After immersion of the hind limb of the rabbit, up to the lower thigh, in a waterbath, at 1°C for 10 to 14 hours under light anesthesia, there was evidence of persistent nerve damage to the tibial nerve, which varied in severity in different animals. Nerve conduction studies, carried out within 24 hours of removal from the bath, showed that in a proportion of the motor and/or afferent fibers, there was conduction failure between the knee and ankle. This was followed, over the next 48 hours, by distal degeneration of the affected fibers. No persistent conduction block was seen. After distal degeneration had occurred, maximal conduction velocity was mildly reduced, suggesting that the fastest-conducting motor and afferent fibers had been particularly affected. Morphological studies confirmed preferential large myelinated fiber degeneration, the earliest lesions being seen in the leg at the level of the upper calf. Limb edema was not seen after cooling, and there was no histological evidence of muscle necrosis or damage to blood vessels. No abnormalities were seen in 4 control animals after hind limb immersion for 12 hours at temperatures of 30 to 35°C. Possible reasons for the proximal site of myelinated nerve fiber damage during hindlimb cooling are discussed. Key words: cold injury, non-freezing injury rabbit tibial nerve conduction studies axonal degeneration MUSCLE & NERVE 14~960-967 1991
NERVE CONDUCTION STUDIES IN EXPERIMENTAL NON-FREEZING COLD INJURY: 11. GENERALIZED NERVE COOLING BY LIMB IMMERSION ROBIN P. KENNETT, BSc, MD, MRCP, and ROGER W. GILLIATT, DM, FRCP
T h e nerve changes in non-freezing cold injury (trench foot, immersion foot) are not well understood. Clinical experience, mainly gained from casualties during two world wars, indicates that prolonged limb cooling in water just above freezing point can produce peripheral nerve damage as a part of a more widespread syndrome, which usually includes limb edema, skin blistering, and vascular damage leading to necrosis and tissue i0ss.23.24 Experimental studies in animals have confirmed that cold exposure can lead to peripheral From the University Department of Clinical Neurology, Queen Square, London, England. Presented in part to the Physiological Society, London, March 1987. Acknowledgments: This work was financially supported by the Action Research for the Crippled Child, and forms the substance of a MD thesis of the University of London (Dr. Kennett) Address reprint requests to Dr. Kennett, The Regional Neurological Centre, Newcastle General Hospital, Westgate Road, Newcastle upon Tyne, NE4 6BE, England. Accepted for publication August 6, 1990.
CCC 0148-639)(/91/0100960-08 $04.00 0 1991 John Wiley & Sons, Inc.
960
Non-Freezing Cold Injury
nerve degeneration with or without changes in other tissues.3,4,12-14,18,20,22 In other experimental studies, local cooling of an exposed nerve trunk has been used to produce the nerve lesion without the risk of accompanyin changes in muscle, skin, or major blood vessels.21,17,2 1 In a previous study, we described the effect on the tibial nerve of the rabbit of local cooling in a metal trough. l 1 Serial nerve conduction studies were used to follow the time-course of the nerve injury, and to assess its severity. Cooling to between 1°C and 5°C for 2 to 4 hours caused a localized lesion which was present at the time of rewarming, and which was followed by Walleriantype degeneration of the distal parts of affected fibers. There was no evidence of persistent conduction block. Histological studies of single teased fibers, and of transverse sections, confirmed that the lesion was axonal. There was no segmental demyelination, and paranodal demyelination was restricted to nodes of Ranvier just proximal to sites of complete degeneration. Three days after cooling, maximal conduction velocity in motor and afferent fibers had fallen by
MUSCLE & NERVE
October 1991
10% to 20% from the preoperative values, suggesting damage to the fastest-conducting axons. This was confirmed by quantitative morphological studies on surviving myelinated fibers, which showed selective loss of the largest axons. Control studies made it clear that the lesion did not occur in nerves which had been placed in the cooling trough for 2 hours, but without cooling. In the present experiments, the intact hindlimb of the rabbit has been cooled in a waterbath at lac,since this more closely approximates the conditions causing non-freezing cold injury in man. 1,18,23,24 As in the previous study, nerve conduction studies have again been used to follow the time-course, and to assess the severity of the resulting nerve damage. Limited histological studies were made for correlation with the conduction findings.
METHODS
New Zealand white rabbits aged 3 to 5 months were used. For the induction of anesthesia, Ketamine 40 mg/kg and Xylazine 5 mg/kg were given intramuscularly. For short procedures, such as nerve conduction studies, anesthesia was continued by half-doses given intramuscularly every 20 to 30 minutes as necessary. For prolonged procedures, glucose-saline was infused into an ear vein, and Ketamine 16 mg/kg and Xylazine 2 mg/kg were added to the infusion at intervals of 15 to 20 minutes. In this way, it proved possible to maintain light anesthesia for periods of up to 14 hours. For whole limb cooling by immersion the animal was positioned prone on a padded frame, with one hind limb shaved and suspended in a waterbath 18 cm deep, which allowed immersion up to the lower thigh 2 cm above the knee joint. A large copper coil attached to a Techne cooling system (Techne FC200 flow cooler and C400 circulator, Techne, Cambridge, Ltd., UK) was used to maintain a bath temperature of 1°C. A stirrer ensured that the temperature was uniform throughout the waterbath. During the prolonged periods of cooling, up to 14 hours in some experiments, the animals were wrapped in cotton wool to conserve body heat. In addition, an electric blanket was used to prevent central body temperature falling below 35°C. At the end of the cooling period, the animal was removed from the bath, the cooled limb was dried, and then allowed to re-warm at room temperature. In 4 animals, thermistors in hypodermic nee-
Non-Freezing Cold Injury
dles (Light Laboratories, Brighton, UK) were used to record the temperature of the calf and plantar muscles during limb cooling for periods of 8 to 12 hours. The techniques used for nerve conduction and histological studies have been described in detail previously. In brief, for conduction studies, steel needle electrodes were used for nerve stimulation and recording. For motor studies, the active recording electrode was placed subcutaneously over the mid-point of the plantar muscle bellies, and a remote electrode at a site distally. The tibial nerve was stimulated with electrodes inserted percutaneously along the course of the nerve in the upper thigh, in the lower thigh, and at the ankle. For studies on afferent fibers, the tibial nerve was stimulated at the ankle, and recording electrodes were inserted close to the nerve at the sciatic notch, the medial plantar nerve in the mid-foot, and the digital nerve at the base of the first toe. Supramaximal stimuli were delivered by a Medelec MS 6 electromyograph. Permanent records of evoked responses were made on photographic paper with a fiber-optic recorder. T h e limb was wrapped in cotton-wool during recordings to prevent cooling, and skin temperatures at the ankle were always between 35" and 38°C. For histological studies, animals were killed at predetermined times with intravenous pentobarbital, and the excised tibial and medial plantar nerves, tied to plastic frames, were fixed by immersion in 4% glutaraldehyde in Sorenson's buffer at pH 7.4. Portions of nerve for teasing were post-fixed in 1% aqueous osmium tetroxide, and softened in increasing concentrations of glycerine. Specimens for transverse sections were post-fixed in osmium tetroxide, dehydrated through alcohol, and embedded in Epon. Transverse sections for light microscopy were stained with toluidine blue, and ultrathin sections for electron microscopy were stained with methanolic uranyl acetate and lead citrate by Reynolds' method. l 9 Quantitative studied were performed on prints from light microscope sections enlarged to a magnification of x1100. Measurements of myelinated fiber diameter were made using a digitizing pad linked to a Research Machines 3802 computer. For histological study of striated muscle, a lumbrical was dissected from the foot, and frozen in liquid nitrogen. Transverse frozen sections were stained by a modified Gomori trichrome method, mounted in DPX, and examined by light microscopy. T o examine neuromuscular junctions and
''
MUSCLE & NERVE
October 1991
961
intramuscular nerve endings, lumbrical muscles were dissected, and immersed in liquid nitrogen. Longitudinal frozen sections, 100-Fm thick, were cut, stained for acetyl cholinesterase and with Sudan Black, mounted in glycerine, and examined by light microscopy. For light microscopy of blood vessels, excised pedal neurovascular bundles were fixed by immersion in 10% formol saline, and conventional paraffin-embedded sections were stained with hematoxylin and eosin.
RESULTS: CLINICAL OBSERVATIONS
The intramuscular temperatures in the calf and in the foot during immersion in water at 1°C were measured in 4 animals. In all cases, the findings were similar, and the results of 1 experiment are shown in Figure 1. It can be seen that the temperature of both the calf and foot fell steadily during the first 30 minutes, but then followed cyclical waves of rewarming which were most pronounced in the foot muscles. Between 2 and 3 hours after the start of cooling, the amplitude of these fluctuations declined, and the temperature of the calf muscles fell to about 5"C, and that of the foot to 1" to 2°C (near bath temperature). In 3 of the 4 animals, the temperature fluctuated by less than 1.5"C during the remaining period of immersion of up to a total of 12 hours. In a fourth animal, a larger cyclical fluctuation of up to 13°C was seen for 30 minutes, 8 hours after the start of cooling.
Animals were examined daily for the first few days after whole limb cooling. It was unusual to detect a neuromuscular deficit clinically. All animals had a normal sitting posture and hop, but occasionally the toe-spreading reflex' was impaired. N o consistent sensory impairment to noxious stmuli was observed, and trophic changes were not seen. Dependent edema of the previously cooled limb did not occur. ELECTROPHYSIOLOGICAL RESULTS
When different periods of hind-limb immersion in water at 1°C were tried, the findings on the next day were as follows. After immersion for 9 hours ( 1 animal), no abnormality of motor conduction was seen. After immersion for 10 hours ( 1 animal), plantar muscle action potential (MAP) amplitude in response to proximal stimulation was reduced by 25%. After immersion for 12 hours, there was marked variation between different animals; in 2 of 12 animals, evoked MAP amplitude was reduced by less than lo%, whereas the others showed reductions ranging from 34% to 98% (mean for whole group 61%). After immersion for 14 hours in 1 animal, MAP amplitude was reduced by 66%. In 4 control animals in which one hind-limb was immersed in water at 30" to 35°C for 12 hours, there was no change in evoked MAP amplitude. After limb cooling, the characteristic early finding in affected nerves was a reduction in amplitude of the MAPS evoked by proximal stimulation, with relative preservation of MAP amplitude on distal stimulation. This was followed by further reduction in the response to distal stimulation on subsequent days. A typical example is shown in Figure 2, from which it should be noted that the
Before,
Immersion
1 Hour
lDay
2 Days
i
i 2 3 Duration of Immersion in Hours
4
FIGURE 1. Diagram showing changes in intramusculartemperature in the calf (a), and in the foot (b), recorded during limb cooling by immersion in water at 1°C. Phasic fluctuations of temperature started 30 minutes after immersion, but diminished in amplitude 2 hours after the onset of cooling. Only minor fluctuations of up to 15°C occurred thereafter in this animal.
962
Non-Freezing Cold Injury
FIGURE 2. Plantar MAPS evoked by nerve stimulation before and 1 hour, 1 day, and 2 days after limb cooling by immersion in water at 1°C for 12 hours. There was conduction failure in the leg 1 hour after the cessation of cooling, followed after 1 to 2 days by failure of conduction distally. Time bar, 10 ms; voltage cal., 10 mV.
MUSCLE & NERVE
October 1991
site of the distal stimulating electrode in the thigh was at or just above the level of cooling, so that changes in MAP amplitude with stimulation at this level were similar to those obtained by proximal stimulation in the thigh. The difference between the effects of proximal and distal stimulation shown in Figure 2 was seen consistently in recordings made on the first day after cooling, and a similar change was already present in 2 animals, which were examined 1 hour after removal of the cooled limb from the bath. Table 1 gives the results of daily motor conduction studies in 6 animals after a cooling period of 12 hours. T h e mean reduction in evoked MAP amplitude on the first post-cooling day was 69% for proximal stimulation, but only 17% for distal stimulation, with a further fall of 55% by the third day. At the end of the 3-day period, the percentage reduction in evoked MAP amplitude was similar for proximal and distal stimulation, and there was no evidence of persistent conduction block between thigh and ankle. Recording of nerve action potentials (NAPs) on the day after limb cooling confirmed the presence of conduction failure in the leg. When the tibia1 nerve at the ankle was stimulated after 10 to 14 hours of cooling, the ascending NAP recorded in the thigh was absent in 2 animals, and reduced in amplitude in 12 animals. It was unchanged in 1 animal after 9 hours of cooling. In 2 of the animals in which MAP amplitude had remained normal after cooling for 12 hours, there was a moderate reduction in NAP amplitude (30% to 40%). In 5 animals which were studied daily and in which medial plantar and digital nerve action potentials were also recorded, the early reduction in amplitude of proximal NAPs in the leg was followed by a progressive reduction in medial plantar and digital NAP amplitudes over the next 2 to 3 days. An example is shown in Figure 3. In this case, the ascending NAP recorded from the sciatic nerve was absent on the first day after cooling for 12 hours (Dl). At this time, the medial plantar
R1
R3
R2
D1
D2 D3
I
5 ms
I
FIGURE 3. NAPs recorded before and on successive days after limb cooling by immersion in water at 1°C for 12 hours. With nerve stimulation at the ankle, the ascending NAP (Rl) was already absent on the first post-cooling day (Dl), whereas medial plantar and digital NAPs (R2, R3) persisted until the third postcooling day (D3). Cal: 5 p V for R1, 50 p V for R2, and 10 p V for R3.
and digital action potentials were only slightly reduced in amplitude compared with those before cooling. On the second post-cooling day (D2), a small action potential could still be recorded from the medial plantar nerve and from the digital nerve; these were no longer present on the third post-cooling day (D3). On the third day after cooling, when loss of excitability in the distal parts of affected fibers was complete, maximal motor conduction velocity (MCV) and the maximal velocity of ascending nerve action potentials (NCV) between thigh and ankle were estimated in surviving fibers. The mean maximal NCV fell from 72 2 7 m/s before cooling to 62 k 9 on the third post-cooling day, and the mean maximal MCV from 62 5 to 56 k 7 m/s. In both cases, the reduction in conduction velocity was statistically significant (t test, P < 0.05). No change was observed after limb immersion in warm water. From these results, it was concluded that after
*
~
~~
Table 1. Evoked MAP amplitude before and on the first and third days after limb immersion in water for 12 hours.
No. of animals
Site of stim.
Immersion at 1" C
6
Immersion at 30" to 35°C
4
Thigh Ankle Thigh Ankle
Before immersion, mean SE 10.8 2 18.1 2 10.5 ? 18.5 2
1.2 1.3 1.7 3.2
Day 1, mean 2 SE (%)
Day 3, mean ? SE (%)
3.3 ? 15.0 ? 10.3 2 17.8 ?
3.0 2 1.2 (28) 5.0 f. 2.0 (28) 9.6 ? 1.4 (91) 17.8 2 2.6 (96)
1.4 (31) 1.3 (83) 1.1 (98) 1.9 (96)
Values in mV, percentage of value before immersion in brackets
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
963
whole limb cooling by immersion, the initial lesion in motor and afferent fibers was not a diffuse one throughout the cooled portion of the tibial nerve, but that it occurred mainly in the leg between knee and ankle, with subsequent degeneration of the distal parts of the affected fibers. The velocity measurements on the third day indicated, that in partial lesions, the fastest-conducting fibers had been affected. There was no evidence of persistent conduction block. When nerve conduction studies were repeated 3 weeks after limb immersion, evoked MAP amplitude was slightly increased compared with the third day after cooling, the mean amplitude for 6 animals having risen from 33% to 52% of that obtained before cooling. There was no change in the relative amplitudes of responses evoked by proximal and distal stimulation. This early increase in amplitude, was not accompanied by gross temporal dispersion of the response, and was thought to be due to collateral reinnervation rather than regeneration from a site in the leg. In support of this was the fact that a similar early increase in amplitude had been seen after local nerve cooling in the thigh," when the known distance between the cooling site and the muscle ruled out the possibility of regeneration and subsequent maturation occurring over such a short time period. MORPHOLOGICAL CORRELATES
In the tibial nerves examined 18 hours after the end of the cooling period, both transverse sections and teased fibers showed abnormalities confined to a region in the calf 2 to 3 cm long, the abnormality being maximal at a level corresponding approximately to the junction of the middle and upper thirds of the tibia. Transverse sections showed that a proportion of the myelinated axons were swollen and often densely stained in toluidine blue preparations. On electron microscopy, the swollen axons were usually filled with membranous organelles, with or without a central core of neurofilaments. These appearances, which were similar to those previously reported after local tibial nerve cooling in a trough," were accompanied by less endoneurial edema than after local cooling, although some was present. The endoneurial capillaries appeared normal in electron micrographs, and no changes were seen in unmyelinated fibers. When single teased fibers were examined 18 hours after cooling, swollen regions separated by discontinuities of the myelin sheath could be seen, the abnormalities extending over a distance which,
964
Non-FreezingCold Injury
in any 1 fiber, did not exceed 5 mm. In most cases, the abnormal regions appeared to start and end at nodes of Ranvier, but this was not always so; in the fiber shown in Figure 4 the proximal end of the lesion is in an internode. These lesions did not occur at precisely the same level in different fibers, but extended over 2 to 3 cm in the whole nerve. At 18 hours, no changes comparable to those described above were seen in the tibial nerve distal to the lesion, or in the medial plantar nerve in the foot. By the end of the first week, however, many of the myelinated fibers in the distal tibial and medial plantar nerves showed Wallerian-type degeneration, with axonal discontinuity, ovoid formation, and the presence of lipid laden macrophages. T h e unmyelinated fibers remained normal. T h e subsequent evolution of the lesion in the tibial and medial plantar nerves was similar to that observed previously after local nerve cooling in a trough. Transverse sections taken at 3 weeks showed myelinated fiber loss of varying severity, with residual myelin debris. At this stage, the upper level of nerve damage could be identified in teased fiber preparations by the presence of "junctional fiber^,"^ with normal internodes proximally and short thinly myelinated internodes (characteristic of early regeneration) distally. Junctional fibers were usually found in a 2- to 4-cm region of the tibial nerve in the upper calf. They were rare in the distal tibial nerve, and were not seen in the medial plantar nerve in the foot, in spite of a careful search. As in our previous experiments using local cooling,' paranodal demyelination was rarely seen except in junctional fibers, where it was commonly present at the first few nodes of Ranvier proximal to the site of complete degeneration. No segmental demyelination was seen.
Proximal
1
-
-. ~-
,,,..
t
t
.
.-
._..
.
.
. .
.
Distal
FIGURE 4. Localized swelling and discontinuity in consecutive portions of a single teased myelinated fiber from the tibial nerve in the calf 18 hours after limb immersion in water at 1°C for 12 hours. The extent of the abnormality (approx. 5 mm) is marked by large arrows. Proximal and distal to this region, the intermodal myelin and the nodes of Ranvier (marked by small arrows) are normal. Bar: 0.5 mm.
MUSCLE & NERVE
October 1991
nal sections of muscle, 7 days after cooling, showed widespread degeneration of preterminal nerve fibers.
In order to study the distribution of fiber loss among myelinated fibers of different diameters, total fiber counts were carried out on the tibial nerve at the ankle and on the medial plantar nerve in the foot 3 weeks after cooling for 12 hours. Measurements were made on the nerves from both lower limbs in 3 animals. In 2 of the 3 cases, physiological studies had shown the nerve lesion to be severe, with reductions in evoked plantar MAP amplitude of 75% and 85%. In the third animal, physiological studies had indicated a relatively mild lesion (evoked MAP amplitude reduced by 40%). Pooled results for affected nerves and for control nerves of the opposite leg are summarized in Table 2. It can be seen that in both the tibial and medial plantar nerves the reduction in fiber numbers on the affected side was greatest for the largest-diameter fibers. There was a greater reduction for fibers of 8 pm and over than for those of 6 pm and over, and a greater reduction for both these categories than for the total population. While early reinnervation might have increased the number of small myelinated fibers, this effect is likely to have been small, since the nerves were taken only 3 weeks after cooling. Early regeneration could not in any case affect fiber numbers for categories greater than 6 pm. Our results, in kee ing with those obtained after local nerve cooling," confirm the relative vulnerability of large-diameter fibers when cooled to temperatures just above freezing point. Transverse sections of lumbrical muscles examined by light microscopy 1 and 7 days after limb cooling for 12 hours showed no pathological abnormalities in muscle fibers, in spite of histological evidence of nerve damage. There was no evidence of endomysial edema, and the pedal arterioles were normal. In contrast, 100-pm thick longitudi-
DISCUSSION
Presistent tissue injury from non-freezing cold injury has been described in man after ex osures varying from 3 hours" to several days?3,' From previous experimental studies,"- 1432022 and from our own work, there is clearly considerable variation between individual animals of the same species in their susceptibility to cold injury. Some of this variation may relate to the temperature fluctuations caused by episodes of cold vasodilatation in an immersed limb. I n 3 of the 4 animals in which we studied this, the periodic increases in deep limb temperature which we observed soon after immersion in the bath seemed to disappear after 2 to 3 hours. In 1 case, these recurred 8 hours after the start of cooling, and it seems from the work of Sayen et a1.20 that, in conscious animals, these small fluctuations in the temperature of an immersed limb may continue for a much longer period. Further experiments are required to establish whether such temperature fluctuations have an effect on the outcome of cooling, in terms of the severity of nerve damage. In contrast to previous animal studies in which limb or tail immersion has given rise to widespread changes in skin, muscle, and blood vess e l ~ , the ~ ~procedure ~ , ~ ~ adopted in the present study has resulted in what might be termed a pure cold neuropathy, without limb swelling and without histological abnormalities of blood vessels or muscles. In this respect, our results are similar to those of Peyronnard et al." who worked with the rat's tail and found that cold immersion for 12 hours produced nerve fiber damage without other
Table 2. Effect of limb immersion (12 hours at 1%) on total myelinated fibers in distal tibial and medial plantar nerves of 3 animals.
Tibia1 nerve
Medial plantar nerve
All fibers
Fibers > 6 Fm
Fibers > 8 Frn
Affected Control (opposite leg) % loss
14,360 18,383
5,992 8,879
2,510 4,753
22
33
47
Affected
11,642
4,595
1,703
Control (opposite leg) % loss
13,855
6,668
3,549
31
52
~~~~
16 ~
~
Specimens taken 3 weeks after cooling
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
965
changes. It is of interest that in the milder cases of human non-freezing cold injury, neurological symptoms appear to predominate, with minor involvement of other tissues.'"' In spite of the earlier suggestion that cooling could cause paranodal or segmental demyelination,6 our own ex eriments confirm the view expressed by others f18,z' that axonal damage is the characteristic feature of non-freezing cold injury. Its selective effect on large myelinated fibers, previously re orted after local nerve cooling in a tro~gh,2~"'~~ and ' ~ after tail i r n r n e r s i ~ n ~ ~has ~~'' also been confirmed. Although, in healthy rabbits, maximal NCV is slightly greater than MCV in the tibial nerve,7 it appears that the fastest-conducting motor and afferent fibers are similar in their susceptibility to injury- as the reduction in velocity we recorded after cooling was not significantly different for the 2 groups (16% for NCV and 15% for MCV). The mechanism of nerve damage by nonfreezing cold injury remains obscure. It has been suggested in the past that it might be due to changes in blood vessels and to nerve ischemia.6 However, for reasons given in our previous paper, l this seems less likely than a direct effect of low temperature on nerve fibers. One argument against an ischemic mechanism in our previous experiments was the short duration of cooling used. For example, cooling for 2 hours in a trough at 1°C was sufficient to produce severe nerve injury, whereas previous experiments in the rabbit have shown that ischemia of this duration was without any long-lasting effect on the tibial nerve.15 This argument cannot be used in relation to whole limb cooling for periods of 12 to 14 hours, but another difference between the effects of ischemia and cooling, which was discussed previously, l 1 also apply to the present experiments. This is the fact that ischemic injury to the tibial nerve in the rabbit has been shown to produce a mixture of Wallerian degeneration and selective demyelinati~n,~ with evidence of conduction block in surviving fiber^.^ The absence of conduction
'
block after cold injury suggests a different mechanism of nerve damage. A clue to etiology may perhaps be derived from our finding that, after hind limb immersion in the rabbit, the main site of damage in both motor and sensory fibers was proximal rather than distal, occurring in the calf rather than in the foot. A possible explanation of this proximal site of nerve damage might be the temperature gradient along the tibial nerve in the first few centimeters below the surface of the bath. At this level, at which the temperature gradient along the nerve is steep, one might expect to find relatively abrupt changes in axonal transport, and also a mismatch between tissue temperature and that of the blood perfusing it. Whether these could lead to changes in membrane permeability and to a rise in intracellular calcium is speculative, but an interpretation of this kind could explain why nerve damage was readily produced within 2 hours by local nerve cooling in a trough, with steep temperature gradients at each edge, whereas whole-limb immersion had to be continued for a much longer period to produce lesions. Alternative explanations for the proximal site of nerve damage during limb cooling seem less cogent. In the normal rabbit hind limb, there is some reduction in the maximal diameter of myelinated fibers in the foot compared with those in the leg, but the difference is small and hardly likely to explain the localized pathology we have found. While a difference in the fatty acid composition of lipid components of membranes at proximal and distal levels is another possible factor, there are no data for rabbit peripheral nerves to indicate that such a difference in chemical composition exists. It is known that, in some arctic mammals, there are measurable differences in the fatty acid composition of lipid between proximal and distal parts of the and that in some coldadapted birds, impulse conduction can persist in distal nerves at temperatures which block conduction p r ~ x i m a l l y but , ~ these cold adaptations are of uncertain relevance to the rabbit hind limb.
REFERENCES 1. Afifi AK, Kimura J , Bell WE: Hypothermia-induced re-
4. Blackwood W, Russell H: Further experiments in the study
versible polyneuropathy: electrophysiologic evidence of axonopathy. Paedzat Neurol 1988;4:49 - 53. 2. Basbaum CB: Induced hypothermia in peripheral nerve: electron microscopic and electrophysiological observations. J Neurocytol 1973;2:171- 187. 3. Blackwood W, Russell H: Experiments in the study of immersion foot. Edin Med J 1943;50:385-398.
of immersion foot. Edin Med J 1945;52:160- 165. 5. Chatfield PO, Lyman CP, Irving L: Physiological adaptation to cold of peripheral nerve in the leg of the Herring Gull (Larus Argentatus). A m J Physiol 1953;172:639-644. 6. Denny-Brown D, Adams RD, Brenner C, Doherty MM: T h e pathology of injury to nerve induced by cold. J Neuropathol Exp Neurol 1945;4:305-323.
966
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
7. Fowler CJ, Gilliatt RW: Conduction velocity and conduction block after experimental ischaemic nerve injury. J Neurol Sci 198 1 ;52:221-238. 8. Gutmann E: Factors affecting recovery of motor function after nerve lesions. J Neurol Neurosurg Psychiatry 1942;5:81-95. 9. Hess K, Eames RA, Darveniza P, Gilliatt RW: Acute ischaemic neuropathy in the rabbit./ Neurol Sci 1979;44: 19-43. 10. Irving L, Schmidt-Nielsen K, Abrahamsen NBS: O n the melting points of animal fats in cold climates. Physiol Zoology 1957;30:93- 106. 1 1 . Kennett RP, Gilliatt RW: Nerve conduction studies on experimental non-freezing cold injury. 1 : Local nerve cooling. Muscle Nerve 1991;14:553-562. 12. Lange K, Weiner D, Boyd LJ: T h e functional pathology of experimental foot immersion. Am Heart J 1948;35:238247. 13. Large A, Heinbecker P: Nerve degeneration following prolonged cooling of an extremity. Ann Surg 1944;120:742-749. 14. Lewis RB, Moen PW: Experimental immersion leg. A m J Med Sci 1952;224:529-539. 15. Lundborg G: Ischaemic nerve injury: experimental studies on intraneural microvascular pathophysiology and nerve function in a limb subjected to temporary circulatory arrest. Scand J Plastic Reconstr Surg 197O;(suppl 6):1 - 113.
Non-Freezing Cold Injury
16. Meng MS, West GC, Irving L: Fatty acid composition of caribou bone marrow. Comp Biochem Physiol 1969;30: 187191. 17. Nukada H, Pollock M, Allpress S: Experimental cold injury to peripheral nerve. Brain 1981;104:779-811. 18. Peyronnard JM, Pedneualt M, Aguayo AJ: Neuropathies due to cold: quantitative studies of structural changes in human and animal nerves, in Neurology. International Congress Series No. 434, Amsterdam, Excerpta Medica, 1978, pp 308-329. 19. Reynolds ES: T h e use of lead citrate at high p H as an electron-opaque stain in electron microscopy. J Cell Biol 1963;17~208-2 12. 20. Sayen A, Meloche BR, Tedeschi G, Montgonery H: Experimental immersion foot: observations in the chilled leg of the rabbit. Clin Sci 1960;19:243-253. 21. Schaumburg H , Byck R, Herman R, Rosengart C: Peripheral nerve damage by cold. Arch Neurol 1967;16:103- 109. 22. Smith JL, Ritchie J, Dawson J: Clinical and experimental observations on the pathology of trench frostbite. J Path Bact 1915;20:159-190. 23. Ungley CC, Blackwood W: Peripheral vasoneuropathy after chilling “immersion foot and immersion hand.” Lancet 1942;ii:447-45 1 . 24. Ungley CC, Channel1 GD, Richards KL: T h e immersion foot syndrome. B r J Surg 1945;33:17-31.
MUSCLE &
NERVE
October 1991
967
NERVE CONDUCTION STUDIES IN EXPERIMENTAL NON-FREEZING COLD INJURY: 11. GENERALIZED NERVE COOLING BY LIMB IMMERSION ROBIN P. KENNETT, BSc, MD, MRCP, and ROGER W. GILLIATT, DM, FRCP
T h e nerve changes in non-freezing cold injury (trench foot, immersion foot) are not well understood. Clinical experience, mainly gained from casualties during two world wars, indicates that prolonged limb cooling in water just above freezing point can produce peripheral nerve damage as a part of a more widespread syndrome, which usually includes limb edema, skin blistering, and vascular damage leading to necrosis and tissue i0ss.23.24 Experimental studies in animals have confirmed that cold exposure can lead to peripheral From the University Department of Clinical Neurology, Queen Square, London, England. Presented in part to the Physiological Society, London, March 1987. Acknowledgments: This work was financially supported by the Action Research for the Crippled Child, and forms the substance of a MD thesis of the University of London (Dr. Kennett) Address reprint requests to Dr. Kennett, The Regional Neurological Centre, Newcastle General Hospital, Westgate Road, Newcastle upon Tyne, NE4 6BE, England. Accepted for publication August 6, 1990.
CCC 0148-639)(/91/0100960-08 $04.00 0 1991 John Wiley & Sons, Inc.
960
Non-Freezing Cold Injury
nerve degeneration with or without changes in other tissues.3,4,12-14,18,20,22 In other experimental studies, local cooling of an exposed nerve trunk has been used to produce the nerve lesion without the risk of accompanyin changes in muscle, skin, or major blood vessels.21,17,2 1 In a previous study, we described the effect on the tibial nerve of the rabbit of local cooling in a metal trough. l 1 Serial nerve conduction studies were used to follow the time-course of the nerve injury, and to assess its severity. Cooling to between 1°C and 5°C for 2 to 4 hours caused a localized lesion which was present at the time of rewarming, and which was followed by Walleriantype degeneration of the distal parts of affected fibers. There was no evidence of persistent conduction block. Histological studies of single teased fibers, and of transverse sections, confirmed that the lesion was axonal. There was no segmental demyelination, and paranodal demyelination was restricted to nodes of Ranvier just proximal to sites of complete degeneration. Three days after cooling, maximal conduction velocity in motor and afferent fibers had fallen by
MUSCLE & NERVE
October 1991
10% to 20% from the preoperative values, suggesting damage to the fastest-conducting axons. This was confirmed by quantitative morphological studies on surviving myelinated fibers, which showed selective loss of the largest axons. Control studies made it clear that the lesion did not occur in nerves which had been placed in the cooling trough for 2 hours, but without cooling. In the present experiments, the intact hindlimb of the rabbit has been cooled in a waterbath at lac,since this more closely approximates the conditions causing non-freezing cold injury in man. 1,18,23,24 As in the previous study, nerve conduction studies have again been used to follow the time-course, and to assess the severity of the resulting nerve damage. Limited histological studies were made for correlation with the conduction findings.
METHODS
New Zealand white rabbits aged 3 to 5 months were used. For the induction of anesthesia, Ketamine 40 mg/kg and Xylazine 5 mg/kg were given intramuscularly. For short procedures, such as nerve conduction studies, anesthesia was continued by half-doses given intramuscularly every 20 to 30 minutes as necessary. For prolonged procedures, glucose-saline was infused into an ear vein, and Ketamine 16 mg/kg and Xylazine 2 mg/kg were added to the infusion at intervals of 15 to 20 minutes. In this way, it proved possible to maintain light anesthesia for periods of up to 14 hours. For whole limb cooling by immersion the animal was positioned prone on a padded frame, with one hind limb shaved and suspended in a waterbath 18 cm deep, which allowed immersion up to the lower thigh 2 cm above the knee joint. A large copper coil attached to a Techne cooling system (Techne FC200 flow cooler and C400 circulator, Techne, Cambridge, Ltd., UK) was used to maintain a bath temperature of 1°C. A stirrer ensured that the temperature was uniform throughout the waterbath. During the prolonged periods of cooling, up to 14 hours in some experiments, the animals were wrapped in cotton wool to conserve body heat. In addition, an electric blanket was used to prevent central body temperature falling below 35°C. At the end of the cooling period, the animal was removed from the bath, the cooled limb was dried, and then allowed to re-warm at room temperature. In 4 animals, thermistors in hypodermic nee-
Non-Freezing Cold Injury
dles (Light Laboratories, Brighton, UK) were used to record the temperature of the calf and plantar muscles during limb cooling for periods of 8 to 12 hours. The techniques used for nerve conduction and histological studies have been described in detail previously. In brief, for conduction studies, steel needle electrodes were used for nerve stimulation and recording. For motor studies, the active recording electrode was placed subcutaneously over the mid-point of the plantar muscle bellies, and a remote electrode at a site distally. The tibial nerve was stimulated with electrodes inserted percutaneously along the course of the nerve in the upper thigh, in the lower thigh, and at the ankle. For studies on afferent fibers, the tibial nerve was stimulated at the ankle, and recording electrodes were inserted close to the nerve at the sciatic notch, the medial plantar nerve in the mid-foot, and the digital nerve at the base of the first toe. Supramaximal stimuli were delivered by a Medelec MS 6 electromyograph. Permanent records of evoked responses were made on photographic paper with a fiber-optic recorder. T h e limb was wrapped in cotton-wool during recordings to prevent cooling, and skin temperatures at the ankle were always between 35" and 38°C. For histological studies, animals were killed at predetermined times with intravenous pentobarbital, and the excised tibial and medial plantar nerves, tied to plastic frames, were fixed by immersion in 4% glutaraldehyde in Sorenson's buffer at pH 7.4. Portions of nerve for teasing were post-fixed in 1% aqueous osmium tetroxide, and softened in increasing concentrations of glycerine. Specimens for transverse sections were post-fixed in osmium tetroxide, dehydrated through alcohol, and embedded in Epon. Transverse sections for light microscopy were stained with toluidine blue, and ultrathin sections for electron microscopy were stained with methanolic uranyl acetate and lead citrate by Reynolds' method. l 9 Quantitative studied were performed on prints from light microscope sections enlarged to a magnification of x1100. Measurements of myelinated fiber diameter were made using a digitizing pad linked to a Research Machines 3802 computer. For histological study of striated muscle, a lumbrical was dissected from the foot, and frozen in liquid nitrogen. Transverse frozen sections were stained by a modified Gomori trichrome method, mounted in DPX, and examined by light microscopy. T o examine neuromuscular junctions and
''
MUSCLE & NERVE
October 1991
961
intramuscular nerve endings, lumbrical muscles were dissected, and immersed in liquid nitrogen. Longitudinal frozen sections, 100-Fm thick, were cut, stained for acetyl cholinesterase and with Sudan Black, mounted in glycerine, and examined by light microscopy. For light microscopy of blood vessels, excised pedal neurovascular bundles were fixed by immersion in 10% formol saline, and conventional paraffin-embedded sections were stained with hematoxylin and eosin.
RESULTS: CLINICAL OBSERVATIONS
The intramuscular temperatures in the calf and in the foot during immersion in water at 1°C were measured in 4 animals. In all cases, the findings were similar, and the results of 1 experiment are shown in Figure 1. It can be seen that the temperature of both the calf and foot fell steadily during the first 30 minutes, but then followed cyclical waves of rewarming which were most pronounced in the foot muscles. Between 2 and 3 hours after the start of cooling, the amplitude of these fluctuations declined, and the temperature of the calf muscles fell to about 5"C, and that of the foot to 1" to 2°C (near bath temperature). In 3 of the 4 animals, the temperature fluctuated by less than 1.5"C during the remaining period of immersion of up to a total of 12 hours. In a fourth animal, a larger cyclical fluctuation of up to 13°C was seen for 30 minutes, 8 hours after the start of cooling.
Animals were examined daily for the first few days after whole limb cooling. It was unusual to detect a neuromuscular deficit clinically. All animals had a normal sitting posture and hop, but occasionally the toe-spreading reflex' was impaired. N o consistent sensory impairment to noxious stmuli was observed, and trophic changes were not seen. Dependent edema of the previously cooled limb did not occur. ELECTROPHYSIOLOGICAL RESULTS
When different periods of hind-limb immersion in water at 1°C were tried, the findings on the next day were as follows. After immersion for 9 hours ( 1 animal), no abnormality of motor conduction was seen. After immersion for 10 hours ( 1 animal), plantar muscle action potential (MAP) amplitude in response to proximal stimulation was reduced by 25%. After immersion for 12 hours, there was marked variation between different animals; in 2 of 12 animals, evoked MAP amplitude was reduced by less than lo%, whereas the others showed reductions ranging from 34% to 98% (mean for whole group 61%). After immersion for 14 hours in 1 animal, MAP amplitude was reduced by 66%. In 4 control animals in which one hind-limb was immersed in water at 30" to 35°C for 12 hours, there was no change in evoked MAP amplitude. After limb cooling, the characteristic early finding in affected nerves was a reduction in amplitude of the MAPS evoked by proximal stimulation, with relative preservation of MAP amplitude on distal stimulation. This was followed by further reduction in the response to distal stimulation on subsequent days. A typical example is shown in Figure 2, from which it should be noted that the
Before,
Immersion
1 Hour
lDay
2 Days
i
i 2 3 Duration of Immersion in Hours
4
FIGURE 1. Diagram showing changes in intramusculartemperature in the calf (a), and in the foot (b), recorded during limb cooling by immersion in water at 1°C. Phasic fluctuations of temperature started 30 minutes after immersion, but diminished in amplitude 2 hours after the onset of cooling. Only minor fluctuations of up to 15°C occurred thereafter in this animal.
962
Non-Freezing Cold Injury
FIGURE 2. Plantar MAPS evoked by nerve stimulation before and 1 hour, 1 day, and 2 days after limb cooling by immersion in water at 1°C for 12 hours. There was conduction failure in the leg 1 hour after the cessation of cooling, followed after 1 to 2 days by failure of conduction distally. Time bar, 10 ms; voltage cal., 10 mV.
MUSCLE & NERVE
October 1991
site of the distal stimulating electrode in the thigh was at or just above the level of cooling, so that changes in MAP amplitude with stimulation at this level were similar to those obtained by proximal stimulation in the thigh. The difference between the effects of proximal and distal stimulation shown in Figure 2 was seen consistently in recordings made on the first day after cooling, and a similar change was already present in 2 animals, which were examined 1 hour after removal of the cooled limb from the bath. Table 1 gives the results of daily motor conduction studies in 6 animals after a cooling period of 12 hours. T h e mean reduction in evoked MAP amplitude on the first post-cooling day was 69% for proximal stimulation, but only 17% for distal stimulation, with a further fall of 55% by the third day. At the end of the 3-day period, the percentage reduction in evoked MAP amplitude was similar for proximal and distal stimulation, and there was no evidence of persistent conduction block between thigh and ankle. Recording of nerve action potentials (NAPs) on the day after limb cooling confirmed the presence of conduction failure in the leg. When the tibia1 nerve at the ankle was stimulated after 10 to 14 hours of cooling, the ascending NAP recorded in the thigh was absent in 2 animals, and reduced in amplitude in 12 animals. It was unchanged in 1 animal after 9 hours of cooling. In 2 of the animals in which MAP amplitude had remained normal after cooling for 12 hours, there was a moderate reduction in NAP amplitude (30% to 40%). In 5 animals which were studied daily and in which medial plantar and digital nerve action potentials were also recorded, the early reduction in amplitude of proximal NAPs in the leg was followed by a progressive reduction in medial plantar and digital NAP amplitudes over the next 2 to 3 days. An example is shown in Figure 3. In this case, the ascending NAP recorded from the sciatic nerve was absent on the first day after cooling for 12 hours (Dl). At this time, the medial plantar
R1
R3
R2
D1
D2 D3
I
5 ms
I
FIGURE 3. NAPs recorded before and on successive days after limb cooling by immersion in water at 1°C for 12 hours. With nerve stimulation at the ankle, the ascending NAP (Rl) was already absent on the first post-cooling day (Dl), whereas medial plantar and digital NAPs (R2, R3) persisted until the third postcooling day (D3). Cal: 5 p V for R1, 50 p V for R2, and 10 p V for R3.
and digital action potentials were only slightly reduced in amplitude compared with those before cooling. On the second post-cooling day (D2), a small action potential could still be recorded from the medial plantar nerve and from the digital nerve; these were no longer present on the third post-cooling day (D3). On the third day after cooling, when loss of excitability in the distal parts of affected fibers was complete, maximal motor conduction velocity (MCV) and the maximal velocity of ascending nerve action potentials (NCV) between thigh and ankle were estimated in surviving fibers. The mean maximal NCV fell from 72 2 7 m/s before cooling to 62 k 9 on the third post-cooling day, and the mean maximal MCV from 62 5 to 56 k 7 m/s. In both cases, the reduction in conduction velocity was statistically significant (t test, P < 0.05). No change was observed after limb immersion in warm water. From these results, it was concluded that after
*
~
~~
Table 1. Evoked MAP amplitude before and on the first and third days after limb immersion in water for 12 hours.
No. of animals
Site of stim.
Immersion at 1" C
6
Immersion at 30" to 35°C
4
Thigh Ankle Thigh Ankle
Before immersion, mean SE 10.8 2 18.1 2 10.5 ? 18.5 2
1.2 1.3 1.7 3.2
Day 1, mean 2 SE (%)
Day 3, mean ? SE (%)
3.3 ? 15.0 ? 10.3 2 17.8 ?
3.0 2 1.2 (28) 5.0 f. 2.0 (28) 9.6 ? 1.4 (91) 17.8 2 2.6 (96)
1.4 (31) 1.3 (83) 1.1 (98) 1.9 (96)
Values in mV, percentage of value before immersion in brackets
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
963
whole limb cooling by immersion, the initial lesion in motor and afferent fibers was not a diffuse one throughout the cooled portion of the tibial nerve, but that it occurred mainly in the leg between knee and ankle, with subsequent degeneration of the distal parts of the affected fibers. The velocity measurements on the third day indicated, that in partial lesions, the fastest-conducting fibers had been affected. There was no evidence of persistent conduction block. When nerve conduction studies were repeated 3 weeks after limb immersion, evoked MAP amplitude was slightly increased compared with the third day after cooling, the mean amplitude for 6 animals having risen from 33% to 52% of that obtained before cooling. There was no change in the relative amplitudes of responses evoked by proximal and distal stimulation. This early increase in amplitude, was not accompanied by gross temporal dispersion of the response, and was thought to be due to collateral reinnervation rather than regeneration from a site in the leg. In support of this was the fact that a similar early increase in amplitude had been seen after local nerve cooling in the thigh," when the known distance between the cooling site and the muscle ruled out the possibility of regeneration and subsequent maturation occurring over such a short time period. MORPHOLOGICAL CORRELATES
In the tibial nerves examined 18 hours after the end of the cooling period, both transverse sections and teased fibers showed abnormalities confined to a region in the calf 2 to 3 cm long, the abnormality being maximal at a level corresponding approximately to the junction of the middle and upper thirds of the tibia. Transverse sections showed that a proportion of the myelinated axons were swollen and often densely stained in toluidine blue preparations. On electron microscopy, the swollen axons were usually filled with membranous organelles, with or without a central core of neurofilaments. These appearances, which were similar to those previously reported after local tibial nerve cooling in a trough," were accompanied by less endoneurial edema than after local cooling, although some was present. The endoneurial capillaries appeared normal in electron micrographs, and no changes were seen in unmyelinated fibers. When single teased fibers were examined 18 hours after cooling, swollen regions separated by discontinuities of the myelin sheath could be seen, the abnormalities extending over a distance which,
964
Non-FreezingCold Injury
in any 1 fiber, did not exceed 5 mm. In most cases, the abnormal regions appeared to start and end at nodes of Ranvier, but this was not always so; in the fiber shown in Figure 4 the proximal end of the lesion is in an internode. These lesions did not occur at precisely the same level in different fibers, but extended over 2 to 3 cm in the whole nerve. At 18 hours, no changes comparable to those described above were seen in the tibial nerve distal to the lesion, or in the medial plantar nerve in the foot. By the end of the first week, however, many of the myelinated fibers in the distal tibial and medial plantar nerves showed Wallerian-type degeneration, with axonal discontinuity, ovoid formation, and the presence of lipid laden macrophages. T h e unmyelinated fibers remained normal. T h e subsequent evolution of the lesion in the tibial and medial plantar nerves was similar to that observed previously after local nerve cooling in a trough. Transverse sections taken at 3 weeks showed myelinated fiber loss of varying severity, with residual myelin debris. At this stage, the upper level of nerve damage could be identified in teased fiber preparations by the presence of "junctional fiber^,"^ with normal internodes proximally and short thinly myelinated internodes (characteristic of early regeneration) distally. Junctional fibers were usually found in a 2- to 4-cm region of the tibial nerve in the upper calf. They were rare in the distal tibial nerve, and were not seen in the medial plantar nerve in the foot, in spite of a careful search. As in our previous experiments using local cooling,' paranodal demyelination was rarely seen except in junctional fibers, where it was commonly present at the first few nodes of Ranvier proximal to the site of complete degeneration. No segmental demyelination was seen.
Proximal
1
-
-. ~-
,,,..
t
t
.
.-
._..
.
.
. .
.
Distal
FIGURE 4. Localized swelling and discontinuity in consecutive portions of a single teased myelinated fiber from the tibial nerve in the calf 18 hours after limb immersion in water at 1°C for 12 hours. The extent of the abnormality (approx. 5 mm) is marked by large arrows. Proximal and distal to this region, the intermodal myelin and the nodes of Ranvier (marked by small arrows) are normal. Bar: 0.5 mm.
MUSCLE & NERVE
October 1991
nal sections of muscle, 7 days after cooling, showed widespread degeneration of preterminal nerve fibers.
In order to study the distribution of fiber loss among myelinated fibers of different diameters, total fiber counts were carried out on the tibial nerve at the ankle and on the medial plantar nerve in the foot 3 weeks after cooling for 12 hours. Measurements were made on the nerves from both lower limbs in 3 animals. In 2 of the 3 cases, physiological studies had shown the nerve lesion to be severe, with reductions in evoked plantar MAP amplitude of 75% and 85%. In the third animal, physiological studies had indicated a relatively mild lesion (evoked MAP amplitude reduced by 40%). Pooled results for affected nerves and for control nerves of the opposite leg are summarized in Table 2. It can be seen that in both the tibial and medial plantar nerves the reduction in fiber numbers on the affected side was greatest for the largest-diameter fibers. There was a greater reduction for fibers of 8 pm and over than for those of 6 pm and over, and a greater reduction for both these categories than for the total population. While early reinnervation might have increased the number of small myelinated fibers, this effect is likely to have been small, since the nerves were taken only 3 weeks after cooling. Early regeneration could not in any case affect fiber numbers for categories greater than 6 pm. Our results, in kee ing with those obtained after local nerve cooling," confirm the relative vulnerability of large-diameter fibers when cooled to temperatures just above freezing point. Transverse sections of lumbrical muscles examined by light microscopy 1 and 7 days after limb cooling for 12 hours showed no pathological abnormalities in muscle fibers, in spite of histological evidence of nerve damage. There was no evidence of endomysial edema, and the pedal arterioles were normal. In contrast, 100-pm thick longitudi-
DISCUSSION
Presistent tissue injury from non-freezing cold injury has been described in man after ex osures varying from 3 hours" to several days?3,' From previous experimental studies,"- 1432022 and from our own work, there is clearly considerable variation between individual animals of the same species in their susceptibility to cold injury. Some of this variation may relate to the temperature fluctuations caused by episodes of cold vasodilatation in an immersed limb. I n 3 of the 4 animals in which we studied this, the periodic increases in deep limb temperature which we observed soon after immersion in the bath seemed to disappear after 2 to 3 hours. In 1 case, these recurred 8 hours after the start of cooling, and it seems from the work of Sayen et a1.20 that, in conscious animals, these small fluctuations in the temperature of an immersed limb may continue for a much longer period. Further experiments are required to establish whether such temperature fluctuations have an effect on the outcome of cooling, in terms of the severity of nerve damage. In contrast to previous animal studies in which limb or tail immersion has given rise to widespread changes in skin, muscle, and blood vess e l ~ , the ~ ~procedure ~ , ~ ~ adopted in the present study has resulted in what might be termed a pure cold neuropathy, without limb swelling and without histological abnormalities of blood vessels or muscles. In this respect, our results are similar to those of Peyronnard et al." who worked with the rat's tail and found that cold immersion for 12 hours produced nerve fiber damage without other
Table 2. Effect of limb immersion (12 hours at 1%) on total myelinated fibers in distal tibial and medial plantar nerves of 3 animals.
Tibia1 nerve
Medial plantar nerve
All fibers
Fibers > 6 Fm
Fibers > 8 Frn
Affected Control (opposite leg) % loss
14,360 18,383
5,992 8,879
2,510 4,753
22
33
47
Affected
11,642
4,595
1,703
Control (opposite leg) % loss
13,855
6,668
3,549
31
52
~~~~
16 ~
~
Specimens taken 3 weeks after cooling
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
965
changes. It is of interest that in the milder cases of human non-freezing cold injury, neurological symptoms appear to predominate, with minor involvement of other tissues.'"' In spite of the earlier suggestion that cooling could cause paranodal or segmental demyelination,6 our own ex eriments confirm the view expressed by others f18,z' that axonal damage is the characteristic feature of non-freezing cold injury. Its selective effect on large myelinated fibers, previously re orted after local nerve cooling in a tro~gh,2~"'~~ and ' ~ after tail i r n r n e r s i ~ n ~ ~has ~~'' also been confirmed. Although, in healthy rabbits, maximal NCV is slightly greater than MCV in the tibial nerve,7 it appears that the fastest-conducting motor and afferent fibers are similar in their susceptibility to injury- as the reduction in velocity we recorded after cooling was not significantly different for the 2 groups (16% for NCV and 15% for MCV). The mechanism of nerve damage by nonfreezing cold injury remains obscure. It has been suggested in the past that it might be due to changes in blood vessels and to nerve ischemia.6 However, for reasons given in our previous paper, l this seems less likely than a direct effect of low temperature on nerve fibers. One argument against an ischemic mechanism in our previous experiments was the short duration of cooling used. For example, cooling for 2 hours in a trough at 1°C was sufficient to produce severe nerve injury, whereas previous experiments in the rabbit have shown that ischemia of this duration was without any long-lasting effect on the tibial nerve.15 This argument cannot be used in relation to whole limb cooling for periods of 12 to 14 hours, but another difference between the effects of ischemia and cooling, which was discussed previously, l 1 also apply to the present experiments. This is the fact that ischemic injury to the tibial nerve in the rabbit has been shown to produce a mixture of Wallerian degeneration and selective demyelinati~n,~ with evidence of conduction block in surviving fiber^.^ The absence of conduction
'
block after cold injury suggests a different mechanism of nerve damage. A clue to etiology may perhaps be derived from our finding that, after hind limb immersion in the rabbit, the main site of damage in both motor and sensory fibers was proximal rather than distal, occurring in the calf rather than in the foot. A possible explanation of this proximal site of nerve damage might be the temperature gradient along the tibial nerve in the first few centimeters below the surface of the bath. At this level, at which the temperature gradient along the nerve is steep, one might expect to find relatively abrupt changes in axonal transport, and also a mismatch between tissue temperature and that of the blood perfusing it. Whether these could lead to changes in membrane permeability and to a rise in intracellular calcium is speculative, but an interpretation of this kind could explain why nerve damage was readily produced within 2 hours by local nerve cooling in a trough, with steep temperature gradients at each edge, whereas whole-limb immersion had to be continued for a much longer period to produce lesions. Alternative explanations for the proximal site of nerve damage during limb cooling seem less cogent. In the normal rabbit hind limb, there is some reduction in the maximal diameter of myelinated fibers in the foot compared with those in the leg, but the difference is small and hardly likely to explain the localized pathology we have found. While a difference in the fatty acid composition of lipid components of membranes at proximal and distal levels is another possible factor, there are no data for rabbit peripheral nerves to indicate that such a difference in chemical composition exists. It is known that, in some arctic mammals, there are measurable differences in the fatty acid composition of lipid between proximal and distal parts of the and that in some coldadapted birds, impulse conduction can persist in distal nerves at temperatures which block conduction p r ~ x i m a l l y but , ~ these cold adaptations are of uncertain relevance to the rabbit hind limb.
REFERENCES 1. Afifi AK, Kimura J , Bell WE: Hypothermia-induced re-
4. Blackwood W, Russell H: Further experiments in the study
versible polyneuropathy: electrophysiologic evidence of axonopathy. Paedzat Neurol 1988;4:49 - 53. 2. Basbaum CB: Induced hypothermia in peripheral nerve: electron microscopic and electrophysiological observations. J Neurocytol 1973;2:171- 187. 3. Blackwood W, Russell H: Experiments in the study of immersion foot. Edin Med J 1943;50:385-398.
of immersion foot. Edin Med J 1945;52:160- 165. 5. Chatfield PO, Lyman CP, Irving L: Physiological adaptation to cold of peripheral nerve in the leg of the Herring Gull (Larus Argentatus). A m J Physiol 1953;172:639-644. 6. Denny-Brown D, Adams RD, Brenner C, Doherty MM: T h e pathology of injury to nerve induced by cold. J Neuropathol Exp Neurol 1945;4:305-323.
966
Non-Freezing Cold Injury
MUSCLE & NERVE
October 1991
7. Fowler CJ, Gilliatt RW: Conduction velocity and conduction block after experimental ischaemic nerve injury. J Neurol Sci 198 1 ;52:221-238. 8. Gutmann E: Factors affecting recovery of motor function after nerve lesions. J Neurol Neurosurg Psychiatry 1942;5:81-95. 9. Hess K, Eames RA, Darveniza P, Gilliatt RW: Acute ischaemic neuropathy in the rabbit./ Neurol Sci 1979;44: 19-43. 10. Irving L, Schmidt-Nielsen K, Abrahamsen NBS: O n the melting points of animal fats in cold climates. Physiol Zoology 1957;30:93- 106. 1 1 . Kennett RP, Gilliatt RW: Nerve conduction studies on experimental non-freezing cold injury. 1 : Local nerve cooling. Muscle Nerve 1991;14:553-562. 12. Lange K, Weiner D, Boyd LJ: T h e functional pathology of experimental foot immersion. Am Heart J 1948;35:238247. 13. Large A, Heinbecker P: Nerve degeneration following prolonged cooling of an extremity. Ann Surg 1944;120:742-749. 14. Lewis RB, Moen PW: Experimental immersion leg. A m J Med Sci 1952;224:529-539. 15. Lundborg G: Ischaemic nerve injury: experimental studies on intraneural microvascular pathophysiology and nerve function in a limb subjected to temporary circulatory arrest. Scand J Plastic Reconstr Surg 197O;(suppl 6):1 - 113.
Non-Freezing Cold Injury
16. Meng MS, West GC, Irving L: Fatty acid composition of caribou bone marrow. Comp Biochem Physiol 1969;30: 187191. 17. Nukada H, Pollock M, Allpress S: Experimental cold injury to peripheral nerve. Brain 1981;104:779-811. 18. Peyronnard JM, Pedneualt M, Aguayo AJ: Neuropathies due to cold: quantitative studies of structural changes in human and animal nerves, in Neurology. International Congress Series No. 434, Amsterdam, Excerpta Medica, 1978, pp 308-329. 19. Reynolds ES: T h e use of lead citrate at high p H as an electron-opaque stain in electron microscopy. J Cell Biol 1963;17~208-2 12. 20. Sayen A, Meloche BR, Tedeschi G, Montgonery H: Experimental immersion foot: observations in the chilled leg of the rabbit. Clin Sci 1960;19:243-253. 21. Schaumburg H , Byck R, Herman R, Rosengart C: Peripheral nerve damage by cold. Arch Neurol 1967;16:103- 109. 22. Smith JL, Ritchie J, Dawson J: Clinical and experimental observations on the pathology of trench frostbite. J Path Bact 1915;20:159-190. 23. Ungley CC, Blackwood W: Peripheral vasoneuropathy after chilling “immersion foot and immersion hand.” Lancet 1942;ii:447-45 1 . 24. Ungley CC, Channel1 GD, Richards KL: T h e immersion foot syndrome. B r J Surg 1945;33:17-31.
MUSCLE &
NERVE
October 1991
967
