<%BANNER%>

Acute Kidney Injury among Trauma Patients

Permanent Link: http://ufdc.ufl.edu/UFE0042973/00001

Material Information

Title: Acute Kidney Injury among Trauma Patients Clinical Predictors, Genomics and Outcomes
Physical Description: 1 online resource (58 p.)
Language: english
Creator: BIHORAC,AZRA
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ACUTE -- GENOMIC -- GLUE -- HOST -- MULTICENTER -- MULTIORGAN -- OUTCOME -- TRAUMA
Clinical Investigation (IDP) -- Dissertations, Academic -- UF
Genre: Medical Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Our objective was to determine clinical and genomic characteristics and in-hospital mortality risk associated with acute kidney injury (AKI) in a multicenter prospective cohort of patients with blunt trauma. Less severe stages of AKI characterized by small changes in serum creatinine (sCr) are inadequately studied among trauma patients. We performed a secondary analysis of subjects enrolled in the Inflammation and the Host Response to Injury (Glue Grant) database who were adult blunt trauma patients without history of kidney disease. AKI was defined by the RIFLE (Risk, Injury, Failure, Loss, and End-stage Kidney) classification, which requires a 50% increase in sCr and stratifies patients into three severity stages: risk, injury, and failure. Association between all stages of AKI and in-hospital mortality was analyzed using a multivariable logistic regression analysis. Genome-wide expression analysis was performed on whole blood leukocytes obtained within 12 hours of trauma. Our results showed that AKI occurred in 26% of 982 patients. The adjusted risk for hospital death was three times higher for patients with AKI compared to patients without AKI (odds ratio OR 3.05 (95% confidence interval CI, (1.73, TO 5.40). This risk was proportional to the severity of AKI and even patients with mild AKI had OR for dying of 2.57 (95% CI, 1.19 to 5.50) compared to patients without AKI. Genome-wide expression analysis failed to show a significant number of genes whose expression could discriminate among patients with and without AKI. We concluded that, in a multi-center prospective cohort of blunt trauma patients, AKI characterized by small changes in sCr was associated with an independent risk of in-hospital death. Early genomic changes of the blood leukocyte transcriptome are not helpful in identifying trauma patients at increased risk of AKI.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by AZRA BIHORAC.
Thesis: Thesis (M.S.)--University of Florida, 2011.
Local: Adviser: Limacher, Marian C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-10-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042973:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042973/00001

Material Information

Title: Acute Kidney Injury among Trauma Patients Clinical Predictors, Genomics and Outcomes
Physical Description: 1 online resource (58 p.)
Language: english
Creator: BIHORAC,AZRA
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ACUTE -- GENOMIC -- GLUE -- HOST -- MULTICENTER -- MULTIORGAN -- OUTCOME -- TRAUMA
Clinical Investigation (IDP) -- Dissertations, Academic -- UF
Genre: Medical Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Our objective was to determine clinical and genomic characteristics and in-hospital mortality risk associated with acute kidney injury (AKI) in a multicenter prospective cohort of patients with blunt trauma. Less severe stages of AKI characterized by small changes in serum creatinine (sCr) are inadequately studied among trauma patients. We performed a secondary analysis of subjects enrolled in the Inflammation and the Host Response to Injury (Glue Grant) database who were adult blunt trauma patients without history of kidney disease. AKI was defined by the RIFLE (Risk, Injury, Failure, Loss, and End-stage Kidney) classification, which requires a 50% increase in sCr and stratifies patients into three severity stages: risk, injury, and failure. Association between all stages of AKI and in-hospital mortality was analyzed using a multivariable logistic regression analysis. Genome-wide expression analysis was performed on whole blood leukocytes obtained within 12 hours of trauma. Our results showed that AKI occurred in 26% of 982 patients. The adjusted risk for hospital death was three times higher for patients with AKI compared to patients without AKI (odds ratio OR 3.05 (95% confidence interval CI, (1.73, TO 5.40). This risk was proportional to the severity of AKI and even patients with mild AKI had OR for dying of 2.57 (95% CI, 1.19 to 5.50) compared to patients without AKI. Genome-wide expression analysis failed to show a significant number of genes whose expression could discriminate among patients with and without AKI. We concluded that, in a multi-center prospective cohort of blunt trauma patients, AKI characterized by small changes in sCr was associated with an independent risk of in-hospital death. Early genomic changes of the blood leukocyte transcriptome are not helpful in identifying trauma patients at increased risk of AKI.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by AZRA BIHORAC.
Thesis: Thesis (M.S.)--University of Florida, 2011.
Local: Adviser: Limacher, Marian C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-10-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042973:00001


This item has the following downloads:


Full Text

PAGE 1

1 ACUTE KIDNEY INJURY AMONG TRAUMA PATIENTS: CLINICAL PREDICTORS, GENOMICS AND OUTCOMES By AZRA BIHORAC A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR T HE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Azra Bihorac

PAGE 3

3 To my father Jusuf Bihorac: my hero, the one to always rely on in the moments of great difficulties, the voice of reason, adviser, moj Tatica.

PAGE 4

4 ACKNOWLEDGMENTS I thank my mentors Doctors Lyle Moldawer, Mark Segal and Joe Layon for their continuous support and for being a hidden force behind my success. I thank Ms. Eve Johnson for invaluable assistance with manuscript preparation and editing. Finally, I thank my hus band Charles E Hobson and my children for patiently waiting for me while I was chasing science.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 6 LIST OF FIGURES .......................................................................................................... 8 ABSTRACT ..................................................................................................................... 9 CHAPTER 1 INTRODUCTION .................................................................................................... 11 2 METHODS .............................................................................................................. 15 Subjects and Data Collection .................................................................................. 15 Outcomes and Covariate Definition ........................................................................ 16 Assessment of Acute Kidney Injury ........................................................................ 17 Genomics Data Analysis ......................................................................................... 17 Statistical Analyses ................................................................................................. 18 3 RESULTS ............................................................................................................... 22 Incidence and Progression of Acute Kidney Injury .................................................. 22 Clinical Characteristics of Patients with Acute Kidney Injury ................................... 22 Acute Kidney Injury and Clinical Outcomes after Trauma ....................................... 23 Acute Kidney Injury and Infectious Complications after Trauma ............................. 24 Genomics Analysis ................................................................................................. 26 4 DISCUSSION ......................................................................................................... 49 LIST OF REFERENCES ............................................................................................... 54 BIOGRAPHICAL SKETCH ............................................................................................ 58

PAGE 6

6 LIST OF TABLES Table page 1 1 Risk, Injury, Failure, Loss, and Endstage Kidney (RIFLE) classification ........... 13 1 2 Values for estimated baseline creatinine using Modificat ion of diet in renal disease equation (CrMDRD) ............................................................................. 13 2 1 Glue Grant inclusion and exclusion criteria for Inflammation and the Host Response to Injury .............................................................................................. 20 2 2 Marshall score from the Inflammation and Host Response to Injury Glue Grant 21 3 1 Comparison between epidemiological cohort (n=982) and genomic cohort (n=158) ............................................................................................................... 28 3 2 Baseline host characteristics, anatomic and physiologic injury severity indicators in the first 24 hours after trauma for patients stratified by RIFLEmax class. .................................................................................................................. 31 3 3 Preexisting host factors and injury description f or patients stratified by RIFLEmax class .................................................................................................... 33 3 4 Severity of illness and clinical outcomes for patient s stratified by RIFLEmax class. .................................................................................................................. 35 3 5 Association between baseline host factors and indicators of anatomic and physiologic injury obtained in the first 24 hours after trauma with the occurrence of acute kidney injury. ...................................................................... 38 3 6 Association between acute kidney injury and hospital mortality. ........................ 39 3 7 Prevalence of nosocomial infections stratified by the occurrence of RIFLE AKI 40 3 8 Characteristics of nosocomial pneumonias and bloodstream infections for patients stratified by RIFLEmax class. ................................................................ 41 3 9 Supervised genomic analysis between trauma patients and control group (uninjured subjects). ........................................................................................... 42 3 10 Class comparison and prediction between patients with no AKI and patients with AKI. Identified were 230 probe sets with a 68% to 77% correct classification rate. ............................................................................................... 42 3 11 Multi class comparison and prediction between patients with no AKI vs. patients with AKI, all stages (RIFLE R, I, and F) ................................................ 43

PAGE 7

7 3 12 Class comparison and prediction between patients with no AKI vs. patients with RIFLE F AKI only ........................................................................................ 43

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Stages of AKI defined by RIFLE criteria and associated mortality risk in the pooled analysis of 71 527 patients. .................................................................... 14 3 1 Probability curves for continuous variables associated with the occurrence of acute kidney injury (AKI).. ................................................................................... 44 3 2 Most common nosocomial infections (NCI ) stratified by severity stages of RIFLE AKI .......................................................................................................... 45 3 3 Correspondence analysis of RifleAKI stages and major types of nosocomial infections, nosocomial pneumonia and bloodstream infections. ......................... 46 3 4 Correspondence analysis of RifleAKI stages and major pathogens. ................. 47 3 5 SAM plot showing 31,165 significant probe sets. ............................................... 48

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ACUTE KIDNEY INJURY AMONG TRAUMA PATIENTS: CLIN ICAL PREDICTORS, GENOMICS AND OUTCOMES By Azra Bihorac May 2011 Chair: Marian C. Limacher Major: Medical Science Clinical and Translational Science Our objective was to determine clinical and genomic characteristics and inhospital mortality risk ass ociated with acute kidney injury (AKI) in a multicenter prospective cohort of patients with blunt trauma. Less severe stages of AKI characterized by small changes in serum creatinine (sCr) are inadequately studied among trauma patients. We performed a sec ondary analysis of subjects enrolled in the Inflammation and the Host Response to Injury (Glue Grant) database who were adult blunt trauma patients without history of kidney disease. AKI was defined by the RIFLE (Risk, Injury, Failure, Loss, and Endstage Kidney) classification, which requires a 50% increase in sCr and stratifies patients into three severity stages: risk, injury, and failure. Association between all stages of AKI and in hospital mortality was analyzed using a multivariable logistic regressi on analysis. Genomewide expression analysis was performed on whole blood leukocytes obtained within 12 hours of trauma. Our results showed that AKI occurred in 26% of 982 patients. The adjusted risk for hospital death was three times higher for patients w ith AKI compared to patients without AKI (odds ratio [OR] 3.05 (95% confidence interval [CI], (1.73, TO 5.40) This risk was proportional to the severity of AKI and even patients with mild AKI had OR for dying of

PAGE 10

10 2.57 (95% CI, 1.19 to 5.50) compared to pat ients without AKI. Genomewide expression analysis failed to show a significant number of genes whose expression could discriminate among patients with and without AKI. W e concluded that, i n a multi center prospective cohort of blunt trauma patients, AKI characterized by small changes in sCr was associated with an independent risk of in hospital death. Early genomic changes of the blood leukocyte transcriptome are not helpful in identifying trauma patients at increased risk of AKI.

PAGE 11

11 CHAPTER 1 INTRODUCTION Although acute kidney injury (AKI) is independently associated with adverse outcomes among critically ill patients, only a few studies have studied AKI among trauma patients.1 2 The majority of these studies have been retrospective, single center reports focusing on severe AKI defined by the need for renal replacement therapy (RRT) or by an increase in serum cre atinine (sCr) above a predefined, usually very high cut off point.3 6 Although a few recent studies have reported the incidence for less severe AKI as high as 31%,7 8 severe AKI defined by the need for RRT had a low incidence (between 0.1% and 8.4%) and high mortality (40% to 70 %).3 5 Hence the clinical importance of AKI with small changes in kidney function after trauma has not been adequately studied or appreciated in clinical practice. With the recent introduction of the RIFLE (Risk, Injury, Failure, Loss, and Endstage Kidney) classification system for AKI, the adverse effects of small changes in sCr level have begun to be recognized and the term AKI has been proposed to encompass the entire spectrum of the syndrome, from minor changes in renal function to the requirement for RRT, replacing the old term of acute renal failure (ARF) 9 10 The RIFLE classification defines three grades of AKI severity (R Risk, I Injury, F Failure) based on changes in sCr level relative to the baseline (Tables 1 1 and 1 2).11 Since the classifications publication in 2004, numerous original investigations using RIFLE have been published and are summarized in recent systematic review of 24 of these studies 12. The majority of the studies included critically ill patients in general or cardiac i ntensive care unit (ICU) settings12 although one study made a populationbased estimate of AKI incidence in Scotland.13 In the analysis of pooled data, increasing AKI

PAGE 12

12 severity was associated with the stepwise increase in relativ e risk for in hospital mortality (Risk, 2.40 ; Injury, 4.15; Failure, 6.37, with respect to nonAKI patients) (Figure 1 1).12 A recent large retrospective study of 120,000 patients evaluated on the first ICU day confirmed the findings of the systematic review.14 In addition, w orse ning RIFLE AKI was associated with longer ICU and hospital st ays and a lower rate of renal recovery.141 6 We have recently demonstrated that not only worse short term outcomes but also worse long term survival is associate d with AKI and is proportional to its severity.17 No study to date has assessed AKI defined by RIFLE criteria among trauma patients. The Inflammation and the H ost Response to Injury is a largescale interdisciplinary research program funded by a Glue Grant award from the National Institute of General Medical Sciences to uncover the biological reasons for different clinical outcomes after traumatic injury. The Tr aumaRelated Data b ase (TRDB), a large multicenter database containing deidentified, prospectively collected clinical and gene expression data from patients with severe blunt trauma, was developed as a part of this program and has greatly facilitated research of clinical outcomes after trauma. The goal of this study was to assess the incidence, clinical predictors, early genomic response of blood leukocytes, and the short term mortality risk associated with RIFLE defined AKI among patients with severe blunt trauma enrolled in the Inflammation and the Host Response to Injury study

PAGE 13

13 Table 11. Risk, Injury, Failure, Loss, and Endstage Kidney (RIFLE) classification Class GFR criteria Urine output criteria Risk Serum creatinine 1.5 < 0.5 ml/kg/hour 6 ho urs Injury Serum creatinine 2 < 0.5 ml/kg/hour 12 hours Failure Serum creatinine 3, or serum creatinine 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 Ren al recovery Complete r ecovery Patient returns to baseline classification within RIFLE criteria Partial r ecovery Persistent change in RIFLE classification but not persistent need for Renal Replacement Therapy Multiply by 88.4 to convert creatinine to m ol/l. Glomerular filtration rate (GFR) criteria are calculated as an increase of serum creatinine above the baseline serum creatinine level. In patients without a history of chronic kidney disease and unknown baseline serum creatinine, it is recommended ca lculating a baseline serum creatinine using the Modification of Diet in Renal Disease equation, assuming a GFR of 75 ml/min/1.73 m2. AKI should be both abrupt (within 1 7 days) and sustained (more than 24 hours) .1 1 Table 12. Values for estimated baseline creatinine using Modification of diet in renal disease equation (CrMDRD) Age (years) Black males (mg/dl) Other males (mg/dl ) Black females (mg/dl) Other females (mg/dl) 20 24 1.5 1.3 1.2 1.0 25 29 1.5 1.2 1.1 1.0 30 39 1.4 1.2 1.1 0.9 40 54 1.3 1.1 1.0 0.9 55 65 1.3 1.1 1.0 0.8 >65 1.2 1.0 0.9 0.8 CrMDRD is calculated by solving the abbreviated Modification of diet in renal disease (MDRD) equation for sCr assuming a glomerular filtration rate (GFR) o f 75 ml/minute/1.73 m2. Estimated glomerular filtration rate = 75 (ml/min per 1.73 m2) = 186 (serum creatinine [SCr]) 1.154 (age) 0.203 (0.742 if female) (1.210 if black) = exp(5.228 1.154 In [SCr]) 0.203 In(age) (0.299 if female) + (0. 192 if black).11

PAGE 14

14 Figure 11. Stages of AKI defined by RIFLE criteria and associated mortality risk in the pooled analysis of 71 527 patients.12

PAGE 15

15 CHAPTER 2 METHODS This study is a secondary analysis of the TRDB and comprises a multicenter prospective cohort of adult severe blunt trauma patients with no previous history of kidney disease. The Steering Committee of the Inflammation and the Host Response to Injury research program and the Institutional Review Board of the University of Florida approved our use of the database, in accordance with the federal requirements for access to protected patient information. Subjects and D ata C ollection Beginning November 2003, the Inflammation and the Host Response to Injury research program enrolled patients with severe blunt trauma in eig ht participating Level I trauma centers (Table 2 1 ). We analyzed completed clinical data for patients older th a n 18 years (age range 1890 years) who lived longer tha n 24 hours following injury and were enrolled between November 2003 and March 2008 (85 pat ients who died in the first 24 hours after injury were excluded) The analysis of infectious complications and AKI was performed in November 2010 and at that time the TRDB had accrued 1952 patients. Although patients with a history of renal disease and s Cr >2 mg/dl were excluded from enrollment, we also excluded an additional nine patients who had sCr >1.5 mg/dl and reported history of chronic kidney disease, as documented in the TRDB. The cohort that was selected for genomic analysis included a subset of these patients whose age was limited to < 55 years and who lived longer then initial 24 hours after trauma. The steering committee of the collaborative research program developed standard operating procedures for clinical management of the patients to mi nimize

PAGE 16

16 variation across centers involved in the data collection.18 As implemented, they were considered the standard of care for patient management and were mandated for all enrolled patients as well as for uniform routine care at each of the participating trauma centers. Trained nurse abstractors prospectively collected clinical data into TRDB, a web based data collecti on platform adapted for this program.19 Outcomes and C ovariate D efinition The Injury Severity Score (ISS) was used as a measure of anatomic injury severity. The ISS is an anatomical scoring system that provides an overall score for patients with multiple injuries. Each injury is assigned an Abbreviated Injury Scale (AIS) score and is allocated to one of six body regions ( h ead f ace chest a bdomen, e xtremities (including p elvis ), e xternal ). The Abbreviated Injury Scale (AIS) is an anatomical scoring system that ranks severity of i njuries on a scale of 1 to 6, with 1 being minor, 5 severe and 6 an survivable injury. Only the highest AIS score in each body region is used. The three most severely injured body regions have their score squared and added together to produce the ISS scor e .20 Clinical outcomes occurring within 28 days of injury were recorded. We used definitions of nosocomial infections and surgical site infections recommended by the Centers for Disease Control.21 The Marshall mul tiple organ dysfunction (MOD) score 3 was used as a cut off point for an organ failure (Table 22 ) .22 For each patient an Acute Physiology and Chronic Health Evaluation (APACHE II) score was calculated for the fir st 24 hours of injury.23 In addition, whenever analyzing effect of AKI we modified APACHE II and MOD scores to exclude renal data (APACHE IInonrenal and MODmax nonrenal ) scores by subtracting renal components from the total scores

PAGE 17

17 Assessment of A cute K idney I njury AKI was defined by the RIFLE classification using the change in sCr during the first 28 days of hospitalization compared to baseline sCr (Tables 1 1 and 1 2).11 For the baseline sCr we used the lower of two values: the lowest measured sCr in the first 24 hours after trauma (85%, 832/982 patients) or the estimated sCr (CrMDRD) (15%, 150/982 patients). CrMDRD was calculated by solving the abbreviat ed Modification of diet in renal disease (MDRD) equation for sCr assuming a glomerular filtration rate (GFR) of 75 ml/minute/1.73 m2.11 Patients with AKI were stratified according to the severity determined by comparing the highest sCr with the baseline sCr. RIFLE R corresponds to a 50% increase in sCr, RIFLE I to a two fold increase in sCr, and RIFLE F to a threefold increase in sCr. Renal outcome was evaluated by comparing the last recorded sCr to the basel ine sCr. Complete renal recovery existed if the sCr returned to a level less than 50% above baseline sCr. Partial renal recovery existed for a persistent increase more than 50% above baseline sCr but no need for RRT. No renal recovery implied a need for RR T at the time of hospital discharge or death (Table 1 1). Genomics Data Analysis Microarray data from whole blood leukocytes for the initial 12 hours of injury were evaluated in a subset of the total trauma population. This subset differed from the total p opulation in that the age distribution was 16 55 years. Microarray data from these patients were normalized using DNA Chip Analyzer 2007 ( developed and maintained by Cheng Li Lab, Harvard School of Public Health) Only patients with complete data for th e RIFLE classification and D NA quality score greater than or equal to 2 were included. Unsupervised analyses were conducted by filtering for probe sets with greater than 50%

PAGE 18

18 coefficient of variation followed by hierarchical cluster analysis. BRB ArrayTools ( developed by Dr. Richard Simon and BRB ArrayTools Development Team ) was used to perform significance analysis of microarray (SAMTM) comparisons between injured and healthy subjects and AKI vs no AKI patients. Class prediction models were used defining a n F test (p<0.001) to identify significant probe sets followed by Leave One Out Cross Validation (LOOCV) analysis using diagonal linear discriminate analysis, K nearest neighbors (for K=1 and 3), and nearest centroid prediction methods. All class compariso ns and prediction models were carried out using 1000 permutations of the data set. Significance for SAMTM analysis was set using a false discovery rate (FDR) of <0.001. Statistical A nalyses Multivariate logistic models were used to assess factors associat ed with AKI and mortality during the hospitalization. Multiple risk factors (patient characteristics, anatomic and physiologic injury indicators) identified on the basis of prior studies of outcomes in the trauma and potential clinical and physiological s ignificance (as determined by the practicing t rauma surgeons and intensivists) were evaluated for univariate association with the primary outcome (twotailed P 0.20) and then entered stepwise into multivariable logistic models, with assessment of the assoc iation between AKI groups and outcome in the presence of the significant covariates. Each variable with a significant association (P<0.05) and additional variables that were not significant but had potential clinical importance were included in the final f ull model. The goodness of fit of the logistic regression model was assessed with the Hosmer Lemeshow test and concordance indices reported as a measure of discriminatory capability of the models. Probit models were used to assess the estimated probabili ty of any in hospital

PAGE 19

19 AKI with specific parameters of interest occurring within 24 hours of hospitalization. Missing parameters were considered missing at random (there were <10% missing values for the wors t lactate levels in the first 24 hours while all other variables had < 1% missing values) and were categorized as a separate level for the purposes of the analytic models. Sensitivity analyses confirmed that the primary study results were consistent with and without inclusion of these cases. All statistic al tests for group comparisons were twotailed. We used c orrespondence analysis an exploratory data analytic technique t o analyze m ultiway tables and provide measure of correspondence between the rows and columns when appropriate. Statistical analyses w ere performed with SAS (version 9.2, Cary, N.C.).

PAGE 20

20 Table 21. Glue Grant i nclusion and exclusion criteria for Inflammation and the Host Response to Injury Criteria Description Inclusion Criteria 1. 2. 3. 4. 5. 6. Patient with blunt trauma without i solated head injury. Emergency department arrival <= 6 hours from time of injury. Blood transfusion within 12 hrs of injury. Base deficit >= 6 OR systolic blood pressure < 90 mmHg within 60 minutes of emergency department arrival. Fully or partially intact cervical spinal cord. AIS ( Abbreviated Injury Scale) head (brain or cranium) 3 OR no AIS head. Exclusion Criteria 1. 2. 3. 4. 5. 6. 7. 8 9. 10. 11. 12. 13. 14. 15. Age < 16 OR age > 55 years. (the upper age limit was applicable for the genomic analysis only) Anticipated survival < 24 hours from injury. Anticipated survival < 28 days due to preexisting medical condition. Inability to obtain first blood draw within first 12 hours after injury. Traumatic brain injury, i.e. GCS (Glasgow Coma Sc ale) less than or equal to 8 after ICU admission AND brain computerized tomography scan abnormality within first 12 hours after injury. Inability to obtain informed consent. Pre existing, ongoing immunosuppression, e.g. Transplant recipient. Pre existing ongoing immunosuppression e.g. Chronic high dose corticosteroids (>20 mg/prednisoneequivalents/day). Pre existing, ongoing immunosuppression Oncolytic drug(s) therapy within the past 14 days. Pre existing, ongoing immunosuppression HIV positive AN D CD4 count <200 cells/mm3. Possible requirement for early immunosuppression e.g. significant likelihood of requiring high dose corticosteroids (e.g. spinal cord injury). Significant pre existing organ dysfunction Lung: currently receiving home oxygen therapy, as documented in medical records. Significant pre existing organ dysfunction Heart: congestive heart failure, as documented in medical records. Significant pre existing organ dysfunction Renal: chronic renal failure (creatinine > 2 mg/dl). Sig nificant preexisting organ dysfunction Liver: cirrhosis with portal hypertension or encephalopathy.

PAGE 21

21 Table 22 Marshall s core from the Inflammation and Host Response to Injury Glue Grant Component Measurem e nt Score 0 1 2 3 4 Repiratory PaO2/FiO2 > 300 (225, 300) (150, 225 ) (75, 150 ) 75 Renal Creatinine < 1.2 ( 1.2 2.4) ( 2.4, 4.0) ( 4.0, 5.7) 5.7 Hepatic Bilirubin < 1.2 ( 1.2, 3.6) ( 3.6, 7.2) ( 7.2. 14.2) 14.2 Cardiovascular PAR 10 (10, 15) (15, 20 ) (20, 30 ) > 30 Hematologic Platelet c ount > 120 (80, 120) (50, 80 ) (20, 50 ) 20 Neurologic Glasgow Coma Score 15 13 14 10 12 7 9 6 PAR (Pressure a djusted h eart rate is computed as The classic Marshall score is the sum of the 6 component scores shown below. For the Inflammation and Host Response to Injury Glue Grant data analysis, the Marshall score is defined as the sum of 5 component scores with the n eurologic component excluded. For the analysis of the data from the initial five years of the study, linearly interpolated component scores were computed. This resulted in a slight over estimate for the renal, hepatic and cardiovascular scores and a slight under estimate for the respiratory and hematologic scores.

PAGE 22

22 CHAPTER 3 RESULTS Incidence and P rogre ssion of A cute K idney I njury We analyzed data for 982 adult patients with severe blunt trauma and no previous history of kidney disease who lived longer tha n 24 hours following injury (85 patients who died in the first 24 hours after injury were excluded) (Table s 3 1 3 2 and 3 3 ). This cohort included severely injured patients as defined by inclusion criteria (82% of patients had ISS score >=25, 80% had an episode of hypotension and 60% required > six units of blood transfusion in the first 24 hours of injury) ( Table 31 3 2 and 3 3 ) Over onefourth of these patients developed AKI in the first 28 days after trauma (26%, 253/982) and twothirds had only mild to moderately severe AKI (RIFLE R and RIFLE I) (Tables 31 and 32) The majority of the patients had the onset of AKI within the first two days of hospital admission (68%) and close to half of the patients progressed from RIFLE class R to RIFLE class I or class F (42%, 77/183). The time to progress from class R to class I was 2 days ( interquartile ran ge 2 5 days) while time to progress to class F was 6 days ( interquartile range 3 10 days ) Only 11% of AKI patients required RRT. Almost half of the patients with AKI failed to complete ly recover their kidney function in the first 28 days after trauma. I ncreasing severity of AKI was associated with less likelihood of renal recovery (Table 34 ). Clinical C haracteristics of P atients w ith A cute Kidney I njury Neither the baseline host factors nor anatomic injury severity score were associated with the risk f or AKI in a multiple logistic regression model that included a number of clinical parameters obtained during the first 24 hours of trauma (Table 35 ). In contrast, several indicators of physiologic injury severity including the lowest body

PAGE 23

23 temperature, the highest lactate level, and the need for packed red blood cells (PRBC) and cryoprecipitate transfusion were independently associated with a higher risk for developing AKI (Table 35 and Figure 31). Interestingly, an absolute value of first measured sCr was not independently associated with the risk for AKI but rather the ratio between first measured sCr and estimated CrMDRD based on patients age, gender and race was independently associated with the risk for developing AKI. Every 10% increase in the ratio between measured sCr on admission and estimated CrMDRD increased risk for developing AKI by 8%. Acute K idney I njury and C linical O utcomes after T rauma Patients with AKI were more severely ill as reflected in the higher MOD scores (Table 3 4 ). Organ dys function as defined by MOD score among AKI patients and the occurrence of organ dysfunction was proportional to the severity of kidney injury. Notably, none of the patients with RIFLE R and RIFLE I and only half of the patients with RIFLE F had kidney dysfunction on the basis of the MOD score. Similarly, only 15% of all AKI patients would be classified as having posttraumatic acute renal failure using the American College of Surgeons Committee on Trauma definition (sCr above 3.5 mg/dl) and all of them were in the RifleF class. Even the American College of Surgeons National Surgical Quality Improvement projects (NSQIP) definition of acute renal dysfunction as sCr above 2 mg/dl does not identify 8 5% of patients with R IFLE defined AKI ( Table 34). Patients with AKI were more likely to develop both infectious and noninfectious complications. They required a longer ICU and hospital stay. The inhospital mortality for patients with AKI was 30% in contrast to 5% for those with no AKI.

PAGE 24

24 AKI was associated with a threefold increase in the risk for hospital mortality in the multivariate logistic model. This model also included host factors, anatomic and physiologic injury indicators in the first 24 hours after trauma and severity scores. This increase in the risk for hospital mortality was proportional to the severity of AKI (Table 36). The least severe yet the most frequent AKI (RIFLE R) was associated with a 2.5fold increase in the risk for mortality even after adjustment for the host characteristics, anatomic and physiologic injury indicators, and severity scores (Table 36). The most severe AKI (RIFLE F) was associated with a fivefold risk for dying in hospital. In addition to AKI, ISS and APACHE II scores on admission, use of vaso pressors in the first 24 hours, and the highest MOD score in the first 28 days were independently associated with inhospital mortality. Acute K idney I njury and Infectious Complications after T rauma The analysis for infectious complications was performed in January 2011; at that time TRDB database had 1952 completed clinical cases. After excluding 213 patients who did not have sCr data beyond day 0, remaining cohort of 1793 patients was analyzed for the association between RIFLE AKI and infectious complica tions. The prevalence of nosocomial infections (NCI) after trauma was 4 9 %. The prevalence of nosocomial pneumonia, bloodstream infections (BSI) and central line related bloodstream infections (CLBSI) was significantly higher among patients with AKI in proportion to AKI severity stages (Figure 32). Patients with AKI had 1.6 times the odds of having NCI compared to patients without AKI (Table 37) This relationship was proportional to the severity of AKI and ranged from 1.3 times the odds for least severe R IFLE R to 2.7 times the odds for most severe RIFLE F when compared to patients without AKI. Among NCI, the odds for pneumonia (1. 55), bloodstream infections ( 2.06 )

PAGE 25

25 and intravenous catheter related infections ( 1.96) were higher for AKI patients. On average, the onset of AKI preceded the onset of pneumonia and bloodstream infections by 5 and 6 days, respectively (Table 38). Interestingly, simple correspondence analysis map revealed a number of clusters: Patients with RIFLE R were associated with isolated pneumonia episodes while patients with RIFLE I and RIFLE F were clustered with BSI and combined episodes of pneumonia and BSI. Since Dimension 1 accounted for 93.77% of the inertia of the map, the severity stage of AKI (RIFLE I and RIFLE F) were the most important determinants of this association ( Figure 33 ). In regard to causative pathogens for nosocomial pneumonias, Staphylococcus aureus was more common gram positive pathogen among patients with AKI (Table 38). The most significant association, however wa s for gram negative bacteria and especially Acinetobacter baumannii : the prevalence of Acinetobacter pneumonia was three fold higher among patients with RIFLE F as compared to patients with no AKI. The simple correspondence analysis map of this matrix reve aled a number of clusters: Patients with RIFLE R were associated with isolated Staphylococcus aureus pneumonia episodes while patients with RIFLE I and RIFLE F were clustered with Acinetobacter pneumonias or combined Staphylococcus aureus and Acinetobacter infections. Dimension 1 accounted for 64.45% of the inertia of the map and RIFLE F (far right quadrant) was the most important determinant of the association. Dimension 2 accounted for 31.68% of the inertia of the map and combined Staphylococcus aureus and Acinetobacter pneumonias as well as isolated Acinetobacter pneumonias (Top quadrants) were the most important determinants of the association ( Figure 34 ).

PAGE 26

26 The gram negative bloodstream infections with Acinetobacter Escherichia coli and Serratia marcesc ens and also with Candida species were more likely to occur among AKI patients (Table 38). The simple correspondence analysis map of this matrix revealed a few interesting clusters: Patients with RIFLE R were associated with coagulase negative Staphylococ cus BSI while patients with RIFLE F were clustered with Acinetobacter Candida and Escherichia coli BSIs. Serratia and Klebsiella BSIs clustered with RIFLE I. Dimension 1 accounted for 65.79% of the inertia of the map and all three AKI stages were contributing to the association. Dimension 2 accounted for 22.21% of the inertia of the map with Serratia Klebsiella Acinetobacter and Candida pathogens being the most important determinants of the association ( Figure 34 ). Genomics Analysis The TRDB contained m icroarray data for 173 patients sampled within 12 hours of traumatic injury along with data obtained from 24 healthy uninjured subjects. Of the 173 trauma patients, 158 met inclusion criteria with 125 subjects having no AKI and 33 with AKI (13 RIFLE R, 9 R IFLE I and 11 RIFLE F) After microarray normalization and model based expression, a supervised analysis between the 173 trauma patients and 24 uninjured subjects incorporating a twoclass comparison (Students t test, P<0.001) and Leave One Out Cross Vali dationTM (LOOCV) identified 30,956 probe sets that were able to correctly classify the healthy subjects from the injured patients in greater than 98% of cases ( Table 39 ). The results were verified using the SAMTM algorithm (FDR<0.001), which identified 31,165 probe sets that distinguished the healthy subjects from the injured patients ( Figure 35 ). The 31,165 probe sets identified by SAMTM analysis were then incorporated into the PAMTM algorithm, which correctly classified the healthy controls and the injured patients 100% of the time (data not shown). These results

PAGE 27

27 indicate that a robust genomic signature is generated from whole blood leukocytes, which can readily discriminate the trauma population early after initial injury from healthy control subjects. Class comparison and prediction between the 125 patients with no AKI and all 33 patients with AKI, identified 230 probe sets with a 68% to 77% correct classification rate ( Table 310). Similarly, a multiclass comparison and prediction between injured patie nts without AKI and those with RIFLE classes R, I, and F, identified 151 probe sets with a 51% to 77% correct classification rate ( Table 311). When the 125 patients with no AKI were compared to the 11 patients in class RIFLE F, 95 probe sets were identifi ed that correctly classified the patients 85% to 91% of the time; however the positive predictive value for patients with RIFLE F was only 0.20 while patients with no AKI carried a positive predictive value of 0.93 ( Table 312). SAMTM analysis was used to verify the class comparison and prediction findings. However, analysis of patients with and without AKI, multiclass analysis between AKI and RIFLE classes, and injured patients with no AKI compared with RIFLE F yielded no significant probe set differences. These findings suggest that although trauma patients with and without AKI may have different patterns of leukocyte gene expression, the genomic changes are modest and lack strong predictive properties. Such findings suggest that there do not appear to be strong genomic signatures associated with AKI in a severely injured cohort in which marked genomic changes are occurring after the injury.

PAGE 28

28 Table 31. Comparison between epidemiological cohort (n=982) and genomic cohort (n=158) Epidemiological cohor t Genomic cohort P Value Range Range Baseline host characteristics Age (years)(mean, 95% CI) 41 (41 44) 18 90 35 (33 37) 18 55 <0.001 Male, No. ( %) 632 (64) 106 (67) 0.45 African American ethnicity, No. (%) 72 (7) 10 (6) 0.63 BMI ( kg/m 2 ) (mean, 95% CI) 28.3 (27.8 28.7) 14.1 67.9 28.6 (27.6 29.7) 17.6 67.9 0.49 Anatomic injury indicators 0 to 24 hours Injury Severity Score, No. (%) Mild injury (<16) Moderate injury (1624) Severe injury (2540) Massive injury (>40) 26 (3) 145 (15) 565 (57) 246 (25) 5 (3) 25 (16) 91 (58) 37 (23) 0.58 Physiologic injury indicators 0 to 24 hours MAP<65 mmHg, No. (%) 787 (80) 123 (78) 0.55 Temperature < 34.5 C, No. ( %) 321 (33) 51 (32) 0.81 Lowest Hct (%)(mean, 95% CI) 23.3 ( 22.9 23.7) 5.0 45.9 23.6 (22.7 24.5) 10.0 37.3 0.54 Apache II (no renal data) (mean, 95% CI) 28 (27 28) 6 47 27 (26 28) 10 37 0.28 Lactate Missing, No. (%) 444 (45) 92 (9) 71 (45) 11 (7) 1.00 Base deficit 10 No. (%) Missing, No. (%) 497 (51) 5 (0.6) 77 (49) 1 (0.5) 0.64 PO 2 /FiO 2 ratio<200, No. (%) Missing, No. (%) 638 (65) 62 (6) 110 (69) 6 (4) 0.33

PAGE 29

29 Table 31. Continued. Epidemiological cohort Genomic cohort P Value Range Range pH < 7.2 No. (%) Missing, No. (%) 412 (42) 6 (0.6) 66 (42) 1 (0.5) 1.00 Blood glucose > 200 mg/dl, No. (%) 423 (43) 58 (37) 0.16 RBC transfusion > 6 U No. (%) 492 (60) 92 (58) 0.63 Platelets transfusion No. (%) 409 (42) 66 (42) 1.00 FFP transfusion No. (%) 645 (66) 118 (71) 0.22 Cryoprecipitate transfusion No. (%) 266 (27) 46 (29) 0.60 Severity of illness (first 28 days) MOD max sco re (mean, 95% CI) 5.7 (5.5 5.9) 1 16 5.5 (5.1 5.9) 0 16 0.41 MOD max nonrenal score (mean, 95% CI) 4.7 (4.5 4.8) 0 14 4.5 (4.2 4.9) 0 12 0.38 Complications (first 28 days) Non infectious complications, No. (%) 449 (46) 81 (51) 0.24 Surgical si te infections, No. (%) 158 (16) 27 (17) 0.75 Nosocomial infections, No. (%) 481 (49) 84 (53) 0.35 Ventilator associated pneumonia, No. (%) 295 (30) 47 (28) 0.61 Renal outcomes (first 28 days) All AKI, No. (%) 253 (26) 33 (21) 0.18 RIFLE R No. (%) 106 (11) 13 (8) 0.25 RIFLE I, No. (%) 79 (8) 9 (6) 0.63 RIFLE F, No. (%) 68 (7) 11 (7) 1.00 Highest sCr in the first 28 days (mg/dl) (mean, 95% CI) 1.38 (1.32 1.44) 0.51 9.2 1.37 (1.18 1.55) 0.6 9.0 0.91

PAGE 30

30 Table 31. Continued. Epide miological cohort Genomic cohort P Value Range Range AKI duration (days) (median, IQR) 5 (2, 13) 1 28 7 (2, 13) 0.69 Renal replacement therapy, No. (%) 29 (3) 5 (14) 0.004 Renal recovery, No. (%) Complete recovery 128 (51) 20 (61) 0.28 Partial recovery 104 (41) 9 (28) 0.15 No recovery 21 (8) 4 (11) 0.56 Outcomes for whole hospital stay ICU length of stay (days) (median, IQR) 10 (5, 19) 0 142 9.5 (5, 18) 0.58 Hospital length of stay (days) (median, IQR) 19 (11, 31) 2 355 20.5 (12, 31) 0.71 Hospital mortality, No. (%) 119 (12) 8 (5) 0.009 Discharge to Home, No. (%) 292 (30) 49 (31) 0.79 Discharge to inpatient rehabilitation, No. (%) 275 (28) 35 (22) 0.12 Discharge to skilled nursing facility, No. (%) 252 (3 0) 56 (35) 0.21 Abbreviations: IQR, interquartile range; 95% CI, 95% confidence interval for the mean; MOD, the Marshall multiple organ dysfunction score. Included minimum maximum values for continuous variables only. Comparing epidemiologic and genomic cohort. Calculated by subtracting renal component from the total MOD score.

PAGE 31

31 Table 32. Baseline host characteristics, anatomic and physiologic injury severity indicators in the first 24 hours after trauma for patients stratified by RIFLEmax clas s. No AKI N=729 AKI P P All AKI patients N=253 RIFLE R N=106 RIFLE I N=79 RIFLE F N=68 Baseline host characteristics Age (years) (mean, 95% CI) 41 (40 42) 47 (45 50) 48 (44 52) 46 (42 50) 47 (20) <0.001 0.79 Male, No. ( %) 455 (62) 1 77 (70) 72 (68) 55 (70) 50 (74) 0.003 0.73 African American ethnicity, No. ( %) 51 (7) 21 (8) 10 (9) 6 (8) 5 (7) 0.49 0.85 BMI (kg/m 2 ) (mean, 95% CI) 27 (27 28) 29 (28, 30) 28 (27 30) 30 (29 32) 30 (28 32) 0.001 0.14 Baseline sCr (mg/dl) (mean, 95% CI) 0.79 (0.77 0.81) 0.81 (0.78 0.84) 0.78 (0.73 0.82) 0.84 (0.78 0.89) 0.84 (0.78 0.89) 0.20 0.16 CrMDRD used for baseline, No. ( %) 80 (11) 70 (28) 21 (20) 28 (35) 21 (31) <0.001 <0.001 Ratio between measured sCr (first 24 hrs) and Cr MDRD (mean, 95% CI) 0.99 (0.97 1.01) 1.25 (1.19 1.32) 1.12 (1.05 1.18) 1.30 (1.19 1.41) 1.41 (1.23 1.59) <0.001 <0.001 Anatomic injury indicators 0 to 24 hours Injury Severity Score No. ( %) Mild injury (<16) Moderate injury (16 24) Severe injury (25 40) Massive in jury (>40) 21 (3) 114 (16) 424 (58) 170 (23) 5 (2) 31 (12) 141 (56) 76 (30) 4 (4) 16 (15) 53 (50) 33 (31) 1 (1) 11 (14) 46 (58) 21 (27) 0 (0) 4 (6) 42 (62) 22 (32) 0.13 0.23 Glasgow Coma Scale Prior to ED admission ED 213 (31) 367 (50) 84 (33) 138 (55) 38 (36) 31 (46) 29 (37) 61 (57) 17 (25) 46 (58) 0.31 0.25 0.68 0.22 Physiologic injury indicators 0 to 24 hours MAP < 65 mm Hg, No. ( %) 565 (78) 222 (88) 88 (83) 71 (90) 63 (93) <0.001 0.09 Temperature < 34.5 C, No. ( %) 200 (2 7) 131 (52) 43 (41) 42 (53) 36 (53) <0.001 0.14

PAGE 32

32 Table 3 2 Continued No AKI N=729 AKI P P All AKI patients N=253 RIFLE R N=106 RIFLE I N=79 RIFLE F N=68 Lowest Hct (%)(mean, 95% CI) 24 (23 24) 22 (21 22) 23 (22 24) 21 (20 23) 20 (19 22) <0 .001 0.07 INR (mean, 95% CI) 1.4 (1.34 1.44) 1.5 (0.7) 1.4 (1.3 1.5) 1.5 (1.3 1.6) 1.8 (1.4 2.1) 0.05 0.05 Apache II (no renal data) (mean, 95% CI) 27 (26 27) 31 (30 32) 29 (28 31) 32 (31 33) 32 (30 33) <0.001 0.005 Lactate mmol/l, No. ( %) 28 8 (39) 156 (62) 55 (52 ) 58 (73 ) 43 (63) <0.001 0.09 Base deficit 10, No. ( %) 327 (45) 170 (67) 66 (62 ) 50 (63 ) 54 (79) <0.001 0.24 PO 2 /FiO 2 ratio<200, No. ( %) 444 (61) 194 (77 ) 69 (65 ) 63 (80 ) 62 (91) <0.001 0.005 pH < 7.2, No. ( %) 269 (37) 143 (57 ) 49 (46 ) 49 (62 ) 45 (66) <0.001 0.15 Blood glucose > 200 mg/dl, No. ( %) 276 (38) 147 (58 ) 60 (57) 43 (54 ) 44 (65) <0.001 0.79 RBC transfusion (U) (median, IQR) 5 (3, 9) 10 (5, 17) 6 (4, 13) 11 (5, 21) 12 (7, 25) <0.001 <0.001 RBC transfu sion > 6 U, No. (%) 322 (44) 170 (67) 58 (55) 56 (71) 56 (82) <0.001 <0.001 Platelet transfusion No. (%) 271 (37) 138 (55) 48 (45) 50 (63) 40 (59) <0.001 0.36 FFP transfusion, No. (%) 448 (62) 197 (78) 75 (71) 64 (81) 58 (85) <0.001 0.06 FFP transfusi on (U) (median, IQR) 2 (0, 7) 6 (2, 14) 5 (0, 10)` 7 (2, 15) 9 (4, 18) <0.001 0.01 Cryoprecipitate transfusion No. (%) 157 (22) 109 (43) 35 (33) 40 (51) 34 (50) <0.001 0.27 Crystalloids (L) (median, IQR) 13 (10, 17) 16 (10, 23) 14 (10.22) 17 (9, 26) 17 (11, 22) <0.001 0.2 Use of vasopressors No. (%) 107 (15) 84 (33) 21 (20) 29 (37) 34 (50) <0.001 0.002 Abbreviations: IQR, interquartile range; CI, 95% confidence interval for the mean; .BMI, body mass index; MAP, mean arterial pressure; Hct, hematocrit; INR, international normalized ratio; PRBC, packed red blood cells; FFP, fresh frozen plasma; U, units. No acute kidney injury (AKI) is those patients without any occurrence of RIFLE criteria. Comparing patients with AKI and patients without AKI. Compa ring patients within the three RIFLEmax subgroups

PAGE 33

33 Table 33. Preexisting host factors and injury description for patie nts stratified by RIFLEmax class All cohort (N=982) No AKI N=729 AKI P P All AKI patients N=253 RIFLE R N=106 RIFLE I N=79 RIFLE F N =68 Preexisting host factors Obesity (BMI>30) 203 (28) 94 (37) 30 (28) 37 (47) 27 (40) 0.001 0.03 Hypertension (n=154, 16%) 100 (14) 54 (21) 20 (19) 14 (18) 20 (29) 0.004 0.16 COPD (n=37, 4%) 22 (3) 15 (6) 4 (4) 6 (8) 5 (7) 0.06 0.46 PVD (n=13, 1%) 9 (1) 4 (2) 2 (2) 1 (1) 1 (1) 0.41 0.67 Heart disease (n=68, 7%) 42 (6) 26 (10) 12 (11) 5 (6) 9 (13) 0.02 0.34 Neurologic disease (n=77, 8%) 51 (7) 26 (10) 9 (8) 7 (9) 10 (15) 0.10 0.39 Diabetes mellitus (n=66, 7%) 39 (5) 27 (11) 11 (10) 4 (5 ) 12 (18) 0.005 0.05 Malignancy (n=38, 4%) 24 (3) 14 (5) 6 (6) 4 (5) 4 (6) 0.11 0.97 Liver disease (n=43, 4%) 22 (3) 21 (8) 4 (4) 8 (10) 9 (13) 0.004 0.07 Smoker (n=303, 31%) 241 (33) 62 (25) 28 (26) 18 (23) 16 (24) 0.01 0.83 Chronic alcohol abuse (n=1 46, 15%) 99 (14) 47 (19) 23 (22) 13 (16 ) 11 (16) 0.06 0.55 Admission medications Antiplatelet agents (n=81, 8%) Coumadin (n=19, 2%) 53 (7) 11 (2) 28 (11) 8 (3) 12 (11) 3 (3) 5 (6) 4 (5) 11 (16) 1 (1) 0.06 0.10 0.16 0.45 Mechanism of injury Pedestrian vs Motor vehicle crash (n=132, 13%) 84 (12) 48 (19) 21 (20) 16 (20) 11 (17) 0.002 0.48 Time from injury to ED (hr) (median, IQR) 1.3 (0.8, 2.5) 1.1 (0.7, 2.1) 1.1 (0.7, 2.2) 1.1 (0.5, 1.8) 1.3 (0.7, 2.6) 0.06 0.39 Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; PVD, peripheral vascular disease;. IQR, interquartile range. Categorical variables are presented as number (percentages). No acute kidney injury (AKI) is those patients without any occurr ence of RIFLE criteria. Comparing patients with AKI and patients without AKI. Comparing patients within the three RIFLEmax subgroups. Heart disease (myocardial infarction, congestive heart

PAGE 34

34 failure, atrial or ventricular tachyarrythmias); Neurologic disease (cerebrovascular disease, dementia, traumatic brain injury, Parkinsonss disease); Malignancy (history of malignancy or metastatic solid tumor); Including NSAIDs, aspirin and all other antiplatelet agents.

PAGE 35

35 Table 34. Severity of illness and c linical outcomes for patients stratified by RIFLEmax class. No AKIN=729 AKI P 1 P 2 All AKI patients N=253 RIFLE R N=106 RIFLE I N=79 RIFLE F N=68 Severity of illness (first 28 days) MODmax score (mean, 95% CI) 4.9 (4.7 5.1) 8.1 (3.4) 6. 0 (5.5 6.5) 8.6 (7.9 9.2) 10.7 (10.0 11.4) <0.001 <0.001 MOD max nonrenal score (mean, 95% CI) 3 4. (3.9 4.2) 6.4 (2.9) 4.9 (4.4 5.4) 7.0 (6.4 7.7) 7.9 (7.3 8.6) <0.001 <0.001 MOD max nonrenal score No. (%) 146 (20) 141 (56) 40 (38) 50 (63) 51 (75) <0.001 <0.001 Day of MOD max score 2 (0, 4) 3 (2, 8) 2 (0, 6) 3 (2, 8) 7 (3, 12) <0.001 <0.001 Organ dysfunction/failure 4 No. (%) Respiratory 67 (9) 78 (31) 20 (19) 25 (32) 33 (49) <0 .001 <0.001 Cardiovascular 270 (37) 184 (73) 59 (56) 65 (82) 60 (88) <0.001 <0.001 Hepatic 26 (4) 45 (18) 7 (7) 16 (20) 22 (32) <0.001 <0.001 Hematology 5 (1) 18 (7) 0 (0) 7 (9) 11 (16) <0.001 <0.001 Renal 0 33 (13) 0 (0) 0 (0) 33 (49) <0.001 <0.001 ICU MV duration (days) (median, IQR) 6 (2,12) 14 (6, 23) 10 (3, 20) 15 (7, 25) 20 (7, 28) <0.001 <0.001 Complications (first 28 days) Non infectious complications, No. (%) 278 (38) 171 (68) 56 (53) 54 (68) 61 (90) <0.001 <0.001 Surgical site inf ections, No. (%) 94 (13) 64 (25) 18 (17) 24 (30) 22 (32) <0.001 0.03 Nosocomial infections No. (%) 325 (45) 156 (62) 54 (51) 56 (71) 46 (68) <0.001 0.01 Ventilator associated pneumonia No. (%) 189 (26) 106 (42) 35 (33) 39 (49) 32 (47) <0.001 0.05 Rena l outcomes (first 28 days) Highest sCr (mg/dl) (mean, 95% CI) 1.08 (1.06 1.10) 2.25 (2.00 2.43) 1.33 (1.25 1.42) 1.92 (1.79 2.00) 4.05 (3.61 4.49) <0.001 <0.001

PAGE 36

36 Table 3 4 Continued No AKI N=729 AKI P1 P2 All AKI patients N=253 RIFLE R N=106 RIFLE I N=79 RIFLE F N=68 NSQIP Definition Highest sCr > 2 mg/dl 5 No. (%) 0 115 (45) 19 (17) 35 (44) 61 (90) <0.001 <0.001 ACSCT Definition Highest sCr > 3.5 mg/dl 6 No. (%) 0 39 (15) 0 0 39 (56) <0.001 <0.001 AKI duration (days) (medi an, IQR) 5 (2, 13) 3 (2, 6) 6 (3, 11) 14 (6, 24) <0.001 Renal replacement therapy No. (%) 29 (11) 2 (2) 5 (6) 22 (32) <0.001 Renal recovery, No. (%) Complete recovery 128 (51) 72 (68) 46 (58) 10 (15) <0.001 Partial recovery 104 (41) 33 ( 31) 30 (38) 41 (60) <0.001 No recovery 21 (8) 1 (1) 3 (4) 17 (25) <0.001 Outcomes for whole hospital stay ICU LOS (days) (median, IQR) 9 (4, 16) 18 (7, 27) 15 (5, 23) 20 (9, 30) 22 (8, 37) <0.001 0.003 Hospital LOS (days) (median, IQR) 18 (11 28) 24 (10, 38) 23 (11, 31) 27 (14, 41) 24 (9, 50) <0.001 0.14 ICU Mortality No. (%) 37 (5) 71 (28) 16 (15) 23 (29) 32 (47) <0.001 <0.001 Hospital mortality No. (%) 39 (5) 80 (32) 20 (19) 25 (32) 35 (51) <0.001 <0.001 Discharge to Home No. (%) 255 (35) 37 (15) 24 (23) 12 (15) 1 (1) <0.001 <0.001 Discharge to inpatient rehabilitation, No. (%) 219 (30) 57 (23) 23 (22) 16 (20) 18 (26) <0.001 <0.001 Discharge to skilled nursing facility, No. (%) 182 (25) 69 (27) 34 (32) 22 (28) 13 (19) <0.001 <0.001 Abbreviations: IQR, interquartile range; 95% CI, 95% confidence interval for the mean; MOD, the Marshall multiple organ dysfunction score; ICU, Intensive care unit; MV, mechanical ventilation; sCr, serum creatinine; LOS length of stay. 1 Comparing patient s with AKI and patients without AKI. 2 Comparing patients within the three RIFLEmax subgroups. 3 Calculated by subtracting renal component from the total MOD score. 4MOD score off point for an organ dysfunction/failure. 5The American College of Surgeons Committee on Trauma (ACSCT) defines acute renal failure

PAGE 37

37 after trauma as sCr above 3.5 mg/dl. 6The American College of Surgeons National Surgical Quality Improvement project (NSQIP) defines acute renal dysfunction in surgical patients as sCr above 2 mg/dl.

PAGE 38

38 Table 35. Association between baseline host factors and indicators of anatomic and physiologic injury obtained in the first 24 hours after trauma with the occurrence of acute kidney injury. Patient Characteristics Occurrence of AKI (a ll severity levels) OR (95% CI) P Host factors Obesity (per kg/m 2) ) 1.02 (0.99, 1.05) 0.16 Hypertension (Reference none) 1.38 (0.80, 2.40) 0.25 Heart disease (Reference none) 1.07 (0.48, 2.37) 0.88 Diabetes mellitus (Reference none) 1.49 (0.69, 3. 26) 0.29 Smoking (Reference none) 0.83 (0.56, 1.24) 0.36 Use of antiplatelets/anticoagulation drugs (Reference none) 1.00 (0.49, 2.07) 0.99 Anatomic injury indicators in the first 24 hours Injury severity Score (per unit change) 0.99 (0.98, 1.00) 0.2 9 Physiologic injury indicators in the first 24 hours Ratio between measured sCr (first 24 hours) and Cr MDRD (per 1 unit change=100% increase) 2.24 (1.05, 4.78) 0.03 Lowest temperature (per degree of Celsius) 0.77 (0.66, 0.89) <0.001 Highest lactate (per mmol/l) 1.08 (1.01, 1.15) 0.02 Lowest hematocrit (per percent change) 0.97 (0.93, 1.01) 0.10 RBC transfusion (per unit lo g transformed) 1.60 (1.12, 2.27) 0.01 Platelets transfusion (Reference none) 0.59 (0.36, 0.96) 0.03 Cryoprecipitate transfusi on (Reference none) 1.80 (1.13, 2.86) 0.01 Highest heart rate (per beats/min) 1.00 (0.99, 1.01) 0.71 Lowest mean arterial pressure (per mmHg) 1.00 (0.99, 1.01) 0.87 Highest serum sodium (per mEq/l) 1.01 (0.99, 1.06) 0.42 INR (per unit) 0.94 (0.76, 1.15 ) 0.49 Worst base deficit (per mEq/l) 0.97 (0.92, 1.00) 0.24 PO 2 /FiO 2 ratio<200 (per unit change) 1.00 (1.00, 1.01) 0.39 Highest blood glucose (per mg/dl) 1.00 (1.00, 1.01) 0.64 Use of insulin (Reference none) 0.95 (0.64, 1.40) 0.72 FFP transfusion ( Reference none) 0.93 (0.55, 1.59) 0.92 Crystalloids infusion (per liter) 1.00 (1.00, 1.01) 0.07 Colloids use (Reference none) 1.19 (0.78, 1.79) 0.41 Use of inotrope (Reference none) 1.25 (0.59, 2.67) 0.72 Use of vasopressors (Reference none) 1.40 (0. 89, 2.20) 0.12 Age, sex and race were excluded from the model as they are already included in the calculation of the ratio of the measured sCr in the first 24 hours and CrMDRD. Components of Apache II (no renal data) score were included in the model as t he score itself was not an important predictor in the model. The odds ratios were calculated with logistic regression analysis (Methods). Abbreviations: CI confidence interval; OR Odds ratio; CrMDRD, estimated serum creatinine for patient age, gender and race (methods); RBC, red blood cells, INR, international normalized ratio; FFP, fresh frozen plasma.

PAGE 39

39 Table 36. Association between acute kidney injury and hospital mortality. Patient Characteristics Hospital mortality C=0.89, P<0.001 Hospital mortality C=0.89, P<0.001 OR (95% CI) P OR (95% CI) P Acute kidney injury (Reference no AKI) 3.05 (1.73, 5.40) <0.001 RIFLE Risk (Reference no AKI) 2.57 (1.19, 5.50) 0.001 RIFLE Injury (Reference no AKI) 2.67 (1.23, 5.83) RIFLE Failure (Reference no AKI ) 4.55 (2.00, 10.36) Host factors Obesity (per kg/m 2) ) 0.97 (0.93, 1.01) 0.18 0.97 (0.93, 1.01) 0.18 Hypertension (Reference none) 1.22 (0.58, 2.55) 0.61 1.19 (0.57, 2.50) 0.65 Heart disease (Reference none) 1.46 (0.49, 4.36) 0.49 1.42 (0.48, 4. 22) 0.53 Diabetes mellitus (Reference none) 0.62 (0.20, 1.89) 0.41 0.55 (0.18, 1.71) 0.29 Smoking (Reference none) 0.52 (0.27, 0.99) 0.05 0.51 (0.26, 0.99) 0.05 Use of antiplatelets/ anticoagulation drugs (Reference none) 1.48 (0.55, 3.97) 0.45 1.48 (0. 55, 3.97) 0.43 Physiologic injury indicators in the first 24 hours Highest lactate (per mmol/l) 1.05 (0.96, 1.14) 0.29 1.05 (0.96, 1.14) 0.29 Worst base deficit (per mEq/l) 1.01 (0.95, 1.08) 0.78 1.01 (0.95, 1.08) 0.70 PO 2 /FiO 2 ratio<200 (per uni t change) 1.00 (1.00, 1.01) 0.45 1.00 (1.00, 1.01) 0.36 Highest blood glucose (per mg/dl) 1.00 (1.00, 1.01) 0.25 1.00 (1.00, 1.01) 0.27 RBC transfusion (per unit) 1.01 (0.98, 1.05) 0.46 1.01 (0.98, 1.05) 0.54 Platelets transfusion (Reference none) 0.55 (0.28, 1.08) 0.08 0.56 (0.28, 1.10) 0.09 Cryoprecipitate transfusion (Reference none) 1.21 (0.63, 2.33) 0.56 1.22 (0.64, 2.35) 0.55 FFP transfusion (per unit) 1.00 (1.00, 1.04) 0.97 1.00 (0.96, 1.04) 0.96 Crystalloids infusion (per liter) 1.00 (1.00, 1 .00) 0.23 1.00 (1.00, 1.00) 0.29 Colloids use (Reference none) 0.90 (0.49, 1.66) 0.72 0.91 (0.50, 1.67) 0.76 Use of vasopressors (Reference none) 2.84 (1.60, 5.06) <0.001 2.75 (1.54, 4.92) <0.00 1 Severity Scores Injury severity score (admission) (per unit change) 1.02 (1.00, 1.05) 0.02 1.03 (1.01, 1.05) 0.02 APACHE II (without renal component, first 24 hours)(per unit) 1.06 (1.00, 1.12) 0.05 1.07 (1.01, 1.13) 0.03 Maximum MOD score (without renal component, first 28 days) (per unit) 1.50 (1.32, 1.71) <0.01 1.48 (1.29, 1.69) <0.00 1 Age, sex and race were excluded from the model as they are already included in the calculation of CrMDRD and RIFLE categories. Age is also included in the calculation of APACHE score. APACHE II and MOD scores were calc ulated without renal components to avoid correlation with AKI. The odds ratios were calculated with logistic regression analysis. Abbreviations: CI confidence interval; OR Odds ratio; RBC, red blood cells; FFP, fresh frozen plasma; MOD, multiple organ dysfunction.

PAGE 40

40 Table 3 7 Prevalence of nosocomial infections stratified by the occurrence of RIFLE AKI All patients (n=1739) AKI (n=712) No AKI (n=1027) P OR (95% CI) Nosocomial infections (all), n (%) 840 (49) 156 (62) 325 (45) <0.001 1.57 (1.29, 1 .9) Pneumonia, n (%) 553 (32) 267 (38) 286 (28) <0.001 1.55 (1.27, 1.91) Bloodstream infections, n (%) 236 (15) 133 (19) 103 (10) <0.001 2.06 (1.56, 2.72) Catheter related infections, n (%) 49 (3) 28 (4) 21 (2) 0.02 1.96 (1.1, 3.48) Urinary tract in fection (UTI), n (%) 286 (17) 136 (19) 150 (15) 0.01 1.38 (1.07, 1.78) Meningitis, n (%) 12 (1) 3 (0) 9 (1) 0.45 1.74 (0.41, 7.32) Pseudomembranous colitis, n (%) 43 (2) 17 (2) 26 (3) 0.85 0.94 (0.51, 1.75) Comparing patients with AKI to patients w ithout AKI. 213 patients did not have data on daily sCr so AKI Rifle was not calculated. Abbreviations: CI confidence interval; OR Odds ratio

PAGE 41

41 Table 38. Characteristics of nosocomial pneumonias and bloodstream infections for patients stratified by RIF LEmax class. No AKI N= 286 1 RIFLE AKI P P All AKI patients N=2 67 RIFLE R N=106 RIFLE I N=79 RIFLE F N=68 Pneumonia onset (days since injury) (median, 25th 75th) 6 (4, 9) 7 (4, 10) 6 (4, 11) 7 (4, 9) 7 (5, 10) 0.21 0.65 Pneumonia onset (days since AKI onset) (median, 25th 75th) 5 (2, 8) 5 (3, 9) 4 (1, 7) 5 (2, 7) 0.25 Bloodstream infection onset (days since injury) (median, 25th 75th ) 7 (5, 13) 6 (4, 13) 8 (6, 10) 8 (5, 14) 0.32 Bloodstream infection onset (days since AKI onset) (me dian, 25th 75th ) 7 (4, 12) 6 (3, 11) 5 (2, 12) 6 (3, 8) 6 (3, 11) 0.16 0.43 Causative Microorganism for Pneumonia Staphylococcus aureus No. (%) 76 (27) 83 (31) 53 (35) 23 (40) 7 (12) 0.24 0.002 Pseudomonas aeruginosa No. (%) 38 (13) 39 (15) 17 (11) 7 (12) 15 (26) 0.65 0.04 Acinetobacter baumannii No. (%) 24 (8) 49 (18) 21 (14) 11 (19) 17 (30) <0.001 <0.001 Escherichia coli No. (%) 13 (5) 22 (8) 17 (11) 2 (4) 3 (5) 0.07 0.04 Serratia marcescens No (%) 16 (6) 8 (3) 2 (1) 1 (2) 5 (9) 0.13 0.04 Causative Microorganism for Pneumonia Staphylococcus aureus No. (%) 24 (23) 23 (17) 12 (23) 7 (19) 4 (9) 0.25 0.26 Coagulase negative Staphylococcus No. (%) 21 (20) 22 (17) 12 (23) 4 (11) 6 (14) 0.45 0.41 Acinetobacter baumannii No. (%) 8 (8) 25 (19) 9 (17) 6 (16) 10 (23) 0.01 0.07 Escherichia coli No. (%) 1 (1) 11 (8) 5 (9) 2 (5) 4 (9) 0.01 0.06 Serratia marcescens No (%) 4 (4) 9 (7) 3 (6) 4 (11) 2 (5) 0.33 0.46 Gram negative organisms (NOS), No. (%) 1 (1) 8 (6) 2 (4) 3 (8) 3 (7) 0.04 0.15 Candida No. (%) 3 (3) 13 (10) 5 (9) 2 (5) 6 (14) 0.03 0.08 All percentages are calculated as proportions of patients with pneumonia episodes only, not the all cohort. Data shown only for microorganisms with different distribution by AKI RI FLE groups.

PAGE 42

42 Table 39. S upervised genomic analysis between trauma patients and control group ( uninjured subjects ) Diagonal linear discriminant analysis 1 nearest neighbor 3 nearest neighbors Nearest centroid Mean percent of correct classification 98 1 00 99 98 Positive predictive value healthy control (n=24) 0.89 1.00 0.96 0.89 Positive predictive value trauma (n=158) 1.00 1.00 1.00 1.00 Analysis incorporated a two class comparison (Students t test, P<0.001) and LOOCV (1000 permutations) 30,956 sig nificant probe sets. Table 310. Class comparison and prediction between patients with no AKI and p atients with AKI. Identified were 230 probe sets with a 68% to 77% correct classification rate Diagonal linear discriminant analysis 1 nearest neighbor 3 nearest neighbors Nearest centroid Mean percent of correct classification 72 71 77 68 Positive predictive value no AKI (n=125) 0.85 0.80 0.81 0.84 Positive predictive value AKI (n=33) 0.36 0.24 0.39 0.32

PAGE 43

43 Table 311. Multi class comparison and predict ion between patients with no AKI vs. patients with AKI, all stages ( RIFLE R, I, and F ) Diagonal linear discriminant analysis 1 nearest neighbor 3 nearest neighbors Nearest centroid Mean percent of correct classification 59 68 77 51 Positive predictive v alue no AKI (n=125) 0.83 0.82 0.78 0.84 Positive predictive value RIFLE R (n=13) 0.05 0.15 0.00 0.05 Positive predictive value RIFLE I (n=9) 0.00 0.00 0.00 0.00 Positive predictive value RIFLE F (n=11) 0.15 0.07 0.00 0.06 F test p<0.001 158 significa nt probe sets. LOOCV, 1000 permutations Table 312. Class comparison and prediction between patien ts with no AKI vs. patients with RIFLE F AKI only Diagonal linear discriminant analysis 1 nearest neighbor 3 nearest neighbors Nearest centroid Mean perc ent of correct classification 85 91 91 81 Positive predictive value no AKI (n=125) 0.93 0.92 0.92 0.94 Positive predictive value RIFLE F (n=11) 0.20 0.00 0.00 0.17 F test p<0.001. 95 significant probe sets. LOOCV, 1000 permutations.

PAGE 44

44 Figure 31. Probability curves for continuous variables associated with the occurrence of acute kidney injury (AKI). A) Ratio between measured serum creatinine in the first 24 hours and Crmord. B) Red blood cells transfusion in the first 24 hours (log transformati on of number of units). C) Lowest temperature (in C in the first 24 hours. D) Worst plasma lactate level (mmol/l) in the first 24 hours A B C D

PAGE 45

45 0% 10% 20% 30% 40% 50% 60% 70% 80% No AKI Rifle R Rifle I Rifle F Nosocomial Infections, all Nosocomial Pneumonia Bloodstream Infections CLBSI Figure 32. Most common nosocomial infections (NCI) stratified by severity stages of RI FLE AKI. P <0.001 when comparing each group for all NCIs, nosocomial pneumonia and bloodstream infections. P=0.009 when comparing groups for central linerelated bloodstream infections (CLBSI).

PAGE 46

46 Figure 33 Correspondence analysis of Rifle AKI stages and major types of nosocomial infections, nosocomial pneumonia and bloodstream infections. The columns represent AKI severity stages while the rows represent type of NCI.

PAGE 47

47 Figure 34 Correspondence analysis of Rifle AKI stages and major pathogens for A) nosocomial pneumonia and B) bloodstream infections. The columns represent AKI severity stages while the rows represent different types of pathogens. A B

PAGE 48

4 8 Figure 35. SAM plot showing 31,165 significant probe sets. Class comparison, SAMTM, FDR>0.001, n 158 trauma vs. n=24 control microarrays.

PAGE 49

49 CHAPTER 4 DISCUSSION In a multi center prospective cohort study of trauma patients with no previous kidney disease, AKI was a common complication associated with an independent risk of hospital death. Close to one fourth of all trauma patients developed AKI and two thirds of them had only mild to moderately severe AKI. Even patients with the least severe AKI had a 2.5fold increase in adjusted risk for death compared to patients without AKI. This risk was indep endent of other indicators of anatomic and physiologic injury severity, and other organ dysfunction. Although the likelihood of developing AKI was associated with changes in several clinical parameters in the first 24 hours after trauma, no such associatio n was observed with the initial genomic characteristics of patients with AKI. Although the adverse effect of all stages of severity of AKI on inhospital mortality is increasingly recognized,9 few studies have exami ned the impact of less severe AKI among trauma patients.3 5 7 8 2426 The American College of Surgeons Committee on Trauma defines acute renal failure after trauma as sCr above 3.5 mg/dl, but only 15% of the AKI patients in our study had a sCr greater than 3 mg/dl. T he average highest sCr among patients with mild AKI was 1.33 mg/dl, a value that often doesnt attract clinical attention in a population of younger trauma patients.6 At the same time, none of the patients with mild and moderate AKI had a renal Marshall score greater or equal to 4, and cons equently would not be classified as having kidney dysfunction by using this traditional scoring system.22 Hence the importance of small changes in sCr early after trauma, indicative of less severe AKI, is likely to be under appreciated and adequate follow up for these patients may not occur in a timely manner. On the other hand, early diagnosis of AKI may help to identify highrisk patients who would benefit from

PAGE 50

50 implementation of goal directed resuscitation therapy, a strategy proven to work for patients with severe sepsis.27 Our findings reiterate the importance of relative changes in sCr rather than absolute sCr value in det ermining the risk for the development of AKI because no single sCr value corresponds to a given GFR across all patients. Instead, the change in sCr compared to baseline determines risk for AKI.11 To estimate bas eline renal function when sCr is unknown, RIFLE advocates use of estimated CrMDRD for the GFR of 75 ml/min per 1.73, a lower end of the normal range for healthy individuals.28 This relative increase in first measured sCr compared to estimated CrMDRD was an independent risk factor for the development of AKI in a multivariate model where every 10% increase in measured sCr increases the ris k for AKI by 8%. Hence in clinical practice an assessment of the relative increase between the first measured sCr and estimated CrMDRD may offer a valuable prognostic tool for the risk of AKI. We demonstrated that less severe AKI was not only common amon g trauma patients but also associated with adverse clinical outcomes. Even mild AKI was associated with a 2.5fold increase in the risk of dying as calculated in a multivariate regression model that included baseline clinical characteristics, indicators of anatomic and physiologic injury severity in the first 24 hours, the severity of other organ dysfunction and details of early resuscitation. All of these parameters carry clinical significance and the results are easily generalized for physicians caring for trauma patients. Although many of the covariates have been previously shown to be associated with in hospital mortality2931 the association of each stage of AKI with increased inhospital mortality is demonstrated for the first time in this study. Future studies should

PAGE 51

51 focus on identifying patients at the highest risk for death even earlier, preferably within the first 12 to 24 hours after trauma. The failure to identify any significant differences in blood leukocyte expression between patients with and without AKI is not surprising. Compared to healthy controls, trauma induced a genomewide reprioritization of leukocyte gene expression, with greater than 60% of t he leukocyte transcriptome changing within the first 12 hours. These changes were comparable to or greater than the changes seen after administration of microbial products to healthy volunteers.32 We found only modest differences in genome wide expression in the trauma patients who experienced AKI, regardless of the RIFLE score. This lack of difference may reflect the overall genomic response to the severity of the traumatic injury, rather than as a predictor for the differenti al physiological responses leading to AKI. Acute kidney injury should no longer be viewed as an indicator of overall severity of illness but instead the injured kidney can exhibit independent effects on other organs, including the lungs.3335 The kidneys receive a higher blood flow per unit mass than other organs but the fraction of extracted oxygen is low due to the presence of abundant diffusional arterial to ven ous (AV) shunting in the cortex and medulla.36 This AV shunting accounts for up to 50% of the oxygen removal from arterial blood before the glomerular capillaries, rendering the kidney very sensitive to conditions of hypoperfusion.36 During experimental hemorrhagic shock, the oxygenation status in kidneys may be impaired in spite of the preserved blood flow.37 At the same time hypoxia may limit renal use of lactate for anaerobic metabolism.38 The oxygen supply to the renal tissue may become impaired even earlier during acute normovolemic

PAGE 52

52 hemodilution, such as during resuscitation wit h large quantities of crystalloid solutions after trauma.39 Once initiated, A KI incites a cascade of inflammatory processes both locally and systemically.40 Because of this susceptibility for hypoxia and early development of kidney injury, the use of more sensitive marker s of kidney injury after trauma may be a useful tool for identifying patients at risk for AKI as well as other organ dysfunctions and dying. Our study has several limitations. This is a retrospective analysis of prospectively collected data from which caus al inference cannot be derived and which is subject to bias from unmeasured factors. Although we attempted to control for selection bias with multivariate statistical method and risk adjustment, we could not completely eliminate the potential for residual confounding. Second, we did not have access to information related to survival after hospital discharge; therefore, our risk estimates for discharged patients might represent the lower limit of the true risk. Finally, our analysis addressed the relationshi p between all cause sCr increase and subsequent adverse events. The etiology of AKI after trauma is usually multifactorial and the strong association of even mild AKI with hospital mortality was demonstrated regardless of the etiology of AKI.9 No conclusive data exist to demonstrate that two traditional etiological categories of AKI, prerenal azotemia and acute tubular necrosis, have meaningful prognostic differences, or, despite common clinical practice, differing r ates of response to therapy.41 A recent report from the Acute Kidney Injury Network has favored the concept of volumeresponsive AKI, and they have suggested a research agenda to address the clinical significance of this type of AKI in future studies.42

PAGE 53

53 In conclusion, in a large multi center prospective cohort of trauma patients with no previous kidney disease, AKI was associated with an independent risk of hospital death. This risk was evident in a dose response manner a nd even patients with the mild AKI had a 2.5fold increase in adjusted risk for death compared to patients without AKI. Future studies need to address whether early identification of patients with less severe AKI will provide a window for therapeutic inter ventions that may reverse adverse clinical outcomes.

PAGE 54

54 LIST OF REFERENCES 1. Ostermann M, Chang RW. Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med 2007;35:183743; quiz 52. 2. Hoste EA, Schurgers M Epi demiology of acute kidney injury: How big is the problem? Crit Care Med 2008;36:S146S51. 3. Regel G, Lobenhoffer P, Grotz M, Pape HC, Lehmann U, Tscherne H. Treatment results of patients with multiple trauma: an analysis of 3406 cases treated between 1972 and 1991 at a German Level I Trauma Center. J Trauma 1995;38:708. 4. Morris JA, Jr., Mucha P, Jr., Ross SE, et al. Acute posttraumatic renal failure: a multicenter perspective. J Trauma 1991;31:158490. 5. Gettings LG, Reynolds HN, Scalea T. Outcome in post traumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Intensive Care Med 1999;25:80513. 6. Trauma ACoS Co. Resources for the Optimal Care of Injured Patient:1999. Chicago: American College o f Surgeons; 1998. 7. Brandt MM, Falvo AJ, Rubinfeld IS, Blyden D, Durrani NK, Horst HM. Renal Dysfunction in Trauma: Even a Little Costs a Lot. J Trauma 2007;62:13624. 8. Vivino G, Antonelli M, Moro ML, et al. Risk factors for acute renal failure in trauma patients. In tensive Care Med 1998;24:80814. 9. Kellum JA. Acute kidney injury. Crit Care Med 2008;36:S141S5. 10. Bagshaw SM, George C, Dinu I, Bellomo R. A multi centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrol Dial Transplant 2008;23:120310. 11. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute renal failure definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8:R20412. Epub 2004 May 24. 12. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: A systematic review. Kidney Int 2008;73:53846. 13. Ali T, Khan I, Simpso n W, et al. Incidence and Outcomes in Acute Kidney Injury: A Comprehensive PopulationBased Study. Journal of the American Society of Nephrology 2007;18:12928.

PAGE 55

55 14. Bagshaw SM, George C, Dinu I, Bellomo R. A multi centre evaluation of the RIFLE criteria fo r early acute kidney injury in critically ill patients. In; 2007. 15. Hoste EA, Clermont G, Kersten A, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care 2006;10:R73. 16. Dasta JF, KaneGill SL, Durtschi AJ, Pathak DS, Kellum JA. Costs and outcomes of acute kidney injury (AKI) following cardiac surgery. Nephrol Dial Transplant 2008;23:19704. 17. Bihorac A, Yavas S, Subbiah S, et al. Long term risk of mortality and ac ute kidney injury during hospitalization after major surgery. Ann Surg 2009;249:8518. 18. Nathens AB, Johnson JL, Minei JP, et al. Inflammation and the Host Response to Injury, a largescale collaborative project: Patient Oriented Research Core-standard operating procedures for clinical care. I. Guidelines for mechanical ventilation of the trauma patient. J Trauma 2005;59:7649. 19. Brandt CA, Deshpande AM, Lu C, et al. TrialDB: A webbased Clinical Study Data Management System. AMIA Annu Symp Proc 2003:7 94. 20. Baker SP, O'Neill B, Haddon W, Jr., Long WB. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974;14:18796. 21. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guid eline for Prevention of Surgical Site Infection, 1999. Centers for Disease Control and Prevention (CDC) Hospital Infection Control Practices Advisory Committee. Am J Infect Control 1999;27:97132; quiz 34; discussion 96. 22. Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995;23:163852. 23. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:81829. 24. Plurad D, Brown C, Chan L, Demetriades D, Rhee P. Emergency department hypotension is not an independent risk factor for post traumatic acute renal dysfunction. J Trauma 2006;61:11207; discussion 7 8. 25. Ala Kokko T, Ohtonen P, Laurila J, Martikainen M, Kaukoranta P. Development of renal failure during the initial 24 h of intensive care unit stay correlates with hospital mortality in trauma patients. Acta Anaesthesiol Scand 2006;50:82832.

PAGE 56

56 26. Nadvi SS, Mokoena T, Gouws E, Haffejee AA. Prognosis in posttraumatic acute renal failure is adversely influenced by hypotension and hyperkalaemia. Eur J Surg 1996;162:1214. 27. Rivers E, Nguyen B, Havstad S, et al. Early goal directed therapy in the treatment of severe seps is and septic shock. N Engl J Med 2001;345:136877. 28. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1266. 29. Durham RMMD, Moran JJR, Mazuski JEMD, Shapiro MJMD, Baue AEMD, Flint LMMD. Multiple Organ Failure in Trauma Patients. Journal of TraumaInjury Infection & Critical Care 2003;55:60816. 30. Sauaia AMDP, Moore FAMD, Moore EEMD, Norris JMP, Lezotte DCP, Hamman RFMDD. Multiple Organ Failure Can Be Predicted as Early as 12 Hours after Injury. Journal of TraumaInjury Infection & Critical Care 1998;45:291303. 31. Ciesla DJ, Moore EE, Johnson JL, Burch JM, Cothren CC, Sauaia A. Obesity Increases Risk of Organ Failure after Severe Trauma. Journal of the American College of Surgeons 2006;203:53945. 32. Calvano SE, Xiao W, Richards DR, et al. A network based analysis of systemic inflammation in humans. Nature 2005;437:10327. 33. Feltes CM, Van Eyk J, Rabb H. Distant organ changes after acute kidney injury. Nephron Physiol 2008;109:p804. 34. Hassoun HT, Grigoryev DN, Lie ML, et al. Ischemic acute kidney injury induces a distant organ functional and genomic response distinguishable from bilateral nephrectomy. Am J Physiol Renal Physiol 2007;293:F3040. 35. Druml W. Long term prognosis of patients with acute Renal Failure: Is intensive care worth it? Intensive C are M edicine 2005;31:11457. 36. Evans RG, Gardiner BS, Smith DW, O'Connor PM. Intrarenal oxygenation: unique challenges and the biophysical basis of homeostasis. Am J Physiol Renal Physiol 2008;295:F125970. 37. Torres LN, Pittman RN, Torres Filho IP. Microvascular blood flow and oxygenation during hemorrhagic hypotension. Microvasc Res 2008;75:21726. 38. Nelimarkka O. Renal oxygen and lactate metabolism in hemorrhagic shock. An experimental study. Acta Chir Scand Suppl 1984;518:144. 39. Johannes T, Mik EG, Nohe B, Unertl KE, Ince C. Acute decrease in renal microvascular PO2 during acute normovolemic hemodilution. Am J Physiol Renal Physiol 2007;292:F796 803.

PAGE 57

57 40. Bonventre JV. Pathophysiology of acute kidney injury: roles of potential inhibitors of inflammation. Contrib Nephrol 2007;156:39 46. 41. Kellum JAMD. Prerenal azotemia: Still a useful concept? Crit Care Med 2007;35:16301. 42. Mehta R, Kellum J, Shah S, et al. Acute Kidney Injury Network (AKIN): report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31.

PAGE 58

58 BIOGRAPHICAL SKETCH Azra Bihorac is an assistant professor in anesthesiology surgery and medicine in the University of Floridas College of Medicine, Department of Anesthesiology. After completing medical school in 1990 she spent several years in Turke y where she trained in internal medicine and nephrology She moved to the U nited S tates in 1999 to continue her research training in nephrology and then completed i nternal medicine, nephrology and critical care medicine training at the University of Florid a. She has a very active research group with a primary interest in clinical and translational research in acute kidney injury, including sepsis and shock. She was awarded an NIH K23 award in 2010, and in 2011 completed her Master of Science with a concentr ation in clinical and translational science while a scholar in the UF CTSIs Advanced Postgraduate Program in Clinical Investigation.