Effects of Aldose Reductase Inhibitors on the Progression of Nerve Fiber Damage in Diabetic Neuropathy Douglas A

1 Diab Camp, 6:35-38

Effects of Aldose Reductase Inhibitors on the Progression of Nerve Fiber Damage in Diabetic Neuropathy Douglas A. Greene

ABSTRACT Aldose reductase inhibitors improve nerve reductase inhibitor revealed a reduction in biochemistry, function, and structure in diabetic intermediates of the polyol pathway. Specific animals and increase nerve conduction in diabetic morphologic lesions that correlate with the degree patients. Nevertheless, it has been difficult to of clinical and electrophysiologic impairment also demonstrate a benefit from these agents in patients were identified. Morphologic evaluation of sural with clinically overt diabetic neuropathy. Direct nerve biopsies obtained after aldose reductase measurement of the nerve tissue penetration and inhibitor treatment suggests that these biochemical and biological potency of these biochemically effective compounds ameliorate compounds is essential to fully understand and clinically relevant structural lesions in patients evaluate their effectiveness. Human sural nerve with diabetic neuropathy. (Journal of Diabetes and biopsies obtained from diabetic neuropathic Its Complications 6;1:35-38, 1992.) patients undergoing treatment with an aldose

INTRODUCTION potassium adenosine triphosphatase (ATPase), a decrease in energy consumption of the nerve tissue, an- Only a decade ago, the long-term neuro- dalterations in nerve conduction. As a result, it ap- logic complications of diabetes mellitus pears that intra-axonal sodium accumulates at the were attributed to nerve degeneration or node of Ranvier and causes localized swelling and the demyelination. More recently, attention 0 characteristic lesion, axo-glial dysjunction. The phar- has focused on a potentially reversible metabolic dis- macologic inhibition of aldose reductase represents an order in which the enzyme, aldose reductase, in re- opportunity to reduce sorbitol accumulation and to sponse to hyperglycemia, promotes increased inter- arrest or reverse nerve damage that occurs as a com- cellular conversion of glucose to sorbitol and de- plication of diabetes mellitus. creased uptake of the cyclic hexanol, myo-inositol.‘,2 Aldose reductase inhibitors block the conversion of The depletion of myo-inositol in nerve tissue appears glucose to sorbitol, the depletion of myo-inositol, and responsible for a variety of metabolic and electrophys- the impairment of sodium-potassium ATPase;2 how- iologic alterations, including a decrease in sodium- ever, their effectiveness in ameliorating clinically overt diabetic neuropathy is uncertain. Short-term studies with aldose reductase inhibitors have not been Reprint requests to be sent to: Dr. Douglas A. Greene, Michigan uniformly effective. 3-9 It may be that the accumula- Diabetes Research and Training Center, University of Michigan, tion of sorbitol contributes little to the chronic nerve 1150 West Medical Center Drive, Ann Arbor, MI 48109, USA. fiber damage that is characteristic of diabetic neuropa- Michigan Diabetes Research and Traininp: Center. Universitv , of Michiga;, Ann Arbor, Michigan, USA ” thy. Long-term studies with aldose reductase inhibi-

0 1992 Journal of Diabetes and Its Complications 10568727/92/$5.00 36 GREENE ] Diab Comp 1992; 151

tom that focus on outcome measures of nerve fiber polyneuropathy from a normal reference population. structure and density are more likely to provide in- Further, diabetics without symptomatic peripheral sights into the etiology and pathology of chronic neuropathy appear to be grouped with the larger nor- symptomatic diabetic neuropathies. Further, detailed mal population, rather than with the patients having measurements of changes in nerve fiber parameters diabetic polyneuropathy. This system allows one to can provide an effective means of comparing different differentiate between neuropathic and non-neuro- aldose reductase inhibitors. pathic diabetics on the basis of a threshold of pa- STRUCTURAL ABNORMALITIES ASSOCIATED thology. WITH DIABETIC NEUROPATHY A similar system has been developed at our institu- tion. It is called the index of normality and is obtained Three structural abnormalities, the so-called triopathy from the product of fiber density and the percentage of diabetic neuropathy, are observed in patients with of fibers with normal appearance.2 This method dif- diabetic neuropathy: fiber loss, fiber atrophy, and fers from that of Dyck et al. lo in that it does not correct specific fiber lesions. Patients with diabetic polyneu- for age and gender. Using this and other quantitative ropathy demonstrate a reduced number of functional measures, we have investigated the effects of differ- myelinated nerve fibers, and those fibers that survive ent aldose reductase inhibitors on nerve fiber mor- are reduced in size. This decrease in axonal size is phology. exhibited as a reduction in cross-sectional area. The three components of the triopathy provide a THE EFFECTS OF ALDOSE REDUCTASE series of measurable and quantifiable morphometric INHIBITORS ON NERVE FIBER MORPHOLOGY parameters which exhibit a relationship to the dura- Three long-term, double-blind, randomized, con- tion of diabetes in patients with diabetic neuropathy. trolled studies of aldose reductase inhibitors are avail- For example, fiber loss can be quantified as a decrease able for the morphologic examination of human nerve in fiber density, measured as the number of fibers per specimens. In each study, sural nerve biopsies were square millimeter of a cross-sectional area of nerve. obtained prior to and after at least 12 months of ther- Fiber atrophy can be characterized several ways; these apy, and morphometric techniques were employed to include a decrease in the mean fiber size, an increase quantify nerve fiber changes. in the myelin endpoint, or a decrease in the ratio of Sorbinil North American Neuropathy Trial. Sural axonal to myelin area. Fiber atrophy in an axon can be nerve biopsy specimens were obtained from 16 dia- demonstrated by changes in the relationship between betics with symptomatic peripheral polyneuropathy the area of the axon and the number of myelin lamella. prior to and 12 months after sorbinil250 mg per day Because myelin assumes a wrinkled appearance when (n = 10) or placebo (n = 6) therapyq2 Although the atrophied, increases in the percentage of wrinkled mean sural nerve sorbitol concentration in sorbinil- myelin is also evidence of fiber atrophy. treated patients was significantly decreased when Specific fiber lesions can be identified by the accu- compared with baseline measurements, no significant mulation of a series of discrete abnormalities on mi- change was noted in placebo-treated patients. Of crographs or as a decrease in the percentage of nor- note, sural nerve sorbitol levels actually increased in mal-appearing fibers. Many of these changes occur one patient who discontinued the drug. This clearly initially at the node of Ranvier, and one of the earliest indicates that sorbinil enters peripheral nerve tissue lesions is the axo-glial dysjunction. This lesion is de- scribed as a loss of tight junction complexes that nor- at the site of aldose reductase and decreases sorbitol mally maintain the integrity of the node of Ranvier by accumulation by inhibiting polyol activity. anchoring terminal loops of myelin to the axolemma. A number of morphometric findings were available More advance abnormalities that occur are swelling from this study. Fiber loss was partially corrected, as of the node and perinodal region and perinodal demy- indicated by a statistically significant increase in nerve elination followed by remyelination. Frequently, in- fiber density in sorbinil-treated patients and no creased segmental demyelination also occurs. In re- change in the placebo-treated patients. The increase sponse to these injuries, increased neurofiber regen- in nerve fiber density was accounted for by a several- eration may be demonstrated with regenerating fold increase in nerve fiber regeneration. It should be sprouts occupying the area derived from the previ- noted that the failure to observe deterioration of nerve ously degenerated axon. fiber density in placebo-treated patients during the It is possible to quantify the overall degree of pa- la-month study is consistent with the slow progres- thology that is present by using a variety of different sion of diabetic polyneuropathy. scoring systems. One method uses a compound pa- The number of specific fiber lesions, including axo- rameter score (density, atrophy, and lesions) that is glial dysjunction, were increased at baseline in both normalized for age and gender.” This approach the placebo and sorbinil groups. During the study, clearly distinguishes patients with diabetic peripheral these returned to normal in the sorbinil group but 1 Diab Comp 1992; 6:l ALDOSE REDUCTASE INHIBITORS 37

were unchanged in the placebo group. Similarly, axo- reached 3.5-5.5 nmol/mg protein in patients whose nal atrophy, as evidenced by either the axon/myelin glucose nerve concentrations were between 250 and ratio or the presence of excessive myelin wrinkling, 350 nmol/mg protein. In contrast, sorbitol concentra- was improved after sorbinil, indicating that aldose re- tions in tolrestat-treated patients reached a plateau at ductase inhibition exerted a beneficial effect on fiber between 1 and 1.5 nmol/mg protein, despite sural atrophy. Sorbinil-treated patients had an increase in nerve glucose concentrations that ranged from less the overall percentage of nerve fibers that showed a than 75 to almost 250 nmol/mg protein. A decrease in normal appearance due to the repair of previously the glucose-mediated rise in fructose also was noted damaged fibers. This increase in normality was cor- with tolrestat treatment but not with placebo, which rected for the increase in regeneration, and thus, ac- indicates a significant inhibition of the polyol pathway tually represents repair of previously damaged by tolrestat. neurons. No statistically significant differences were revealed When this data was combined into the overall score in a comparison of nerve morphometry in patients for index of normality, an improvement in the index receiving open-label tolrestat who were randomly of normality was noted with sorbinil, as compared reassigned to double-blind treatment with tolrestat or with no change in the placebo group. This indicates placebo. This is expected based on the slow rate of that biochemically effective aldose reductase inhibi- progression of diabetic neuropathy. There was, how- tors appear to produce consistent improvement in the ever, a statistically significant relationship between structural abnormalities associated with diabetic neu- the duration of treatment with tolrestat and the extent ropathy. of nerve fiber regeneration. Furthermore, a negative correlation was observed between the duration of Statil North American Neuropathy Trial. In this treatment with tolrestat and axo-glial dysjunction, double-blind, controlled study, sural nerve biopsy which suggests that prolonged treatment with tolres- specimens were obtained from 47 neuropathic dia- tat stimulated fiber regeneration and prevented wors- betics prior to and 18 months after treatment with ening or improved axo-glial dysjunction. statil600 mg per day (n = 21) or placebo (n = 26). *I A more-extensive evaluation of nerve fiber mor- No significant change was observed in nerve sorbitol phology was conducted using tolrestat-treated pa- concentrations or in the index of normality in either tients from the United States study and comparing statil- or placebo-treated patients. This suggests that them with two reference populations of untreated dia- statil did not achieve tissue levels in the peripheral betic patients entering other aldose reductase inhibi- nerve that were adequate to inhibit aldose reductase. tor trials. Recently, this reference population was ex- Consistent with the lack of a biochemical effect, there panded by the addition of younger patients with a was no improvement in any morphometric param- shorter duration of diabetes. eters with statil compared with placebo. Three statistical approaches were used to correct for Tolrestat United States Trial. This was an open- differences in age and duration of diabetes between label study in which single sural nerve biopsies were the groups. The data for the reference populations obtained after 3.5-4 years of treatment with tolres- were adjusted with age and nondiabetic controls for tat. l2 Patients were then entered into a 12-month dou- each of the morphometric parameters. Data also were ble-blind, randomized continuation or withdrawal adjusted for age and duration of diabetes. Finally, a treatment period. A sural nerve biopsy was obtained complex analysis of the expanded diabetic reference at the end of the 12-month period. Eighteen neuro- population was performed using a quadratic surface pathic patients with long-standing diabetes (duration adjustment paradigm to look at within group reIation- of approximately 25 years) who were treated for 3-5 ships regarding both age and duration of diabetes. years with tolrestat 200-400 mg per day in the open- A significant increase in fiber regeneration was ob- label study were randomized to continue receiving served in untreated diabetics when compared with tolrestat (n = 9) or were switched to placebo treat- nondiabetic controls. This increase would be expected ment (n = 9). These patients tended to be older and from any patient with diabetic peripheral polyneurop- predominantly insulin-dependent diabetics; thus, it athy. However, tolrestat treatment was associated can be assumed that they had advanced peripheral with a fourfold increase in fiber regeneration com- neuropathy prior to entering the study. pared with untreated diabetics. Axo-glial dysjunction Sorbitol and glucose concentration measurements was significantly increased in untreated diabetics with from sural nerve biopsies obtained prior to and after polyneuropathy, as compared with nondiabetic con- 12 months of therapy showed that tolrestat inhibited trols, and was essentially normal after tolrestat treat- sural nerve sorbitol accumulation. In placebo-treated ment. This indicates that tolrestat had corrected the patients, there was a glucose-mediated rise in sural increase in axo-glial dysjunction that was seen in the nerve sorbitol levels. Peak sorbitol concentrations other reference populations. SimiIarIy, segmental de- 38 GREENE J Diab Comp 1992; 63

myelination, which was increased in the reference MB, Lattimer SA, Greene DA: Regeneration and repair population compared with nondiabetic controls, was of myelinated fibers in sural-nerve biopsy specimens from patients with diabetic neuropathy treated with normalized in patients who were treated with tolres- sorbinil. N Engl J Med 319:548-555, 1988. tat for 3.5-4 years. 3. Fagius J, Brattber A, Jameson S, Berne C: Limited ben- Analysis of the effects of tolrestat on fiber density efit of treatment of diabetic polyneuropathy with an was more complex because fiber density is a reflection aldose reductase inhibitor: A 24-week controlled trial. of the course of diabetes over many years, which Diabefologia 28323-329, 1985. makes adjustments for age and duration of diabetes 4. Young RJ, Ewing DJ, Clarke BF: A controlled trial of difficult. Compared with untreated diabetics, there sorbinil, an aldose reductase inhibitor, in chronic pain- was a trend for improvement in fiber density in the ful diabetic neuropathy. Diabetes 32:938-942, 1983. tolrestat group. But when the larger reference popula- 5. Lewin IG, O’Brien IAD, Morgan MH, Corral1 RJM: tion-which included many patients with less severe Clinical and neurophysiological studies with the aldose reductase inhibitor, sorbinil, in symptomatic dia- neuropathy-was used for comparison, there ap- betic neuropathy. Diabetologia 26:445-448, 1984. peared to be a decrease in fiber density in the tolrestat 6. Lehtinen JM, Hyvonen SK, Uusitpa M, Pahakainen E, group; thus, on the basis of these patient populations, Halonen T, Kilpelainen H: The effect of sorbinil treat- it is not possible to make a conclusion about the effect ment on red cell sorbitol levels and clinical and electro- of tolrestat on nerve fiber density. physiologic parameters of diabetic neuropathy. 1 Neu- rol 223:174-177, 1986. CONCLUSION 7. Christensen JEJ, Varnek L, Gregersen G: The effect of an aldose reductase inhibitor (Sorbinil @) on diabetic These data support the contention that nerve fiber neuropathy and neural function of the retina: A dou- regeneration is a sensitive and consistent marker of ble-blind study. Acta Neural Stand 71:164-167, 1985. tissue-specific biological activity of aldose reductase 8. Fagius J, Jameson S: Effects of aldose reductase inhibi- inhibitors in clinical and overt diabetic periphera1 tor treatment in diabetic polyneuropathy-a clinical polyneuropathy. Data from long-term studies with and neurophysiological study. J Neural Neurosurg Psy- two of the three aldose reductase inhibitors presented chiatry 44:991-1001, 1981. herein demonstrate a broad spectrum of morphomet- 9. Handelsman DJ, Turtle JR: Clinical trial of an aldose ric improvements that are indicative of overall tissue reductase inhibitor in diabetic neuropathy. Diabetes 30:459-464, 1981. repair. This data strongly argues for the role of the polyol pathway in the pathogenesis of clinically overt 10. Dyck PJ, Karnes JL, Daube J, O’Brien I’, Service FJ: Clinical and neuropathological criteria for the diagno- diabetic peripheral polyneuropathy and maintains the sis and staging of diabetic polyneuropathy. Brain hope that treatment with aldose reductase inhibitors 108:861-880, 1985. will reduce the progression or improve and partially 11. Greene DA, Sima AAF, Statil North American Painful reverse neuropathy even at late stages when the pa- Neuropathy Study Group: Efficacy of aldose reductase tient is clinically impaired. inhibitors in human sural nerve (abstract), presented at symposium, Controversies in Etiology & Treatment REFERENCES of Diabetic Neuropathy. New York, June 1991. 12. Greene DA, Bochenek W, Harati Y, Sima AAF, Hoh- 1. Greene DA, Lattimer SA, Sima AAF: Sorbitol, phos- man T, Hicks D, Beg M, Conen B, Tolrestat Study phoinositides, and sodium-potassium-ATPase in the Group: Biochemical and morphometric response to tol- pathogenesis of diabetic complications. N Engl J Med restat in human diabetic nerve (abstract), presented at 316:599-606, 1987. European Association for the Study of Diabetes. CO- 2. Sima AAF, Bril V, Nathaniel V, McEwen TAJ, Brown penhagen, September 1990.