Pathological, morphometric, and teased fiber studies of sural nerve from 36 diabetic patients with (n = 32) and without (n = 4) neuropathy and from 47 healthy subjects provide evidence that in diabetic polyneuropathy: (1) fiber loss is primary; (2) demyelination and remyelination with or without onion bulb formation are secondary; (3) remaining fibers, on average, have the same ratio of small to large fibers as in healthy individuals, but with a greatly increased variability; and (4) the spatial distribution of fiber loss is both diffuse and multifocal. Criteria developed during the study of experimental models of ischemic neuropathy were employed to assess whether ischemic nerve damage had occurred in diabetic polyneuropathy. We conclude that there is increasing evidence that microvascular pathological abnormality and ischemia may be involved in the pathogenesis of human diabetic polyneuropathy. Cases with selective loss of small or large afferent fibers are probably extremes of a normal distribution and not different disorders.

Characterization and quantitation of the spatial distribution of pathological abnormalities along the length of nerves may be helpful in understanding the underlying mechanisms of diabetic polyneuropathy. To this end, by examining transverse sections of nerve roots and proximal-to-distal levels of lower limb nerves in 9 controls and 15 diabetic patients with polyneuropathy, we have determined the myelinated fiber (MF) number, size distribution, median diameter, and variability of density (MFs/mm2) among frames and among fascicles. Even in cases with mild polyneuropathy, fiber loss, a decrease in the median diameter, and an increase in the variability of density among frames and among fascicles began in proximal nerve and extended to distal levels. Multifocal fiber loss along the length of nerves and sprouting provide the best explanation for these findings. The pattern is dissimilar from that observed in diffuse metabolic disease of Schwann cells, neuronal degeneration, and dying-back neuropathy, but like that found in experimental ischemic neuropathy induced by embolization of nerve capillaries.

Whether compression nerve injury is due to ischemia, direct mechanical injury, or both remains unsettled. To assess structural changes of nerve during compression, peroneal nerves of rats were compressed at various pressures for different times, and the structural alterations were stopped by simultaneous in situ and perfusion fixation. The structural changes observed during a few minutes of compression cannot be explained by ischemic injury because the pathologic alterations characteristic of ischemia take many hours to develop and in any case are different from the ones found here. The pressure-and time-related structural changes observed in the present study under the cuffwere (i) decrease in fascicular area and increase in fiber density due to expression of endoneurial fluid; (i) compression and expression of axoplasm, sometimes to the point of fiber transection; (iii) lengthening of internodes; and (iv) obscuration of nodes of Ranvier due to cleavage and displacement of myelin and overlapping of nodes by displaced loops of myelin. At the edges of the cuff the changes were (i) increase of fascicular area probably from expressed endoneurial fluid; (ii) widening of nodal gaps, perhaps mainly from translocated axonal fluid; and (iii) disordered structure of axoplasm. We suggest that the process of paranodal demyelination and axonal transection are linked, occur during the act of compression, and are due to shear forces. The initial event is expression of endoneurial fluid, followed by compression and expression of axoplasm and cleavage and displacement oflayers of myelin. Conceivably, with prolonged cuff compression ischemic injury might be found to be superimposed on mechanical injury.In humans, nerve compression injury may follow use of a tourniquet at too high a pressure, for too long a time, or from improper application; lying in one position without moving for a long time (as may occur during anesthesia, inebriation, or coma) with a limb nerve compressed against bone by a protruding ridge or hard surface; or prolonged sitting with the legs crossed or prolonged leaning on the elbows (1, 2). Excellent recovery is expected after compression injury, whereas it is usually delayed and faulty after nerve transection. This difference in outcome is usually explained by conduction block and demyelination in the former and complete fiber degeneration and faulty regeneration in the latter (3-5).The mechanisms underlying nerve compression injury have usually been attributed to ischemia (6-8), to mechanical forces (9), or to both.In the present study the structural alterations of nerve during compression were stopped by simultaneous in situ and perfusion fixation, usually during short periods of compression. We found nodal lengthening and other acute structural changes after only a few minutes of compression, which appear to explain the paranodal demyelination and axonal degeneration that are characteristic of nerve compression injury. The structural alterations that develop during acute compression are different...

Permanent axotomy in cats produced by hind limb ablation results in sequential pathological alterations of myelinated fibers of the proximal nerve stump. The changes are like those previously described for human uremic neuropathy and such system atrophies as Friedreich's ataxia: axonal atrophy + myelin wrinkling + nodal lengthening and internodal demyelination + remyelination. [16, 371. The alteration is due to breakdown of the terminal parts of internodes and subsequent formation of short intercalated internodes [27, 291, or breakdown of the entire old internode and formation of several short new internodes. Segmental absence of myelin (from nodal lengthening or partial or complete internodal demyelination) is, however, only rarely encountered in health [ 101.Segmental absence of myelin and increased variability of myelin thickness between internodes are encountered in several peripheral nerve diseases [lo]. Such changes occur as the result of application of external pressure [8,9,321, at points of entrapment [l], from endoneurial deposition of amyloid [14] Figure 1.To test this proposed cellular sequence of axonal atrophy and to study the underlying molecular mechanisms, an experimental model of chronic axonal atrophy was required. X-irradiation of lumbosacral spinal cord and ganglia, with careful shielding of limb nerves and their Schwann cells for later studies, was tried but provided equivocal results. Administration of various neurotoxins to inhibit perikaryal protein synthesis was considered unsuitable since Schwann cell synthesis might also be affected. In the cat, permanent axotomy by hind limb removal was shown to result in axonal atrophy of MFs of the ventral and dorsal roots [5]. We now report that axonal atrophy develops in the proximal peripheral nerve stump and that it leads sequentially to myelin wrinkling, nodal lengthening, internodal demyelination, and, finally, remy elination. Materials and MethodsUnder aseptic conditions and pentobarbital anesthesia, the left hind limb of mongrel cats was amputated at the hip. Cats were housed in large cages with solid floors so they could walk about freely. The cage contained a styrofoamcovered ledge for sleeping. Comparable levels of the seventh lumbar segmental nerve just distal to the spinal ganglion (proximal segment) and of the sciatic nerve 1 cm above the amputation level were removed after perfusion From the Peripheral Nerve Laboratory, Mayo Medical School and Foundarion, Rochester, MN 55901.

The peripheral axons of lower motor and spinal ganglion neurons were permanently transected and not allowed to regrow to target tissue in adult cats by amputation of the hind limb at the hip. The number and sizes of L-7 lower motor neurons at two levels (cell bodies of lateral group motor neurons and myelinated fibers [MFs] of ventral root) and of L-7 spinal ganglion neurons at two levels (cell bodies of L-7 spinal ganglion and MFs of dorsal root) were morphometrically evaluated in groups of cats at 3 months, 9 months, and 18 months after amputation and compared with the number and sizes of neurons in controls or with those on the opposite side. The number of neurons decreased only minimally after amputation. The diameter of neuron cell bodies was only equivocally reduced. By contrast, the median diameter and the peak diameter of both large and small MFs of dorsal and ventral nerve roots were significantly (approximately 30%) less than those of controls. This reduction in diameter of MFs is judged to be related to chronic axonal atrophy rather than to selective loss of large fibers. Permanent transection of distal axons should therefore prove to be a good model of chronic axonal atrophy.