ISCHAEMIA-REPERFUSION INJURY OF THE PERIPHERAL NERVE:

AN EXPERIMENTAL STUDY
AYDIN SARAY, M.D.,1* BELGIN CAN, M.D.,2
FILIZ AKBIYIK, M.D.,3 and IBRAHIM ASKAR, M.D.4

Although the neuropathology of ischaemic fibre degeneration more severe and prominent with reperfusion, indicating that
is relatively well known, its pathogenesis is poorly under- oxidative stress damages the BNB and causes IFD. Results
stood. One of the presumed mechanisms is oxidative stress, of functional testing by the sciatic function index correlated
causing the breakdown of the blood-nerve barrier (BNB) and with other parameters as walking track analysis results got
ending in lipid peroxidation. We evaluated the effect of isch- worse as reperfusion periods increased. Impairment of walk-
aemia and reperfusion on the sciatic-tibial nerve of the rat ing patterns was more striking after the first day and contin-
and investigated the biochemical, pathological, and func- ued up to the third week. These data indicate that severe
tional evidence of BNB disruption and lipid peroxidation. The ischaemia of the peripheral nerve results in reperfusion in-
distal portion and trifurcation of the sciatic nerve were ren- jury, functional impairment, and disruption of the BNB. Mi-
dered ischaemic by clamping the femoral vessels for 3 h and crovascular events, which may occur during reperfusion,
followed by varying durations of reperfusion. Reperfusion re- may be important in amplifying the nerve fibre degeneration
sulted in an increase in lipid peroxidation beginning from the that initiated during ischaemia.
first hour and increasing until the seventh day, followed by a
gradual decline over the following weeks. Nerve oedema and © 1999 Wiley-Liss, Inc.
ischaemic fibre degeneration (IFD) consistently became MICROSURGERY 19:374–380 1999

Studies in recent years have shown that restoration of the eral nerve. The goal was to facilitate the creation of a re-
blood supply after a period of ischaemia not only permits producible model in the rat sciatic nerve in which previous
survival of cells which have survived despite ischaemia but attempts are modified.
may also result in further tissue injury.1,2 Many organ sys-
tems including the brain, heart, viscera, and muscle that
have been subjected to ischaemia and reperfusion have MATERIALS AND METHODS
demonstrated wide variations in sensitivity.3–5 Although
much is known, the precise pathophysiology of ischaemia- A total of 220 male Albino rats weighing 250–275 g
reperfusion injury in the peripheral nerve remains to be were used in a series of experiments. Rats were divided into
elucidated. 11 groups: two control groups (positive and negative con-
A few studies have used several methods such as liga- trols) and nine experimental reperfusion groups. Each group
tion or embolization of the abdominal aorta, common iliac, included 20 rats. The rats were anaesthetized with intraper-
and femoral arteries to create an ischaemia-reperfusion itoneal pentobarbital (50 mg/kg). Right femoral vessels
model of the peripheral nerve in small animals.6–9 The rat were exposed through an inguinal incision and were dis-
sciatic nerve is a versatile model for studying the effects of sected free from the femoral nerve under a microscope. The
ischaemia and reperfusion. The present study describes the
trifurcation of the right sciatic nerve into the peroneal, tibial,
histopathological, biochemical, and functional conse-
and sural branches was rendered almost completely isch-
quences of the ischaemia-reperfusion injury of the periph-
aemic by occluding the femoral artery and vein with a vas-
cular clamp for 3 h.10 Three hours were selected as the
1
Department of Plastic Surgery, Mersin University, Mersin, Turkey critical ischaemia duration in accordance with a previous
2
Department of Histology and Embryology, Medical Faculty of Ankara Univer-
study.11 In reperfusion groups, the vascular clamp was re-
sity, Ankara, Turkey moved after 3 h of ischaemia and the sciatic nerves were
3
Department of Biochemistry, Hacettepe University School of Medicine, An- allowed to reperfuse with increasing durations (Table 1). In
kara, Turkey control groups, group I was the normal nerve with no isch-
4
Department of Plastic and Reconstructive Surgery, Medical Faculty of Dicle aemia and no reperfusion; group II was the ischaemia-only
University, Diyarbakir, Turkey
group with no reperfusion. The left sciatic nerves in all
*Correspondence to: Aydin Saray, M.D., Mersin University, Department of
Plastic Surgery, Eski Otogar Yani, 33070, Mersin-ICEL, Turkey
groups served as controls of the contralateral sciatic nerve
E-mail: [email protected] for functional evaluation.
© 1999 Wiley-Liss, Inc.
 Ischaemia-Reperfusion of Peripheral Nerve 375

Table 1. Experimental Groups 5 min. Following this, 3 mL of 0.5% cadmium acetate was
Duration of ischaemia added, mixed, and centrifuged at 2,500 rpm for 10 min. The
Groups (3 h) Reperfusion top layer was read at an absorbence of 353 nm. Lipid hy-
I No ischemia No reperfusion droperoxides were expressed as Abs353/g wet weight and
II + No reperfusion compared to a standard curve (cumene hydroperoxide).
III + +1 h
IV + +2 h Functional Evaluation
V + +3 h
VI + +12 h The animals were observed two to three times a week to
VII + +24 h identify any wound problems or foot ulcers. Animals were
VIII + +48 h followed by serial gait testing preoperatively and at 1-week
IX + +7 days intervals throughout the course of the experiment. Exami-
X + +14 days
nation of the walking patterns was performed, as has been
XI + +21 days
previously described as a method of assessing nerve lesions
and modified by de Medinaceli et al. was performed.15,16
Histological Preparation Animals were allowed conditioning trials on a 8.2 × 42-cm
walking track darkened at one end. Original technique was
Nerves were fixed in situ using 2.5% glutaraldehyde in
slightly modified at this point and a blotting paper cut to the
0.1 mol/L phosphate buffer (pH 7.2) at +40°C for 2–4 h.
appropriate dimensions was placed in the bottom of the
The entire length of the sciatic nerve with its branches was
track; the animals’ hind feet were dipped in Indian ink and
harvested; the distal sciatic and its trifurcation into tibial,
the rat was permitted to walk down the track. The film was
peroneal, and sural branches were separately harvested.
coded by animal number with no reference of the experi-
Nerve segments were postfixed in 1% osmium tetroxide,
mental group to allow for blind measurement of the walking
dehydrated in serial alcohols, and embedded in Araldite
track. For a permanent record, the tracing was photo-
6005 (Ciba-Geigy, Summit, NJ). Transverse sections of 1
graphed.
␮m were cut from the distal sciatic nerve, stained with
Measurements of print length (PL), toe spread (TS), and
toluidine blue-azur II, and examined under a Zeiss Ax-
intermediary toe spread (IT) were measured from the coded
ioskop photomicroscope (Thornwood, NY). Ultrathin sec-
films for both experimental (prefix E) and normal (prefix N)
tions were cut on an LKB III ultramicrotome (Bromma,
sides by a blinded investigator. The sciatic function index
Sweden), stained with uranyl acetate-lead citrate, and ex-
(SFI) was calculated according to the following formula16:
amined on a JEOL 100 transmission electron microscope
(Peabody, MA). SFI ⳱ −38.3 [(EPL-NPL)/NPL] +
109.5 [(ETS-NTS)/NTS] + 13.3 [(EIT − NIT)/NIT] − 8.8.
Biochemical Preparation
Tracks from unmeasurable cases, resulting from fixed joint
Lipid extracts for hydroperoxides were prepared by the contractures and self-mutilation, were excluded from analy-
method adapted by Nagamatsu et al.9 from the method of sis. All rats underwent walking track assessment preopera-
Folch et al.12 for peripheral nerve tissue. The un-desheathed tively for baseline and then at 1, 2, and 3 weeks.
neural tissue was homogenized in 0.5 mL ice-cold 0.9 %
sodium chloride. A 0.4-mL aliquot of homogenate was Statistics
transferred to 4.00 mL chloroform-methanol (2:1 v/v). This Statistical evaluations were carried out using the re-
mixture was vortexed for 1 min then centrifuged at 2,500g. peated measured analysis of variance with multiple com-
The lower layer was removed for evaporation. The upper parisons of means used for statistical analysis. A Bonferoni
layer was washed with chloroform-methanol-H2O (86:14:1 adjustment of the alpha level was done to take into account
v/v/v) and centrifuged. The lower layers were combined and the number of multiple comparisons.
evaporated under nitrogen. The lipids were then dissolved
in 0.5 mL chloroform-methanol (2:1, v/v) and stored at
−80°C until analysis. RESULTS
Lipid Peroxides
Determination of Hydroperoxides
Ischaemia was maintained for 3 h followed either with-
Lipid hydroperoxides of trifurcation of the sciatic nerve out reperfusion or with reperfusion times of 1, 2, 3, 12 h, 1
were measured by the modified iodometric method of day, 2 days, and 1, 2, or 3 weeks. Nerve hydroperoxides
Buege and Aust13 and Low and Nickander.14 The lipid ex- were measured and expressed as nanomoles per milligram
tract was dried under nitrogen, mixed with 1 mL acetic wet weight (Fig. 1). Reperfusion resulted in a large and
acid-chloroform (3:2, v/v), and then 50 L 1.2 g/mL potas- significant increase in hydroperoxides when compared with
sium iodide. The tube was capped and kept in the dark for ischaemic-nonreperfused and normal nerves. Maximal val-
376 Saray et al.

Figure 2. Variation of SFI with time in the experimental groups. The

value 0 denotes normal function and minus scores indicate func-
Figure 1. Nerve hydroperoxide levels of the experimental groups. tional deterioration.

ues were seen after 1 h of reperfusion and elevated levels

were observed in the following 3 weeks. Hydroperoxide
levels were higher in the first week and then tended to
decrease gradually.

SFI Determinations
All reperfusion groups and ischaemia-only groups
showed statistical deterioration in sciatic nerve function (P
< 0.001) as a function of time. The serial changes in SFI are
depicted in Figure 2. Preoperatively, the SFI in all groups
was approximately 0, indicating normal function. For ref-
erence purposes, the control group is compared with each
experimental group. Within each group, the left, unoperated
foot is referred to as the normal foot, whereas the right foot
is designated as the experimental group. Within the control Figure 3. Group I. Normal nerve. Toluidin blue-Azur II. Scale bar =
group, all print parameters for both feet were normal and 100 µm ×250
tended to increase. Following 3 h of ischaemia in group II,
the SFI decreased from normal function (SFI near 0) to −24
± 3.0 at first week, 19 ± 3.7 at 2 weeks, and 16 ± 2.2 at 3 changes such as myelin breakdown appeared in group III
weeks. The most deteriorated functional results were ob- (Fig. 5). Related changes of myelin, such as irregularities,
served in group III (3 h ischaemia:1 h reperfusion) with −29 vacuolisation, appearance of lamellar bodies, and mild to
± 4.1 at the first week. Elevated values of 27 ± 2.2 and 23 moderate demyelination, were observed in groups V to XI
± 1.3 remained at the second and third weeks, respectively. (Figs.6–11). Also, group V nerves showed erythrocyte plug-
The SFI in the remaining groups (IV to XI) demonstrated ging in perineurial and endoneurial vessels (Fig. 6). Ultra-
deteriorated function and minimal recovery, with SFI rang- structural examination of group V showed axonal degenera-
ing from −-28 ± 4.2 to −20 ± 2.1. All groups showed an tion, shrunken axons, vacuolization of axons, endoneurial
improvement in sciatic nerve function by 3 weeks. Improve- oedema, and erythrocyte plugging in endoneurial vessels
ment in function reached a plateau by 3 weeks and was (Figs. 7, 8). The constant pathological features in reperfu-
maintained at values near −20. sion groups were axonal degeneration showing swollen wa-
tery axons, dark axons, and shrunken axons which were
Histological Assessment noted alongside myelin loss (Figs. 5–11). After 1 week of
Normal nerve exhibited normal findings indifferent reperfusion, degenerative changes were observed in
from routine examinations (Fig. 3). Ischaemic nerves with- Schwann cells while axonal degeneration persisted (Figs.
out reperfusion showed endoneurial oedema and many fi- 9–11). Group XI demonstrated removal of necrotic debris,
bres had undergone ischaemic changes such as axonal exudate, remnants of degenerated reticulum fibres, periaxo-
shrinking and swelling (Fig. 4). Histological findings were nal coagulated materials, and capillary dilatation (Fig.
not uniform and with increasing reperfusion, pathological 10A,B). Leukocyte plugging was not noted in epineurial or
 Ischaemia-Reperfusion of Peripheral Nerve 377

Figure 4. Group II. Ischaemic nerves without reperfusion. Endoneu- Figure 7. Group V. Ultrathin section of ischaemic nerves after reper-
rial oedema, axonal shrinking (arrow), and swelling (arrowheads) are fusion of 3 h. Axonal degeneration, shrunken axons (arrow), dark axons
seen. Toluidin blue-Azur II. Scale bar = 100 µm ×250. (arrowheads), vacuolisation of axons (v), and endoneurial oedema
(star) are seen. Uranyl acetate-lead citrate. Scale bar = 5 µm ×3,600.

Figure 5. Group III. Ischaemic nerves after reperfusion of 1 h. En-

doneurial oedema, axonal degeneration, and myelin breakdown (ar- Figure 8. Group V. Ultrathin section of ischaemic nerves after reper-
rows) are seen. Toluidin blue-Azur II. Scale bar = 100 µm ×250. fusion of 3 h. Erythrocyte plugging in vessels (arrow) and perivas-
cular oedema (star) are seen. Uranyl acetate-lead citrate. Scale bar
= 5 µm ×4,800.

Figure 6. Group V. Ischaemic nerves after reperfusion of 3 h. Myelin

breakdown (arrows), axonal vacuolisation (arrowhead), and erythro- Figure 9. Group IX. Ischaemic nerves after reperfusion of 1 week.
cyte plugging (double arrow) in vessels are seen. Toluidine blue- Axonal degeneration and endoneurial oedema persist. Toluidine
Azur II. Scale bar = 100 µm ×250. blue-Azur II. Scale bar = 100 µm ×250.
378 Saray et al.

Figure 11. Group X. Ischaemic nerves after reperfusion of 2 weeks.

A mast cell is seen with its granules (arrow). Toluidine blue-Azur II.
Scale bar = 100 µm ×250.

duced functional, biochemical, and histological evidence of

ischaemic injury following 3 h of ischaemia. Reperfusion
failure was evident in the peripheral nerve after 1 h and
especially after 3 h of ischaemia. With increasing duration
of reperfusion, indirect evidence of inadequate restoration
of nerve blood flow was encountered as this finding was
shown in previous studies.9,18
The large increase in lipid hydroperoxides and histo-
logical evidence of ischaemic axonal degeneration and sig-
nificant endoneurial oedema strongly suggest that oxidative
stress leads to breakdown of the blood-nerve barrier (BNB),
resulting in endoneurial oedema. Indirect evidence of BNB
damage by oxidative stress, secondary to reperfusion, was
Figure 10. A,B: Group XI. Ischaemic nerves after reperfusion of 3 previously shown as an increase in the permeability surface
weeks. Necrotic debris, exudate (star), periaxonal coagulate material area product (PA).18 PA is a product of the permeability
(arrowheads), and capillary dilatation (arrow) are seen. Toluidine coefficient and the surface area of the capillary. Since nerve
blue-Azur II. Scale bar = 100 µm ×250. blood flow, an indicator of capillary surface area, was fur-
ther reduced due to reperfusion, the increase in PA was a
endoneurial vessels but only mast cells were observed in the true reflection of an increase in the permeability coeffi-
14-day reperfusion group (Fig. 11). cient.18 Nagamatsu et al.9 later showed that nerve blood
flow was reduced in postischaemic reperfused nerves which
persisted until 7 days of reperfusion. They have also pro-
DISCUSSION
vided complementary biochemical and histological evi-
There are inherent problems in the study of peripheral dence on the detrimental effects of ischaemia-reperfusion.
nerve ischaemia. Unlike the brain and heart, where effects Our neuropathological findings were similar to those
of ischaemia are easily produced, peripheral nerve is rela- observed by Nagamatsu et al.9 but were different in several
tively resistant because of its low energy demands and ex- aspects from those reported by Nukada et al.8,9 Nukada et
tensive anastomoses.17 Thus, early attempts to create ex- al. noted severe demyelination. According to Nagamatsu et
perimental ischaemia were mainly unsuccessful. Recently, al., the earliest changes in the peripheral nerve were axonal.
several models of ischaemia have been developed.6–9 Our However, the preliminary and predominant change in our
proposed model can be a versatile and simple modification study was endoneurial oedema which was later accompa-
of similar studies based on Bell’s study.11 Our modified nied by axonal degeneration and eventual appearance of
model of severe nerve ischaemia, produced by occluding mild to moderate demyelination. Nagamatsu et al. noted
femoral vessels with a vascular clamp, consistently blocked increased axonal degeneration with prolonged duration of
nerve blood flow as confirmed by the results. Following the ischaemia, rather than myelin changes. Experimental de-
footsteps of previous studies, this model consistently pro- signs were different but this cannot be the unique explana-
 Ischaemia-Reperfusion of Peripheral Nerve 379

tion. In Nukada et al.’s study, nerves were rendered isch- injury may be due to generation of oxygen-free radicals that
aemic for 7 h, which is longer than our ischaemia time and damage the endothelial barrier and, subsequently, endoneu-
that of Nagamatsu et al. However, 75% of the nerves sub- rial contents. The mechanisms are likely to be similar to
jected to 5 h of ischaemia in Nagamatsu et al.’s study dem- those involved in the heart, gut, and brain but should be
onstrated axonal degeneration without manifesting demy- modified in terms of the threshold of each tissue to isch-
elination. Moreover, Nukada et al. utilized a longer period aemic and reperfusion damage. In this study, we aimed to
of reperfusion after longer ischaemia so they might have create a peripheral nerve ischaemia-reperfusion model in
observed the longstanding changes that took time to evolve. which the exact mechanisms of injury can be elucidated and
Nagamatsu et al. have studied reperfusion until 7 days and strategies for overcoming the adverse effects of reperfusion
have not observed demyelination. After 3 h of ischaemia, injury can be developed. The peripheral nerve is more re-
we allowed reperfusion with different durations, also be- sistant to ischaemia and ischaemic injury ensues after many
yond 1 week, and we observed demyelination in varying hours, thus permitting the amelioration of detrimental ef-
degrees. This may explain why Nagamatsu et al. have not fects if correct treatment is instituted on time.
observed similar changes during reperfusion for 1 week.
Direct and indirect evidence of ischaemic conduction
failure in the peripheral nerve has been demonstrated fol-
lowing occlusion of the vasa nervorum.19 Again, Schmelzer REFERENCES
et al.18 have shown that persistent conduction failure on
1. Weisfeldt ML. Reperfusion and reperfusion injury. Clin Res 1987;35:
proximal stimulation and the progressive reduction in com- 13–20.
pound muscle action potential on distal stimulation indicate 2. Kerrigan CL, Stotland MA. Ischemia reperfusion injury: a review.
that the electrophysiological abnormalities were likely due Microsurgery 1993;14:165–175.
3. Heiss WD, Rosner G. Functional recovery of cortical neurons related
to axonal degeneration rather than transient conduction to degree and duration of ischaemia. Ann Neurol 1983;14:294–301.
block. A prominent deterioration of function, using the SFI, 4. Shine KI, Douglas AM. Low calcium reperfusion of ischemic myo-
was noted. Analysis of walking tracks using the SFI has, in cardium. J Mol Cell Cardiol 1983;15:251–260.
5. Hickey MJ, Hurley JV, Angel MF, O’Brien BMcC. The response of
the past, been a highly useful method for determining re- the rabbit rectus femoris muscle to ischemia and reperfusion. J Surg
covery of function following crush injury and surgical re- Res 1992;53:369–377.
pair of rat sciatic nerves.16,20–22 The near-normal return of 6. Korthals JK, Wisniewski HM. Peripheral nerve ischaemia. Part 1.
Experimental model. J Neurol Sci 1975;24:65–76.
function in group II (ischaemic-only) after 4 weeks was also 7. Hess K, Eames RA, Darveniza P, Gilliat RW. Acute ischemic neu-
not surprising. This is a neurapraxia-type injury and proved ropathy in the rabbit. Neurol Sci 1979;44:19–43.
that the endoneurial tube maintains physical continuity and 8. Nukada H, Powell HC, Myers RR. Spatial distribution of nerve injury
after occlusion of major vessels in the rat sciatic nerves. J Neuropathol
that near-normal return of function is possible. In the pre- Exp Neurol 1993;52:452–459.
sent study, the SFI indicated functional deterioration in the 9. Nagamatsu M, Schmelzer JD, Zollman PJ, Smithson IL, Nickander
ischaemic nerve and functional results worsened as dura- KK, Low PA. Ischemic reperfusion causes lipid peroxidation and fiber
degeneration. Muscle Nerve 1996;19:37–47.
tions of reperfusion increased. SFI values in groups III–XI 10. Sladky JT, Greenberg JH, Brown MJ. Regional perfusion in normal
deteriorated to values near −29 and showed improvement by and ischemic rat sciatic nerves. Ann Neurol 1985;17:191–195.
reaching values near −20 by the third week. Functional 11. Bell MA, Weddell AGM. A descriptive study of the blood vessels of
the sciatic nerve in the rat, man and other mammals. Brain 1984;107:
correlations of histopathological peripheral nerve injury 871–898.
were also not surprising. At 3 weeks, there was a lack of 12. Folch J, Lees M, Sloane-Stanley GA. A simple method of total lipid
statistical difference between the experimental groups III– extraction and purification. J Biol Chem 1957;226:497–509.
13. Buege JA, Aust SD. Microsomal lipid peroxidation. Methods Enzymol
XI in which reperfusion was allowed. It seems reasonable to 1978;52:302–310.
comment that detrimental effects of reperfusion sustained 14. Low PA, Nickander KK. Oxygen free radical effects in sciatic nerve in
beyond the third week and studies with prolonged observa- experimental diabetes. Diabetes 1991; 40:873–877.
15. Hruska RE, Kennedy S, Silbergeld O. Quantitative aspects of normal
tion durations are warranted. locomotion in rats. Life Sci 1979;25:171–180.
The effects of ischaemia are amplified in several tissues 16. De Medinaceli L, Freed WJ, Wyatt RJ. An index of the functional
during reperfusion, a phenomenon referred to as reperfusion condition of rat sciatic nerve based on measurements made from walk-
ing tracks. Exp Neurol 1982;77:634–643.
injury or reduced flow.23–25 Our results suggest that a reper-
17. Lundborg G. Intraneural circulation. Orthop Clin North Am 1988;19:
fusion phenomenon also exists in the peripheral nerve if 1–11.
ischaemia is near-total and of long duration, since this con- 18. Schmelzer JD, Zochodne DW, Low PA. Ischemic and reperfusion
dition occurs after 3 h of occlusion. A few studies have injury of rat peripheral nerve. Proc Natl Acad Sci USA 1989;86:1639–
1642.
reported similar histopathological results.8,26–28 The mo- 19. Parry GJ, Linn DJ. Transient focal conduction block following experi-
lecular mechanisms underlying reperfusion injury in the pe- mental occlusion of the vasa nervorum. Muscle Nerve 1986;9:345–
ripheral nerve remain to be elucidated. Based on the pro- 348.
20. Chen LE, Seaber AV, Glisson RR, Davies H. The functional recovery
gressive and sustained increase in lipid peroxidation prod- of peripheral nerves following defined acute crush injuries. J Orthop
ucts and histopathological results, we propose that nerve Res 1992;10:657–664.
380 Saray et al.

21. De Medinaceli L, Seaber AV. Experimental nerve reconnection: im- 25. Hyslop AP, Hinshaw DB, Halsey WA, Schraufstatter IU, Sauerheber
portance of initial repair. Microsurgery 1989;10:56–70. RD. Mechanisms of oxidant mediated cell injury. J Biochem 1988;
22. Bain JR, Mackinnon SE, Hunter DA. Functional evaluation of com- 263:1655–1675.
plete sciatic, peroneal, and tibial nerve lesions in the rat. Plast Reconstr 26. Nukada H, Dyck PJ. Microsphere embolization of nerve capilleries
Surg 1989;83:129–138. and fiber degeneration. Am J Pathol 1984;115:275–287.
23. McCord JM. Oxygen-derived free radicals in post-ischemic tissue in-
jury. N Engl J Med 1985;312:159–163. 27. Nukada H, McMorran PD. Perivascular demyelination and intramy-
24. Angel MF, Ramasastry SS, Swartz WM, Basford RE, Futrell JW. Free elinic edema in reperfusion injury. J Anat 1994;185:259–266.
radicals: basic concepts concerning their chemistry, pathophysiology, 28. Nukada H, Powell HC, Myers RR. Perineural window: demyelination
and relevance to plastic surgery. Plast Reconstr Surg 1987;79:990– in nonherniated endoneurium with reduced nerve blood flow. J Neu-
997. ropathol Exp Neurol 1992;51:523–530.