.85 0.53 1.37 0.502 Primary insurance (Medicare) 1.08 0.60 1.94 0.803 Primary insurance (Medicaid) 0.79 0.28 2.18 0.644 Primary insurance (none) 0.26 0.04 2.01 0.199 Primary insurance (VA) 2.58 0.74 9. 02 0.137 Primary insurance (private) Referent BMI(kg/m 2 ) 1.01 0.97 1.06 0.513 Smoking status (never smoker) Referent Smoking status (current smoker) 0.79 0.36 1.74 0.563 Smoking status (former smoker) 1.02 0.55 1.90 0.945 Baseline MDRD eGFR (ml /min/1.73m 2 ) 1.00 0.99 1.01 0.749 Congestive Heart Failure 1.65 0.59 4.66 0.341 Myocardial Infarction 1.21 0.50 2.91 0.676 Peripheral Vascular Disease 1.08 0.34 3.39 0.895 Cerebrovascular Disease 1.26 0.37 4.28 0.705 Chronic Pulmonary Disease 1.03 0.5 8 1.83 0.927

PAGE 119

119 Table 3 8. Continued Hazard Ratio 95%CL p value Diabetes 1.27 0.67 2.38 0.456 Hypertension 1.04 0.61 1.76 0.892 UTI 0.99 0.52 1.85 0.963 Nephrolithiasis 1.86 0.46 7.62 0.376 Urinary Obstruction and Hydronephrosis 1.65 0.95 2 .90 0.078 Clinical staging of the bladder cancer according to TNM classification Primary tumor (Ta, Tis) Referent Primary tumor (T1) 0.87 0.26 2.92 0.816 Primary tumor (T2) 1.46 0.51 4.17 0.481 Primary tumor (T3) 1.47 0.45 4.76 0.521 Primary tu mor (T4) 0.88 0.16 4.90 0.886 Primary tumor (TX) 1.31 0.43 3.95 0.636 Regional lymph nodes (N0) Referent Regional lymph nodes (N1) 2.74 0.83 8.99 0.097 Regional lymph nodes (N2) 5.53 1.96 15.60 0.001 Regional lymph nodes (N3) 2.50 0.60 10.35 0.207 Regional lymph nodes (NX) 1.51 0.91 2.52 0.114 Distant metastasis (M0) Referent Distant metastasis (M1) 1.58 0.22 11.48 0.650 Distant metastasis (MX) 0.95 0.54 1.65 0.850 AJCC stage group (0, 0is) Referent AJCC stage group (1) 0.90 0.23 3.45 0. 877 AJCC stage group (2) 1.82 0.55 6.09 0.329 AJCC stage group (3) 1.41 0.37 5.39 0.620 AJCC stage group (4) 2.52 0.66 9.64 0.178 AJCC stage group (unknown) 1.52 0.44 5.24 0.511

PAGE 120

120 Table 3 8. Continued Hazard Ratio 95%CL p value Intravesical anticarcinogenic 0.67 0.27 1.66 0.384 Chemotherapy 1.88 1.17 3.05 0.010 Radiotherapy 2.25 0.97 5.21 0.058 ACE inhibitor 1.34 0.84 2.16 0.219 Analgesics 1.37 0.66 2.87 0.400 Aminoglycoside 1.04 0.65 1.64 0.880 Angiotensin II receptor blocker 0.80 0.4 4 1.45 0.469 Antineoplastic 1.68 1.06 2.66 0.026 Antiviral 1.86 0.51 6.79 0.342 ASA 1.44 0.91 2.28 0.122 Beta blocker 2.14 1.12 4.07 0.021 Beta lactam antibiotics 3.49 0.84 14.50 0.086 Diuretic 1.41 0.88 2.26 0.154 Fluoroquinolone 1.87 1.07 3.29 0.0 29 NSAID 1.29 0.81 2.04 0.283 Platinum based antineoplastic 1.50 0.93 2.41 0.093 Statin 1.19 0.76 1.87 0.453 Vancomycin 2.06 1.24 3.44 0.006 Vasodilator 1.07 0.62 1.83 0.816 Xanthine oxidase inhibitor 1.90 0.87 4.16 0.108 BMI, body mass index ; UD urinary diversion

PAGE 121

121 Table 3 9 30% in patients with radical cystectomy plus urinary diversion using m ulitvariable Cox proportional hazards model Hazard Ratio 9 5%CL p value UD (ileal conduit) Referent UD (cutaneous or orthotopic continent diversion) 1.45 0.68 3.10 0.332 UD (other types of reconstruction) 2.16 0.75 6.16 0.152 Age (years) 1.02 0.99 1.05 0.178 Gender (male) 1.75 0.81 3.75 0.152 Race (white) Referent Race (black) 1.59 0.48 5.24 0.444 Race (other) 3.30 1.46 7.49 0.004 Baseline MDRD eGFR (ml/min/1.73m 2 ) 1.01 1.00 1.02 0.281 Diabetes 1.10 0.56 2.14 0.790 Hypertension 1.08 0.59 1.98 0.789 Urinary Obstruction and Hydronephrosis 1.05 0.51 2.19 0.891 Regional lymph nodes (N0) Referent Regional lymph nodes (N1) 2.23 0.53 9.31 0.271 Regional lymph nodes (N2) 3.11 0.92 10.52 0.067 Regional lymph nodes (N3) 1.67 0.35 7.95 0.518 Regional lymph nodes (NX) 1.85 1.03 3.32 0.038 Chemotherap y 1.96 1.14 3.37 0.016 Radiotherapy 1.42 0.55 3.68 0.472 Beta blocker 2.11 0.99 4.53 0.054 Beta lactam antibiotics 2.19 0.49 9.74 0.305 Fluoroquinolone 1.74 0.91 3.35 0.096 Vancomycin 1.98 1.10 3.55 0.022 UD, urinary diversion

PAGE 122

122 Table 3 10. A sso ciation between clinical variables and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) in patients with radical cystectomy plus urinary diversion using univariable logistic regression models Odds Ratio 95%CL p value UD (ileal condu it) Referent UD (cutaneous or orthotopic continent diversion) 1.53 0.44 5.31 0.665 UD (other types of reconstruction) 1.31 0.37 4.72 0.931 Age (years) 1.01 0.96 1.05 0.826 Gender (male) 1.40 0.55 3.53 0.481 Race (white) Referent Race (black) 3. 15 0.31 31.55 0.374 Race (other) 1.05 0.14 7.82 0.651 Marital status (married) 0.57 0.22 1.46 0.242 Primary insurance (Medicare) 0.82 0.27 2.54 0.966 Primary insurance (Medicaid) 0.66 0.11 4.00 0.962 Primary insurance (none) 1.31 0.17 10.26 0.972 Pri mary insurance (VA) >999.99 <0.001 >999.99 0.967 Primary insurance (private) Referent BMI(kg/m 2 ) 1.02 0.98 1.07 0.315 Smoking status (never smoker) Referent Smoking status (current smoker) 1.31 0.63 2.76 0.563 Smoking status (former smoker) 1.19 0.65 2.18 0.945 Baseline MDRD eGFR (ml/min/1.73m 2 ) 1.00 0.98 1.01 0.631 Congestive Heart Failure 0.82 0.33 2.00 0.657 Myocardial Infarction 1.04 0.42 2.57 0.926 Peripheral Vascular Disease 1.01 0.40 2.59 0.981 Cerebrovascular Disease 0.93 0.32 2.72 0 .892 Chronic Pulmonary Disease 1.14 0.63 2.04 0.666

PAGE 123

123 Table 3 10. Continued Odds Ratio 95%CL p value Diabetes 1.31 0.69 2.48 0.400 Hypertension 0.89 0.57 1.39 0.603 UTI 0.93 0.52 1.65 0.799 Nephrolithiasis 1.00 0.25 3.98 1.000 Urinary Obstruction and Hydronephrosis 0.87 0.49 1.54 0.632 Clinical staging of the bladder cancer according to TNM classification Primary tumor (Ta, Tis) Referent Primary tumor (T1) 0.51 0.16 1.58 0.242 Primary tumor (T2) 1.50 0.64 3.52 0.353 Primary tum or (T3) 2.53 0.79 8.14 0.119 Primary tumor (T4) 0.63 0.08 4.97 0.664 Primary tumor (TX) 1.48 0.58 3.74 0.411 Regional lymph nodes (N0) Referent Regional lymph nodes (N1) N/A N/A N/A N/A Regional lymph nodes (N2) 5357.74 0.00 >999.99 0.952 Regional lymph nodes (N3) 0.04 0.00 >999.99 0.945 Regional lymph nodes (NX) 0.07 0.00 >999.99 0.956 Distant metastasis (M0) Referent Distant metastasis (M1) 1777.28 0.00 >999.99 0.967 Distant metastasis (MX) 0.02 0.00 >999.99 0.967 AJCC stage group (0, 0is ) Referent AJCC stage group (1) 0.52 0.18 1.47 0.218 AJCC stage group (2) 1.48 0.63 3.49 0.367 AJCC stage group (3) 1.56 0.48 5.06 0.463 AJCC stage group (4) 1.38 0.36 5.32 0.637 AJCC stage group (unknown) 0.88 0.38 2.06 0.773

PAGE 124

124 Table 3 10. Continued Odds Ratio 95%CL p value Intravesical anticarcinogenic 0.73 0.15 3.48 0.695 Chemotherapy 1.88 0.81 4.36 0.140 Radiotherapy <0.001 <0.001 >999.99 0.971 ACE inhibitor 0.80 0.51 1.27 0.352 Analgesics 1.17 0.58 2.35 0.658 Aminoglycoside 0.76 0.50 1.15 0.195 Angiotensin II receptor blocker 0.90 0.56 1.45 0.664 Antineoplastic 0.93 0.60 1.43 0.729 ASA 0.75 0.48 1.19 0.223 Beta blocker 0.79 0.47 1.32 0.358 Beta lactam antibiotics 1.39 0.54 3.54 0.488 Diuretic 0.79 0.52 1.20 0.259 Fluo roquinolone 1.02 0.64 1.62 0.930 NSAID 1.25 0.81 1.92 0.322 Platinum based antineoplastic 1.03 0.65 1.63 0.891 Statin 0.90 0.58 1.38 0.623 Vancomycin 0.92 0.58 1.44 0.706 Vasodilator 0.82 0.50 1.34 0.420 Xanthine oxidase inhibitor 72.25 0.00 >999.99 0.976 BMI, body mass index; UD, urinary diversion

PAGE 125

125 Table 3 11. Association between clinical variables and rapid decline in renal function (MDRD eGFR >3 ml/min/1.73m 2 /year) in patients with radical cystectomy plus urinary diversion using multi variable logistic regression model Odds Ratio 95%CL p value UD (ileal conduit) Referent UD (cutaneous or orthotopic continent diversion) 1.03 0.38 2.81 0.958 UD (other types of reconstruction) 1.47 0.48 4.50 0.503 Age (years) 1.01 0.96 1.06 0.780 Gender (male) 1.20 0.68 2.11 0.524 Baseline MDRD eGFR (ml/min/1.73m 2 ) 1.00 0.98 1.01 0.661 Diabetes 1.38 0.70 2.74 0.351 Hypertension 0.81 0.50 1.33 0.408 UD, urinary diversion

PAGE 126

126 Figure 3 1. Plot of predicted mean of MDRD eGFR over time in blad der cancer patients according to the type of urinary diversion

PAGE 127

127 Figure 3 2. Plot of predicted mean of MDRD eGFR over time in bladder cancer patients according to the type of urinary diversion

PAGE 128

128 Figure 3 3 Plot s of predicted mean of MDRD eGFR over time in bladder cancer patients with radical cystectomy plus urinary diversion according to (A) gender, (B) smoking status, (C) diabetes mellitus, and (D) chemotherapy

PAGE 129

129 Figure 3 4 Plots of predicted mean of MDRD eGFR over time in bladder cancer pati ents with radical cystectomy plus urinary diversion according to (A) angiotenesin II receptor blockers, (B) diuretics, (C) platinum based antineoplastic, and (D) statins

PAGE 130

130 Figure 3 5 Kaplan Meier survival curves of renal deterioration free (MDRD eGFR 30) survival of bladder cancer patients according to the type of urinary diversion

PAGE 131

131 Figure 3 6 Kaplan Meier survival curves of renal deterioration 20) survival of bladder cancer patients according to the type of urinary diversion

PAGE 132

132 Figure 3 7 Kaplan Meier survival curves of renal deterioration free ( 25 ) survival of bladder cancer patients according to the type of urinary diversion

PAGE 133

133 Figure 3 8 Kaplan Meier survival curves of renal deterioration 40) sur vival of bladder cancer patients according to the type of urinary diversion

PAGE 134

134 Figure 3 9 Kaplan Meier survival curves of renal deterioration 57) survival of bladder cancer patients according to the type of urinary diversion

PAGE 135

135 Figure 3 10 Cumulative hazards curves of renal deterioration 30 ) in patients with radical cystectomy plus urinary diversion according to predictive factors (Male, age 68 years, race white, baseline MDRD eGFR 7 2.5 ml/min/1.73m 2 regional lymph nodes (cl inical N) N0, radiotherapy no, beta blocker no, fluoroquinolone no) VANCO, vancomycin; CTX, chemotherapy; DM, diabetes mellitus; HTN, hypertension; B lactam, beta lactam antibiotics; HYDRO, hydronephrosis

PAGE 136

136 Figure 3 11 ROC analysis of the fitted logi stic regression model to predict rapid decline in renal function (MDRD eGFR > 3 ml/min/1.73m 2 /year) in patients with radical cystectomy plus urinary diversion (AUC: 0.596, 95%CL: 0.478 0.714)

PAGE 137

137 CHAPTER 4 ACUTE KIDNEY INJURY AFTER UROLOGIC SURGERIES Backgro und Acute kidney injury (AKI) is common among hospitalized patients and is associated with substantial morbidity, mortality, and high health care costs (Bihorac et al., 2009; Case et al., 2013; Chertow et al., 2005; U chino et al., 2005; Xue et al., 2006) The incidence of AKI has been increasing, in part as a result of the expansion of invasive medical and surgical procedures (Xue et al., 2006) Thus far, the majority of stud ies that have investigated the incidence and risk factors for postoperative AKI examined populations other than urologic surgery (e.g. cardiothoracic surgery or general surgery). Little is known about frequency, risk factors, and renal outcome of AKI after surgeries in patients with urologic disorders. Given the fact that AKI is associated with high mortality, increased cost of care, prolonged hospitalization, and increased likelihood of chronic kidney disease ( CKD ) it is important to examine incidence and predictive factors of AKI in patients with urologic surgery. Therefore, we conducted a retrospective study of patients who were admitted to a tertiary referral hospital with for urologic surgeries between the years 2000 to 2010 with the following aims: 1) to identify the predictive factors for AKI within 30 days after urologic surgery; and 2) To identify the predictive factors for partial or no renal recovery after AKI within 30 days after urologic surgery. We hypothesized that in patients with urologic disorders who undergo surgery, the odds of postoperative AKI and the odds of partial or no renal recovery versus complete recovery after AKI episode would significantly d iffer across different patient characteristics and co morbid conditions We believe the res ults from this study will 1) inform patients and providers about the risk of

PAGE 138

138 AKI and likelihood of recovery from AKI after urologic surg ery and 2) i dentify predictive factors that lead to increased risk of AKI and partial or non recovery of AKI in this population, thus leading to targeted risk factor reduction prior to urologic surgery. Ultimately, prevention of AKI after urologic surgery would reduce negative health consequences and cost of care related to AKI. Methods Study Participants This a single center retrospective study of adult patients who were admitted to a tertiary referral hospital and underwent urological surgery procedure s between 2000 and 2010. We included patients who were hospitalized for more than 2 days (48 hours) and less than 90 days, and had three or more measurements of serum creatinine (SCr) within 90 days after admission date We excluded p atients who had CKD s tage 5 on admission date (i.e. established kidney failure, glomerular filtration rate (GFR) <15 mL/min/1.73 m 2 or a need for permanent renal replacement therapy (RRT)) and patients with hospital le ngth of stay more than 90 days. We excluded urologic proce dures with case counts less than 15. A total of 1,557 adult patients met our in clusion and exclusion criteria and formed our analytic cohort. We used a comprehensive database created by integrating the administrative, clinical, pharmacy, laboratory, and su rgery data The study was approved by the University of Florida Institutional Review Board. A cute Kidney Injury (AKI) AKI was defined by percentage of change in measured SCr levels during hospitalization compared to the baseline SCr level using the RIFLE classification (R renal risk, I injury, F failure, L loss of kidney function, E end stage renal disease) of

PAGE 139

139 the Acute Dialysis Quality Initiative Group (Bellomo et al., 2004) Patients who experienced a change in measured SCr level greater than 50% compared to the baseline SCr level were categorized as AKI. Patients with AKI were further class ified as RIFLE R, corresponding to a greater than 50% increase in SCr value, RIFLE I, corresponding to a 2 fold increase in SCr value, and RIFLE F, corresponding to a 3 fold increase in SCr value (Table 4 1 ). Renal Outcome after AKI We determined the renal outcome after AKI episode by comparing the last recorded SCr value within 90 days to the baseline SCr value. Renal outcome was classified into three categories: complete renal recovery, partial renal recovery, and no renal recovery. Complete renal recover y was defined as a SCr value that returned to a level less than 50% above the baseline SCr value. Partial renal recovery was defined by a persistent increase in SCr value with more than 50% above the baseline SCr value, but without any need for RRT No ren al recovery was defined as a need for RRT at the time of hospital discharge or death (Bihorac et al., 2010) In this study, we grouped partial or non recovery of renal function together compared to complete ren al recovery and we analyzed renal recovery after AKI as a dichotomous outcome. Predictors Clinical relevance and the previous literature guided the selection of covariates. in formation on age, gender, race, and primary insurance. The surgical procedures information was obtained according to the International Classification of Disease, Ninth Revision, Clinical Modification (ICD 9 CM) procedure codes. We excluded urologic procedu res with case counts less than 15. We grouped the remaining urologic

PAGE 140

140 procedures into five major groups as follows: 1) nephroureterectomy/total nephrectomy, 2) partial nephrectomy, 3) radical cystectomy, 4) radical prostatectomy, and 5) endoscopic/other typ e of urologic procedures. The radical prostatectomy group was considered as the referent in our analysis because based upon prior studies p atients undergoing radical prostatectomy experience relatively low rate of post operative AKI (Joo et al., 2016) The ICD 9 CM procedure codes that were used to group above pro cedures are listed in Appendix D Primary diagnosis and co morbid conditions were obtained according to ICD 9 CM diagnosis codes. We grouped the primary dia gnosis into 5 groups as follows: 1) malignant neoplasm of kidney, ureter, or renal pelvis, 2) bladder cancer, 3) prostate cancer, 4) nephrolithiasis, and 5) miscellaneous which included all remaining primary diagnosis codes. Co morbid conditions at the bas eline included congestive heart failure (CHF), peripheral vascular disease (PVD), myocardial infarction (MI), cerebrovascular disease, chronic pulmonary disease, connective tissue disease rheumatic disease, diabetes mellitus (DM), dementia, HIV/AID, hypert ension (HTN), liver disease, urinary obstruction/hydronephrosis, and urinary tract infections (UTIs). Additionally, Charlson comorbid ity index (Charlson, Pompei, Ales, & Mackenzie, 1987; Deyo, Cherkin, & Ciol, 1992) adapted for use with medical records, was used. We included baseline estimated glomerular filtration rate (eGFR) as calculated by the Chronic Kidney Disease Epidemiology Collaboration (CKD EPI) equation (Levey et al., 2009) The CKD EPI equation is expressed as follows: 1.209 X 0.993Age X 1.018 [if female] X 1.159 [if black] (Levey et al., 2009) We also obtained information on medications administered during hospitalization.

PAGE 141

141 Statistical Analysi s The baseline demographics and clinical characteristics were examined in patients with and without AKI and statistical di fferences were assessed using one way analysis of variance (ANOVA) or chi squared tests for continuous and categorical variables, respectively. We used univariate logistic regression to identify predictive factors, such as patient characteristics, primary diagnosis, urological surgery procedures, co mor bid conditions, and medications for AKI following urologic surgery. The multivariable logistic model was fitted using a backward elimination procedure. Candidate variables included in the backward elimination procedure were age, baseline eGFR, and urological surgery procedures a priori plus other clinically relevant variables associated with AKI in the univariate analysis at statistical significance of p<0.1. The statistical significance of interaction terms b etween variables in the model was examined. The goodness of fit of the statistical model was assessed using Hosmer Lemeshow test (Hosmer & Lemeshow, 2000) In the next step, predictive performance of the fitted models was assessed using area under the receiver operating characteristics curve (AUC). In patients with AKI, we used univariate logistic regression model to identify predictive factors, such as patient characteristics, co morbid conditions, type of the procedure, and medications, for partial or no renal recovery following AKI episode, using univariate logistic regression models. The multiv ariable logistic model was fitted using a backward elimination procedure. Candidate variables included in the backward elimination procedure were age, baseline eGFR, and urological surgery procedures a priori plus other clinically relevant variables associ ated with partial or no renal recovery in the univariate analysis at statistical significance of p<0.1. The statistical significance

PAGE 142

142 of interaction terms between variables in the model was examined. The goodness of fit of the statistical model was assessed using Hosmer Lemeshow test (Hosmer & Lemeshow, 2000) In the next step, predictive performanc e of the fitted models was assessed via receiver operat ing characteristics (ROC) curve analysis using leave one out cross validation method (Gonen, 2007; Molinaro, Simon, & Pfeiffer, 2005) In the leave one out cros s validation method, the data on one subject is set aside and a prediction model is built based upon the rest of the data, and then the fitted model is used to compute the predicted probability for the left out observation, and this process is repeated for each subject (PROC LOGISTIC, PREDPROBS=CROSSVALIDATE) (Gonen, 2007; Molinaro et al., 2005) The statistical significance level was set at two sided alpha=0.05. All statistical analyses were performed using Statisti cal Analysis Software (SAS) version 9.4 (SAS Institute Inc, Cary, North Carolina). Results Characteristics of Patients The mean age of the 1,557 men and women who were included in the study was 6015 years. The majority of patients were males (66.7%), Cau casian (82.0%), and lived in an urban area (69.5%). The primary urologic procedure group (%) distribution was: endoscopic/other type of urologic procedures (29.4%), nephroureterectomy/ total nephrectomy (27.9%), partial nephrectomy (12.6%), radical cystecto my (17.3%), and radical prostatectomy (12.8%). The primary diagnosis group (%) distribution was: malignant neoplasm of kidney, ureter, or renal pelvis (29.0%), bladder cancer (19.0%), prostate cancer (13.4%), and nephrolithiasis (12.7%), and miscellaneous (26.0%).

PAGE 143

143 The median (IQR) length of stay was 5(3 9) days. The median (IQR) length of stay by procedure category was: 5(3 8) days for endoscopic/other type of urologic procedures, 5(3 9) days for nephroureterectomy/total nephrectomy, 4(3 6) days for partia l nephrectomy, 9(7 13) days for radical cystectomy, and 3(2 4) days for radical prostatectomy. The median (IQR) number of SCr measurements was 5(4 9). The median (IQR) number of SCr measurements by procedure category was: 5(4 9) for endoscopic/other type o f urologic procedures, 6(4 10) for nephroureterectomy/total nephrectomy, 5(4 6) for partial nephrectomy, 10(7 13) for radical cystectomy, and 3(3 4) for radical prostatectomy. Table 4 2 compares the sociodemographics and clinical characteristics of patient s between according to the urologic procedure group. Patients who underwent radical cystectomy were older (6613 yrs), and those who underwent endoscopic/other type of urologic procedures were younger (5617 yrs). In our cohort, 20% and 10% of patients had emergency admission type and weekend admission day, respectively. Patients in endoscopic/other type of urologic procedure group were more likely to have an emergency admission type (46%, p value <0.001) and patients who underwent radical cystectomy were m ore likely to be admitted on a weekend (20%, p value<0.001). The mean baseline eGFR was higher in patients who underwent radical prostatectomy (8220 ml/min/1.73m 2 ) and was lower in radical cystectomy (7023 ml/min/1.73m 2 ) and nephroureterectomy/ total nep hrectomy (7024 ml/min/1.73m 2 ) group (Table 4 2). Table 4 3 lists the frequency and distribution of co morbid conditions according to urologic procedure group. Approximately, 45%, 19%, 18%, and 10% of patients had hypertension, diabetes mellitus, urinary o bstruction/hydronephrosis, and UTIs

PAGE 144

144 respectively. Patients in the endoscopic/other type of urologic procedures and the radical cystectomy group were more likely to have urinary obstruction/hydronephrosis and UTIs (Table 4 3). Table 4 4 lists the medicati on use among patients according to urologic procedure group. Approximately, 53%, 42%, 41%, 26%, and 18% of patients had beta blocker, diuretics, aminoglycoside, NSAIDS, and vancomycin use, respectively. Patients in endoscopic/other type of urologic procedu res and radical cystectomy group were more likely to have aminoglycoside and vancomycin use (Table 4 4). Distribution of AKI Of 1,557 patients included in this study, 601(39%) patients developed AKI. The incidence of AKI by procedure category was: 173(38%) for endoscopic/other type of urologic procedures, 200(46%) for nephroureterectomy/total nephrectomy 49(25%) for partial nephrectomy 142(53%) for radical cystectomy, and 37(19%) for radical prostatectomy (Table 4 5). Among 601 patients who developed AKI, the AKI RIFLE classification (%) distribution was: Risk (67%), Injury (20%), and Failure (13%). The percentage of patients with AKI RIFLE Injury and Failure were higher in radical cystectomy, endoscopic/other type of urologic procedures, and nephrouretere ctomy / total nephrectomy group (Table 4 5). Predictors of AKI Table 4 6 presents the unadjusted association between several parameters and AKI using univariable logistic regression models. In unadjusted models, older age, lower baseline CKD EPI eGFR, havi ng Medicare, Medicaid, or no insurance as compared to having private insurance, emergency admission type, and weekend admission day were associated with higher odds of AKI. Patients who underwent

PAGE 145

145 endoscopic/other type of urologic procedures, nephroureterec tomy/ total nephrectomy, or radical cystectomy had greater odds of AKI, as compared to those who underwent radical prostatectomy (Table 4 6). Patients in the primary diagnosis group of malignant neoplasm of kidney, ureter, or renal pelvis, bladder cancer, n ephrolithiasis, or miscellaneous had greater odds of AKI, as compared to those with primary diagnosis of prostate cancer. Also having congestive heart disease, myocardial infarction, chronic pulmonary disease, cancer, UTIs, and urinary obstruction/hydronep hrosis was associated with gr eater odds of AKI. Higher Charlson comorbidity index was associated with higher odds of AKI. Finally, a cetylsalicylic acid (ASA), beta blockers, diuretics, inotropes, vasopressors, and vancomycin use were associated with greate r odds of AKI, using univariable logistic regression models (Table 4 6). We included all parameters that were associated with AKI using univariate analysis in a multivariable logistic regression model. We used a backward elimination procedure to construct the final multivariable logistic regression model. The result from this fully adjusted model is presented in Table 4 7. After adjusting for covariates, patients who underwent radical cystectomy (OR=3.69; 95%Cl: 2.34 5.82; p value< 0.001), nephroureterectom y/ total nephrectomy (OR=3.42; 95%Cl: 2.24 5.24; p value< 0.001), endoscopic/other type of urologic procedures (OR=2.43; 95%Cl: 1.52 3.88; p value< 0.001), had greater odds of developing AKI, as compared to those who underwent radical prostatectomy. Patient s who underwent partial nephrectomy had higher odds of AKI, as compared to those who underwent radical prostatectomy, but this association it was not statically significant (OR=1.62; 95%Cl: 0.98 2.66; p value= 0.058). In the adjuste d model, worse baseline eGFR increased the odds of AKI, but this

PAGE 146

146 association was no longer statistically significant (OR=0.96; 95%Cl: 0.99 1.01; p value=0.749). Patients admitted on a weekend were 48% more likely to develop AKI than patients admitted on weekdays (OR=1.48; 95%Cl: 1.03 2.13; p value= 0.035). In the adjusted model, other predictive factors for higher odds of AKI were CHF (OR=1.86; 95%Cl: 1.13 3.04; p value= 0.014), cancer (OR=1.79; 95%Cl: 1.34 2.40; p value< 0.001), UTIs (OR=1.50; 95%Cl: 1.03 2.18; p value= 0.033), u rinary obstruction/hydronephrosis (OR=1.75; 95%Cl: 1.30 2.36; p value< 0.001), and vasopressors (OR=1.82; 95%Cl: 1.09 3.04; p value= 0.022), and vancomycin use (OR=1.78; 95%Cl: 1.32 2.41; p value< 0.001) (Table 4 7). The predictive performance of the fitt ed logistic regression models to predict AKI was assessed using ROC analysis. Four models were constructed and analyzed, and the result is reported in Table 4 8. The ROC analysis of these models are as follows: Model 1, age and baseline eGFR ( area under th e receiver operat ing characteristics curve (AUC) : 0.574; 95%Cl: 0.544 0.604); Model 2; age, baseline eGFR, and procedure type (AUC: 0.574; 95%Cl: 0.544 0.604); Model 3, age, baseline eGFR, procedure type, admission day, CHF, cancer, UTIs, and urinary obstr uction/hydronephrosis (AUC: 0.574; 95%Cl: 0.544 0.604); and Model 4, age, baseline eGFR, procedure type, admission day, CHF, cancer, UTIs, and urinary obstruction/hydronephrosis, vasopressors, and vancomycin (AUC: 0.574; 95%Cl: 0.544 0.604) (Table 4 8 and Figure 4 1 ). The assessment of the fitted prediction model (Model 4) via ROC analysis using leave one out cross validation method indicated that this model has a poor discriminative ability (AUC: 0.671; 95%Cl: 0.644 0.698) (Figure 4 2).

PAGE 147

147 Predictors of Par tial or Non Recovery of Renal Function after AKI Among 601 patients with AKI, 462(77%), 124(21%), and 15(2%) patients had complete, partial, and non recovery of renal function. The distribution (%) of partial or non recovery of renal function after AKI acc ording to urologic procedure group were as follows: endoscopic/other type of urologic procedures (17%), nephroureterectomy/total nephrectomy (43%), partial nephrectomy (16%), radical cystectomy (8%), and radical prostatectomy (14%) (Table 4 9). Among 601 patients who developed AKI following urologic procedures, the median (IQR) length of stay was 8(5 14) days. T he median (IQR) length of stay by procedure category was: 8(5 14 ) days for endoscopic/other type of urologic procedures, 8(4 13 ) days for nephroure terectomy/total nephrectomy 7(5 9 ) days for partial nephrectomy 11(8 17) days for radical cystectomy, and 4(3 7 ) days for radical prostatectomy. The median (IQR) number of SCr measurements was 9(6 14). The median (IQR) number of SCr measurements by proce dure category was: 9(6 15 ) for endoscopic/other type of urologic procedures, 8(5 14 ) for nephroureterectomy/total nephrectomy 8(6 11 ) for partial nephrectomy 11(8 18) for radical cystectomy, and 4(4 8 ) for radical prostatectomy. Table 4 10 shows the una djusted association between several factors and partial recovery or non recovery of renal function after AKI in patients, using univariable logistic regression models. Patients who underwent nephroureterectomy/ total nephrectomy were more likely to have par tial or non recovery of renal function after AKI than those who underwent radical prostatectomy. In unadjusted analysis, other variables associated with higher odds of partial or non recovery of renal function after AKI were greater baseline CKD EPI eGFR, AKI RIFLE Injury and Failure, as compared to AKI

PAGE 148

148 RIFLE Risk, PVD and cancer. Routine/elective admission type, UTIs, and urinary obstruction/hydronephrosis were associated with decreased odds of partial or non recovery of ren al function after AKI (Table 4 10 ). We included all parameters that were associated with partial or non recovery of renal function after AKI, using univariate analysis in a multivariable logistic regression model. We used a backward elimination procedure to construct the final multivar iable logistic regression model. The result from this fully adjusted model is presented in Table 4 11 .In the adjusted model, patients who underwent nephroureterectomy/ total nephrectomy were 7.15 times more likely than patients who underwent radical prostat ectomy to experience partial or non renal recovery after AKI (OR=7.15; 95%Cl: 2.43 21.07; p value< 0.001). As compared to patients with AKI RIFLE class Risk, patients with AKI RIFLE class of Injury and Failure were more likely to experience partial or non renal recovery. The adjusted analysis showed that increasing age and eGFR were associated with greater odds of partial or non recovery of renal function (OR=1.02; 95%Cl: 1.00 1.03; p value=0.049 and OR=1.17; 95%Cl: 1.06 1.28; p value= 0.001 for age and eGF R, respectively). UTIs was associated with smaller odds of partial or non recovery of renal function after AKI (OR=2.43; 95%Cl: 1.52 3 .88; p value< 0.001) (Table 4 11 ). The predictive performance of the fitted logistic regression models to predict partial or non recovery following AKI was assessed using ROC analysis. Four models were constructed and analyzed, and the result is reported in Table 4 12 The ROC analysis of these models are as follows: Model 1, age and baseline eGFR (AUC: 0.584; 95%Cl: 0.530 0 .638); Model 2; age, baseline eGFR, and RIFLE AKI Class (AUC: 0.683;

PAGE 149

149 95%Cl: 0.632 0.734); Model 3, age, baseline eGFR, RIFLE AKI Class, and type of urologic procedure (AUC: 0.801; 95%Cl: 0.761 0.842); and Model 4, age, baseline eGFR, RIFLE AKI Class, and type of urologic procedure, and UTI s (AUC: 0.816; 95%Cl: 0.777 0.855) (Table 4 12 and Figure 4 3 ). The assessment of the fitted prediction model (Model 4) via ROC analysis using leave one out cross validation method indicated that this model has a fair to good discriminative ability (AUC: 0.794; 95%Cl: 0.752 0.836) (Figure 4 4). Discussion Among 1,557 patients who underwent urologic procedures in a tertiary referral hospital, the incidence AKI was high. We found that AKI developed in 39% of patients inclu ding 38%, 46%, 25%, 53%, and 19% of patients in endoscopic/other type of urologic procedures, nephroureterectomy/total nephrectomy, partial nephrectomy, radical cystectomy, and radical prostatectomy group, respectively. Multivariate analysis demonstrated t hat radical cystectomy, nephroureterectomy/total nephrectomy, and endoscopic/other type of urologic procedures, as compared to radical prostatectomy, worse baseline eGFR, weekend admission, CHF, cancer, UTIs, urinary obstruction/hydronephrosis, and vasopre ssors, and vancomycin use were independently associated with greater odds of AKI. Furthermore, we found that among 601 patients with AKI, 139(23%) patients experienced partial or non recovery of renal function. Multivariate analysis showed that nephrourete rectomy/total nephrectomy, as compared to radical prostatectomy, AKI RIFLE Injury and Failure, as compared to AKI RIFLE Risk, and increasing age and baseline eGFR were independently associated with greater odds of partial or non recovery of renal function after AKI. Also, UTIs was associated with decreased odds of partial or non recovery of renal function after AKI.

PAGE 150

150 The incidence of AKI in our cohort is more than that reported for other non cardiac surgeries and comparable to what is reported in the literat ure for similar urologic procedures (Carmichael & Carmichael, 2003; Joo et al., 2016; Joung et al., 2016; Kheterpal et al., 2007; Kim, Bae, Ma, Kweon, & Kim, 2014) However, to date few studies have investigated th e incidence for AKI in urologic procedures, and the majority of these studies assessed AKI risk following only limited urologic procedures. For example, in a retrospective study of 557 patients with renal cell carcinoma (RCC), authors reported that the inc idence of AKI after partial nephrectomy and radical nephrectomy was similar to what we report for partial nephrectomy and somewhat higher than that we report for nephroureterectomy/total nephrectomy at 24% and 70%, respectively (Kim et al., 2014) Another retrospective study of 1,340 patients who underwent radical prostatectomy, reported that incidence AKI in robotic assisted laparoscopic radical prostatectomy (10.4%) was higher than in retropubic radical prostatecto my (5.5%) (Joo et al., 2016) Finally, Joung et al. retrospectively studied 202 patients who underwent radical cystectomy and reported 31% developed postoperative AKI (Joung e t al., 2016) Few studies have gone further to investigate the predictors of AKI after urologic procedures. In a recently published retrospective study of 233 patients with RCC, the authors showed that percent functional volume preservation and warm isch emia time can predict AKI after partial nephrectomy (Ding et al., 2016) Joung et al found that, among 202 patients with radical cystectomy, the AKI incidences did not differ between lieal conduit and neobladder gr oups (Joung et al., 2016) A nother r etrospective study reported an incidence of 6.7 % for AKI occurred in the hospitalized urology population

PAGE 151

151 and suggested urinary obstructio n and sepsis or a combination of both as e tiology of AKI (Caddeo, Williams, McIntyre, & Selby, 2013) The authors added that complete re nal recovery achieved in 57.7% of patients and found that non elective patients were more likely to have complete renal recovery (Caddeo et al., 2013) A key the strength of our study is comparative assessment of AKI risk across different type of urologic procedures. We examined the associations of several demographics, co morbid conditions and medications use variables an d odds of AKI in this cohort. Finally, w e also investigated the incidence and predictors of renal recovery after AKI episodes. Our study has several limitations. First, due to retrospective nature of this study several important clinical variables includi ng operation time, ischemia time, blood transfusion rate, crystalloid and colloid infused volume, and radiocontrast agent use were not assessed in our analysis. Second, the final multivariable regression models, which were built using backward elimination procedure, may not be optimal because variables are removed from the model one at a time and some variables may be missed. The removal of less statistically significant variables may increase the statistical significance of remaining variables, leading to exaggeration of significance of these variables. In order to address this, we compared fitted models using stepwise and forward selection methods to backward elimination procedure and we did not find any differences in the set of the selected variables. An other approach for the model selection would be criterion based procedures such as Akaike Information Criterion (AIC) and Bayes Information Criterion (BIC); however, we did not use these procedures in the present study. Finally, although we assessed the ac curacy of the fitted prediction

PAGE 152

152 models via ROC analysis using cross validation, these fitted prediction models must be externally validated in other populations. In summary, we demonstrated that the incidence of AKI is high in patients undergoing urolog ic procedures, ranging from 19% in patients undergoing radical prostatectomy to as high as 53% in patients undergoing radical cystectomy. We showed that radical cystectomy, nephroureterectomy/ total nephrectomy, and endoscopic/other type of urologic procedu res, as compared to radical prostatectomy, worse baseline eGFR, weekend admission, CHF cancer, UTIs, urin ary obstruction/hydronephrosis, and vasopressors and vancomycin use were independently associated with greater odds of AKI. Additionally, we found th at nephroureterectomy/ total nephrectomy as compared to radical prostatectomy, AKI RIFLE Injury and Failure, as compared to AKI RIFLE Risk, and increasing age and baseline eGFR were independently associated with greater odds of partial or non recovery of r enal function after AKI. We believe our findings will help inform urologic patients and providers about ris k and predictive factors of AKI and partia l or non recovery of renal function following AKI. Furthermore, t his insight can guide the management of th is patient populat ion in clinical settings such that modifiable pre, intra or post operative factors may be addressed to help prevent development of AKI and related negative health outcomes. For example, hemodynamic optimization, avoidance of nephrotoxic drugs such as vancomycin or appropriate drug adjustment can be implemented, particularly in patients who are at higher risk fo r developing post operative AKI, such as those with CHF and cancer. In the next step, the AKI and partial and non recovery predi ction models resulted from this study must be externally validated in other populations. If these models prove to be accurate, and with help of

PAGE 153

153 additional prospective or controlled trials which identify other clinically relevant parameters not addressed in this study, an AKI risk assessment tool can be constructed, tailored specifically for patients undergoing urologic procedures.

PAGE 154

154 Table 4 1 Risk, Injury, Failure, Loss, and End stage Kidney (RIFLE) classification of AKI (Bagshaw, George, Bellomo, et al., 2008) Class Serum creatinine/GFR criteria Urine output criteria Ri sk Serum creatinine 1.5 or decrease in < 0.5 ml/kg/hour 6 hours Injury Serum creatinine 2 or decrease in < 0.5 ml/kg/hour 12 hours Failure Serum creatinine 3, or decrease in mg/dl with an acute rise > 0.5 mg/dl < 0.3 ml/kg/hour 24 hours, or anuria 12 hours Loss Complete loss of kidney function > 4 weeks End stage kidney disease Complete loss of kidney function > 3 months

PAGE 155

155 Table 4 2 Sociodemographic and clinical character istics of patients according to urologic procedure gr oup between 2000 and 2010 All(n=1,557) Endoscopic/other type of urologic procedures(n=458) Nephroureterectomy/total nephrectomy(n=434) Age ,mean (SD) years 60(15) 56(17) 60(16) Male, n (%) 1039(66.73 ) 274(59.83) 236(54.38) Race/ethnicity, n(%) White 1238(81.99) 357(80.41) 353(84.25) African American 161(10.66) 57(12.84) 45(10.74) Hispanic 51(3.38) 14(3.15) 9(2.15) Other 60(3.97) 16(3.60) 12(2.86) Urban residency, n(%) 1080(69.45) 303(66.16) 3 05(70.60) Primary insurance, n(%) Medicare 716(45.99) 204(44.54) 199(45.85) Medicaid 151(9.70) 64(13.97) 47(10.83) Private 593(38.09) 150(32.75) 156(35.94) No insurance 97(6.23) 40(8.73) 32(7.37) Admission type (emergency), n(%) 308(19. 78) 209(45.63) 46(10.60) Admission type (routine/elective), n(%) 1249(80.22) 249(54.37) 388(89.40) Admission day (weekday), n(%) 1405(90.24) 396(86.46) 399(91.94) Admission day (weekend), n(%) 152(9.76) 62(13.54) 35(8.06) Primary Diagnosis Malignant neoplasm of kidney, ureter, or renal pelvis 451(28.97) 16(3.49) 296(68.20) Prostate cancer 209(13.42) 5(1.09) 0(0.00) Bladder cancer 296(19.01) 86(18.78) 0(0.00) Nephrolithiasis 197(12.65) 156(34.06) 11(2.53) Misc. 404(25.95) 19 5(42.58) 127(29.26) Baseline CKD_EPI eGFR, mean (SD) 73.4(25.0) 73.3(28.6) 70.4(24.2)

PAGE 156

156 Table 4 2 Continued Partial nephrectomy(n=196) Radical cystectomy(n=270) Radical prostatectomy(n=199) p value Age ,mean (SD) years 58(13) 66(13) 61(8) <0. 001 Male, n (%) 124(63.27) 206(76.30) 199(100.00) <0.001 Race/ethnicity, n(%) White 155(82.89) 237(89.43) 136(69.74) <0.001 African American 20(10.70) 7(2.64) 32(16.41) Hispanic 10(5.35) 8(3.02) 10(5.13) Other 2(1.07) 13(4.91) 17(8.72) Urban area residency, n (%) 141(71.94) 192(71.11) 139(69.85) 0.470 Primary insurance, n(%) Medicare 72(36.73) 173(64.07) 68(34.17) Medicaid 15(7.65) 18(6.67) 7(3.52) Private 101(51.53) 66(24.44) 120(60.30) No insurance 8(4.08) 13(4.81) 4(2.01) Admission type (emergency), n(%) 4(2.04) 46(17.04) 3(1.51) <0.001 Admission type (routine/elective), n(%) 192(97.96) 224(82.96) 196(98.49) Admission day (weekday), n(%) 195(99.49) 216(80.00) 199(100.00) <0.001 Admission day (weeke nd), n(%) 1(0.51) 54(20.00) 0(0.00) Primary Diagnosis Malignant neoplasm of kidney, ureter, or renal pelvis 139(70.92) 0(0.00) 0(0.00) N/A Prostate cancer 0(0.00) 5(1.85) 199(100.00) Bladder cancer 0(0.00) 210(77.78) 0(0.00) Nephrolithi asis 1(0.51) 29(10.74) 0(0.00) Misc. 56(28.57) 26(9.63) 0(0.00) Baseline CKD_EPI eGFR, mean (SD) 76.4(22.7) 70.1(22.9) 82.0(19.7) <0.001 CKD EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR, estimated glomerular filtration rate ; Misc., mis cellaneous

PAGE 157

157 Table 4 3. Co morbid conditions of patients accord ing to urologic procedure group between 2000 and 2010 All(n=1,557) Endoscopic/other type of urologic procedures(n=458) Nephroureterectomy/total nephrectomy(n=434) Co morbid Conditio ns n(%) Congestive Heart Failure 80(5.14) 29(6.33) 28(6.45) Myocardial Infarction 77(4.95) 22(4.80) 23(5.30) Chronic Pulmonary Disease 215(13.81) 62(13.54) 69(15.90) Peripheral Vascular Disease 70(4.50) 15(3.28) 36(8.29) Cerebrovascular Disease 31 (1.99) 11(2.40) 9(2.07) Diabetes 289(18.56) 67(14.63) 70(16.13) Hypertension 698(44.83) 180(39.30) 191(44.01) Liver Disease 37(2.38) 16(3.49) 7(1.61) Cancer 1027(65.96) 144(31.44) 315(72.58) UTI s 159(10.21) 101(22.05) 21(4.84) Urinary Obstruction/Hy dronephrosis 286(18.37) 168(36.68) 52(11.98) Charlson comorbidity index, median(IQR) 2(1 3) 1(0 2) 2(2 4)

PAGE 158

158 Table 4 3. Continued Partial nephrectomy(n=196) Radical cystectomy(n=270) Radical prostatectomy(n=199) p value Co morbid Condit ions n(%) Congestive Heart Failure 8(4.08) 13(4.81) 2(1.01) 0.035 Myocardial Infarction 6(3.06) 22(8.15) 4(2.01) 0.025 Chronic Pulmonary Disease 22(11.22) 44(16.30) 18(9.05) 0.095 Peripheral Vascular Disease 9(4.59) 9(3.33) 1(0.50) <0.001 Cerebrov ascular Disease 3(1.53) 7(2.59) 1(0.50) 0.496 Diabetes 51(26.02) 60(22.22) 41(20.60) 0.002 Hypertension 105(53.57) 136(50.37) 86(43.22) 0.004 Liver Disease 6(3.06) 6(2.22) 2(1.01) 0.232 Cancer 144(73.47) 225(83.33) 199(100.00) <0.001 UTI s 4(2.04) 33( 12.22) 0(0.00) <0.001 Urinary Obstruction/Hydronephrosis 4(2.04) 59(21.85) 3(1.51) <0.001 Charlson comorbidity index, median(IQR) 2(2 3) 3(2 8) 2(2 3) <0.001 UTI s urinary tract infection

PAGE 159

159 Table 4 4. Medication use in patients accord ing to urologic procedure group between 2000 and 2010 All(n=1,557) Endoscopic/other type of urologic procedures(n=458) Nephroureterectomy/total nephrectomy(n=434) ACEIs 394(25.31) 103(22.49) 101(23.27) ASA 101(6.49) 48(10.48) 28(6.45) Aminoglycoside 644 (41.36) 224(48.91) 161(37.10) Beta blockers 832(53.44) 217(47.38) 258(59.45) Diuretics 654(42.00) 153(33.41) 218(50.23) Inotropes 14(0.90) 7(1.53) 4(0.92) NSAIDs 403(25.88) 102(22.27) 111(25.58) Statins 327(21.00) 93(20.31) 83(19.12) Vasopressors 108 (6.94) 33(7.21) 38(8.76) Vancomycin 285(18.30) 100(21.83) 83(19.12)

PAGE 160

160 Table 4 4. Continued Partial nephrectomy(n=196) Radical cystectomy(n=270) Radical prostatectomy(n=199) p value ACEIs 57(29.08) 80(29.63) 53(26.63) 0.125 ASA 7(3.57) 17(6.30) 1(0. 50) <0.001 Aminoglycoside 65(33.16) 134(49.63) 60(30.15) <0.001 Beta blockers 103(52.55) 190(70.37) 64(32.16) <0.001 Diuretics 126(64.29) 124(45.93) 33(16.58) <0.001 Inotropes 0(0.00) 3(1.11) 0(0.00) 0.218 NSAIDs 37(18.88) 63(23.33) 90(45.23) <0.001 Statins 47(23.98) 56(20.74) 48(24.12) 0.512 Vasopressors 7(3.57) 28(10.37) 2(1.01) <0.001 Vancomycin 19(9.69) 69(25.56) 14(7.04) <0.001 ACE s, inhibitors, angiotensin converting enzyme inhibitors; ASA, acetylsalicylic acid; NSAIDs, nonsteroidal anti infl ammatory drugs

PAGE 161

161 Table 4 5. Frequency and distribution of AKI in patients accord ing to urologic procedure group between 2000 and 2010 All(n=1,557) Endoscopic/other type of urologic procedures (n=458) Nephroureterectomy/ total nephrectomy ( n=434) Partial nephrectomy (n=196) Radical cystectomy (n=270) Radical prostatectomy (n=199) p value AKI, n(%) 601(38.60) 173(37.77) 200(46.08) 49(25.00) 142(52.59) 37(18.59) <0.001 AKI RIFLE Class Risk 402(25.82) 102(22.27) 145(33.41) 37(18.88) 9 4(34.81) 24(12.06) <0.001 Injury 123(7.90) 45(9.83) 33(7.60) 7(3.57) 30(11.11) 8(4.02) Failure 76(4.88) 26(5.68) 22(5.07) 5(2.55) 18(6.67) 5(2.51) AKI, acute kidney injury; RIFLE, risk, injury, failure, loss, end stage

PAGE 162

162 Table 4 6. Association between predictive factors and AKI after urologic procedures using univariable logistic regression models OR 95%CL p value Age (years) 1.01 1.00 1.02 0.002 Gender (male) 1.04 0.84 1.29 0.736 Race/ethnicity White Referent Blac k 0.98 0.70 1.37 0.431 Hispanic 0.70 0.38 1.28 0.368 Other 0.82 0.48 1.41 0.812 Baseline CKD EPI eGFR (10ml/min/1.73m 2 ) 0.92 0.88 0.95 <0.001 Residency Area (rural vs. urban) 1.12 0.90 1.40 0.307 Primary Diagnosis Prostate cancer Referent Mal ignant neoplasm of kidney, ureter, or renal pelvis 2.79 1.90 4.09 <0.001 Nephrolithiasis 1.77 1.13 2.78 0.013 Bladder cancer 4.13 2.75 6.20 <0.001 Misc. 2.38 1.61 3.52 <0.001 Surgical Procedure Radical prostatectomy Referent Nephroureterect omy /total nephrectomy 3.74 2.50 5.61 <0.001 Partial nephrectomy 1.46 0.90 2.36 0.124 Radical cystectomy 4.86 3.16 7.46 <0.001 Endoscopic/other type of urologic procedures 2.66 1.77 3.98 <0.001 Primary Insurance Medicare 1.82 1.45 2.29 <.0001 Medicai d 1.54 1.07 2.23 0.021 Private Referent No insurance 1.67 1.07 2.59 0.023

PAGE 163

163 Table 4 6. Continued OR 95%CL p value Admission Type Routine Elective Referent Emergency 1.36 1.05 1.74 0.018 Admission Day Weekday Referent Weeken d 2.06 1.47 2.89 <0.001 Congestive Heart Failure 2.51 1.58 3.97 <0.001 Myocardial Infarction 1.63 1.22 2.18 0.001 Chronic Pulmonary Disease 1.63 1.22 2.18 0.001 Peripheral Vascular Disease 1.44 0.89 2.33 0.135 Cerebrovascular Disease 1.50 0.74 3.07 0.261 Diabetes 1.16 0.90 1.51 0.258 Hypertension 1.16 0.95 1.42 0.156 Liver Disease 0.97 0.49 1.90 0.924 Cancer 1.35 1.09 1.68 0.007 UTI s 1.80 1.30 2.51 <0.001 Urinary Obstruction and Hydronephrosis 1.84 1.42 2.38 <0.001 Charlson comorbidity index 1.12 1.08 1.17 <0.001 ACEIs 1.17 0.92 1.49 0.197 ASA 1.99 1.25 3.17 0.004 Aminoglycoside 1.13 0.91 1.39 0.266 Beta blockers 1.46 1.19 1.79 <0.001 Diuretics 1.25 1.01 1.54 0.038 Inotropes 2.74 1.10 0.003* NSAIDs 0.91 0.72 1.16 0.463 Statins 0.93 0 .72 1.21 0.600 Vasopressors 2.96 1.86 4.73 <0.001 Vancomycin 2.14 1.62 2.83 <0.001 *Exact conditional analysis was performed

PAGE 164

164 Table 4 7. Association between predictive factors and AKI after urologic procedures using multivariable logistic regression mo dels OR 95%CL p value Age (years) 1.00 0.99 1.01 0.749 Baseline CKD EPI eGFR (10ml/min/1.73m 2 ) 0.96 0.91 1.01 0.116 Surgical p rocedure Radical prostatectomy Referent Nephroureterectomy/total nephrectomy 3.42 2.24 5.24 <0.001 Partial nephr ectomy 1.62 0.98 2.66 0.058 Radical cystectomy 3.69 2.34 5.82 <0.001 Endoscopic/other type of urologic procedures 2.43 1.52 3.88 <0.001 Admission d ay Weekday Referent Weekend 1.48 1.03 2.13 0.035 Congestive heart f ailure 1.86 1.13 3.04 0.014 Cancer 1.79 1.34 2.40 <0.001 UTI s 1.50 1.03 2.18 0.033 Urinary obstruction /h ydronephrosis 1.75 1.30 2.36 <0.001 Vasopressors 1.82 1.09 3.04 0.022 Vancomycin 1.78 1.32 2.41 <0.001 AKI, acute kidney injury; OR, odds ratio, CL, confidence limits; UTI, u rinary tract infection s

PAGE 165

165 Table 4 8. ROC analysis of models to predict AKI after urologic procedures AUC 95%CL Model 1 0.574 0.544 0.604 Model 2 0.638 0.610 0.667 Model 3 0.675 0.648 0.702 Model 4 0.688 0.661 0.715 Model 1 age and baseli ne CKD EPI eGFR Model 2 age, baseline CKD EPI eGFR, and procedure type Model 3 age, baseline CKD EPI eGFR, procedure type, admission day, CHF, cancer, UTI s and urinary obstruction/hydronephrosis Model 4 age, baseline CKD EPI eGFR, procedure type, admi ssion day, CHF, cancer, UTI s urinary obstru ction/hydronephrosis, vasopresso res, and vancomycin

PAGE 166

166 Table 4 9. Frequency and distribution of renal outcome after AKI in patients with AKI according to urologic procedure groups All(n=601) Endoscopic/ other type of urologic procedures (n=173) Nephroureterectomy/ total nephrectomy (n=200) Partial nephrectomy (n=49) Radical cystectomy (n=142) Radical prostatectomy (n=37) p value Renal outcome, n(%) Complete recovery 462(76.87) 144(83 .24) 114(57.00) 41(83.67) 131(92.25) 32(86.49) <0.001 Partial recovery 124(20.63) 26(15.03) 81(40.50) 6(12.24) 6(4.23) 5(13.51) Non recovery 15(2.50) 3(1.73) 5(2.50) 2(4.08) 5(3.52) 0(0.00) Partial or non recovery, n(%) 139(23.13) 29(16.76) 8 6(43.00) 8(16.33) 11(7.75) 5(13.51) <0.001

PAGE 167

167 Table 4 10 Association between predictive factors and partial recovery or non recovery of renal function following AKI in patients with urologic procedures using univariable logistic regres sion models OR 95%CL p value Age (years) 1.00 0.99 1.01 0.920 Gender (male) 0.85 0.57 1.26 0.408 Race/ethnicity White Referent Black 0.83 0.44 1.58 0.905 Hispanic 1.45 0.49 4.26 0.187 Other 0.34 0.08 1.46 0.136 Baseline CKD EPI eGFR (1 0ml/min/1.73m 2 ) 1.11 1.03 1.19 0.004 Residency Area (rural vs. urban) 1.13 0.75 1.69 0.565 Primary Diagnosis Prostate cancer Referent Malignant neoplasm of kidney, ureter, or renal pelvis 3.80 1.53 9.44 0.004 Nephrolithiasis 1.05 0.34 3.19 0.93 5 Bladder cancer 0.87 0.33 2.35 0.790 Misc. 1.68 0.65 4.33 0.281 Surgical Procedure Radical prostatectomy Referent Nephroureterectomy/total nephrectomy 4.83 1.81 12.91 0.002 Partial nephrectomy 1.25 0.37 4.19 0.719 Radical cystectomy 0.54 0.1 7 1.66 0.280 Endoscopic/other urologic procedures 1.29 0.46 3.59 0.627 Primary Insurance Medicare 0.82 0.54 1.25 0.358 Medicaid 0.77 0.39 1.55 0.467 Private Referent No insurance 0.69 0.30 1.60 0.389 Admission Type (Routine Elective) Referent Admission Type (Emergency) 0.43 0.25 0.73 0.002

PAGE 168

168 Table 4 10 Continued OR 95%CL p value Admission Day (Weekday) Referent Admission Day (Weekend) 0.70 0.39 1.27 0.242 AKI RIFLE Class (Risk) Referent AKI RIFLE Class (Injury) 1.17 0.71 1 .93 0.024 AKI RIFLE Class (Failure) 4.36 2.61 7.30 <0.001 Congestive Heart Failure 1.57 0.83 2.99 0.167 Myocardial Infarction 0.50 0.19 1.31 0.160 Chronic Pulmonary Disease 0.92 0.55 1.53 0.744 Peripheral Vascular Disease 2.61 1.27 5.36 0.009 Cereb rovascular Disease 2.27 0.79 6.50 0.126 Diabetes 0.80 0.49 1.30 0.364 Hypertension 1.14 0.78 1.67 0.492 Liver Disease 1.34 0.41 4.34 0.626 Cancer 1.66 1.06 2.59 0.026 UTI s 0.23 0.10 0.54 0.001 Urinary Obstruction and Hydronephrosis 0.49 0.29 0.81 0. 005 Charlson comorbidity index 1.00 0.94 1.08 0.921 ACEIs 0.84 0.55 1.28 0.416 ASA 1.25 0.70 2.23 0.442 Aminoglycoside 0.87 0.60 1.28 0.484 Beta blockers 0.82 0.55 1.22 0.325 Diuretics 1.13 0.77 1.65 0.533 Inotropes 1.34 0.41 4.34 0.626 NSAIDs 0.84 0.54 1.29 0.421 Statins 0.73 0.45 1.18 0.198 Vasopressors 1.15 0.66 1.98 0.621 Vancomycin 0.85 0.56 1.30 0.460 *Exact conditional analysis was performed ; AKI, acute kid ney injury; Misc. miscellaneous ; ACEIs, angiotensin co nverting enzyme inhibitors; ASA, acetylsalicylic acid; NSAIDs, nonsteroidal anti inflammatory drugs; UTI, urinary tract infection s

PAGE 169

169 Table 4 11 Association between predic tive factors and partial recovery or non recovery of renal function following AKI in patients with urologic proc edures using multivariable logistic regression models OR 95%CL p value Age (years) 1.02 1.00 1.03 0.049 Baseline CKD EPI eGFR (10ml/min/1.73m 2 ) 1.17 1.06 1.28 0.001 Surgical Procedure Radical prostatectomy Referent Nephroureterectomy/tota l nephrectomy 7.15 2.43 21.07 <0.001 Partial nephrectomy 1.78 0.48 6.58 0.388 Radical cystectomy 0.64 0.19 2.16 0.467 Endoscopic/other urologic procedures 1.98 0.64 6.10 0.233 AKI RIFLE Class Risk Referent Injury 1.59 0.90 2.80 0.010 Failure 11.95 6.02 23.73 <0.001 UTI s 0.17 0.06 0.46 <0.001 AKI, acute kidney injury; OR, odds ratio, CL, confidence limits; UTI, urinary tract infection s

PAGE 170

170 Table 4 12 ROC analysis of models to predict partial recovery or non recovery of renal funct ion following AKI in patients with urologic procedures AUC 95%CL Model 1 0.584 0.530 0.638 Model 2 0.683 0.632 0.734 Model 3 0.801 0.761 0.842 Model 4 0.816 0.777 0.855 Model 1 age and baseline CKD EPI eGFR Model 2 age, baseline CKD EPI eGFR, and RIFLE AKI Class Model 3 age, baseline CKD EPI eGFR, RIFLE AKI Class, and type of urologic procedure Model 4 age, baseline CKD EPI eGFR, RIFLE AKI Class, and type of urologic procedure, and UTI s

PAGE 171

171 Figure 4 1. Performance of the fitted models to p r edict AKI after after urologic procedures using ROC analysis

PAGE 172

172 Model 1 age and baseline CKD EPI eGFR Model 2 age, baseline CKD EPI eGFR, and procedure type Model 3 age, baseline CKD EPI eGFR, procedure type, admission day, CHF, cancer, UTI s and urinary o bstruction/hydronephrosis Model 4 age, baseline CKD EPI eGFR, procedure type, admission day, CHF, cancer, UTI s urinary obstruction/hydronephrosis, vasopressures, and vancomycin

PAGE 173

173 Figure 4 2. Cross validation of the fitted m odel (Model 4) to predict AKI after urologic procedures using ROC analysis Model (Model 4) age, baseline CKD EPI eGFR, procedure type, admission day, CHF, cancer, UTIs, urinary obstruction/hydronephrosis, vasopressures, and vancomycin ROC1 (Cross validati on)

PAGE 174

174 Figu re 4 3 Performance of the fitted models to predict partial recovery or non recovery of renal function following AKI in patients with urologic procedures using ROC analysis

PAGE 175

175 Model 1 age and baseline CKD EPI eGFR Model 2 age, baseline CKD EPI e GFR, and RIFLE AKI Class Model 3 baseline CKD EPI eGFR, RIFLE AKI Class, and type of urologic procedure Model 4 baseline CKD EPI eGFR, RIFLE AKI Class, and type of urologic procedure, and UTIs

PAGE 176

176 Figure 4 4. Cross validation of the fitted m odel (Model 4) to predict partial recovery or non recovery of renal function following AKI in patients with urologic procedures using ROC analysis Model (Model 4) baseline CKD EPI eGFR, RIFLE AKI Class, and type of urologic procedure, and UTIs ROC1 (Cros s validation)

PAGE 177

177 CHAPTER 5 CONCLUSIONS Summary of Findings In the first study of this dissertation, we retrospectively studied 384 bladder cancer patients including 172 patients who underwent radical cystectomy (RC) plus urinary diversion (UD) and 212 pati ents who did not undergo RC plus UD between 2000 and 2014 with aim of comparing renal function decline between bladder cancer patients with and without RC plus UD. We used three well recognized definitions of renal function decline: (1) change in MDRD eGFR over time (eGFR slope), (2) time to 30%, and (3) odds of rapid decline in renal function, defined by a decrease in the MDRD eGFR >3 ml/min/1.73m2/year. We found that patients who underwent RC plus UD experienced a faster d ecline in renal function over time, as measured by MDRD eGFR slope, compared to patients who did not undergo the procedure, despite adjustment for age, baseline characteristics, co morbid conditions, and the stage of bladder cancer through propensity score and medications use. Patients with RC plus UD had a mean MDRD eGFR decline of 3.8 5 ml/min/1.73m2/year, whereas those without RC plus UD had a mean eGFR decline of 0.69 8 ml/min/1.73m2/year. Furthermore, we found that bladder cancer patients who underwen 30% and greater odds of rapid decline in eGFR compared to controls; this persisted despit e adjustment for age but was no longer statistically significant after adjustment for propensity score, co morbid cond itions and medication use. In the second study of this dissertation, we retrospectively studied 172 bladder cancer patients who underwent RC plus UD between 2000 and 2014 in order to

PAGE 178

178 investigate whether the type of urinary diversion affects the renal func tion and to identify the predictive factors for renal function decline after RC plus UD for bladder cancer. Similar to the first study, we used three definitions of renal function decline: (1) change in eGFR over time (eGFR slope), (2) time to percentage c 30%, and (3) odds of rapid decline in renal function, defined by defined by eGFR >3 ml/min/1.73m 2 /year. We found that patients who underwent cutaneous or orthotopic continent diversion experienced a faster decline in renal function over tim e as measured by eGFR slope, as compared to those with ileal conduit diversion. In other words, patients with cutaneous or orthotopic continent diversion had a mean eGFR decline of 6.775 ml/min/1.73m 2 /year, whereas those with ileal conduit diversion had a mean eGFR decline of 3.161 ml/min/1.73m 2 /year. Patients with other type of reconstruction had a mean eGFR decline of 1.931 ml/min/1.73m 2 /year, even though this was not statically significant compared to those with ileal conduit diversion. Multivariate analysis demonstrated that cutaneous or orthotopic continent diversion was no longer associated with faster decline in eGFR; h owever, clinical N1, N2, NX, and use of xanthine oxidase inhibitor were independently associated with faster decline in mean eGFR We did not find any statically significant differences in rates of renal deterioration, defined by a decrease in the eGFR of 30% or greater, or odds of rapid decline in renal function, defined by eGFR >3 ml/min/1.73m 2 /year, across different types of UD. Multivariate analysis showed that race (other, as compared to white), clinical NX, chemotherapy, and vancomycin use were associated with higher rates of renal deterioration throughout the study period. Multivariate analysis failed to identify any

PAGE 179

179 predictiv e factors for rapid decline in renal function, defined by defined by eGFR >3 ml/min/1.73m 2 /year. In the third study of this dissertation we conducted a retrospective study of 1,557 patients who underwent urologic procedures between the years 2000 to 2010 with the objectives of identifying predictive factors for AKI within 30 days following urologic procedures and identifying prognostic actors for partial or no renal recovery after AKI. We found that AKI occurred in 39% of patients with the following distri bution (%) according to the type of procedure: endoscopic/other type of urologic procedures (38%), nephroureterectomy/total nephrectomy (46%), partial nephrectomy (25%), radical cystectomy (53%), and radical prostatectomy (19%). Multivariate analysis demon strated that radical cystectomy, nephroureterectomy/total nephrectomy, and endoscopic/other type of urologic procedures, as compared to radical prostatectomy, worse baseline eGFR, weekend admission, CHF, cancer, UTIs, urinary obstruction/hydronephrosis, an d vasopressors, and vancomycin use were independently associated with higher odds of AKI. Additionally, we found that among 601 patients with AKI, 139(23%) patients had partial or non recovery of renal function. Multivariate analysis demonstrated that neph roureterectomy/total nephrectomy, as compared to radical prostatectomy, AKI RIFLE Injury and Failure, as compared to AKI RIFLE Risk, and increasing age and baseline eGFR were independently associated with higher odds of partial or non recovery of renal fun ction after AKI. Finally, UTIs decreased the odds of partial or non recovery of renal function after AKI.

PAGE 180

180 Discussion Our findings from the first study which indicate that bladder cancer patients with RC plus UD experience steep decline in renal function a re consistent with the majority of the prior studies that have examined this relationship. For example, a retrospective study of 111 patients with orthotopic neobladder substitution and 50 patients with ileal conduit diversion found deterioration of renal function, defined by a decrease in the MDRD eGFR>10ml/min/1.73m 2 before and 10 years after the surgery, in 36% and 21% of patients who underwent ileal conduit urinary diversion and orthotopic neobladder, respectively (Jin et al., 2012) However, the authors investigated a heterogeneous cluding those who died within 10 years after the UD from the analysis. Also, a decrease in the eGFR>10ml/min/1.73m 2 after 10 years might occur in some patients due to the normal aging process (Jonsson et al., 2001; L indeman et al., 1985) Another retrospective study of 169 patients with RC plus UD reported renal deterioration, defined by a decrease in the eGFR of 20% or greater, in 46% of patients with a mean follow up of 104 months (Nishikawa et al., 2014) Osawa et al. reported, that mean eGFR decreased from 74.6 ml/min/1.73m 2 before surgery to 63.9 ml/min/1.73m 2 and 34% of patients showed reduced renal function, defined by a greater than 25% decrease in eGFR from the baseli ne, in a retrospective study of 70 patients with RC plus UD with a median follow up period of 34.5 months (Osawa et al., 2013) The main differences between our study and prior studies are the inclusion of bladder cancer patients without RC plus UD as a control group, use of three widely acceptable definitions of renal function decline, and appropriate statistical adjustment for potentially confounding

PAGE 181

1 81 variables using propensity score and multivariable regr ession models. Moreover, prior studies considered small changes in eGFR as a renal deterioration event. For example, 25% and 20%, respectively, as a renal deterioration event. We defined renal deterioration as a decrease in the eGFR or 30% 30% was chosen as an alternative outcome because it is strongly and consistently associated with the risk of end stage renal disease (ESRD) and mortality (Coresh et al., 2014) In our sensitivity analysis, we 20%, 25%, and 40% as cutoff values, patients with RC plus UD were more at risk of a renal deterioration event 20% and 25% in the multivariate analysis. It is possible that multivariate analysis might have shown an independent association between RC plus UD and a renal deterioration event, defined by a decrease in eGFR of 20% or 25%, throughout the study period. In our second study, consistent with previous studies, we did not find an independent association between the type of UD and decline in renal function after appropriate adjustment (Gershman et al., 2015; Gilbert et al., 2013; Nishikawa et al., 2014; Zabell et al., 2015) Prior studies found that sepsis, hypertension (Nishikawa et al., 2014; Samuel et al., 2006) dia betes, urinary obstruction (Jin et al., 2012) acute pyelonephritis and chemotherapy (Nishikawa et al., 2014; Osawa et al., 2013) are independently associated with a decline in renal function following RC plus US. It should be noted that these studies were limited by at least one of the followings: inadequate ascertainment of the renal outcome, small sample size, heterogeneous population, inadequate adjustment for potentially confounding factors, and lack of repeated

PAGE 182

182 measures of renal function. Osawa et al. and Nishikawa et al. used the standard Japanese formula for eGFR (Nishikawa et al., 2014; Osawa et al., 2013) Our study was differe nt for the reason that we used three different widely acceptable methods to evaluate renal impairment in patients after RC plus UD. In our study, we found that chemotherapy was strongly associated with renal deterioration. Also multivariate analysis demons trate d that clinical N1, N2, NX, and use of xanthine oxidase inhibitor and v ancomycin were independently associated with decline in renal function. Furthermore, in the present study, we did not have data to compare the effect of refluxing versus non reflu xing orthotopic neobladder, but previous studies (Harraz et al., 2014) did not find any difference in terms of renal outcomes between these two surgical techniques. Also, we did not have any data to measure the expe rience of urologists who performed the surgical procedures. It is reasonable to assume that a urologist ex perienced in performing the procedures may have better surgical outcomes. A key importation limitation of our first and second study, and also several other previous studies, was not accounting for death as a competing risk. It is likely that some patients with concomitant co morbid conditions might have died before they developed renal deterioration. In conventional Cox proportional hazards model, whic h we used in this study, the censoring mechanism is assumed to be noninformative. But, this is not true In the presence of competing risks. Therefore, we should have used methods designed for competing risks analysis, such as the cause specific proportiona l hazards model (Lunn & McNeil, 1995) the s ubdistribution proportional hazards model (Fine & Gray, 1999) or parametric mixture model (Lau et al., 2011) But, we did not have data on the death of the patients; as a result, we were limited to use conventional methods

PAGE 183

183 for time to event analysis. In fut ure research, in order to address this shortcoming, mortality data should be obtain and appropriate competing risk analysis should be utilized to correctly estimate the risk of renal deterioration among bladder cancer patients undergoing RC plus UD. In the third study, we found that 39% of patients who underwent urologic procedures developed AKI which was somewhat comparable to what is reported in the literature for similar urologic procedures (Joo et al., 2016; Jo ung et al., 2016; Kim et al., 2014) However, it should be noted that the majority of these studies assessed AKI risk following only limited urologic procedures, and very few of them investigated the risk factors of AKI or examined the renal outcome afte r AKI. For example, a retrospective study of 222 patients with RC plus UD reported a postoperative AKI incidence of 31% and no association between the AKI risk and type of UD (Joung et al., 2016) Another retrospect ive study of 233 patients with RCC demonstrated that the percent functional volume preservation and warm ischemia time can predict AKI after partial nephrectomy (Ding et al., 2016) In our study, we found that radic al cystectomy, nephroureterectomy/total nephrectomy, and endoscopic/other type of urologic procedures, as compared to radical prostatectomy, worse baseline eGFR, weekend admission, CHF, cancer, UTIs, urinary obstruction/hydronephrosis, and vasopressors, an d vancomycin use were independently associated with higher odds of AKI, using multivariate analysis. However, it should be noted that the assessment of this fitted prediction model via ROC analysis using cross validation indicated that this model has a poo r discriminative ability (AUC: 0.671; 95%Cl: 0.644 0.698). Therefore, we think there were several key prognostic clinical parameters, including but not limited to operation

PAGE 184

184 time, ischemia time, blood transfusion rate, crystalloid and colloid infused volume and radiocontrast agent use, that we did not include in this analysis. We think the association between these parameters and AKI risk should also be examined in order to develop a useful AKI risk prediction tool for patients undergoing urologic procedure Another subject that might raise a question is our choice of GFR estimation equation. In the first and second study, we used the MDRD study equation (Levey et al., 1999) to estimate GFR; however, in the third stud y, we used the CKD EPI equation (Levey et al., 2009) to estimate GFR. Both MDRD and CKD EPI questions are widely used in clinical practice and research. The MDRD formula performs well in older populations and those with CKD. However, the MDRD equation may not accurately estimate renal function in the healthy individuals and at the early stages of renal impairment (Stevens et al., 2006) In this population, the MDRD equation te nds to underestimate the GFR when directly measured GFR (i.e. by use of an exogenous filtration marker such as inulin or 125I iothalamate) is less than 90 ml/min/1.73m 2 leading to a false positive diagnosis of kidney disease (Stevens et al., 2006) In comparison, the CKD EPI formula (Levey et al., 2009) was developed using both CKD and non CKD patients to address the shortcoming of the MDRD equation, which was developed in p atients with CKD (Levey et al., 2009) The CKD EPI equation is more precise and accurate than the MDRD equation, particularly at higher GFR values; however, it offers lower sensitivity for detecting renal impairment (Levey et al., 2009; Murata et al., 2011; Stevens et al., 2010) The CKD EPI equation has a sensitivity and specificity of 91% and 87% to detect a measured GFR<60 ml/min/1.73 m 2 using iothalamate as gold standard (Levey et al., 2009) The use of the CKD EPI equation,

PAGE 185

185 compared to the MDRD equation, misclassifies fewer low risk individuals as having kidney disease; hence, reducing false positive diagnosis (Levey et al., 2009; Stevens et al., 2010) It is important to note that the elderly and non Caucasians were underrepresented in the sample population used to develop the CKD EPI equation (Levey et al., 2009) We used the MDRD equation in the first and second study because it is reasonably accurate for the assessment of renal function in the population under study (i.e. older individuals with bladder cancer). We used the CKD EPI equation in the thi rd study because study population in the third study were likely to start with more preserved renal function, unlike the bladder cancer patients with RC plus UD, where the CKD EPI formula is more reliable. Future Directions Many gaps exist in the litera ture regarding factors that are associated with renal function decline in bladder cancer following RC plus UD. Also, poorly designed retrospective studies using non standard definitions for renal function decline have been unable to draw a solid conclusion We attempted to address many shortcomings of prior research in the present studies by using three well recognized definitions of renal function decline and including a control group for comparison. However, the retrospective nature of the present studies has limitations that can be overcome only by well designed prospective observational studies or controlled trials. With respect to AKI after urologic surgeries, AKI and partial and non recovery prediction models resulting from this study should be externa lly validated in other populations. Future prospective or controlled trials should identify other clinically relevant parameters (e.g. operation time, ischemia time, blood transfusion rate,

PAGE 186

186 crystalloid and colloid infused volume, and radiocontrast agent us e) not addressed in this study. In the next step, an AKI risk assessment tool can be constructed and tailored specifically for use in patients undergoing urologic procedures. Furthermore, while our study focused on AKI and short term outcome of renal funct ion, another important issue is the association between the AKI occurrence and long term renal function decline. It is necessary to understand the role of postoperative AKI in long term trajectory of renal function decline among patients undergoing urologi c procedures. To sum, the existing body of literature in conjunction with the findings from our study suggests that bladder cancer patients undergoing RC plus UD, irrespective of the type of UD and other prognostic factors, are at increased risk of renal function decline. In addition, AKI occurrence is common after urologic procedures and approximately a quarter of patients with AKI do not achieve complete renal recovery. Future prospective studies or controlled trials are needed, however, to further investigate these associations, longer term renal outcomes, and determine whether appropriate intervention programs (e.g. pre and post operative nephrology consultation and close monitoring of renal function), tailored to this group of patients, can decr ease the rate of long term renal impairment.

PAGE 187

187 APPENDIX A ICD 9 CM DIAGNOSIS CODES FOR BLADDER CANCER 188 Malignant neoplasm of bladder 188.0 Malignant neoplasm of trigone of bladder 188.1 Malignant neoplasm of dome of bladder 188.2 Malignant ne oplasm of lateral wall of bladder 188.3 Malignant neoplasm of anterior wall of bladder 188.4 Malignant neoplasm of posterior wall of bladder 188.5 Malignant neoplasm bladder neck 188.6 Malignant neoplasm of ureteric orifice 188.7 Malignant neoplasm of urachus 188.8 Malignant neoplasm of other specified sites of bladder 188.9 Malignant neoplasm of bladder part, unspecified 189.3 Malignant neoplasm of urethra 233.7 Carcinoma in situ of bladder

PAGE 188

188 APPENDIX B PROCEDURE CODES IN MANAGEMENT OF BLADDER CANCER CPT procedure codes 51020 Cystotomy or cystostomy; with fulguration and/or insertion of radioactive material 51530 Cystotomy or cystostomy; with fulguration and/or insertion of radioactive material 51570 Cystectomy; complete 51575 Cystectomy; complete, with bilateral pelvic lymphadenectomy, including external iliac hypogastric, and obturator nodes 51580 Cystectomy; complete, with uretereosigmoidostomy or ureterocutaneous transplantations 51585 Cystectomy; complete, with uretere osigmoidostomy or ureterocutaneous transplantations; with bilateral pelvic lymphadenectomy, includingexternal iliac, hypogastric, and obturator nodes 51590 Cystectomy; complete, with ureteroileal conduit or sigmoid bladder, including intestine anastomosi s 51595 Cystectomy; complete, with ureteroileal conduit or sigmoid bladder, including intestine anastomosis with bilateral pelvic lymphadenectomy, including external iliac, hypogastric, and obturator nodes 51596 Cystectomy, complete, with continent diversi on, any open technique, using any segment of small and/or large intestine to construct neobladder 51597 Pelvic exenteration, complete, for vesical, prostatic or urethral malignancy, with removal of bladder and ureteral transplantations, with or without hy sterectomy and/or abdominoperineal resection of rectum and colon and colostomy, or any combination thereof 51720 Bladder instillation of anticarcinogenic agent (including detention time) 52224 Cystourethroscopy, with fulguration (including cryosurgery o r laser surgery) or treatment of MINOR (less than 0.5 cm) lesion(s) with or without biopsy 52234 Cystourethroscopy, with fulguration (including cryosurgery or laser surgery) and/or resection of; SMALL bladder tumor(s) (0.5 to 2.0 cm) 52235 Cystourethro scopy, with fulguration (including cryosurgery or laser surgery) and/or resection of; MEDIUM bladder tumor(s) (2.0 to 5.0 cm) 52240 Cystourethroscopy, with fulguration (including cryosurgery or laser surgery) and/or resection of; LARGE bladder tumor(s) 52 250 Cystourethroscopy with insertion of radioactive substance, with or without biopsy or fulguration ICD 9 procedure codes 57.4 Transurethral excision or destruction of bladder cancer 57.49 Other transurethral excision or destruction of lesion or tissu e of bladder 57.5 Other excision or destruction of bladder 57.59 Open excision or destruction of other lesion or tissue of bladder 57.6 Partial cystectomy 57.71 Radical cystectomy 57.79 Other total cystectomy 57.87 R econstruction of urinary bladd er

PAGE 189

189 APPENDIX C ASSESSMENT OF PROPORTIONAL HAZARDS ASSUMPTION

PAGE 190

190

PAGE 191

191

PAGE 192

192 APPENDIX D ICD 9 CM PROCEDURE CODES Radical prostatectomy 605 Nephroureterectomy/total nephrectomy 5551 Partial nephrectomy 554 Rad ical cystectomy 5771 560 5651 Endoscopic/other urologic procedures 5504 5503 5749 598 5779 6029 5587 5501 5732 576 8774 5674 5641 5718

PAGE 193

193 LIST OF REFERENCES ACS, A. C. S. (2014). Survival rates for bladder cancer by stage. Retrieved from http://www.cancer.org/cancer/bladdercancer/detailedguide/bladder cancer survival rates Aschengrau, A., & Seage III, G. R. (2008). Essentials of E pidemiology in Public Health (2nd ed.). Babjuk, M., Burger, M., Zigeuner, R., Shariat, S. F., van Rhijn, B. W. G., Comperat, E., Roupret, M. (2013). EAU Guidelines on Non Muscle invasive Urothelial Carcinoma of the Bladder: Update 2013. European Uro logy, 64 (4), 639 653. doi:10.1016/j.eururo.2013.06.003 Bagshaw, S. M., George, C., Bellomo, R., & Comm, A. D. M. (2008). A comparison of the RIFLE and AKIN criteria for acute kidney injury in critically ill patients. Nephrology Dialysis Transplantation, 2 3 (5), 1569 1574. doi:10.1093/ndt/gfn009 Bagshaw, S. M., George, C., Dinu, I., & Bellomo, R. (2008). A multi centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrology Dialysis Transplantation, 23 (4), 1203 1210. doi:10.1093/ndt/gfm744 Bellomo, R., Ronco, C., Kellum, J. A., Mehta, R. L., Palevsky, P., & Workgroup, A. (2004). Acute renal failure definition, outcome measures, animal models, fluid therapy and information technology needs: the Second Internati onal Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Critical Care, 8 (4), R204 R212. doi:10.1186/cc2872 Bihorac, A., Delano, M. J., Schold, J. D., Lopez, M. C., Nathens, A. B., Maier, R. V., Moldawer, L. L. (2010). Incide nce, Clinical Predictors, Genomics, and Outcome of Acute Kidney Injury Among Trauma Patients. Annals of Surgery, 252 (1), 158 165. doi:10.1097/SLA.0b013e3181deb6bc Bihorac, A., Yavas, S., Subbiah, S., Hobson, C. E., Schold, J. D., Gabrielli, A., Sega l, M. S. (2009). Long Term Risk of Mortality and Acute Kidney Injury During Hospitalization After Major Surgery. Annals of Surgery, 249 (5), 851 858. doi:10.1097/SLA.0b013e3181a40a0b Blais, L., Ernst, P., & Suissa, S. (1996). Confounding by indication and channeling over time: The risks of beta(2) agonists. American Journal of Epidemiology, 144 (12), 1161 1169. Caddeo, G., Williams, S. T., McIntyre, C. W., & Selby, N. M. (2013). Acute kidney injury in urology patients: incidence, causes and outcomes. Nephr ourol Mon, 5 (5), 955 961. doi:10.5812/numonthly.12721

PAGE 194

194 Campbell, S. C., Novick, A. C., Streem, S. B., Klein, E., & Licht, M. (1994). C omplications of Nephron Sparing Surgery for Renal Tumors Journal of Urology, 151 (5), 1177 1180. Canter, D., Viterbo, R., Kutikov, A., Wong, Y. N., Plimack, E., Zhu, F., Boorjian, S. A. (2011). Baseline Renal Function Status Limits Patient Eligibility to Receive Perioperative Chemotherapy for Invasive Bladder Cancer and Is Minimally Affected by Radical Cystectomy. Urol ogy, 77 (1), 160 165. doi:10.1016/j.urology.2010.03.091 Carmichael, P., & Carmichael, A. R. (2003). Acute renal failure in the surgical setting. Anz Journal of Surgery, 73 (3), 144 153. doi:10.1046/j.1445 2197.2003.02640.x Case, J., Khan, S., Khalid, R., & Khan, A. (2013). Epidemiology of acute kidney injury in the intensive care unit. Crit Care Res Pract, 2013 479730. doi:10.1155/2013/479730 Charlson, M. E., Pompei, P., Ales, K. L., & Mackenzie, C. R. (1987). A new method of classifying prognostic co mor bidity in longitudinal studies: development and validation Journal of Chronic Diseases, 40 (5), 373 383. doi:10.1016/0021 9681(87)90171 8 Chawla, L. S., Eggers, P. W., Star, R. A., & Kimmel, P. L. (2014). Acute Kidney Injury and Chronic Kidney Disease as Interconnected Syndromes. New England Journal of Medicine, 371 (1), 58 66. doi:10.1056/NEJMra1214243 Chertow, G. M., Burdick, E., Honour, M., Bonventre, J. V., & Bates, D. W. (2005). Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. Journal of the American Society of Nephrology, 16 (11), 3365 3370. doi:10.1681/asn.2004090740 Cleveland. (2013). Urinary Reconstruction and Diversion. Retrieved from http://my.clevelandclinic.org/health/treatments_and_procedures/hic_Urinary_Reconstru ction_and_Diversion Coresh, J., Selvin, E., Stevens, L. A., Manzi, J., Kusek, J. W., Eggers, P., Levey, A. S. (2007). Pre valence of chronic kidney disease in the United States. Jama Journal of the American Medical Association, 298 (17), 2038 2047. doi:10.1001/jama.298.17.2038 Coresh, J., Turin, T. C., Matsushita, K., Sang, Y., Ballew, S. H., Appel, L. J., Consortium, C K. D. P. (2014). Decline in Estimated Glomerular Filtration Rate and Subsequent Risk of End Stage Renal Disease and Mortality. Jama Journal of the American Medical Association, 311 (24), 2518 2531. doi:10.1001/jama.2014.6634 Crainiceanu, C. M., Ruppert, D ., & Wand, M. P. (2005). Bayesian analysis for penalized spline regression using WinBUGS. Journal of Statistical Software, 14 (14), 24. de Boer, I. H., Katz, R., Fried, L. F., Ix, J. H., Luchsinger, J., Sarnak, M. J., Kestenbaum, B. (2009). Obesity and Change in Estimated GFR Among Older Adults. American Journal of Kidney Diseases, 54 (6), 1043 1051. doi:10.1053/j.ajkd.2009.07.018

PAGE 195

195 Deyo, R. A., Cherkin, D. C., & Ciol, M. A. (1992). A dapting a c linical c omorbidity i ndex for use with ICD 9 CM administra tive databases Journal of Clinical Epidemiology, 45 (6), 613 619. doi:10.1016/0895 4356(92)90133 8 Ding, Y. Z., Kong, W., Zhang, J., Dong, B. J., Chen, Y. H., Xue, W., Huang, Y. R. (2016). Spherical cap surface model: A novel method for predicting r enal function after partial nephrectomy. International Journal of Urology, 23 (8), 667 672. doi:10.1111/iju.13126 Eckardt, K. U., Coresh, J., Devuyst, O., Johnson, R. J., Kottgen, A., Levey, A. S., & Levin, A. (2013). Evolving importance of kidney disease: from subspecialty to global health burden. Lancet, 382 (9887), 158 169. doi:10.1016/s0140 6736(13)60439 0 Eisenberg, M. S., Thompson, R. H., Frank, I., Kim, S. P., Cotter, K. J., Tollefson, M. K., Boorjian, S. A. (2014). Long Term Renal Function Out comes after Radical Cystectomy. Journal of Urology, 191 (3), 619 625. doi:10.1016/j.juro.2013.09.011 Elyasi, S., Khalili, H., Dashti Khavidaki, S., & Mohammadpour, A. (2012). Vancomycin induced nephrotoxicity: mechanism, incidence, risk factors and special populations. A literature review. European Journal of Clinical Pharmacology, 68 (9), 1243 1255. doi:10.1007/s00228 012 1259 9 Endre, Z. H., Pickering, J. W., & Walker, R. J. (2011). Clearance and beyond: the complementary roles of GFR measurement and inju ry biomarkers in acute kidney injury (AKI). American Journal of Physiology Renal Physiology, 301 (4), F697 F707. doi:10.1152/ajprenal.00448.2010 Fine, J. P., & Gray, R. J. (1999). A proportional hazards model for the subdistribution of a competing risk. Journal of the American Statistical Association, 94 (446), 496 509. doi:10.2307/2670170 Fox, C. S., Larson, M. G., Leip, E. P., Culleton, B., Wilson, P. W. F., & Levy, D. (2004). Predictors of new onset kidney disease in a community based population. Jama Journal of the American Medical Association, 291 (7), 844 850. doi:10.1001/jama.291.7.844 Freedman, B. I., Soucie, J. M., & McClellan, W. M. (1997). Family history of end stage renal disease among incident dialysis patients. Journal of the American Society of Nephrology, 8 (12), 1942 1945. Gershman, B., Eisenberg, M. S., Thompson, R. H., Frank, I., Kaushik, D., Tarrell, R., Boorjian, S. A. (2015). Comparative impact of continent and incontinent urinary diversion on long term renal function after radic al cystectomy in patients with preoperative chronic kidney disease 2 and chronic kidney disease 3a. International Journal of Urology, 22 (7), 651 656. doi:10.1111/iju.12770

PAGE 196

196 Gilbert, S. M., Lai, J., Saigal, C. S., Gore, J. L., & Project, U. D. i. A. (2013). Downstream complications following urinary diversion. J Urol, 190 (3), 916 922. doi:10.1016/j.juro.2013.03.026 Go, A. S., Chertow, G. M., Fan, D. J., McCulloch, C. E., & Hsu, C. Y. (2004). Chronic kidney disease and the risks of death, cardiovascular even ts, and hospitalization. New England Journal of Medicine, 351 (13), 1296 1305. doi:10.1056/NEJMoa041031 Gonen, M. (2007). Analyzing Receiver Operating Characteristic Curves with SAS Cary, NC: SAS Institute Inc. Gooley, T. A., Leisenring, W., Crowley, J. & Storer, B. E. (1999). Estimation of failure probabilities in the presence of competing risks: New representations of old estimators. Statistics in Medicine, 18 (6), 695 706. doi:10.1002/(sici)1097 0258(19990330)18:6<695::aid sim60>3.3.co;2 f Grams, M. E., Coresh, J., Segev, D. L., Kucirka, L. M., Tighiouart, H., & Sarnak, M. J. (2012). Vascular Disease, ESRD, and Death: Interpreting Competing Risk Analyses. Clinical Journal of the American Society of Nephrology, 7 (10), 1606 1614. doi:10.2215/cjn.0346041 2 Gratzke, C., Seitz, M., Bayrle, F., Schlenker, B., Bastian, P. J., Haseke, N., Bachmann, A. (2009). Quality of life and perioperative outcomes after retroperitoneoscopic radical nephrectomy (RN), open RN and nephron sparing surgery in patients wit h renal cell carcinoma. Bju International, 104 (4), 470 475. doi:10.1111/j.1464 410X.2009.08439.x Greene FL, P. D., Fleming ID, Fritz A, Balch CM, Haller DG. (2002). AJCC Cancer Staging Manual (6th ed.). Chicago, Illinois: Springer Verlag. Greenland, S., & Neutra, R. (1980). CONTROL OF CONFOUNDING IN THE ASSESSMENT OF MEDICAL TECHNOLOGY. International Journal of Epidemiology, 9 (4), 361 367. Harraz, A. M., Mosbah, A., Abdel Latif, M., El Assmy, A., Gad, H., & Shaaban, A. A. (2014). Impact of the type of ur eteroileal anastomosis on renal function measured by diuretic scintigraphy: long term results of a prospective randomized study. Bju International, 114 (2), 202 209. doi:10.1111/bju.12511 Hautmann, R. E., Volkmer, B. G., Schumacher, M. C., Gschwend, J. E., & Studer, U. E. (2006). Long term results of standard procedures in urology: the ileal neobladder. World Journal of Urology, 24 (3), 305 314. doi:10.1007/s00345 006 0105 z Hemmelgarn, B. R., Zhang, J., Manns, B. J., Tonelli, M., Larsen, E., Ghali, W. A., Culleton, B. F. (2006). Progression of kidney dysfunction in the community dwelling elderly. Kidney International, 69 (12), 2155 2161. doi:10.1038/sj.ki.5000270

PAGE 197

197 Heung, M., & Chawla, L. S. (2012). Predicting progression to chronic kidney disease after recovery from acute kidney injury. Current Opinion in Nephrology and Hypertension, 21 (6), 628 634. doi:10.1097/MNH.0b013e3283588f24 Hosmer, D. W., & Lemeshow, S. (2000). Applied Logistic Regression New York: Wiley. Howlader, N. N., AM. Krapcho, M. Gars hell, J. Neyman, N. Altekruse, SF. Kosary, CL. Yu, M. Ruhl, J. Tatalovich, Z. Cho, H. Mariotto, A. Lewis, DR. Chen, HS. Feuer, EJ. Cronin, KA. (2013). SEER Cancer Statistics Review, 1975 2010 Retrieved from http://seer.cancer.gov/archive/csr/1975_2010/results_merged/sect_27_urinary_bladder. pdf Hunsicker, L. G., Adler, S., Caggiula, A., England, B. K., Greene, T., Kusek, J. W., Teschan, P. E. (1997). Predictors of the progression of renal disease in the Modification of Diet in Renal Disease Study. Kidney International, 51 (6), 1908 1919. doi:10.1038/ki.1997.260 Inker, L. A., Astor, B. C., Fox, C. H., Isakova, T., Lash, J. P., Peralta, C. A., Berns, J. S. (2014). KDOQI US Commentary on the 2012 KDIGO Clinical Practice Guideline for the Evaluation and Management of CKD. American Journal of Kidney Diseases, 63 (5), 713 735. doi:10.1053/j.ajkd.2014.01.416 Ishani, A., Xue, J. L., Himmelfarb, J. Eggers, P. W., Kimmel, P. L., Molitoris, B. A., & Collins, A. J. (2009). Acute Kidney Injury Increases Risk of ESRD among Elderly. Journal of the American Society of Nephrology, 20 (1), 223 228. doi:10.1681/asn.2007080837 Jin, X. D., Roethlisberger, S., Burkhard, F. C., Birkhaeuser, F., Thoeny, H. C., & Studer, U. E. (2012). Long term Renal Function After Urinary Diversion by Ileal Conduit or Orthotopic Ileal Bladder Substitution. European Urology, 61 (3), 491 497. doi:10.1016/j.eururo.2011.09.004 Jonsson, O., Olofsson, G., Lindholm, E., & Tornqvist, H. (2001). Long time experience with the Kock heal reservoir for continent urinary diversion. European Urology, 40 (6), 632 640. doi:10.1159/000049849 Joo, E. Y., Moon, Y. J., Yoon, S. H., Chin, J. H., Hwang, J H., & Kim, Y. K. (2016). Comparison of Acute Kidney Injury After Robot Assisted Laparoscopic Radical Prostatectomy Versus Retropubic Radical Prostatectomy A Propensity Score Matching Analysis. Medicine, 95 (5), 1 6. doi:10.1097/md.0000000000002650 Joung, K. W., Kong, Y. G., Yoon, S. H., Kim, Y. J., Hwang, J. H., Hong, B., & Kim, Y. K. (2016). Comparison of postoperative acute kidney injury between ileal conduit and neobladder urinary diversions after radical cystectomy: A propensity score matching analysi s. Medicine, 95 (36), 5. doi:10.1097/md.0000000000004838

PAGE 198

198 Jurkovitz, C., Franch, H., Shoham, D., Bellenger, J., & McClellan, W. (2002). Family members of patients treated for ESRD have high rates of undetected kidney disease. American Journal of Kidney Dise ases, 40 (6), 1173 1178. doi:10.1053/ajkd.2002.36866 Kheterpal, S., Tremper, K. K., Englesbe, M. J., O'Reilly, M., Shanks, A. M., Fetterman, D. M., Swartz, R. D. (2007). Predictors of postoperative acute renal failure after noncardiac surgery in pati ents with previously normal renal function. Anesthesiology, 107 (6), 892 902. Kim, C. S., Bae, E. H., Ma, S. K., Kweon, S. S., & Kim, S. W. (2014). Impact of partial nephrectomy on kidney function in patients with renal cell carcinoma. Bmc Nephrology, 15 8. doi:10.1186/1471 2369 15 181 Kirkali, Z., Chan, T., Manoharan, M., Algaba, F., Busch, C., Cheng, L., Weider, J. (2005). Bladder cancer: Epidemiology, staging and grading, and diagnosis. Urology, 66 (6A), 4 34. doi:10.1016/j.urology.2005.07.062 K ristjansson, A., & Mansson, W. (2004). Renal function in the setting of urinary diversion. World Journal of Urology, 22 (3), 172 177. doi:10.1007/s00345 004 0431 y Lantz, A. G., Saltel, E., & Cagiannos, I. (2010). Renal and functional outcomes following cy stectomy and neobladder reconstruction. Cuaj Canadian Urological Association Journal, 4 (5), 328 331. Lau, B., Cole, S. R., & Gange, S. J. (2009). Competing Risk Regression Models for Epidemiologic Data. American Journal of Epidemiology, 170 (2), 244 256. doi:10.1093/aje/kwp107 Lau, B., Cole, S. R., & Gange, S. J. (2011). Parametric mixture models to evaluate and summarize hazard ratios in the presence of competing risks with time dependent hazards and delayed entry. Statistics in Medicine, 30 (6), 654 665. doi:10.1002/sim.4123 Levey, A. S., Bosch, J. P., Lewis, J. B., Greene, T., Rogers, N., Roth, D., & Modification Diet Renal Dis Study, G. (1999). A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equatio n. Annals of Internal Medicine, 130 (6), 461 +. Levey, A. S., & Coresh, J. (2012). Chronic kidney disease. Lancet, 379 (9811), 165 180. doi:10.1016/s0140 6736(11)60178 5 Levey, A. S., Coresh, J., Balk, E., Kausz, A. T., Levin, A., & Steffes, M. W. (2003). National Kidney Foundation practice guidelines for chronic kidney disease (vol 139, pg 137, 2003). Annals of Internal Medicine, 139 (7), 605 605. Levey, A. S., Coresh, J., Balk, E., Kausz, A. T., Levin, A., Steffes, M. W., Eknoyan, G. (2003). Natio nal kidney foundation practice guidelines for chronic kidney disease:

PAGE 199

199 Evaluation, classification, and stratification. Annals of Internal Medicine, 139 (2), 137 147. Levey, A. S., Coresh, J., Greene, T., Stevens, L. A., Zhang, Y. P., Hendriksen, S., Chronic Kidney Dis Epidemiology, C. (2006). Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Annals of Internal Medicine, 145 (4), 247 254. Levey, A. S., Stev ens, L. A., Schmid, C. H., Zhang, Y. P., Castro, A. F., Feldman, H. I., Chronic Kidney Dis Epidemiology, C. (2009). A New Equation to Estimate Glomerular Filtration Rate. Annals of Internal Medicine, 150 (9), 604 U607. Lewis, E. J., Hunsicker, L. G. Bain, R P., & Rohde, R. D. (1993). The Effect of Angiotensin Converting Enzyme Inhibition on Diabetic Nephropathy New England Journal of Medicine, 329 (20), 1456 1462. doi:10.1056/nejm199311113292004 Li, L., Astor, B. C., Lewis, J., Hu, B., Appel, L. J ., Lipkowitz, M. S., Greene, T. H. (2012). Longitudinal Progression Trajectory of GFR Among Patients With CKD. American Journal of Kidney Diseases, 59 (4), 504 512. doi:10.1053/j.ajkd.2011.12.009 Lin, D. Y. (1997). Non parametric inference for cumula tive incidence functions in competing risks studies. Statistics in Medicine, 16 (8), 901 910. doi:10.1002/(sici)1097 0258(19970430)16:8<901::aid sim543>3.0.co;2 m Lin, D. Y., Wei, L. J., & Ying, Z. (1993). C hecking the Cox model with cumulative sums of mar tingale based residuales Biometrika, 80 (3), 557 572. doi:10.1093/biomet/80.3.557 Lindeman, R. D., Tobin, J., & Shock, N. W. (1985). L ongitudinal studies on the rate of decline in renal function with age Journal of the American Geriatrics Society, 33 (4), 278 285. Lunceford, J. K., & Davidian, M. (2004). Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study. Statistics in Medicine, 23 (19), 2937 2960. doi:10.1002/sim.1903 Lunn, M., & McNeil, N. (1995). A pplying cox regression to competing risks Biometrics, 51 (2), 524 532. doi:10.2307/2532940 Madersbacher, S., Schmidt, J., Eberle, J. M., Thoeny, H. C., Burkhard, F., Hochreiter, W., & Studer, U. E. (2003). Long term outcome of ileal conduit di version. Journal of Urology, 169 (3), 985 990. doi:10.1097/01.ju.0000051462.45388.14 Manoharan, M., Ayyathurai, R., & Soloway, M. S. (2009). Radical cystectomy for urothelial carcinoma of the bladder: an analysis of perioperative and survival outcome. Bju International, 104 (9), 1227 1232. doi:10.1111/j.1464 410X.2009.08625.x

PAGE 200

200 Matsushita, K., van der Velde, M., Astor, B. C., Woodward, M., Levey, A. S., de Jong, P. E., Chronic Kidney Dis Prognosis, C. (2010). Association of estimated glomerular filtrati on rate and albuminuria with all cause and cardiovascular mortality in general population cohorts: a collaborative meta analysis. Lancet, 375 (9731), 2073 2081. doi:10.1016/s0140 6736(10)60674 5 McClellan, W. M., Newsome, B. B., McClure, L. A., Howard, G., Volkova, N., Audhya, P., & Warnock, D. G. (2010). Poverty and Racial Disparities in Kidney Disease: The REGARDS Study. American Journal of Nephrology, 32 (1), 38 46. doi:10.1159/000313883 Molinaro, A. M., Simon, R., & Pfeiffer, R. M. (2005). Prediction er ror estimation: a comparison of resampling methods. Bioinformatics, 21 (15), 3301 3307. doi:10.1093/bioinformatics/bti499 Murata, K., Baumann, N. A., Saenger, A. K., Larson, T. S., Rule, A. D., & Lieske, J. C. (2011). Relative Performance of the MDRD and C KD EPI Equations for Estimating Glomerular Filtration Rate among Patients with Varied Clinical Presentations. Clinical Journal of the American Society of Nephrology, 6 (8), 1963 1972. doi:10.2215/cjn.02300311 National Institutes of Health, N. I. o. D. a. D a. K. D., Bethesda, MD (2013). U.S. Renal Data System, USRDS 2013 Annual Data Report: Atlas of Chronic Kidney Disease and End Stage Renal Disease in the United States U.S. Renal Data System, USRDS 2013 Annual Data Report: Atlas of Chronic Kidney Disease and End Stage Renal Disease in the United States. Retrieved from http://www.usrds.org/2013/view/Default.aspx Nieuwenhuijzen, J. A., de Vries, R. R., Bex, A., van der Poel, H. G., Meinhardt, W. Antonini, N., & Horenblas, S. (2008). Urinary diversions after cystectomy: The association of clinical factors, complications and functional results of four different diversions. European Urology, 53 (4), 834 844. doi:10.1016/j.eururo.2007.09.008 Nishika wa, M., Miyake, H., Yamashita, M., Inoue, T., & Fujisawa, M. (2014). Long term changes in renal function outcomes following radical cystectomy and urinary diversion. International Journal of Clinical Oncology, 19 (6), 1105 1111. doi:10.1007/s10147 014 0661 y Noordzij, M., Leffondre, K., van Stralen, K. J., Zoccali, C., Dekker, F. W., & Jager, K. J. (2013). When do we need competing risks methods for survival analysis in nephrology? Nephrology Dialysis Transplantation, 28 (11), 2670 2677. doi:10.1093/ndt/gft3 55 Odden, M. C., Whooley, M. A., & Shlipak, M. G. (2006). Depression, stress, and quality of life in persons with chronic kidney disease: The Heart and Soul Study. Nephron Clinical Practice, 103 (1), C1 C7. doi:10.1159/000090112

PAGE 201

201 Osawa, T., Shinohara, N., Maruyama, S., Oba, K., Abe, T., Maru, S., Nonomura, K. (2013). Long Term Renal Function Outcomes in Bladder Cancer After Radical Cystectomy. Urology Journal, 10 (1), 784 789. Pepe, M. S., & Mori, M. (1993). Kaplan meier, marginal or conditional prob ability curves in summ a rizing competing risks failure time data Statistics in Medicine, 12 (8), 737 751. doi:10.1002/sim.4780120803 Perimenis, P., Burkhard, F. C., Kessler, T. M., Gramann, T., & Studer, U. E. (2004). Ileal orthotopic bladder substitute co mbined with an afferent tubular segment: Long term upper urinary tract changes and voiding pattern. European Urology, 46 (5), 604 609. doi:10.1016/j.eururo.2004.07.009 Pettitt, D. J., Saad, M. F., Bennett, P. H., Nelson, R. G., & Knowler, W. C. (1990). F am ilial predisposition to renal disease in two generations of Pima Indians with type 2 (non insulin dependent) diabetes mellitus Diabetologia, 33 (7), 438 443. doi:10.1007/bf00404096 Pintilie, M. (2007). Analysing and interpreting competing risk data. Stati stics in Medicine, 26 (6), 1360 1367. doi:10.1002/sim.2655 Ploeg, M., Aben, K. K. H., & Kiemeney, L. A. (2009). The present and future burden of urinary bladder cancer in the world. World Journal of Urology, 27 (3), 289 293. doi:10.1007/s00345 009 0383 3 P ycha, A., Comploj, E., Martini, T., Trenti, E., Mian, C., Lusuardi, L., Palermo, S. (2008). Comparison of Complications in Three Incontinent Urinary Diversions. European Urology, 54 (4), 825 834. doi:10.1016/j.eururo.2008.04.068 Ries, F., & Klastersk y, J. (1986). N ephrotoxicity induced by cancer chemotherapy with special emphasis on cisplatin toxicity American Journal of Kidney Diseases, 8 (5), 368 379. Rifkin, D. E., Shlipak, M. G., Katz, R., Fried, L. F., Siscovick, D., Chonchol, M., Sarnak, M. J. (2008). Rapid Kidney Function Decline and Mortality Risk in Older Adults. Archives of Internal Medicine, 168 (20), 2212 2218. doi:10.1001/archinte.168.20.2212 Rouanne, M., Perreaud, A., Letang, N., Yonneau, L., Neuzillet, Y., Herve, J. M., Leb ret, T. (2015). Trends in Renal Function After Radical Cystectomy and Ileal Conduit Diversion: New Insights Regarding Estimated Glomerular Filtration Rate Variations. Clinical Genitourinary Cancer, 13 (3), E139 E144. doi:10.1016/j.clgc.2014.11.003 Sakaguch i, Y., Hatta, T., Hayashi, T., Shoji, T., Suzuki, A., Tomida, K., Tsubakihara, Y. (2013). Association of Nocturnal Hypoxemia with Progression of CKD. Clinical Journal of the American Society of Nephrology, 8 (9), 1502 1507. doi:10.2215/cjn.11931112

PAGE 202

202 S amuel, J. D., Bhatt, R. I., Montague, R. J., Clarke, N. W., & Ramani, V. A. C. (2006). The natural history of postoperative renal function in patients undergoing ileal conduit diversion for cancer measured using serial iotopic glomerular filtration rate an d (99m)trechnetium mercaptoacetyltriglycine renography. Journal of Urology, 176 (6), 2518 2522. doi:10.1016/j.juro.2006.07.146 Satagopan, J. M., Ben Porat, L., Berwick, M., Robson, M., Kutler, D., & Auerbach, A. D. (2004). A note on competing risks in surv ival data analysis. British Journal of Cancer, 91 (7), 1229 1235. doi:10.1038/sj.bjc.6602102 Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7 (2), 147 177. doi:10.1037//1082 989x.7.2.147 Schis terman, E. F., Cole, S. R., & Platt, R. W. (2009). Overadjustment Bias and Unnecessary Adjustment in Epidemiologic Studies. Epidemiology, 20 (4), 488 495. doi:10.1097/EDE.0b013e3181a819a1 Schmitt, R., Coca, S., Kanbay, M., Tinetti, M. E., Cantley, L. G., & Parikh, C. R. (2008). Recovery of kidney function after acute kidney injury in the elderly: A systematic review and meta analysis. American Journal of Kidney Diseases, 52 (2), 262 271. doi:10.1053/j.ajkd.2008.03.005 Shankar, A., Klein, R., & Klein, B. E. K. (2006). The association among smoking, heavy drinking, and chronic kidney disease. American Journal of Epidemiology, 164 (3), 263 271. doi:10.1093/aje/kwj173 Siegel, R., Ma, J. M., Zou, Z. H., & Jemal, A. (2014). Cancer Statistics, 2014. Ca a Cancer Jou rnal for Clinicians, 64 (1), 9 29. doi:10.3322/caac.21208 Stein, J. P., Lieskovsky, G., Cote, R., Groshen, S., Feng, A. C., Boyd, S., Skinner, D. G. (2001). Radical cystectomy in the treatment of invasive bladder cancer: Long term results in 1,054 pa tients. Journal of Clinical Oncology, 19 (3), 666 675. Stevens, L. A., Coresh, J., Greene, T., & Levey, A. S. (2006). Medical progress Assessing kidney function Measured and estimated glomerular filtration rate. New England Journal of Medicine, 354 (23 ), 2473 2483. doi:10.1056/NEJMra054415 Stevens, L. A., Schmid, C. H., Greene, T., Zhang, Y. P., Beck, G. J., Froissart, M., Levey, A. S. (2010). Comparative Performance of the CKD Epidemiology Collaboration (CKD EPI) and the Modification of Diet in Renal Disease (MDRD) Study Equations for Estimating GFR Levels Above 60 mL/min/1.73 m(2). American Journal of Kidney Diseases, 56 (3), 486 495. doi:10.1053/j.ajkd.2010.03.026 Steyerberg, E. (2009). Clinical Prediction Models: A Practical Approach to Develo pment, Validation, and Updating New York: Springer.

PAGE 203

203 Tanner, R. M., Brown, T. M., & Muntner, P. (2012). Epidemiology of Obesity, the Metabolic Syndrome, and Chronic Kidney Disease. Current Hypertension Reports, 14 (2), 152 159. doi:10.1007/s11906 012 0254 y Tareen, N., Zadshir, A., Martins, D., Pan, D. Y., Nicholas, S., & Norris, K. (2005). Chronic kidney disease in African American and Mexican American populations. Kidney International, 68 S137 S140. doi:10.1111/j.1523 1755.2005.09723.x Thoeny, H. C., S onnenschein, M. J., Madersbacher, S., Vock, P., & Studer, U. E. (2002). Is ileal orthotopic bladder substitution with an afferent tubular segment detrimental to the upper urinary tract in the long term? Journal of Urology, 168 (5), 2030 2034. doi:10.1097/01 .ju.0000034328.26362.0d Uchino, S., Bellomo, R., Goldsmith, D., Bates, S., & Ronco, C. (2006). An assessment of the RIFLE criteria for acute renal failure in hospitalized patients. Critical Care Medicine, 34 (7), 1913 1917. doi:10.1097/01.ccm.0000224227.70 642.4f Uchino, S., Kellum, J. A., Bellomo, R., Doig, G. S., Morimatsu, H., Morgera, S., Investigators, B. K. (2005). Acute renal failure in critically ill patients A multinational, multicenter study. Jama Journal of the American Medical Associatio n, 294 (7), 813 818. doi:10.1001/jama.294.7.813 Vikse, B. E., Irgens, L. M., Leivestad, T., Hallan, S., & Iversen, B. M. (2008). Low birth weight increases risk for end stage renal disease. Journal of the American Society of Nephrology, 19 (1), 151 157. doi :10.1681/asn.2007020252 Vupputuri, S., Soucie, J. M., McClellan, W., & Sandler, D. P. (2004). History of kidney stones as a possible risk factor for chronic kidney disease. Annals of Epidemiology, 14 (3), 222 228. doi:10.1016/s1047 2797(03)00126 1 Walker, A. M. (1996). Confounding by indication. Epidemiology, 7 (4), 335 336 Witjes, J. A., Comperat, E., Cowan, N. C., De Santis, M., Gakis, G., Lebret, T., Sherif, A. (2014). EAU Guidelines on Muscle invasive and Metastatic Bladder Cancer: Summary of t he 2013 Guidelines. European Urology, 65 (4), 778 792. doi:10.1016/j.eururo.2013.11.046 Xue, J. L., Daniels, F., Star, R. A., Kimmel, P. L., Eggers, P. W., Molitoris, B. A., Collins, A. J. (2006). Incidence and mortality of acute renal failure in Med icare beneficiaries, 1992 to 2001. Journal of the American Society of Nephrology, 17 (4), 1135 1142. doi:10.1681/asn.2005060668 Yoon, P. W., Bastian, B., Anderson, R. N., Collins, J. L., & Jaffe, H. W. (2014). Potentially Preventable Deaths from the Five L eading Causes of Death United States, 2008 2010. Mmwr Morbidity and Mortality Weekly Report, 63 (17), 369 374.

PAGE 204

204 Zabell, J. R., Adejoro, O., Konety, B. R., & Weight, C. J. (2015). Risk of End Stage Kidney Disease after Radical Cystectomy According to Urin ary Diversion Type. Journal of Urology, 193 (4), 1283 1287. doi:10.1016/j.juro.2014.10.103

PAGE 205

205 BIOGRAPHICAL SKETCH Shahab Bozorgmehri received his doctorate degree in the field of epidemiology, Ph D from the College of Public Health and Health Professions at the University of Florida. He received his doctorate degree in medicine, M D from Shahid Beheshti University of Medical Sciences, Tehran, Iran; and his Master in Public Health, with a concentration in epidemiology, from the College of Public Health an d Health Professions, University of Florida. He has worked as a graduate research assistant in the Division of Nephrology, Hypertension and Renal Transplantation, Department of Medicine, and as a graduate teaching assistant at the Department of Epidemiolog y, University of Florida. Prior to joining the doctoral program, he has worked as a clinician in Iran. His research interests are in kidney diseases, urological disorders, and cancer.