Magnetic resonance imaging studies of fetal spinal cord transplants / by Edward D. Wirth, III


Material Information

Magnetic resonance imaging studies of fetal spinal cord transplants / by Edward D. Wirth, III
Physical Description:
x, 151 leaves : ill. ; 29 cm.
Wirth, Edward D., 1964-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Spinal Cord -- transplantation   ( mesh )
Fetal Tissue Transplantation   ( mesh )
Cats   ( mesh )
Magnetic Resonance Imaging   ( mesh )
Spinal Cord Injuries -- diagnosis   ( mesh )
Department of Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Neuroscience -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1992.
Bibliography: leaves 139-150.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001751950
oclc - 28488061
notis - AJG4889
System ID:

This item is only available as the following downloads:

Full Text







Digitized by the Internet Archive
in 2012 wif ur Mrfirom
University of Florida, George A. Smathers Libraries with support from LYRASIS and the Sloan Foundation


This undertaking was a collaborative project in the

truest sense--it could not have been completed without the

support and input of many people. Foremost were the expertise

and astute guidance bestowed by my mentor, Dr. Paul Reier, and

by Dr. Thomas Mareci. I am particularly grateful to Dr. Reier

for the opportunity to participate in the fascinating field of

neural injury and transplantation, and to Dr. Mareci for the

chance to pursue my passion for physics through the wonderful

technology of nuclear magnetic resonance. I am also extremely

thankful to Dr. Dan Theele, who unselfishly dedicated many

hours of his time to this project. Additional appreciation is

extended to each member of my supervisory committee--Drs.

Floyd Thompson, Jeff Fitzsimmons, Bruce Hunter, and Roger

Reep, for their constructive criticisms and helpful

suggestions regarding my work. I would also like to thank Dr.

Douglas Anderson for contributing cats with clinically

relevant spinal cord injuries, so that the studies described

herein may be more directly applicable to magnetic resonance

imaging of human subjects.

Invaluable assistance in tackling many of the technical

problems encountered during the course of this project was

provided by Barbara Beck, Bill Brey, Randy Duensing and Dr.


Chris Mladinich. I am also thankful for the fine technical

and organizational support of Barbara O'Steen, Minnie Smith,

Regina Reier, and Teresa Lyles. A considerable amount of

computer programming support was provided by David Brown and

Dan Williams.

I am particularly thankful for the emotional support and

patience of my wife, Lori, who endured many days and nights

without my company. She provided encouragement when I needed

it most, and also helped with editorial suggestions and

technical support. I would also like to acknowledge the

support of my fellow grad students, especially Greg

Schrimsher, who provided superb technical assistance and

editorial suggestions.

Financial support for this research was provided by NIH

grant 27511 to Dr. Reier, the Mark F. Overstreet Fund for

Spinal Cord Regeneration Research, NIH grant NS-29362 to Dr.

Mareci, the University of Florida NMR Resource (NIH grant P41

RR-02278), American Paralysis Association grant TC8704 to Dr.

Mareci, the Veteran's Administration, the University of

Florida Division of Sponsored Research Graduate Research

Assistantship Program, and the University of Florida Center

for Neurobiological Sciences (NIMH grant MH15737). Additional

travel support was provided by the American Paralysis




ACKNOWLEDGMENTS......................................... iii

LIST OF ABBREVIATIONS....... ........................... vii

ABSTRACT................................................. viii


1 INTRODUCTION AND BACKGROUND ........................ 1

Magnetic Resonance Imaging of Spinal Cord Injury... 1
Fetal Neural Transplants and Spinal Cord Repair.... 6
In Vivo MRI of Fetal Neural Transplants............. 7
Experimental Goals................................... 9
Notes .................................... ........... 11

2 GENERAL METHODS .................... ...... ........... 12

Experimental Animals................................ 12
Spinal Cord Injury and Transplantation Protocols... 13
Postoperative Care.........................................16
Magnetic Resonance Imaging Studies................. 17
Histological Follow-Up............................. 18
Notes .................................... ........... 19

ADULT CAT SPINAL CORD ......................... 20

Introduction. ............................. ......... 20
Materials and Methods............................... 21
Results ............................................. 24
Discussion.......................................... 44
Notes .................................... ........... 50

IMAGING........................................ 51

Introduction................................. ....... 51
Materials and Methods.............................. 53
Results ............................................. 56
Discussion ......................................... 85

CORD ............................................ 93

Introduction........ ... .............................. 93
Materials and Methods.............................. 96
Results .................................. ........... 112
Discussion.......................................... 121

6 SUMMARY AND CONCLUSIONS............................ 125

Predictive Value of Spin-Echo MR Images for
Determination of Graft Survival............... 125
NMR and Immunological Rejection ................... 127
Additional Roles for MRI in Neural
Transplantation.............................. 130
Strategies for Optimal NMR Imaging of the Spinal
Cord .......................................... 132
Future Directions.................................. 134
Conclusions........................................ 137

REFERENCES...... ......... ............................... 139

BIOGRAPHICAL SKETCH.................................... 151


BSt brainstem

CNS central nervous system

FSC fetal spinal cord

MRI magnetic resonance imaging

NCx neocortex

NMR nuclear magnetic resonance

Q quality factor

RF radiofrequency

SCI spinal cord injury

SNR signal-to-noise ratio

T longitudinal (spin-lattice) relaxation time

T2 transverse (spin-spin) relaxation time

TE echo time

TR repetition time


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



Edward D. Wirth, III

August, 1992

Chairman: Dr. Paul J. Reier
Major Department: Neuroscience

The application of magnetic resonance imaging (MRI) to

the study of spinal cord transplantation has not been

explored, but holds great promise for the observation of

neural grafts in living subjects. Accordingly, the following

series of experiments demonstrate that conventional MR imaging

methods can be used to visualize fetal transplants in the

injured cat spinal cord and that specialized MR imaging coils

offer the potential for observing more subtle anatomical

details of these fetal grafts.

Standard surface coils and spin-echo imaging methods were

used initially to determine whether MRI could illustrate the

presence of fetal CNS transplants in the adult cat spinal

cord. Following euthanasia, histological sections were

obtained from spinal levels that were in register with the MR

images. These specimens established that chronically


surviving graft tissue consistently exhibited a hyperintense

appearance relative to the host spinal cord on intermediate

T,-weighted images.

A controlled, longitudinal study was then performed to

determine what postoperative delay, if any, is required before

a fetal transplant can be reliably distinguished from

potentially overlapping pathology (e.g., edema) that could

yield comparable signals on T,- and T2-weighted images.

Although the transplants in graft recipients and injury sites

in control animals presented similar signals on T1-weighted

images, surviving grafts exhibited a significantly lower

signal intensity on T2-weighted images than injured-only

controls by 8 weeks postoperative.

One shortcoming observed in the above studies was that

anatomical details of the transplants (e.g., central canal-

like structures) were frequently obscured due to volume

averaging artifacts from insufficient spatial resolution.

This limitation was addressed by comparing the performance of

several optimized imaging coils. It was found that a large

increase in signal-to-noise and spatial resolution over

conventional surface coils could be achieved by mounting

implanted imaging coils directly onto the spine.

Collectively, these results indicate that MRI can be a

reliable diagnostic tool for monitoring intraspinal neural

transplants in living subjects. These findings also address

an important prerequisite (in situ verification of transplant

survival) of any future clinical trials.


Magnetic resonance imaging (MRI) has emerged over the

last few years as the definitive modality for diagnosing many

disorders of the central nervous system (CNS), including

trauma to the brain and spinal cord. The key advantage of MRI

is the ability to observe noninvasively many of the

pathophysiological manifestations of CNS injury and disease.

This offers the clinician or investigator the capacity for

studying pathological processes as part of a dynamic


Magnetic Resonance Imaging of Spinal Cord Injury

Acute Spinal Cord Injury

Although clinical spinal cord injury (SCI) has been

evaluated traditionally with plain film roentgenograms,

myelography and, more recently, contrast-enhanced computed

tomography (CT), MRI is the only imaging technique that allows

direct visualization of the spinal cord (Kalfas et al., 1988).

This advantage is of fundamental importance because the

primary concern in evaluating acutely spinal injured patients

is recognition of spinal cord damage that could result in a

permanent loss of motor and sensory function below the level

of injury (Masaryk, 1989).

Several studies have attempted to determine which types

of pathology resulting from spinal cord injuries in either

humans or experimental animals are demonstrable on T,- and T2-

weighted spin-echo MR images (Hackney et al., 1986; Kalfas et

al., 1988; Bondurant et al., 1990; Kulkarni et al., 1987;

Flanders et al., 1990). Typically, damage to the spinal cord

is visible as either disruption of extrinsic structures (e.g.,

vertebrae), which may impinge on the spinal cord, and/or

intramedullary lesions. Concerning extrinsic structures, MRI

is limited in its ability to define bony fractures accurately,

but has proven to be superior to CT and roentgenograms for

visualization of both acute disc herniation and spinal cord

deformity secondary to extrinsic compression (e.g., by bone

fragments; Kalfas et al., 1988).

MRI is unique, however, in its ability to demonstrate

either extrinsic or intrinsic soft tissue pathology. Examples

of gross and more subtle lesions that have been observed on MR

images include overt transaction, focal enlargement or

swelling of the spinal cord, hemorrhage and edema. The gross

morphological disturbances, such as transaction, deformity and

swelling, have been depicted better on T1-weighted images,

whereas T2-weighted images are generally more useful for

illustrating intrinsic pathology, such as hemorrhage and edema

(Hackney et al., 1986; Bondurant et al., 1990; Kulkarni et

al., 1987).


The distinction between hemorrhage and edema has

important functional consequences and, accordingly, has

received considerable attention. Studies of acute SCI in rats

or dogs following either compression with an aneurysm clip or

a weight-drop contusion injury have demonstrated that focal

intraparenchymal hemorrhage exhibits a very low signal

intensity on T2-weighted images and is isointense to slightly

hypointense relative to the undamaged spinal cord on T -

weighted images (Hackney et al., 1986; Chakeres et al., 1987).

Similar findings have been obtained in clinical human studies

(Bondurant et al., 1990; Kulkarni et al., 1987; Flanders et

al., 1990) in which low signal intensity on both Ti- and T.-

weighted images within the first several hours after injury

indicated hemorrhage. These investigations also have noted

that in the subacute-to-chronic phases (i.e., from

approximately 1 week [subacute] to greater than several months

[chronic] postinjury), hemorrhage may present an increased

signal on T1-weighted images and then become hyperintense on

both T, and T2 images. These three patterns of signal

intensity are attributable to the various physiological forms

of iron (i.e., deoxyhemoglobin and intracellular methemoglobin

within red blood cells, and extracellular methemoglobin that

is released following lysis of red blood cells; Gomori and

Grossman, 1987). In contrast, edema consistently presents

with high signal intensity on T2-weighted images (Clasen,


The emphasis on distinguishing hemorrhage and edema stems

from the fact that they represent irreversible versus

potentially reversible forms of tissue damage, respectively

(Hackney et al., 1986). Thus, in the clinical setting, MRI

can be instrumental in determining whether the neurological

loss following SCI will be complete and can help determine

what type of surgical intervention should be performed. The

severity of injury has great impact on functional outcome:

extensive injuries with intraparenchymal hemorrhage have a

poor prognosis; with moderate damage and incomplete

dysfunction, aggressive treatment (e.g., early decompression)

may be rewarding, and low-velocity injuries produce transitory

dysfunction which is self-limited and rapidly clears (i.e., a

spinal "concussion") (Masaryk, 1989). Accordingly, MRI may

eventually aid in monitoring the effects of various forms of

therapeutic intervention, both surgical and pharmacological.

One potential caveat (Kalfas et al., 1988), however, is that

a small percentage of patients present with normal MR images

but clinically complete SCI (i.e., no motor or sensory

function below the injury level).

Chronic Spinal Cord Injury

In the clinical setting, the first week following spinal

cord injury is generally considered to be the acute stage,

when most of the edema, vascular compromise, and metabolic

disruptions occur (Quencer, 1988). Following this period is

the subacute phase, which lasts from one to several weeks

postinjury and is generally not distinguished radiologically

from the chronic stage.

In chronically injured patients, it is important to

differentiate potentially treatable from untreatable forms of

ongoing pathology. In this regard, MRI has proven to be

superior to contrast-enhanced CT for defining intramedullary

and subarachnoid cysts, myelomalacia (scarring, neuronal

degeneration and microcyst formation), scarring/fibrosis, and

compression (Quencer, 1988). Thus, MR images can be used to

explain a clinically declining picture and provide a more

accurate determination of whether surgical intervention is

warranted (Quencer et al., 1986).

Several types of chronic pathology that may be corrected

or minimized through surgical procedures have been detected on

MR images. For example, MRI can distinguish an inoperable

myelomalacic cord from a well-defined, shuntable, cyst. MR

images can also aid in preoperative assessment for

decompression of spinal cord tissue and/or nerve roots,

removal of a herniated disc, and determination of whether an

anterior or posterior approach for spinal stabilization is

warranted (Quencer, 1988; Quencer et al., 1986).

Other clinical investigations of chronic SCI have

correlated signal intensities on MR images with prognoses of

functional recovery. The combination of high signal intensity

on T2-weighted images with low signal on T1 images has been

associated with little chance for recovery, especially if

chronic compression of the spinal cord is evident (Yamashita

et al., 1990; Takahashi et al., 1989). By comparison, normal

signal intensity on TI images or both T, and T2 images

corresponded to a much higher probability for functional


Fetal Neural Transplants and Spinal Cord Repair

Although damaged neuronal tissue in the CNS has

traditionally been considered irreplaceable, there have been

many recent advances in CNS grafting methods directed toward

the promotion of functional recovery (reviewed in Bjorklund et

al., 1987; Gage and Buzsaki, 1989; Reier et al., 1992a).

Whereas many of these investigations have focused on

intracerebral transplantation, others have begun to explore

the potential of fetal CNS tissue to foster functional

improvement in the injured spinal cord. Accordingly- several

studies have demonstrated the ability of embryonic spinal

cord, brainstem and neocortex to survive and establish axonal

connections with adjacent host tissue in both the acutely and

chronically injured rat spinal cord (Reier et al., 1986; Hould

and Reier, 1989; Nornes et al., 1983; Anderson et al., 1991;

Jakeman and Reier, 1991; Reier et al., 1992b); also reviewed

in Das, 1987). It has also been reported that fetal spinal

cord and brainstem grafts can contribute to some degree of

functional recovery in neonatal (Kunkel-Bagden and Bregman,

1990) and adult rats (Stokes and Reier, 1990; Buchanan and

Nornes, 1986). In addition, recent studies have shown the


feasibility of implanting regions of the embryonic neuraxis

into acutely and chronically injured spinal cords of adult

cats (Anderson et al., 1991; Reier et al., 1992b).

One fundamental limitation pertaining to most studies

involving fetal CNS or paraneural (e.g., adrenal medullary

tissue) grafts to the brain or spinal cord is that

illustration of the dynamic features of fetal grafts (e.g.,

growth, survival, myelination, and host-graft approximation)

are usually limited to inferences derived from longitudinal

postmortem analyses. This issue becomes increasingly more

restrictive as transplantation studies are extended to higher

mammals (e.g., cat or subhuman primate) since it is not always

practical or desirable to use a large number of such animals.

In addition, the long-range desire to extend CNS grafting

methods to humans (Lopez-Lozano et al., 1989; Marx, 1990;

Lindvall et al., 1990; Allen et al., 1989; Jiao et al., 1989),

for whom noninvasive evaluations of the growth, integration,

survival and/or potential rejection of a neural graft are

equally essential, further underscores the need to adopt

alternative ways for obtaining such information from living


In Vivo MRI of Fetal Neural Transplants

Accordingly, a relatively benign approach, such as

magnetic resonance imaging offers several potential

advantages. Foremost is the capacity for repeated in vivo

evaluation of transplants with minimal risk to the subject.

In addition, MRI is the imaging modality of choice for

diagnosing the variety of pathologies, including posttraumatic

sequelae, that affect the spinal cord (Hackney et al., 1986;

Hagenau et al., 1987; Miller et al., 1987; Kalfas et al.,

1988; Quencer, 1988; Modic et al., 1989). Considerations that

underscore the advantages of MRI for studying the spinal cord

include a capacity for multiplanar imaging with high spatial

resolution and excellent soft tissue discrimination (e.g.,

between gray and white matter).

Recently, MRI has been used to evaluate fetal CNS

transplants in the kainic acid-lesioned striatum or thalamus

of the adult rat (Peschanski et al., 1988; Norman et al.,

1989; Norman et al., 1990). Some general MRI descriptions of

the survival of intracerebral CNS grafts in cats (Villablanca

et al., 1990) and primates (Miletich et al., 1988) have also

been published recently. In contrast to spinal cord trauma,

however, there is no consensus in the literature regarding the

signal intensities exhibited by fetal CNS transplants

(Miletich et al., 1988; Peschanski et al., 1988; Smith et al.,

1988; Norman et al., 1988; Villablanca et al., 1990; Wirth et

al., 1989; Norman et al., 1989; Norman et al., 1990).

Although Norman et al. (1990) reported fetal tissue to be

roughly isointense relative to the host brain on both T1- and

T2-weighted images, other investigators have described

transplants that were markedly hypointense on T,-weighted

images (Peschanski et al., 1988). Peschanski and coworkers


propose several possible mechanisms that could account for the

observed reduction in the T, values of the transplant: (1)

increased presence of free radicals, (2) elevation of blood-

related ferric ions, and/or (3) increased vascularization.

However, as the authors note, all of these explanations were

ruled out by careful histological examination. An alternative

interpretation is that the T2 reduction was due to a decrease

in the overall mobility of part of the water associated with

macromolecules (Wehrli et al., 1984).

Experimental Goals

The application of MRI to studies of fetal neural tissue

grafts is still in its infancy. Thus, it is unclear to what

degree imaging protocols used to evaluate one region of the

CNS would optimally apply to another. For example,

differences in the volume of host tissue alone, as well as the

nature of the associated pathology, present challenges that

are unique to experiments involving the spinal cord. In

addition, MRI has yet to be used in investigations of

intraspinal grafts in any animal model.

Therefore, the present series of studies examined whether

viable fetal cat neural tissue transplanted into the injured

adult cat spinal cord could be visualized in vivo by high-

field MR imaging (Chapter 3). It should be noted that

although considerable success has been achieved in

transplanting fetal rat CNS tissue, several logistical factors

favored the cat as the animal model for these MRI


investigations. Specifically, these considerations included:

the larger size of the cat spinal cord relative to the rat,

the more limited availability of fetal cat tissue from timed-

pregnant donors, and reduced graft-host histocompatibility

between cats relative to the inbred rat strains that are

generally used for fetal CNS grafting. In addition,

transplantation of fetal CNS tissue in the cat represents an

initial step in the extension of these neural grafting

approaches toward higher mammalian species and, ultimately,


Another series of experiments was then conducted to

assess the predictive value of MRI for differentiating fetal

grafts from both subacute and chronic spinal cord pathology on

a dynamic basis through repeated imaging of individual

subjects (Chapter 4). In addition to standard subjective

evaluation of signal intensities, quantitative approaches

were explored to provide observer-independent indices of graft

viability and to obtain estimates of the T, and T2 relaxation

times for transplanted embryonic tissue (Chapter 4). Lastly,

another series of experiments compared optimized

radiofrequency coil geometries for obtaining MR images of the

spinal cord with much higher spatial resolution than is

currently available (Chapter 5). In summary, the collective

goal of this work is to explore and develop the potential of

MRI as a reliable diagnostic tool for monitoring intraspinal

fetal transplants in living subjects. Such approaches have


immediate relevance to the experimentalist and may ultimately

be directly translated to the clinical setting. Thus, it is

anticipated that MRI will become an integral component of

long-term behavioral and electrophysiological studies aimed at

assessing recovery of function following CNS grafting.


Portions of this chapter are reprinted with permission
from Wirth, E.D.,III, D.P. Theele, T.H. Mareci, D.K. Anderson,
S.A. Brown, and P.J. Reier (1992) In vivo magnetic resonance
imaging of fetal cat neural tissue transplants in the adult
cat spinal cord. J.Neurosurq. 76:261-274. Copyright 1992 by
Journal of Neurosurgery.


Experimental Animals

Adult, female, random-source cats were used for all of

these studies. The cats that had received hemisection lesions

were housed individually in the University of Florida Health

Center Animal Resources Department (accredited by the American

Association of Laboratory Animal Caretakers [AALAC]),

according to the Guidelines established by the National

Institutes of Health.

Cats that had received compression injuries were housed

at the Cincinnati Veterans Administration Animal Facility

(also accredited by AALAC) for approximately two months

postoperatively. Upon examination by a veterinarian, these

cats were transported to the University of Florida via

commercial airline for the transplantation and magnetic

resonance imaging portion of these studies.

All cats were examined daily by a veterinarian or

veterinary technician for general health conditions and for

any complications following either the spinal cord injury or

the MRI experiments. All surgical procedures were performed

using sterile technique.


Spinal Cord Iniury and Transplantation Protocols

Hemisection Injury Model

Donor tissue preparation. For each recipient, donor

tissue was obtained by Cesarean-section from mature, random-

source, gravid female cats. In the absence of naturally

occurring estrus, these cats were induced into sexual

receptivity with the administration of follicle stimulating

hormone-pituitary extract (FSH-P, Schering) at a dosage of 2

mg/day until overt signs of estrus were observed or, for a

maximum of 5 days (Wildt, 1986). Mating times and crown-rump

measurements (Nelson and Cooper, 1975) were recorded for

determination and verification of fetal ages, respectively.

Following Cesarean-section each fetus was immediately

placed in cold Hank's balanced salt solution. Under aseptic

conditions either fetal cat brainstem (BSt), neocortex (NCx),

or spinal cord (FSC) was removed between embryonic days 21 and

38 (E21-E38). The surrounding meninges (except pia mater)

were stripped free. In the case of fetal spinal cord and

brainstem grafts, the attached dorsal root ganglia or cranial

ganglia, respectively, were also removed. The dissected fetal

tissue was then cut into pieces closely approximating the

length of the hemisection cavity.

Surgical lesion and transplantation. The hemisection

lesion of the spinal cord was performed on adult (8 mos. 3

yrs.) female cats weighing 2.5 5.0 kg. Cyclosporine A

(Sandimmune, Sandoz) was administered at a dosage of 10 mg/kg


beginning 24 hours prior to surgery and then continued daily

thereafter until the end of the experiment.

Following induction of anesthesia with ketamine HC1 (10

mg/kg), xylazine (2 mg/kg), and glycopyrollate (15 gg/kg), the

cats were maintained on 1-3% halothane. The eyes were coated

with sterile ophthalmic lubricant to prevent corneal

desiccation. Lactated Ringer's solution (10 mg/kg/hr) was

administered through an indwelling catheter placed in the

cephalic vein.

Each cat was placed in sternal recumbency on a

circulating hot water heating pad. Strict attention to

aseptic technique was maintained throughout the surgical

procedure. A skin incision was made on the dorsal midline;

the muscle and fascia were dissected from the dorsal spinous

processes and laminae to provide adequate exposure of a one-

segment hemisection/transplantation site at T,,-L2. The

laminectomy was performed using Rongeur forceps and bone wax

was employed to maintain hemostasis. The dura mater was

incised near the midline, and reflected. The pia-arachnoid

was then incised and carefully separated from the surface of

the cord, preserving the integrity of the major vessels. The

hemisection lesion, which sometimes extended over the midline,

was then made with the aid of gentle aspiration and

microdissection. Once bleeding was arrested with bovine

thrombin-soaked gelfoam and CSF filling of the cavity

controlled, donor tissue was introduced into the lesion.


Control animals had similar lesions but did not receive any

donor tissue. The leptomeninges were then sutured separately

with 8-0 Prolene microsuture. Durafilm (Codman & Shurtleff)

was placed over the dura and the muscle, fascia, subcutis, and

skin were then closed in layers.

Compression Injury Model

Donor tissue preparation. Pieces of either feline E23

BSt or E23 FSC were obtained as described in the previous

section. In one case, a cell suspension of E23 FSC was

prepared from fetal tissue. A slurry of cells was made by

mincing the tissue into small pieces in a 0.6% glucose-saline

solution. The pieces were next dissociated by mechanical

trituration using fire-polished pipettes of progressively

decreasing diameter. A suspension of approximately 1.6 x 106

cells/cc was obtained and cell viability of >90% was

determined using acridine orange and ethidium bromide (Brundin

et al., 1985).

Surgical lesion and transplantation. The procedures for

static-load compression injuries of the feline spinal cord

have been described in detail elsewhere (Anderson et al.,

1980; Anderson et al., 1976). Briefly, cats were

anesthetized with intraperitoneal (i.p.) pentobarbital sodium

(Nembutal, 30 mg/kg) and immobilized in a stereotaxic frame.

A one-segment laminectomy was performed in the upper lumbar

region (L). A guide tube was positioned directly over and

perpendicular to the center of the spinal cord. Compression

was produced by passing a weighted steel rod (170-190 g; tip,

6 mm in diameter) through the guide tube until it rested

directly on the dura overlying the center of the spinal cord

for 5 minutes.

Five to seven weeks after the compression injury, the

cats were begun on a daily regimen of cyclosporine A (10

mg/kg). One or 2 days later, the original surgery/trauma site

was reexposed and the location of the contusion was determined

by identifying a necrotic region and an area of cystic

cavitation within the spinal cord. The meninges were incised

at this location, the necrotic tissue was aspirated, and the

cavity was then flushed with physiological saline. Whole

pieces of either feline E23 BSt or E23 FSC were placed into

these cavities. The remainder of the surgical procedure was

identical to that described for the hemisection lesion.

In one case, the dura was incised at the lesion site, and

120 Al of a cell suspension of feline E23 BSt was injected

through the pia-arachnoid into a small, pulsating cavity. A

Hamilton syringe, fitted with a 30 gauge needle, was used for

the suspension injections. Surgical closure and postoperative

management were as described in the hemisection model.

Postoperative Care
Intravenous fluids and supplemental heat were continued

until the animals were awake, at which time they were

transferred to a heated recovery cage. Prophylactic

antibiotics (Amoxicillin, 15 mg/kg B.I.D.) were administered


for 5 days postoperatively. Temperature, pulse, respiration,

neurological functions and responsiveness to handling were

monitored closely as indicators of infection or pain. Urinary

bladder and bowel functions were monitored and, when

necessary, the bladders were manually expressed three times

daily, until normal micturition was restored. The cats were

maintained on a balanced, low ash, acidic urine-inducing diet

(C/D, Feline maintenance, Hill's Science Diet) to aid in the

prevention of cystitis. An exercise/socialization area was

provided for the cats to enhance recovery. All animals

recovered uneventfully, though with a mild neurological

deficit as a result of the spinal lesion.

Magnetic Resonance Imaging Studies

MR images were obtained at one or more time points

ranging from immediately postoperative to two years

posttransplantation (or control lesion). The animals were

injected with glycopyrollate (Robinul-V, 15 ig/kg) and

anesthesia was induced with a mixture of ketamine (Ketaset, 11

mg/kg) and xylazine (Rompun, 2.2 mg/kg). An intravenous

catheter was placed in a cephalic vein for constant infusion

of lactated Ringer's and 5% dextrose. The surgical site was

determined by identifying the missing dorsal spinous

processes) and was marked for placement on the MR imaging

coil with adhesive tape.

Following endotracheal intubation, anesthesia was

maintained with isoflurane (Forane, 1-3% in 100% 02) via a


Bain coaxial circuit nonrebreathing apparatus. Body

temperature was monitored with an esophageal probe (Luxtron

Fluoroptic Thermometry System) and maintained by a circulating

water heating pad (GRI Medical Products). To further

stabilize body temperature, the ambient temperature in the

magnet bore was increased by the periodic use of a 1500 watt

hot-air blower connected to a 10 foot length of 6 inch

diameter corrugated pipe. Heart rate (Space Labs Patient

Monitor), expired CO, and respiratory rate (Allegheny

International Medical Technology) were also monitored

continuously during the imaging process.

The cats were placed in the supine position (to reduce

breathing motion artifacts) in a 12 cm diameter Plexiglas

half-cylinder holder and centered over a curvilinear surface

coil attached to the external surface of the holder (except

where indicated in Chapters 4 and 5). MRI was performed on a

Spectroscopy Imaging Systems, 2.0 Tesla, VIS imaging

spectrometer by acquiring T1- and T2-weighted multislice spin-

echo images in both the transverse and sagittal planes.

Histological Follow-Up

For histological verification of the MR views, the cats

were deeply anesthetized with intravenous (i.v.) pentobarbital

sodium (Nembutal, 30 mg/kg) immediately following MRI and

perfused intracardially with 0.9% saline followed by either

fixative for electron microscopy (5% glutaraldehyde, 4%

paraformaldehyde in 0.1M Sorenson's phosphate buffer) or


immunocytochemistry fixative (4% paraformaldehyde in 0.1M

Sorenson's phosphate buffer). The spinal cords were removed,

visually inspected, and cut transversely into 1 mm blocks in

register with the previously obtained transverse MR images.

Blocks then were embedded in epon, from which 2 1m thick

sections were obtained and then stained with toluidine blue

and examined by light microscopy. These sections were

evaluated to determine how closely the NMR images paralleled

changes in white and gray matter and in the fetal transplants,

as observed in sections of fixed tissue.


Portions of this chapter are reprinted with permission
from Wirth, E.D.,III, D.P. Theele, T.H. Mareci, D.K. Anderson,
S.A. Brown, and P.J. Reier (1992) In vivo magnetic resonance
imaging of fetal cat neural tissue transplants in the adult
cat spinal cord. J.Neurosurq. 76:261-274. Copyright 1992 by
Journal of Neurosurgery.



Recent studies have shown successful and reproducible

application of intraspinal grafting methods to a variety of

spinal lesions in rats, including acute and chronic resection

and contusion injuries (Reier et al., 1986; Houle and Reier,

1988; Reier et al., 1988). These methods also have been

applied to acute and advanced lesions of the cat spinal cord

(Anderson et al., 1989; Anderson et al., 1991; Reier et al.,

1992). However, determination of transplant survival has

relied exclusively on light and electron microscopic

examination of post-mortem material from these valuable animal

models. Thus, it would be desirable to have a technique

whereby these embryonic grafts could be monitored

noninvasively in living subjects.

Among the various radiological modalities for studying

the central nervous system, nuclear magnetic resonance imaging

(MRI) and spectroscopy (MRS) have proven to be the most useful

in documenting the many anatomical and metabolic derangements

resulting from CNS trauma (Hackney et al., 1986; Kalfas et

al., 1988; Quencer, 1988; Takahashi et al., 1987; Vink et al.,

1989; Mafee, 1989; Vink et al., 1987; Chakeres et al., 1987;



Masaryk, 1989). However, the ability of MRI to resolve

implants of fetal CNS tissue from the adjacent host spinal

cord has not been tested.

Accordingly, this study explored whether MRI could allow

direct observation of chronically surviving fetal CNS grafts

in the injured adult cat spinal cord. To keep the

interpretation of the MR images as simple as possible, these

experiments focused on the chronic phase of spinal cord injury

(SCI), by which time many of the pathological sequelae

following SCI have subsided (Quencer, 1988). In addition, the

cats were imaged immediately prior to sacrifice to facilitate

the comparison of the MR images and postmortem specimens.

It is reported here that, as confirmed by correlative

postmortem gross and histological analyses, areas of medium to

moderately high signal intensity on in vivo intermediate T,-

weighted MR images consistently corresponded with the presence

of intraspinal fetal grafts in the hemisected or compression-

injured feline spinal cord.

Materials and Methods

Animals and Surgical Procedures

The surgical procedures are described in detail in

several recent reports (Anderson et al., 1991; Reier et al.,

1992; Wirth et al., 1992) and Chapter 2. Briefly, eleven

adult female cats and one control animal were anesthetized

with ketamine and xylazine, intubated and maintained under

anesthesia with halothane. Following incision of the


overlying skin and fascia, the paraspinal musculature was

retracted and a laminectomy was performed at the T,3-L,

vertebral level. In seven cats and the control animal, the

meninges were then incised and a hemisection lesion was made

with gentle aspiration. Alternatively, in four cats the

meninges were left intact and a static-load compression injury

was made. The details of this procedure have been described

previously (Anderson et al., 1976; Anderson et al., 1980;

Chapter 2).

Donor CNS tissue was then prepared for all cats except

the control animal using methods previously described (Reier

et al., 1986; Wirth et al., 1992; Chapter 2). Immediately

following the hemicordotomy or seven weeks after a compression

injury, pieces of fetal feline CNS tissue were introduced into

the spinal cord injury site. The leptomeninges were then

sutured and covered with a piece of synthetic dura, followed

by closure of the musculature and overlying incision.

Magnetic Resonance Imaging Studies

The general experimental protocol for the MRI studies is

described in detail in Chapter 2. MR images were obtained at

one time point from five months to two years

posttransplantation (or control lesion) in all of the animals

except one (cat H2, Table 1, Results), which was examined on

multiple occasions (6 & 8 months) prior to its last imaging

session at 10 months.

Following induction of anesthesia with ketamine and

xylazine, the cats were intubated and anesthesia was

maintained with isoflurane via a Bain coaxial circuit

nonrebreathing apparatus. The cats were placed in the supine

position (to reduce breathing motion artifacts) in a 12 cm

diameter Plexiglas half-cylinder holder and centered over a

curvilinear 6 cm x 6 cm single-turn surface coil attached to

the external surface of the holder. MRI was performed on a

Spectroscopy Imaging Systems, 2.0 T, VIS imaging spectrometer

by acquiring multislice spin-echo images (TR/TE= 1000/30

msec.) in both the transverse and sagittal planes. The

repetition and echo times were empirically chosen to optimize

gray-white matter contrast, signal-to-noise ratio, and

acquisition time. After centering the graft site over the

surface coil, survey measurements with slice thicknesses of 2-

3 mm and 2 signal averages (12 min. acquisition time) were

performed to determine the precise location of the graft site.

Imaging experiments with 1 mm slices and 4 signal averages (24

min. acquisition time) were then performed to study the graft

area in greater detail and to reduce the partial volume

effects seen in thicker slices. All experiments had a read-

out gradient of 1 G/cm, a 6 cm x 6 cm field of view and a 256

x 256 acquisition matrix and, therefore, an in-plane

resolution (i.e., pixel size) of 0.23 mm x 0.23 mm.

Histological Follow-Up

Immediately following MRI, cats were deeply anesthetized

and perfused-fixed intracardially (see Chapter 2 for details).

Spinal cord segments representative of the MR images were cut

into 1 mm blocks and embedded in epon for subsequent

sectioning and light microscopy. Some tissue blocks were

prepared for immunocytochemistry and were cut into 40 gm

sections on a vibratome and incubated with various antisera

for companion studies (Anderson et al., 1991).


Intact Cord Regions

At levels of the host spinal cord rostral or caudal to

the lesion, minimal pathology was seen and the gray matter was

clearly defined. On intermediate T1-weighted images, intact

gray matter exhibited a high signal, whereas normal white

matter typically yielded a less intense image (Fig. 3-1 a).

This contrast also provided excellent differentiation of gray

and white matter on sagittal MR images (e.g., Fig. 3-6 a).

Although the general integrity of the spinal cord was not

usually disturbed at levels beyond 1-2 cm from the

transplants, areas of spongiform degeneration and

demyelination (e.g., in the dorsal funiculus, Fig. 3-1 b)

presented lower levels of signal intensity than normal white

matter (Fig. 3-1 a). The MR images also provided good

visualization of structures surrounding the spinal cord

Figure 3-1. Magnetic resonance images and representative
histological sections at intact and transplanted spinal cord

a) Intermediate T1-weighted magnetic resonance (MR) image
(TR/TE = 1000/30 ms) through a spinal cord region
approximately 1.5 cm caudal to the level of the graft site in
Cat Cl. Imaging was performed six months after implantation
of fetal feline (E23) brainstem (BSt) in a compression lesion
at the L, spinal cord level. The gray matter appears as an
H-shaped region of moderately high signal intensity
(asterisk). Other visible structures include the very bright
epidural fat (e) and the ventral spinal artery (arrow). Note
that the dorsal columns have a lower signal intensity
(arrowhead) than the ventral and lateral white matter; bar =
3 mm.
b) A toluidine blue stained 2 gm histological section from
approximately the same spinal level as Fig. 3-1 a. The gray
matter (gm) and white matter (wm) are intact, except for some
degeneration in the dorsal columns (arrowhead); bar = 1 mm.
c) Transverse MR image acquired six months posttransplantation
through the center of the bright area "g" in Fig. 3-4 b. This
animal had received a hemimyelotomy followed immediately by a
graft of E37 fetal neocortex. The right side of the spinal
cord appears homogeneously hyperintense (g) in relation to the
host white matter (h); bar = 3 mm.
d) A 40 pm vibratome section, immunostained with an antibody
against myelin basic protein, from approximately the same
spinal level as Fig. 3-1 c. Fetal graft tissue (G) and host
white matter (HW) correspond to the hyperintense (g) and
hypointense (h) regions, respectively, in Fig. 3-1 c; bar = 1
e) MR image of Cat H3 six months posttransplantation showing
an area of low signal (gc) within the hyperintense transplant
region on the right side of the spinal cord.
f) Transverse toluidine blue stained 2 1m section from
approximately the same spinal level as Fig. 3-1 e. A greatly
expanded central canal-like structure (gc) is present in the
upper left portion of the graft (G). The apparent disparity
between the size of the canal in Fig. 3-1 e and 3-1 f
exemplifies the problem of volume averaging in which multiple
structures, such as graft tissue and canal-like structures,
are contained within the same 1 mm thick MR image slice. Note
that the MRI signals from the graft and the remaining host
gray matter (arrowhead) are very comparable and that both are
significantly brighter than the host white matter (HW).

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosurq.
76:261-274. Copyright 1992 by Journal of Neurosurgery.




including: the ventral spinal artery, epidural fat pad and

dural venous sinuses (Fig. 3-1 a).

Transplants in Hemisection Cavities

Six months posttransplantation, five of the seven graft

recipients (cats H2-H6, Table 3-1) that had received fetal

neural tissue grafts in hemisection lesions exhibited areas of

medium-to-high signal intensity in the transplantation region.

A less defined signal was seen at graft levels in Cat H7

whereas only hypointense zones were observed in Cat HI and at

the injury site in the lesion-only control.

The hyperintense zones in Cats H2-H6 had signal levels

similar to that of normal gray matter and were generally

restricted to the side of the spinal cord where the graft had

been placed (Fig. 3-1 c,e). Histological examination

subsequently verified the presence of large viable transplants

in each of these five cases (e.g., Fig. 3-1 d,f). High-power

light microscopic examination of the transplanted fetal tissue

revealed healthy, well-differentiated neuronal and glial

populations (Fig. 3-2 a,b). Multiple central canal-like

structures were also often observed, as were areas of heavy

myelination. The grafts were highly vascularized and

cytologically resembled mature CNS tissue with an abundance of

densely packed neuritic processes (for additional details of

graft cytology, see Anderson et al., 1991).

In some cases, the predominantly high signal zones seen

on MR images also contained central and/or peripheral areas of


Cat Observations on MR images Histological findings

HI Large hypointense region
H2 Hyperintense region on
right side of cord
bordered rostral and
caudal by hypointense
H3 Large hyperintense area
with very hyperintense
boundary containing some
medium & low signal areas
bordered by rostral
H4 Medium signal intensity on
left side of cord with
hyperintense medial border
H5 Medium signal intensity on
left side of cord, center
of cord is hyperintense

H6 Right half of cord is
H7 Center of cord is
hypointense, surrounded by
a thin rim of slightly
hyperintense signal

Large cyst, no graft
Graft bordered by
rostral and caudal cysts

Graft in left half of
cord bordered by rostral

Graft in left half of
cord bordered by rostral
Graft in right half of
cord; gliosis and
degenerating tissue in
center of cord
Graft in right half of
Very small graft
dorsolateral to a large
central cyst

Figure 3-2. High power light micrographs of transplanted
fetal cat tissue.

a) High power view of the 2 pm section shown in Fig. 3-3 d.
Two prominent central canal-like structures (gc) lined with
ependymal cells (arrowheads) are present in the middle of the
field. Also conspicuous are many viable, mature neurons, glia
and myelinated axons (arrows); bar = 50 Am.
b) A high power view of a different region in the same section
as Fig. 3-2 a. The graft is highly vascularized (asterisks)
and many neuronal cell bodies of different sizes (arrows) with
prominent pale nuclei and dark nucleoli are visible. Numerous
tightly packed, heavily myelinated axons (M) are also evident;
bar = 50 Mm.

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosura.
76:261-274. Copyright 1992 by Journal of Neurosurgery.


(Z ~9...
,. .. .
t'r,e .t'9 P, ^f.t''->



either low or heterogeneous signal intensity (Fig. 3-1 e). It

was found that this signal heterogeneity was due to structural

diversity within the graft (Fig. 3-1 f), and that these

distinct tissue formations frequently appeared smaller on

corresponding MR images due to volume averaging artifact.

While the transplants shown in Fig. 3-1 were hyperintense

relative to the host side of the spinal cord, some variability

in the contrast between host and graft was observed in other

subjects (Fig. 3-3 a,c). For example, in some cases a modest

contrast between host and graft was due to some high signal

intensity regions in the host neuropil. Subsequent

histological analysis demonstrated that in these cases the

elevated signal on the host side was in register with a

significant amount of necrotic host tissue (Fig. 3-3 b,d),

probably due to a progressive degeneration resulting from

inadvertent vascular compromise during preparation of the

lesion. On the other hand, MR images of Cat H7 did not

exhibit any differences in signal intensity between the host

and graft sides of the spinal cord (Fig. 3-3 e). Postmortem

observation of tissue specimens from this cat, two years

posttransplantation, revealed only a small graft in the left

dorsolateral quadrant of the spinal cord (Fig. 3-3 f).

Although the contrast between host and graft was

occasionally poor, portions of the host-graft interface were

often marked by moderately to extremely hyperintense signals

(Fig. 3-3 a,c). These boundary segments usually presented in

Figure 3-3. Transverse MR images and representative
histological sections of fetal cat CNS tissue grafts in
hemisection lesions.

a) Transverse MR image of the graft site in Cat H5 eleven
months posttransplantation exhibiting an area of medium signal
intensity on the right side of the spinal cord (g,) adjacent
to markedly (2) and slightly (3) hyperintense regions. At the
contralateral edge of the spinal cord, intact host white
matter (4) has noticeably less signal; bar = 3 mm.
b) Toluidine blue stained 2 gm section from approximately the
same spinal level as Fig. 3-3 a. Graft tissue occupies only
the left half of the spinal cord (g,), whereas the central
portion of the cord (corresponding to areas "2" and "3" in
Fig. 3-3 a contains degenerating host neuropil. Intact host
white matter (HW) on the left matches the hypointense area "4"
in Fig. 3-3 a; bar = 1 mm.
c) Transverse MR image of Cat H4 seven months
posttransplantation showing an area of medium signal intensity
on the left side of the spinal cord (arrow) with an adjacent
hyperintense band (asterisk); bar = 3 mm.
d) Toluidine blue stained 2 im section from approximately the
same spinal level as Fig. 3-3 c. Graft tissue (g) occupies
the left side of the spinal cord and is next to a region of
extensively degenerated gray matter. Also note that in Fig.
3-3 c, the contrast between transplant and host is much less
than in the previous figures. The loss of contrast may
reflect the coexistence of an extensively myelinated
transplant with degeneration and edema in the host ventral
(Vwm) and lateral white matter; bar = 1 mm.
e) Transverse MR image of Cat H7 two years posttransplantation
showing an hyperintense signal defining the periphery of the
spinal cord (arrowheads). The center of the cord appears as
a region of decreased signal (c); bar = 3 mm.
f) Toluidine blue stained 2 Am section from approximately the
same spinal level as Fig. 3-3 e. Encompassing a large,
central cyst (C) is a thin rim of host white matter (HW) and
a small region of graft tissue (g). Note that the
corresponding graft and host areas in Fig. 3-3 e do not have
appreciably different intensities. Hence, in this animal the
transplant was not distinguished from the remaining host
spinal cord on the MR images. It should also be noted that
the size of this graft was approximately 0.5 mm diameter and
thus approached the limit of spatial resolution on the MR
images; bar = 1 mm.

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosurq.
76:261-274. Copyright 1992 by Journal of Neurosurgery.

t ..


the sagittal plane as regions of high intensity situated along

portions of the rostral and/or caudal borders of the graft

site and along parts of the medial border of the transplanted

side on transverse image slices. Examination of histological

sections from corresponding spinal levels showed that these

areas consisted of gliotic and/or degenerating or edematous

host tissue (Fig. 3-3 b,d).

In six of the seven cats studied (cats H1-5 & H7),

sharply demarcated regions of low signal intensity were also

observed on the sagittal MR images at or near the

histologically identified transplantation site. Cats H2-5 and

H7 exhibited these hypointense areas adjacent to the

hyperintense regions described above (e.g., Fig. 3-4 b-d). On

the other hand, both a control animal that had received a

lesion without an implant of fetal neural tissue (Fig. 3-4 a)

and Cat H1, which had received a graft, presented only large

zones of relatively homogeneous and low signal intensity.

Postmortem gross and histological examination of specimens

from spinal cord segments representative of the MR images

revealed that these hypointense territories were in register

with cavities that were filled with cerebrospinal fluid (CSF)

and were largely devoid of tissue elements. In Cats H2-5 and

H7, these cavities were found next to the viable grafts

described above, whereas both the control animal and Cat HI

failed to show any hyperintense areas on the MR images. Also,

Figure 3-4. Sagittal and transverse MR images of viable graft
tissue and regions of cavitation in transplant recipients and
a lesion-only control.

a) Sagittal MR image of a control animal that received a
hemisection lesion but no implant of fetal tissue. Note that
only a hypointense signal is observed at the injury site (c)
and that the intensity is similar to that of the cysts in Fig.
3-4 b-d. The size of this cyst reflects a smaller surgical
lesion than that shown in Fig. 3-4 b-d; bar = 4 mm.
b) Sagittal MR image of Cat H2 at the level indicated by (1)
in Fig. 3-4 d. An hyperintense area (g), identified as graft
tissue postmortem (see Fig. 3-1 d), is flanked by very
hypointense rostral and caudal zones (c), which were found to
be CSF-filled cysts. The level of the axial section shown in
Fig. 3-4 d is indicated by "a"; bar = 4 mm.
c) Sagittal MR image of Cat H2 at the level indicated by (2)
in Fig. 3-4 d. On this side of the spinal cord the rostral
and caudal cysts (c) are confluent. The level of the axial
section shown in Fig. 3-4 d is indicated by "a"; bar = 4 mm.
d) Axial MR image at the level indicated by "a" in Figs. 3-4
b,c. A surviving transplant (g) is bordered by a hypointense
cyst (c). The lines (1 and 2) indicate the planes of section
shown in Fig. 3-4 b and 3-4 c, respectively; bar = 4 mm.
These images of Cat H2 were acquired 1i months prior to the
image shown in Fig. 3-1 c. Since Figs. 3-1 c, 3-1 d and 3-4
d are from the same level of the transplant, it appears that
the graft expanded to fill in part of the cyst (c) evident in
Fig. 3-4 d. However, the rostral and caudal cysts in Fig. 3-4
b remained at the time of sacrifice (data not shown).

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosura.
76:261-274. Copyright 1992 by Journal of Neurosurgery.



Cat HI did not exhibit any viable graft tissue on postmortem

histological examination.

Parenthetically, the region of hypointensity in the one

cat (H2) imaged on multiple occasions showed progressive

changes. Thus, a large area of cavitation was seen between

host and graft at earlier postgraft intervals (Fig. 3-4 d),

whereas the volume of this cavity was much smaller when the

last MRI was obtained (see Fig. 3-1 c,d).

Transplants in Compression Lesions

MRI was also tested for its ability to demonstrate

surviving fetal grafts in compressed spinal cords. All four

animals in this group (cats C1-C4, Table 3-2) exhibited

regions of hyperintense signal at the transplantation site.

In Cats Cl and C2, the zone of high signal intensity

encompassed up to 90% of the cross-sectional area of the

spinal cord (e.g., Fig. 3-5 a). The hyperintense areas in

these two cats were similar to those observed on MR images of

the transplantation site in cats with grafts inserted

immediately following hemimyelotomies (see Fig. 3-1 c,e).

However, in contrast to animals in the hemisection group, in

which both the hyperintense areas and histologically

identified grafts were largely restricted to one half of the

spinal cord, postmortem analyses of Cats Cl and C2

demonstrated that the hyperintense areas corresponded to

transplanted fetal tissue that had grown to fill very large

central defects (Fig. 3-5 b). Further histological inspection


Cat Observations on MR images Histological findings
Cl Homogeneous medium-to-high Large graft filling
signal in dorsal 90% of dorsal 90% of cord
C2 Hyperintense area Large graft bordered by
occupying up to 90% of rostral cyst
cord diameter bordered
rostrally by a hypointense
C3 Mixed-to-hyperintense Graft in center of cord
signal in center of cord with adjacent small
with three adjacent focal cysts
(< 0.5 mm diam.)
hypointense areas
C4 Dorsum of cord compressed Graft in center of cord
down to level of central
canal; diffuse
hyperintense appearance
ventral and rostral to
compressed area

Figure 3-5. MR images and representative histological
sections of surviving fetal cat tissue in compression

a) Transverse MR image of the graft site in Cat C1 (approx.
1.5 cm rostral to Fig. 3-1 a). An uniform, hyperintense
signal is present from nearly the entire spinal cord (g)
except for a thin band of low signal along the ventromedial
(large arrowhead) and ventrolateral borders (small
arrowheads); bar = 3 mm.
b) Toluidine blue stained 2 Am section from approximately the
same spinal level as Fig. 3-5 a. A very large fetal graft (g)
occupies about 90% of the cross-sectional area of the spinal
cord. The only host tissue is the ventromedial and
ventrolateral white matter (wm) on each side of the spinal
cord; bar = 1 mm.
c) Transverse MR image of Cat C3 five months after injection
of a 120 Al suspension of E23 BSt into a compression injury
cavity. A small central core of medium signal is bounded by
three focal areas of markedly reduced signal (arrowheads) and
a zone of hyperintense signal; bar = 3 mm.
d) Toluidine blue stained 2 Am section from approximately the
same spinal level as Fig. 3-5 c. The center of the spinal
cord contains graft tissue (g) and three small cysts
(arrowheads). Note that the central core of the MR image
(Fig. 3-5 c) exhibits medium intensity, signifying some volume
averaging of the hyperintense signals from the transplant and
the hypointense cysts. The perimeter of the spinal cord
consists of intact host white matter (w); bar = 1 mm.

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosurq.
76:261-274. Copyright 1992 by Journal of Neurosurgery.

'~ .'- ;,


of these transplants showed features of differentiated

neuropil identical to what was seen in grafts to hemisection

lesions (Fig. 3-2 a,b).

On MR images of Cat C3, however, the zone of high signal

intensity was confined to a circular-to-elliptical central

location in the cord surrounded by tissue yielding a signal

similar to that of normal host white matter. This MRI

appearance was also observed on images of the animal that had

received an injection of a fetal cell suspension into the

lesion cavity (cat C4; Fig. 3-5 c). The signal intensity of

the graft area was comparable to both the MR images of Cats Cl

and C2 and host gray matter (refer to Fig. 3-1), but the

dorsal and ventral horns observed in Fig. 3-1 were absent.

Postmortem analyses revealed that the hyperintense areas

matched the locations of surviving neural grafts (Fig. 3-5 d).

Similar to MR images of animals in the hemisection group,

Cats C2 and C3 also displayed some areas of low signal

intensity neighboring the hyperintense zones (Fig. 3-6 a,b).

The locations of the low signal areas coincided with large

cystic regions in postmortem specimens (Fig. 3-6 c). The

correlation between the hypointense zones on the MR images and

the areas of cyst development observed in fixed tissue samples

is consistent with the findings obtained from the hemisection


Figure 3-6. MR images and representative histological section
from a compression-injured spinal cord containing both viable
graft tissue and a substantial cyst.

a) Sagittal MR image of Cat C2 five months after implantation
of E23 FSC into a compression injury. A rostral hypointense
region (C) is easily distinguished from an adjacent
hyperintense zone (g) at the graft site. Note that the caudal
graft-host interface is marked by the interruption of the dark
and light white and gray matter bands, respectively; bar = 4
b) Transverse MR image slice through the middle of the
hypointense area (C) in Fig. 3-6 a. This image reveals that
the hypointense area is located in the center of the spinal
cord (c) while some peripheral host tissue remains (w). Note
that Fig. 3-6 a and 3-6 b provide excellent rostrocaudal and
mediolateral information, respectively, about the location and
size of the hypointense region; bar = 3 mm.
c) Toluidine blue stained 2 gm section from approximately the
same level as Fig. 3-6 b. The center of the cord contains a
partitioned, large cyst (C) enclosed by a rim of host white
matter (w); bar = 0.5 mm.

Reprinted with permission from Wirth, E.D.,III, D.P. Theele,
T.H. Mareci, D.K. Anderson, S.A. Brown, and P.J. Reier (1992)
In vivo magnetic resonance imaging of fetal cat neural tissue
transplants in the adult cat spinal cord. J.Neurosurq.
76:261-274. Copyright 1992 by Journal of Neurosurgery.



The results of this investigation demonstrate a strong

correlation between areas of medium to moderately high signal

intensity on intermediate T1-weighted, spin-echo MR images and

the presence of long-term surviving transplants in lesions of

the adult cat spinal cord. This MRI appearance of fetal

transplants contrasted with the lower signal intensity

exhibited by the intact host white matter.

As one control for this study, an animal was prepared

which had received only a partial spinal resection.

Subsequently, the site of injury yielded a very hypointense

signal quite unlike that derived from viable graft tissue. In

addition to this control animal, markedly hypointense areas

were also found on MR images of the graft sites where

intraspinal transplants were either absent or poorly

integrated. Consistent with observations of standard T -

weighted images, these hypointense areas on the intermediate

T,-weighted images corresponded to cysts characterized by long

relaxation times (Wehrli et al., 1984). When such low signal

zones approximated hyperintense areas representative of graft

tissue, the contrast aided in defining the borders of the

transplant where host-graft integration was not optimal. In

the more extreme situation, a region of weak signal intensity

at the graft site correlated with the absence of any surviving

donor tissue as had occurred in one case (cat HI) in this

investigation. Thus, at these advanced postgraft stages, a


hyperintense signal on MR images was reasonably predictive of

the presence of some viable graft tissue. On the other hand,

regions of low signal intensity were more indicative of cysts

or wounds in which no graft or other tissue masses were seen.

Imaging Parameters

The imaging protocols used in this study were optimized

for signal-to-noise and contrast between gray and white

matter. The intermediate T1-weighted (TR=1000 msec, TE=30

msec) spin-echo images presented are similar in appearance to

standard T1-weighted images as evidenced by the high signal

from the epidural fat and very low signal from CSF. However,

the intermediate T1-weighted images have a TR slightly longer

than the measured T, for gray matter (Wehrli et al., 1984).

This increased the contrast between gray and white matter over

T1-weighted images while still maintaining a superior signal-

to-noise ratio. In addition, T2-weighted images were not

acquired because of their inherently low signal-to-noise and

the long acquisition times required, though the capacity of

T2-weighting for highlighting CSF is recognized. Overall, the

advantage of excellent signal-to-noise allowed for acquisition

of very thin slices (1 mm thick) with high in-plane resolution

(234 pm x 234 gm). This proved to be extremely important for

studying in vivo spinal cord transplants in cats with graft

sizes typically approximating 2 mm x 2 mm x 6 mm. Experience

from this study suggests that the maximum useful slice

thickness is 2-3 mm in the transverse plane and 1-2 mm in the

sagittal plane. Our imaging system is currently limited to a

minimum slice thickness of approximately 1 mm.

Comparison with MRI Studies of Fetal Brain Transplants

The MRI signals obtained from the graft sites in this

study are consistent with MRI results of fetal striatal tissue

transplants in the kainic acid-lesioned striatum (Norman et

al., 1989; Norman et al., 1990). Moreover, another study

(Peschanski et al., 1988) found that the T2 relaxation time of

grafted fetal striatal neurons was decreased to 30-50 ms, as

compared to 70-80 ms for normal gray and white matter.

However, the T2 of the grafts was not measured in the present

study because of the observed partial volume artifacts. It

was also observed that these hyperintense areas may contain

subregions of heterogeneous signal due to volume averaging of

graft tissue with extramedullary elements (e.g. CSF) and/or

other components (e.g. an expanded central canal) within the

graft parenchyma.

Host-Graft Approximation

In some animals, portions of the interface between graft

and host were delineated by segments of very high signal

intensity separating normal-appearing host spinal cord from

the moderately hyperintense transplant areas (refer to Fig. 3-

3 a,c). The high signal intensity at the graft boundary may

indicate residual spinal edema (Hackney et al., 1986),

degeneration prior to actual cavitation (e.g., Fig. 3-3 a,c),

or the presence of gliosis between graft and host (Barnes et


al., 1988). Glial scars often form along areas of the CNS

that become exposed to non-CNS environments (Reier and Houle,

1988). In the absence of an hyperintense border, however, it

was frequently difficult to ascertain whether the host and

donor tissue were confluent or separated by a more subtle

intervening scar. This limitation is due to both volume

averaging artifacts from insufficient spatial resolution and

inadequate contrast between the graft and adjoining host

tissue. Emphasis on the ability to resolve the extent of

host-graft integration can be relevant to the potential for

forming axonal connections that may be vital to graft-mediated

functional recovery (Kunkel-Bagden and Bregman, 1990).

Present Limitations of MRI

Although the available sensitivity is adequate for gross

visualization of fetal grafts, increased spatial resolution

would help to reduce the volume averaging artifacts, provide

more information regarding host-graft apposition and permit

the observation of very small pieces of graft tissue whose

sizes approach the limit of available spatial resolution.

Increased resolution could be achieved through further

optimization of radiofrequency (rf) coil geometry to match the

anatomy of the spine, imaging at higher field strengths,

acquisition of more signal averages, or volume imaging. The

former two considerations require minor and major hardware

upgrades, respectively, whereas the latter two options only

demand increased imaging time. Accordingly, the potential for


increased signal-to-noise through rf coil optimization is the

focus of Chapter 5.

Future Directions

Finally, for MRI to become a more fully reliable

diagnostic tool for the evaluation of CNS transplants, future

investigations must consider MRI's ability to assess the

viability and metabolic status of the graft at both subacute

and chronic posttransplantation intervals, as well as for

providing indices of incipient graft rejection. For example,

at the advanced postgrafting intervals sampled in this

investigation, both degenerating and gliotic regions of the

host spinal cord exhibit MRI signals very similar to those

yielded by viable graft tissue. Thus, if a transplant is

rejected and subsequently replaced by fibrotic tissue it is

conceivable that large hyperintense areas might still appear

at graft sites that actually contain little or no donor

tissue. It is also possible that evolving lesions may in some

cases present signals that could be interpreted as graft

tissue. These issues are addressed in Chapter 4, with the

overall goal of providing a temporal continuum from subacute

to chronic phases upon which the potential of MRI for

determining graft survival can be tested.

Although a good correlation was observed between certain

MRI signals and the presence of a healthy fetal CNS transplant

in this investigation, standard spin-echo imaging techniques

alone cannot be used to assess the viability of a graft


directly. It is possible that this may be achieved by

prelabeling the grafts with an MRI contrast agent prior to

transplantation (Smith et al., 1988). For example,

suspensions of fetal rat basal forebrain neurons that were

prelabeled with the paramagnetic contrast agent colloidal gold

and transplanted into the basal forebrain of adult rats showed

a markedly distinct signal from the host brain on MRI images

(Byrne et al., 1989). Destruction of the graft cells

containing the paramagnetic contrast agent would presumably

result in decreasing contrast between graft and host due to

loss of the contrast agent from the graft site, provided the

agent was not taken up by neighboring cells.

In order to measure the metabolic status of a graft,

however, magnetic resonance spectroscopy (MRS) and/or

metabolite-specific imaging is required. One such

investigation has reported the ability of MRS to measure

changes in lactic acid, intracellular pH and high energy

phosphates, which are associated with the bioenergetic status

of cells, in the injured rabbit spinal cord (Vink et al.,


Concluding Comment

Although several issues remain to be resolved in order to

optimize MRI in relation to experimental neural tissue

transplantation, these initial observations are encouraging

and provide a framework for future technical improvements and

strategies. By providing in vivo evidence indicative of graft


survival or demise, MRI, alone or in conjunction with other

diagnostic methods, can certainly help to enhance the use of

laboratory animals required for fundamental and applied

research. In conjunction with histological, behavioral, and

electrophysiological studies, MRI can also facilitate the

study of transplants in traditional resection and more

clinically relevant models of spinal cord injury, such as

provided by static-load compression. It is also recognized

that in situ verification of graft survival by MRI will be an

important part of any potential clinical trials involving

intraspinal neural tissue grafting.

Portions of this chapter are reprinted with permission
from Wirth, E.D.,III, D.P. Theele, T.H. Mareci, D.K. Anderson,
S.A. Brown, and P.J. Reier (1992) In vivo magnetic resonance
imaging of fetal cat neural tissue transplants in the adult
cat spinal cord. J.Neurosurq. 76:261-274. Copyright 1992 by
Journal of Neurosurgery.



Recent investigations have demonstrated the capacity of

magnetic resonance imaging for visualization of transplants of

fetal CNS tissue in the injured brain (Miletich et al., 1988;

Peschanski et al., 1988; Smith et al., 1988; Norman et al.,

1988; Villablanca et al., 1990; Norman et al., 1989; Norman et

al., 1990) and for obtaining images of surviving long-term

intraspinal grafts (Wirth et al., 1989; Wirth et al., 1992;

Chapter 3). These studies have shown that signals on T,- and

T2-weighted MR images can indicate whether grafted embryonic

striatal tissue is undergoing a neurodegenerative process

(Norman et al., 1990). It has also been shown that

chronically surviving transplants (> 5 months) in the spinal

cord exhibit a hyperintense signal on intermediate T,-weighted

images that is readily distinguished from hypointense areas of

cavitation (Wirth et al., 1992; Chapter 3).

The majority of these reports, however, have provided

only general descriptions of grafted embryonic tissue

immediately prior to sacrifice. Thus, the ability of MRI to

dynamically probe the development of neural grafts in situ has

not been investigated. Further, it remains unclear whether



the MRI signals from developing fetal neural tissue and

evolving pathology due to traumatic spinal cord injury are

distinguishable at earlier postgrafting intervals (i.e., < 5


Rapid demonstration of graft survival could be of

considerable benefit in long-term behavioral evaluations of

functional loss due to spinal cord injury (Schrimsher et al.,

1990) as well as potential transplant-mediated functional

recovery (Stokes and Reier, 1990; Buchanan and Nornes, 1986;

Kunkel-Bagden and Bregman, 1990; Kunkel-Bagden et al., 1991).

For example, damage to specific sectors of the rat cervical

spinal cord, as determined from postmortem specimens, has been

shown to produce distinct qualitative and quantitative

deficits on a forelimb reaching task (Schrimsher, 1992).

Hence, accurate visualization of graft integration with

various spinal cord regions in vivo could allow for

correlation of anatomical repair with behavioral performance

in living subjects.

Therefore, this study explored whether MRI could

differentiate the complex and potentially overlapping signals

presented by developing fetal cat CNS grafts and ongoing

pathology in the injured cat spinal cord from acute to chronic

intervals (0-5 months) following transplantation. MR images

of the graft site were evaluated by both subjective appraisal

of pixel intensities and with more quantitative approaches.

With regard to the quantitative assessments, two different


numerical methods were employed. The first procedure involved

comparing the normalized mean pixel intensity from the

transplant or injury location in graft recipients with the

mean intensity of the injury site in lesion-only control

animals. Alternatively, the T, and T2 relaxation times of the

graft epicenter were measured and compared with known values

of normal CNS regions and various types of pathology

(Vanderknaap and Valk, 1990; Holland et al., 1986; Haacke,


Portions of this investigation have been summarized

previously (Wirth et al., 1991).

Materials and Methods

Animals and Surgical Procedures

The surgical procedures are described in detail in

several recent reports (Anderson et al., 1991; Reier et al.,

1992b; Wirth et al., 1992; Chapter 2). Briefly, twelve adult

female cats were anesthetized with ketamine and xylazine and

a laminectomy was performed at the L,2 vertebral level.

Following incision of the meninges, either a dorsolateral

funiculotomy or hemiresection lesion was made using gentle

aspiration, followed immediately by implantation of fetal cat

CNS tissue into the cavity. Eight of the animals received

oral cyclosporine (10 mg/kg) one day prior to transplantation

and daily thereafter. Three additional cats received

hemicordotomies but no implant of donor tissue and served as

controls. As part of a companion study (Chapter 5), two graft


recipients were fitted with a chronically implanted radio-

frequency coil that was attached to the spine around the

laminectomy site.

Magnetic Resonance Imaging

MRI sampling was performed immediately postoperative and at

1, 2, 4, 8, 12, 16, and 20 weeks postgrafting. Specimens for

correlative histology were obtained at 8, 12, 16, and 20 weeks

after transplantation. A similar paradigm was utilized for

the lesion-only control animals. The imaging protocol was the

same as that described in Chapter 2, which consisted of

intermediate TI- and T2-weighted multislice, spin-echo


Ouantitative Analyses

To validate the subjective interpretations of signal

intensities on the MR images (Table 4-1, Results), a more

quantitative approach also was explored. For intermediate T -

weighted images, this approach entailed recording the mean

pixel value (i.e., signal intensity) of a homogeneous region

(approx. 12 pixels [0.63mm2]) located near the epicenter of

the graft/lesion site. The mean pixel value from the

transplant/injury site was then normalized to the mean signal

of fat (within the same image) for comparison across time

points and animals. The signal from fat was used as a

reference because it uniformly presented the highest intensity

within these images and, thus, allowed for variations in

either imaging coil performance or scaling of the displayed


image data. This technique was validated by imaging several

normal, unoperated cats and comparing the normalized signal

intensities from gray matter, white matter and paraspinal

muscles. A similar approach was used for T2-weighted images,

except that cerebrospinal fluid (CSF) was used as a reference

instead of fat. On these images, the CSF uniformly presented

the highest intensity.

Estimates of T1 and T2 Relaxation Times

In five cats, estimates of the T, and T2 relaxation times

for the transplant/lesion site also were made. For measuring

T1, spin-echo images with a fixed TE (23 msec) and an array of

TRs (140, 200, 400, 800, 1600, 3200 msec) were acquired for a

single transverse image (2 mm thickness) through the epicenter

of the graft/injury site. Images for determining T2 were

collected from the same location with a fixed delay, dl (3000

msec), between consecutive pulse sequences (dl = TR (TE +

at/2); at = acquisition time) and an array of TE's (23, 30,

60, 120, 240 msec).

Since the time required to collect these data was

considerable (2.5 hours total per session), the Ti and T2 times

only were estimated for a single 2mm slice through the

graft/lesion epicenter. Consequently, after the first

sampling interval for a given animal, it became necessary to

locate the original spinal level accurately for subsequent

imaging sessions so that the Ti and T2 times could be followed

longitudinally. At each subsequent imaging interval, the


original spinal level for TI and T, was determined within 0.2

mm by using the compact bone portion of the epiphyses of the

adjacent vertebral body. This region is visible on sagittal

MR images as a thin line (approx. 0.2 mm thickness) of very

low signal intensity separating the vertebral bone marrow

cavity from the adjoining intervertebral disk. The overall

length of the vertebral body remained constant throughout the

experiment for each of the five cats and, thus, it was

possible to obtain images from precisely the same spinal level

at intervals spanning several months. The mean pixel

intensity (measured from the raw image data) was then recorded

for a 1mm2 box centered over the graft site, and the T1 and

T2 values were computed from two-parameter exponential (non-

linear least-squares) fits using the computer program

Mathematica (Wolfram Research). For these curve fits, a

monoexponential model was assumed.


Qualitative Image Assessments

The surgery site was visible in all subjects as an area

in which the normal H-shaped region of gray matter and

adjacent white matter were replaced by a mass of relatively

homogeneous signal (e.g., Fig. 4-1; refer to Fig. 3-1 a for an

MR image of an intact spinal cord). This abnormal signal zone

was observed within hours of grafting or injury on the

operated side of the spinal cord and was present in all

animals throughout the duration of the experiment.


In order to obtain an accurate assessment of graft

survival, both T2-weighted images and intermediate T1-weighted

images were acquired (e.g., Fig. 4-1) at every sampling

interval. The T2-weighted images clearly demonstrated CSF-

filled compartments whereas the intermediate T1-weighted

images provided a superior signal-to-noise ratio and detailed

visualization of spinal cord anatomy.

As shown in Fig. 4-1 (top row), the transplant location

in four of the six cats (0A28, 9J24, 1L7, 1117, Table 4-1)

with surviving grafts at the time of sacrifice exhibited a

medium-to-hyperintense signal on intermediate T1-weighted

images for every postgrafting interval studied. In contrast,

the two remaining cats in this group (1L10, 1116) always

presented moderate-to-hypointense transplant zones.

Interestingly, none of these six cats exhibited a shift from

high to low signal intensity or vice versa during the entire


As a group, the six cats with viable fetal tissue also

presented a broad range of signal intensities on T2-weighted

images. However, the majority of cats in this group (four of

six) exhibited a signal intensity from the graft site on T,

images that was slightly brighter than the contralateral host

spinal cord (Fig. 4-1, bottom row). Postmortem histological

sections from equivalent spinal levels to those observed in

the MR images confirmed the presence of viable grafts (Fig.


V-O ,-4 a)W (D a >

44 :1- 14 4J "V (
4.1 0 4i-i M a) 10 a)

r_ o a ) -4

o *d 4' 0) a
C. 4 .0p .,>
m- a) 4j 4

4- 4j C W 4

0 II S-5,
a 0 > 4J .4 M

Wa ) (0
> ) 4 a u 4.I 4.).4c
-4 id

V 4J .4 -C 4- t. >
C 4 -Ht (0 40

0 4J W
o4 oo a) o
HU 4 c a)- a-4

-H00( f

C. 4 Z 0 4 U ) tM

t 0 ,11 1 E4JW5

o oW CID 0 4
4J 9 .,, m 4.)4
MOl E-4 p -

4 a- 44j c
.a 4 W 0 4 V 4. ) 0
a ) U4) 4 4 ( g
| 0)ao a 4)
: 4.) 0: 0 .Ja
(a ,"4 -A eq V M 3

o W.H a) g) .3H (0

4.4 44 4)-. .0.
0' %1a l, a rc)

0 1 0. O)
- 01, .0 tp w

H 45 *-' c c(v" (-0 4

4 a u 0 W = iw

.t CL to CL>oC

., owheo o ooxw
P4 P r 14> la M -4 4) A




op 7

c c >.+
*.H *HO q0 04
0.0. C *-

044 '

0) 0



0 t % O

CO .4

.5 tp C Li


4) .3)^
) 1 ''0 -
CA C 01

4O 0 14 CCO
> ra rd c 4


P> 0.0
) 0o 0 1 .3

Li r-4 *-4 10 4J
0 O' 0C
~..4c ) '4 C

1 i




In contrast to animals with surviving transplants, the

injured portion of the spinal cord in the remaining fetal

tissue recipients consistently exhibited a very bright

appearance on T,-weighted images (Fig. 4-3, bottom row). By

twelve weeks postgrafting, the signal from these surgical

sites was uniformly on par with the very intense image

exhibited by CSF. Postmortem gross and histological

examination of comparable spinal cord segments from this group

revealed that the bright lesion areas corresponded to cavities

that were filled with CSF and were largely devoid of neuronal

tissue elements (Fig. 4-4).

On intermediate T1-weighted images, the cats without

surviving grafts presented a progressive decrease in signal

intensity at the later postgrafting intervals (> 8 weeks; Fig.

4-3, top row). For example, during the first four weeks

postgrafting, the transplant region in five of six cats in

this group exhibited medium-to-hyperintense signal on TI

images (Fig. 4-3, top left). By 12 weeks, however, the graft

site in five of the six cats demonstrated a noticeable

decrease in signal intensity (Fig. 4-3, top right).

In comparison to both the cats with and without surviving

transplants, the lesion site in the control group also

exhibited a predominantly medium-to-hyperintense appearance on

intermediate T1-weighted images during the first eight

postoperative weeks (Fig. 4-5, top row). Although no changes

were observed in the control animals from eight to twelve

4- 0 1 (0 C M

0 a) i0 0
-r O I0 0

-4 0
*HC C00

0 4J0 0

~4 -, 0

oo P

So .~
40 ,4 .0
r04 4J 4
-H O4

'-- 4 )..

u.-0 0*

EH 0

r-4m .0

C 0 00-14 .

4- 0
4'O,-4 -4

4* -) a 0 -

O Nc 4 04 0
-4aCO 0g 2

4m' 0 .0

a.) 0.0 > 0
C 4) Vj C

s oo 0( o
0 C -4 H
r 4., 0 -4 J-4 44
o-4 >4 n c
4.) 00 01
r.- 4 jr
H r-4 -4
0 4.)41
4-1 4 to r


a 4-1





0 4
.m ,

4 1

C-H 4J


0) 0

41$ m0.

.0 C
$B4 .


on 5 1
SiN 8
O^i f

? 11

l< .
eP a A



-1 a



0 0)
'0 0

41 ,.
,H U ,



*H 00

S4-) M

4o 4o

> Q

-4 -ls 0
5-l 4

*^t +,1


*H a)4344
e r- M



weeks, MR images of the injury epicenter in the one control

animal that was studied for longer than 12 weeks appeared

hypointense relative to the host spinal cord at the 16 and 20

week postoperative intervals (data not shown).

In contrast to the intermediate T1-weighted images, the

three control animals exhibited a remarkably consistent

appearance on T2-weighted images. As shown in Fig. 4-5

(bottom row), all three cats presented an extremely bright

signal that was comparable to the intensity of CSF at all

postoperative intervals. At time points beyond eight weeks

postinjury, these markedly hyperintense images were similar to

those from cats with nonsurviving transplants, and were

noticeably brighter than T2 images of viable grafts (compare

bottom rows of Figs. 4-1, 4-3, & 4-5). Postmortem

histological sections from the lesion epicenters (e.g., the

spinal level shown in Fig. 4-5), verified the presence of

cysts in all control animals (Fig. 4-6).

Quantitative Analyses

To facilitate comparison of the MR images from transplant

recipients and control animals, the signal intensities

exhibited by the graft or lesion epicenters relative to

several anatomical landmarks (e.g., gray and white matter)

were estimated (Table 4-1). As shown in the far right column

of the table, six of the twelve graft recipients exhibited

viable transplants on postmortem gross and histological

examinations whereas no evidence of surviving fetal tissue was

40 n '0

o c
0 -4,
4J M 0)4V


0 .

oo to
S 4 4
5 0)

S0 0)4

00,c .

0 0

C, 4, V

0 20


-H .0
SgC0 0
"-4 0

41H'0 0

V 4 000)
0 0V
- U'.
4)U(a a-

E- 44 .

' 4-)
a) c
a m

4J 4-4
*^ ^



r 0


(0 0*

0) 0

tHo M
43 C Ul
0 0 W)

'-4-H X

10 0jr
W H"
0B >..0

.C0 44
0 0,


0 E '0
M 4-1
.0H a4-)


*J 40 0

42J0 40

0 0 t
a f



*H 4-I
-H 0'
"^l oi

4-1 0 4.)
0'-H 0B
-H 04
I 0n

^ as
0 (


o eH.0
m-r ,





0) ,4


0 t


9 4-'

4 *H )

Q) a-4

U U)
V0 0

.. >0


0a o

m .I- o
, 4

WU 0

Z )

) .0



3m 00 00 BO IwN 00 wci 0 ON 00 MN 1N
Sa Z N Z N 2 | >-4 H N >4 r-4 ZW iz-4 N rq. 1

O 'W N N l N ) NIn


Ln N Ln Ln NUm o I I mLn

. N ClW)N W) rNNW HN H N W)
H E -E4E-4E- E-4 E-4 E-4

r-l 4
O N to NW) 1N) n I.rm NWn cnq) 4 NW) NW) N WI N w
.. .. .... .... .
N .r N *4N W.-4r4l EI)N .4N 4NI -4N l"I N lN r.4N .lN

cw r mlWm )Wm I mW ) in mn cn NW) n lw NW N -

.-N N -4 B .-N .-N 4N -4N .N .-4 ,-..N
CE- E-4E E4 E- H4 E- P E E E4 E-4E4 H- H HE H H E3-43

%D EE-E4- E4 & 3 E4 -

aI 4r Wm Win Lo -d NW r4 W L (n mn nC

4 3-43 r-4 C43.3.
N! *-

Z4 4I N

m E-I H E EE-4N H H E- trN E-4 E-4
N E- 4EW 3-

4) l 10 .-4 N N-l N

4 = M -l (= = = = r-l -
"lfl (K H
u o n o' o o e -

$l -- 1
D CD $P ~ 1C4

N )


HN n


~E- E ,E

Q.N HN Hv N Ho N H)'d


E EE E-E P E-E-4

,- U "

,r r N

0 ,-I I oE
I --- --
2' H
I' M W

.- H C) h N
H H _^ 10 Qs H

. C 11

4J 'I I'C

) ,PH r-

4 1 1. 0
> oN
pM 0a
-4J V -

0 C 4 4 r.

r t
o >. 0

-A w

)> 1} 1 S
E r. 4 .,I
r-i I P-i-i

04J 3. n- ,

) 0
r. )
>-- C 4%.')

VdICd 14
O4-) >4 9

4 a) k
on -dO

C0 C0 UM a4
C 4 ) 1 4J &

N Uc) N t
uj 0 0) u


$4 4 WC
4uH Dd 1
L in *I 4 J E4 Cl

S4JMic -- 4
3-i c o *
( 'D) 0) a 0) *

I-4C *-ai
Bf >4J J

observed in the remaining six cats. Note that one animal

(0A28) developed two clearly distinct regions of signal

intensity on the MR images, which were subsequently identified

as a transplant and a cyst. Hence, these two regions were

treated separately when comparing cats with surviving (N=6)

versus failed (N=7) transplants.

To help identify any patterns or trends in appearance on

MR images among the three groups of animals (i.e., surviving

grafts, nonsurviving grafts and controls), the estimates of

signal intensity presented in Table 4-1 were plotted on a

graph as a function of the postoperative delay (Fig. 4-7).

The graphs in Fig. 4-7 appear to be consistent with the images

in Figs. 4-1, 4-3, and 4-5, in that there is considerable

overlap of all three groups on intermediate T1-weighted images

(upper graph), whereas cats with surviving transplants

generally appear less bright than the other two groups on Tz-

weighted images (lower graph). However, it is difficult to

visualize any temporal trends in signal intensity from the

scatter plots in Fig. 4-7. Hence, the average score (i.e.,

signal intensity) for each group was plotted as a function of

the postoperative delay (Fig. 4-8).

In order to validate the subjective image assessments

presented above and to provide an observer-independent index

of graft viability, a more quantitative analysis of the images

was explored (see Methods for details). Briefly, the

normalized mean pixel intensities were measured from the same

Figure 4-7. Scatter plots of the estimated transplant/lesion
site intensity.

Top graph) A plot of the intensity scores (from Table 4-1)
based on observations of intermediate T1-weighted images
through the transplant or injury epicenter. There is
considerable overlap between all three groups through the
first 12 weeks postoperative. At 16 and 20 weeks, injury
sites in controls and cats with failed transplants appear to
demonstrate a more hypointense image than at earlier
postgrafting intervals.

Bottom graph) A plot of the intensity scores (from Table 4-1)
based on observations of T2-weighted images through the
transplant or injury epicenter. The graft/lesion sites in
controls and in cats with failed transplants are predominantly
very bright (a score of 5) whereas the signals from the group
with surviving grafts appear to have more scattered scores.

T -weighted Images (TR=1000 ms, TE=30 ms)

So7 p7 7

- 8 8B

I I I I t I I I I I

0 2 4 6 8 10 12 14
Weeks Postoperative

16 18 20

T -weighted Images (TR=2000 ms, TE=90 ms)

5 -1E 1 f 8, o 7

4 4 k |0 80 0


2 0

0 2

0 0 0

4 6 8

O Surviving Grafts (N=6)
Nonsurviving Grafts (N=7)
V Controls (N=3)

10 12 14 16 18 20

0 Surviving Grafts (N=6)
* Nonsurviving Grafts (N=7)
V Controls (N=3)

* *


Figure 4-8. Average intensity scores for each group of
animals as a function of postoperative delay.

Top graph) The scores each of the three groups shown in the
top graph in Fig. 4-7 (i.e., for T1-weighted images) were
averaged and plotted as a function of postoperative delay. On
this graph, the overlap between all three groups through the
first 12 weeks postoperative is easily observed. Consistent
with Fig. 4-3 (top right) and with the chronic cysts observed
in Chapter 3, at 16 and 20 weeks the injury sites in controls
and cats with failed transplants present a more hypointense
image than at earlier postgrafting intervals.

Bottom graph) A plot of the average intensity scores for cats
in each of the three groups shown in the bottom graph in Fig.
4-7 (i.e., for T2-weighted images). Both the control cats and
graft recipients with failed transplants exhibited much
brighter images than the group with viable fetal tissue.

Ti-weighted Images (TR=1000 ms, TE=30 ms)

0 0 Surviving Grafts (N=6)
4.5 Non-Surviving Grafts (N=7)
V Lesion-Only Controls (N=3)

0 2 4 6 8 10 12 14 16 18 20
Weeks Postoperative

T2-weighted Images (TR=2000 ms, TE=90 ms)

5.0 -7- ------



3.5 -/0 -
SO Surviving Grafts (N=6)
3.0 Non-Surviving Grafts (N=7)

2.5 V Lesion-Only Controls (N=3)



0 2 4 6 8 10 12 14 16 18 20

Weeks Postoperative


areas presented in Table 4-1 and are summarized in Table 4-2.

Importantly, all of the subjective estimates were completed

prior to the pixel measurements to avoid any bias due to a

priori knowledge of the measured signal intensities. To allow

for a visual comparison between the subjective interpretations

and quantitative data from the MR images, the average mean

intensity for each group of cats is plotted in Fig. 4-9. By

comparing Figs. 4-8 and 4-9, it is evident that the average

estimated and measured signals on both T,- and T2-weighted

images are in general agreement for all three groups of cats.

In addition, statistical comparisons of the mean measured

pixel intensities were made for sampling intervals at which

four or more cats from each transplant group were imaged

(i.e., 4, 8, and 12 weeks postoperative). Utilizing a two-

tailed t-test, the signals from the two groups of graft

recipients on intermediate T,-weighted images were not

significantly different at any of these three time points.

However, the signal exhibited by viable graft tissue on T2-

weighted images was significantly lower than that of a failed

transplant at both 8 weeks (t= -6.72, p< 0.001) and 12 weeks

(t= -4.029, p< 0.005).

Another quantitative NMR analysis of the transplant or

lesion epicenters involved acquiring sequential images with an

array of either TE or TR, from which the T1 and T2 relaxation

times could be computed (see Methods for details). The

results of this experiment for five fetal tissue recipients

*' W H W W M W W M M

S00 00 0 W N 00 MN 0 ON 00 MN MN
m o ZN ZN lZ aJ q ZN 4 8 N 4> 4 x4
So00 If o
mm (LO n -WQ r. Ch i e0

t 00 00 00 OHI
0 Cl1N H N H N H- 4N
N E-4 P P E- P __ p p__ __
,nf 00n mr v o

S00 00 00 O0-

H4 E4E44i ______ P ___ P EP ___p __p
to %D NH v 0 in0 ( li W
%G c N0V0 oo o4 0 to t

S00 00 00 000 0000 OHi OHio Oo
H P &4 PE E- SEE4 < PPE-4 P PE H H 4E'H H EU4 P &4E
NO0 m'D N M. H mUn mMO Ho. o, o00 vJ mm im ON

W 00000 *OO oo oo 00 OH OH 000 00 00
rHN HN HIN ; r9-N HrNN HN HN rlN rlN rN
to Hr E E- H Z P P EP H H H4Z EE4 H EA P E1 P P
o\t: o o o

Ir 00 00H % w 00

0 O; C; C r o8o C C; C
4 .. .. .. .
M co o 0 oo oo
Hr. N Hr 4N HNMN N r HN
P E __ __E-4__ _1p E_ p_ p _P&__ EH_4E & E4 E4

nio >loH 'o O co 'ON DO 'Oa OW i NVO
U! 0 .0 .. a .

S00 wo 00 00 OH H00 0oz o00
? .. 4. 4J -- : .. .. .. ..4

H0 Ha rl N H N q HN fl
- EIE-4 __ p_ Ee4 E-PI EA E E E- EE
HID ).i H(1'D ifN O OWN

w< 00 -400 WOO 00 05 00

Nt N Hm P f Cn.

Hol N

WJ4 ID H( N N N N H lH H
UH Oi n o 0 On 0 o r0 cH H

on o W o
S w O I O I 0
__> 01 IU C C_
0M O C 0 N 0
x mC co U 1- V CMi U co

= = *

Figure 4-9. Mean measured pixel intensities for each group of
animals as a function of postoperative delay.

Top graph) The average normalized pixel intensity for each
group (summarized in Table 4-2) on intermediate T1-weighted
images is plotted as a function of postoperative delay. At
intervals where four or more animals from at least two groups
were imaged, no significant differences in signal between
groups were observed. Bars are one standard deviation.

Bottom graph) The average normalized pixel intensity for each
group (summarized in Table 4-2) on T,-weighted images was
plotted as a function of postoperative delay. The difference
in mean signal intensity between animals with surviving grafts
and those with failed transplants is significant at 8 weeks
(p< 0.001) and at 12 weeks (p< 0.005) postoperative. Note
that 4, 8 and 12 weeks were the only time points at which four
or more animals from both transplant groups were imaged. Bars
are one standard deviation.

T1-weighted Images (TR=1000 ms, TE=30 ms)

O Surviving Grafts (N=6)
* Non-Surviving Grafts (N=7)
V Lesion-Only Controls (N=3)


0 2 4 6 8 10 12 14 16

Weeks Postoperative

18 20

T2-weighted Images (TR=2000 ms, TE=90 ms)

1 .0

0.9 -



0.6 O Surviving Grafts (N=6)
0.5 Non-Surviving Grafts (N=7)
0.4 -7 Lesion-Only Controls (N=3)

0.3 -



0.0 1
0 2 4 6 8 10 12 14 16 18 20

Weeks Postoperative

*r4 1a km
> 0 M
--14 4-) M to M m 0A
14 m ac 9mA m> 1A
SM W N WN W w 0

Imm mm N
) I 3n f I'

0H 1 a a 1 61f
mn mn m mm mm mm

rI-l oun oV l 0 0I 0- 0
1r-i ONH NQ N r -I

C P E E P-4 E-4 P P4 E-P

m 0 am am m
to w o r oFA
;I 9n rn v N 141
a I ion 0l 'r-I ONH
00 00 H QQ O
z HHzH

E- P E-1 E-4 E P &4E-4 H E

r (MM O 101
Hco i-H H 0 M
M R ri r 0 rg a r
E-4 Pf EH EA E P EE^l P H

go a H 0 0 I o

N HEi & PiE E,

0 0
I 0 10 r 0

U H1 Hl H H-l H


and one injured-only control animal are summarized in Table 4-

3. Note that, except for a few instances, the T, times for

both the control animal and the cats with grafts were

predominantly between 1000 ms and 2000 ms. No obvious

differences were observed in T2 times either, with values

ranging from 48 ms to 174 ms in the transplant group and from

71 ms to 122 ms in the control cat.


These results indicate a strong correlation between

transplant survival and a moderately hyperintense signal on

T2-weighted images. Further, the signals on T2-weighted images

of surviving and nonsurviving fetal transplants are

significantly different by 8 weeks postgrafting and may be

differentiated as early as 4 weeks postgrafting. However, it

should be noted that these results were based upon groups of

animals and that there was some overlap, in which a surviving

graft might be mistaken as a cyst (i.e., a false negative).

Importantly, there was no overlap in the other direction,

i.e., no false positives. The one false negative that was

observed was found to be due to small cysts (< Imm diameter)

situated adjacent to the transplant. Thus, they were volume

averaged into the same 2 mm thick MRI slice with transplant

tissue, which resulted in a very bright image since T2 images

are very sensitive to CSF.

In contrast to the results presented for chronically

surviving grafts in Chapter 3, intermediate T1-weighted images


were of little use in distinguishing surviving grafts from

either failed transplants or lesion-only controls during the

first 12 postoperative weeks. Since these images are more

sensitive to tissue elements than T2 images, these results

suggest that inflammation and phagocytic debridement of CNS

injuries occurs over an extended period of time (up to 3

months). Collectively, these data are consistent with the

known longitudinal histopathology of spinal cord injury. For

example, areas of cavitation are present at roughly 3 weeks

following a traumatic injury and phagocytosis may be observed

in the adjacent tissue stumps for up to 10 weeks postinjury

(Das, 1989; Wagner et al., 1978; Noble and Wrathall, 1985;

Bresnahan, 1978). Further, the trend toward a more

hypointense image in animals with failed transplants and

controls at 16 and 20 weeks postoperative are consistent with

the hypointense chronic cysts observed in Chapter 3. Thus, a

dynamic continuum has now been established whereby MRI can be

used reliably to detect the survival of transplanted fetal CNS

tissue at virtually all time points from 4 weeks to over 2

years postgrafting.

This study also highlighted the capacity of MRI for

repeated in vivo observations of individual subjects. For

example, cats in this study were imaged as many as 9 times

over the 20 week duration of the experiment. Significantly,

it was feasible to relocate the lesion epicenter to within 0.2

mm at subsequent sampling intervals spanning several months.

The only adverse component of the MR imaging paradigm for

experimental animals is the necessity for administration of

general anesthesia to ensure that the cats remain motionless

(aside from breathing) during imaging. However, anesthesia

was maintained in this study with isoflurane, which has a high

degree of safety and is rapidly expired following cessation of

delivery to the subject. Thus, animals regained consciousness

within minutes after removal from the magnet and required a

minimum of postimaging care.

Quantitative Analyses

The quantitative approaches presented in this study

represent an important first attempt to validate the

subjective ranking of signal intensities presented by

transplanted neural tissue. Although only one observer (EDW)

evaluated the images, the results illustrated in figures 4-8

and 4-9 show a strong link between the subjective and

objective evaluations. Thus, these data suggest that graft

survival can be determined reliably through visual inspection

of the MR images.

Two potential caveats, however, should be noted for the

observed signals on the MR images. First, the estimates in

Fig. 4-8 show more variability from one sampling interval to

the next than the measured mean pixel intensities,

particularly at time points where only one animal from a given

group was imaged (e.g., nonsurviving grafts at 6 weeks).

Second, subjective interpretations of the MR images are highly


observer dependent. Thus, evaluations of graft survival and

overall lesion anatomy from a multiple slice MR image set

could generate considerable discrepancies between observers.

In this regard, measures of mean pixel intensities could

provide an observer-independent index of transplant survival.

These measurements also allow a group of subjects to be

analyzed with parametric statistical methods. Hence, a preset

range (e.g., two standard deviations) could be used to semi-

automate the process of determining transplant survival.

However, the regions) used for these measurements would still

have to be visually identified at least once, i.e., at the

first sampling interval. It should also be noted that

although signal intensity is fairly stable over time, it may

be significantly decreased when multisection and/or multiecho

imaging methods are used (Fitzsimmons and Googe, 1985; Slone

and Fitzsimmons, 1987).

T, and T Relaxation Times

An attempt was made in this study to measure the T1 and

T2 relaxation times from the graft/lesion site to assess

whether these parameters could provide an additional

quantitative method for determining transplant survival. For

example, fetal CNS tissue has very long relaxation times,

hence, if a graft survives and matures it is expected that the

relaxation times would converge with those of adult gray

matter (Vanderknaap and Valk, 1990; Holland et al., 1986).

Alternatively, if the transplant perishes, then the resulting

dissolution should produce increased T1 and T2 values that

would more closely approximate those of cerebrospinal fluid.

However, no difference in the TI and T2 times between

surviving transplants and a spinal cord injury without fetal

tissue was observed in this study. The absence of a

significant difference in TI times is consistent with the

similar signal intensities exhibited by viable graft tissue

and cysts on the intermediate T1-weighted images. In

contrast, the extremely bright signal presented by cysts on

T2-weighted images was not matched by a large measured T2 from

the one control animal for which this type of data was

collected. The most probable explanation for this paradoxical

result is that adjacent areas of cerebrospinal fluid and

neural tissue were volume averaged in the same 2 mm thick MR

image, which has been proposed as the cause of large

variability in the reported T1 and T2 times of various CNS

regions (Kjos et al., 1985). Indeed, the histological

specimens showed that this animal had a much smaller lesion

than those presented in Figs. 4-4 and 4-6.

Present Limitations of MR Imaging

The primary limitation observed in this study was the

available signal-to-noise and, hence, spatial resolution.

Thus, determination of graft survival was hampered by volume

averaging artifacts, in which signals from both transplanted

fetal tissue and adjacent structures (e.g., CSF-filled

cavities) were superimposed in the same MR image. These

finding are consistent with the study in Chapter 3, in which

MRI allowed gross visualization of chronically surviving

intraspinal transplants. However, there was insufficient

sensitivity to observe more subtle anatomical details, such as

host-graft approximation.

Accordingly, the feasibility of implanting rf coils

around the transplant site was tested in two animals as part

of the study in Chapter 5. It was found that the MRI coils

could be easily affixed to the spine immediately following the

transplantation procedure without disturbing the grafted

region of the spinal cord. Although the coils failed after 2

and 8 weeks postoperative due to broken solder joints (see

Chapter 5 for details), they generated much higher quality

images than the single-turn surface coil used for this study.

Future Directions

One aspect of the imaging paradigm that should be

addressed is the optimal spin-echo pulse sequence to use for

imaging the spinal cord. As observed in this study and in

Chapter 3, T1-weighted images provide excellent illustrations

of spinal cord anatomy whereas T2-weighted images were more

sensitive to regions with large quantities of free water.

Although T,- and T2-weighted images have their unique

advantages, proton density weighted images also provide good

visualization of spinal cord anatomy (these images were

obtained as part of the series for measuring the T1 and T2

relaxation times). In addition, it is fairly simple to

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E4TZTBJB8_UB1A5G INGEST_TIME 2012-09-13T17:55:41Z PACKAGE AA00011806_00001