Revie" Article

Drugs 39 (2): 190-202, 1990 0012-6667/90/0002-0190/$06.50/0 © ADlS Press Limited All rights reserved. DRUG03310

Aldose Reductase Inhibitors in the Treatment of Diabetic Neuropathy

A Review of the Rationale and Clinical Evidence E.A. Masson and AJ.M. Boulton Manchester Royal Infirmary. Manchester, England

Contents

Summary ................................................................................................................................... 190 l. Diabetic Neuropathy - History and Definitions ............................................................... 190 2. Aldose Reductase, the Polyol Pathway and the Myoinositol Hypothesis ....................... 192 3. Clinical Trials of Aldose Reductase Inhibitors in Human Diabetic Neuropathy .......... 194 4. Toxicity of Aldose Reductase Inhibitors ............................................................................ 198 5. Discussion ............................................................................................................................. 198 6. Conclusions ........................................................................................................................... 200

Summary

This review considers the definition of clinical diabetic neuropathy and the theoretical basis for the use of aldose reductase inhibitors in the treatment of distal sensorimotor neuropathy, the most common clinical problem. Myoinositol depletion is related to hyperglycaemia-induced pol yo I activity, changes which are associated with early functional deficits in acute experimental diabetes. These changes are reversible by the administration of aldose reductase inhibitors, and this provides the rationale for the treatment of human diabetic neuropathy with these agents. Many early trials of these drugs have produced some evidence of clinical benefit in patients with diabetic neuropathy, but interpretation of data is difficult as patient selection and neuropathy definition are not yet standardised. In addition, it is possible that once the neuropathic process is initiated, there is a point where it becomes irreversible, and treatment with aldose reductase inhibitors may therefore be of more relevance in early neuropathy. Long term double-blind multicentre trials are in progress, and preliminary data from some of these are reasonably encouraging. In conclusion, the results from clinical trials of the aldose reductase inhibitors in this difficult area are sufficiently encouraging to lead us to be optimistic about their future development, and continuing work should clarify their potential role with respect to the prophylaxis and treatment of diabetic neuropathy.

1. Diabetic Neuropathy - History and Definitions Polyneuropathy has been said to be one of the

most common of the long term complications of diabetes mellitus, although estimates of its prevalence have varied widely (Bruyn & Garland 1970; Clements & Bell 1982), depending on the minimal

Aldose Reductase Inhibitors in Diabetic Neuropathy

Table I. Clinical classification of the diabetic neuropathic syndromes. Chronic distal sensorimotor neuropathy is the most common clinical problem (based on Boulton & Ward 1986)

Mononeuropathies

Polyneuropathies

Cranial Entrapment (e.g. carpal tunnel syndrome) Radiculopathy (root lesions) Multiplex

Proximal motor (amyotrophy) Autonomic Acute sensory Chronic distal sensorimotor

criteria for diagnosis, and the population involved. When standardised criteria are applied, estimates of prevalence usually vary between 12 and 50% (Boulton et al. 1985; Thomas & Eliasson 1975). It is generally agreed, however, that incidence and severity vary with age, duration of diabetes, and possibly with the quality of glycaemic control. Although the symptoms typical of diabetic sensory neuropathy were first described by Pavy in 1887, it is now well recognised that many diabetic patients have peripheral nerve abnormalities without symptoms, and also that those with severe symptoms may have little or no clinical or electrophysiological evidence of nerve dysfunction (Archer et al. 1983). It is clear, therefore, that there is no single entity which can be described as 'diabetic neuropathy'; rather, the term encompasses a heterogeneous clinical, and possibly pathological, group. Many attempts have been made to provide a rational classification based on aetiological (Clements & Bell 1982) or clinical criteria (Boulton & Ward 1986) [table I). In addition, it must be remembered that diabetic patients may present with an unrelated neuropathy, that they are prone to entrapment neuropathies (such as carpal tunnel syndrome) [Mulder et al. 1961], and that different diabetic neuropathic syndromes which are recognised as clinical entities are not pure and may overlap (Clements & Bell 1982). Recent work (Dyck 1988) and the San Antonio Conference (Consensus Statement 1988) have suggested that a combination of symptom scores, objective tests of sensibility, conventional electrophysiology, and autonomic function assessments are required for a full characterisation of diabetic neuropathy. Most studies of drug

191

treatments in peripheral neuropathies adopt a pragmatic approach to the definition of peripheral sensorimotor neuropathy, in which objective parameters of nerve dysfunction, such as reduced conduction velocities and objective measures of sensory function, are combined with symptom assessments and clinical measures. Various features of diabetic neuropathy make a metabolic background likely: the morphological picture is nonspecific, with axonopathic degenerative changes affecting all fibre types, and segmental demyelination and remyelination. Neuropathy occurs in both major types of diabetes, the prevalence increasing with the duration of the diabetes (Pirart 1978). Energy metabolism has been shown to be disturbed in various ways, and abnormalities of many metabolic processes are established, and in some cases interlinked. Many workers have speculated that there may be a single underlying metabolic mechanism for the development of diabetic complications, and the polyol pathway has been an established candidate for some time. However, reduced nerve blood supply and endoneurial hypoxia have been demonstrated in experimental diabetes (Tuck et a1. 1984), a state which has been confirmed in established human diabetic neuropathy (Newrick et al. 1986). Experiments with rats reared in hypoxic conditions (Low et al. 1986) have produced functional nerve abnormalities similar to those seen in experimental diabetes, and similar Table II. Demonstrated sites of aldose reductase activity and

suggested relationship to diabetic complications Target tissue

Complication

Eye - cornea Eye - retina Eye - lens Kidney (glomerular podocytes) Peripheral nerve Arterial wall

Keratopathy Retinopathy Cataracta

a

Nephropathy Sensorimotor neuropathy Accelerated degenerative vascular disease

The definitely sorbitol-related 'sugar cataract' is actually quite rare in human diabetes. More common is an accelerated 'senile' cataract.

192

Drugs 39 (2) 1990

• Endoneunal t -...---i Endoneurial vascular hypoxia resistance

Hyperglycaemia

.... .....

.........

...........

....

----~---

Structural neuropathy

Fig.·1. Aetiologieal factors implicated in the pathogenesis of diabetic neuropathy. Solid lines represent experimentally established links. There may also be a genetic predisposition.

phenomena are seen in humans with chronic hypoxia secondary to lung disease (Masson et al. 1988). This has led to the suggestion that capillary and rheological changes as yet undefined have an important role in the development of tissue damage. Morphological studies of later diabetic neuropathy have shown considerable microvascular pathology (Yasuda & Dyck 1987), and the spatial distribution of fibre loss has been said to suggest ischaemia {Dyck et al. 1986}, although this point has been disputed {Llewelyn et at. 1988}. In addition, the formation of advanced glycosylation end products has been associated with the development of diabetic complications {Brownlee et at. 1988}. It seems likely that there are a variety of metabolic insults to peripheral nerves as a consequence of insulin deficiency and hyperglycaemia whose relative importance is not yet established {Low 1987}. The authors' view of the aetiological factors involved is summarised in figure I.

2. Aldose Reductase, the Polyol Pathway and the Myoinositol Hypothesis It is now more than 30 years since the first demonstration of sorbitol accumulation in the lens of

diabetic rats with 'sugar' cataracts (Van Heyningen 1959), an observation which .led to the investigation of polyol metabolism and its rate-limiting enzyme aldose reductase as a possible mechanism for the induction of the late diabetic complications. Aldose reductase has been demonstrated in all the important target tissues for diabetic damage {table II} although its physiological function is not known {Flynn 1983}. It was demonstrated in nervous tissue over 20 .years ago (Gabbay et al. 1966; Stewart et at. 1967). The enzyme has a low affinity for glucose, but when there is excess substrate for the normal metabolic pathways it is activated to produce sorbitol which is then reduced to fructose. These and other metabolites accumulate and reach equilibrium at high concentrations in tissues which are insulin-independent for glucose transport, where intracellular glucose levels closely reflect plasma concentrations (Gabbay 1973). Polyol effects can be modelled in several ways (fig. 2): diabetes can be induced in animal species by the administration of pancreatic toxins such as alloxan or streptozocin {streptozotocin}, and spontaneously diabetic strains of rodents have been bred. If non-diabetic animals are fed excessive galactose dramaticpolyol effects are seen, as this sugar is an alternative substrate for aldose reductase, but the

Aldose Reductase Inhibitors in Diabetic Neuropathy

193

product (galactitol) has no pathway for further reduction, and escapes only by diffusion, so that considerable accumulation occurs (Stewart et al. 1967). Such animal models develop characteristic nerve abnormalities of conduction slowing, and resistance to ischaemic conduction block, which have been described as the 'cardinal electro physiological abnormalities' of diabetic neuropathy (Low et al. 1986). Resistance to ischaemic conduction block is a curious and somewhat paradoxical phenomenon, where nerve impulse propagation continues for an unusually long time after local blood flow is occluded, despite the fact that diabetic nerves are susceptible to pressure palsy (Mulder et al. 1961), and are vulnerable to longer term ischaemic insult (Nakuda 1988). In animal models of diabetes, polyol pathway overactivity has been linked to depletion of endoneurial myoinositol (Greene & Lattimer 1984), and is associated with reduced sodium potassium adenosine triphosphate (ATP)-ase activity and reduction of axonal transport of proteins (Tomlinson et al. 1984), both of which have been prevented by the administration of aldose reductase inhibitors. The various stages of deranged metabolism in peripheral nerves are represented in figure 3, and evidence from animal experiments supports this hypothesis in various ways (Greene et al. 1988). .

NADPH

The central role of myoinositol depletion, which has an observed but not directly proven link with polyol overactivity, is supported by the observation that myoinositol normalisation by insulin administration, myoinositol supplementation (Greene et al. 1975), or treatment with aldose reductase inhibitor (Mayer & Tomlinson 1983) restores reduced nerve conduction velocities in diabetic animals. However, studies in these experimental models assess functional changes in acute diabetes, which are also observed in human diabetes at diagnosis (Terkildsen & Christensen 1971; Ward et al. 1971), but are not associated with major structural change, although some pathological lesions have been described in animal models (Sima et al. 1986). In most studies of human diabetic neuropathy considerable structural changes are present in both nerve fibres and endoneurial microvessels (Malik et al. 1989; Timperley et al. 1985). The link from early and reversible functional changes to later structural neuropathy is not, as yet, definitely established. Myoinositol supplementation in human diabetes has nof conclusively .altered function (Gregerson et al. 1983), although one report has suggested improvement in electrophysiological parameters (Salway et al. 1978). Data from studies ofbiopsied human nerves are

NADP+

'----/ Atdose

d·Glucose

NAD+~NADP+

•

reductase

Sorbitol

Sorbitol" ____ _______ • d-Fructose dehydrogenase (potyot dehydrogenase)

Galactose _ _ _A_t_do_s_e _ _ _ __ reductase

Galactitol

II -------71

No metabolic pathway available

~

NADH

NADPH

Fig. 2. The aldose reductase pathway, and the basis of the galactose-fed animal model of polyol-related .abnormalities.

Drugs 39 (2) 1990

194

Hyperglycaemia

Polyol flux

Impaired intracellular signalling

Decreased regenerative response

Structural damage

Conduction slowing

Fig. 3. Simplified summary of the 'myoinosito] hypothesis' of the aetiology of diabetic neuropathy (based on Greene et al. 1988).

confusing in relation to polyol activity and myoinositol depletion. Dyck et al. (1980) and Hale et al. (1987) did not find myoinositol depletion in large series of nerve biopsies. Other reports have demonstrated reduction of myoinositol concentrations (Mayhew et al. 1983), and all found increased poIyol flux, although this may not relate directly to nerve abnormality (Dyck et al. 1980) or to nerve glucose concentration (Malik et al. 1989). Another study (Dyck et al. 1988), of 21 diabetic patients and 11 controls, led to a firm conclusion that myoinositol depletion was not a feature of human diabetic nerve, although polyol concentration related to loss of myelinated fibres. Aldose reductase inhibitor treatment for 1 year in 6 patients decreased sorbitol concentrations, but did not change the myoinositol content of repeat biopsies. However, some authorities (Winegrad et al. 1989) argue that such information is irrelevant to the myoinositol

hypothesis for 2 reasons. Firstly, the insulin independent nature of glucose transport into nerves means that glucose levels and hence polyol products will show wide 24-hour fluctuations, and secondly, it is probably a decrease in extracellular myoinositol uptake which is associated with alterations of cellular energy metabolism rather than total nerve myoinositol.

3. Clinical Trials 0/ Aldose Reductase Inhibitors in Human Diabetic Neuropathy (table III) Early reports of the use of aldose reductase inhibitors in diabetic neuropathy appeared in 1981, and involved the use of the compound alrestatin. Handelsman and Turtle (1981) reported a nonrandomised crossover study of the short term treatment of 9 patients with severe neuropathic symp-

195

Aldose Reductase Inhibitors in Diabetic Neuropathy

toms. Although symptomatic improvement was seen, no objective change could be documented. Fagius and Jameson (1981) reported a double-blind placebo-controlled trial of 30 patients with mild asymptomatic nerve dysfunction, and concluded that several objective parameters were improved by alrestatin. Another study (Culebras et al. 1981) suggested that sensory conduction velocity could be improved by short term (5 days) treatment with intermittent intravenous boluses of alrestatin. This improvement, however, occurred in only 4 of 10 patients treated, with 4 more having an apparent deterioration, and the remaining 2 unchanged. There was no control group and the changes reported were not statistically significant. Toxic effects, particularly the occurrence of skin rashes, were common with the use of this compound. Most subsequent reports have concerned the effects of the compound sorbinil, which although less toxic and more potent than alrestatin, is also commonly associated with skin rash, and occasional se-

rious hypersensitivity reactions have been reported (Pitts et al. 1986). Trials of sorbinil were first reported in 1983 and most have involved small numbers of patients treated for short periods of time. Definitions of neuropathy and study design have varied considerably. Two double-blind placebo-controlled crossover trials (Judzewitsch et al. 1983; Young et al. 1983) have suggested some beneficial effect of80rbinil. The former involved 39 patients with no symptoms of neuropathy, the latter 15 patients with chronic pain. Both reported the effect of short term (4 or 9 weeks) treatment on electrophysiological measures. Judzewitsch and colleagues reported improvements of about I mls in peroneal motor conduction velocities with sorbinil administration, while Young et al. were unable to confirm this although sural nerve action potential amplitude improved. Pain and tendon reflex scores also improved in the latter study, but clinical sensory scores deteriorated. A further study (Jaspan et al. \983) of II patients with painful neuropathy unrespon-

Table III. Summary of clinical aldose reductase inhibitor studies in diabetic neuropathy (see text for details). Large studies with ponalrestat and ONO-2235 are in progress; results are awaited

Reference

Drug

Culebras et al. (1981) Fagius & Jameson (1981) Handelsman & Turtle (1981) Christiensen et al. (1985) Fagius et al. (1985)

Alrestatin Alrestatin Alrestatin Sorbinil Sorbinil

Guyet al. (1988) Jaspan et al. (1983) Judzewitsch et al. (1983) Lewin et al. (1984) O'Hare et al. (1988) Sima et al. (1988)

Sorbinil Sorbinil Sorbinil Sorbinil Sorbinil Sorbinil

Young et al. (1983) Boulton et al. (1986) Boulton et al. (1989) Ryder et al. (1986) Florkowski et al. (1989)

Sorbinil Tolrestat Tolrestat Tolrestat Ponalrestat

Abbreviations: uc

No. of patients

Study design

Duration (months)

Comments

10 30 9 37 55

uc db, co nr, co db, c db, c

5 days 3 2 1 6

39 11 39 13 27 16

db, c uc db, co db, co db, c db, c with biopsy db, co db, c db, c db, c db, c

12 1 2

No effect Minor improvements No objective improvement No effect Electrophysiological improvements No effect Symptomatic improvement Some objective improvement No changes No effect Morphological improvements reported Some objective improvement Concordant improvement in 28% Some objective Improvement Minor changes only No effect

15 550 190 260 55

12 12

12 6 2 6

= uncontrolled; db = double-blind; co = crossover; nr = nonrandomised; c = controlled .

196

sive to conventional treatments reported marked improvements in symptoms and some improvement in muscle power and sensation. However, in this uncontrolled trial the patients treated were a heterogeneous group including 4 with proximal motor neuropathy (diabetic amyotrophy), which makes interpretation of the data even more difficult than usual. A further double-blind crossover trial was reported in 1984 (Lewin et al. 1984). 13 patients ,:;"ith symptomatic neuropathy had 16 weeks of treatment with sorbinil, given randomly during weeks 5 to 8 or weeks 13 to 16 of the study period. One patient was withdrawn following a toxic reaction. Measurements were taken of vibration perception thresholds, motor and sensory conduction velocities; symptoms were assessed using a visual analogue scale. Autonomic function was assessed in 7 patients who were not taking cardioactive medication. No changes were apparent in any of these parameters after sorbinil treatment. It was suggested that this might be due to the advanced age of the patients studied (mean age 60 years), their long duration of diabetes and the severity of the neuropathy. A longer treatment period of 24 weeks was used in a cohort of 55 patients JFagius et al. 1985). This study was also of double-blind design in patients with symptomatic neuropathy. Significant improvements were reported in electrophysiological measures and one of several autonomic function assessments. A sophisticated analysis of the results led to the conclusion that there was a high probability of sorbinil being efficacious in the study group. In the same year, another group reported a negative result in a blinded study of 37 patients who were asymptomatic but had abnormalities of sensory perception thresholds (Christiensen et al. 1985). There are several recent reports of long term trials of sorbinil therapy. One involved 27 patients completing the study protocol, 19 of whom received sorbinil (O'Hare et al. 1988). Patients were selected on the basis of symptoms and signs with objective evidence of nerve dysfunction, either abnormal vibration perception thresholds or nerve

Drugs 39 (2) 1990

conduction velocity. Neuropathy was present for more than 6 months at entry and varied from 'mild' to 'severe'; glycaemic control did not change during the study. After a 2-month placebo run-in period patients were randomised to 12 months of treatment, with full neurological assessments every 3 months. Measurement was made of autonomic function and vibration perception thresholds as well as clinical assessment and conventional electrophysiological parameters. Visual analogue scales were used for self assessment of symptoms. Not only was no difference shown in any parameter measured in the study group, but 4 of the treatment group and 1 of the placebo group developed neuropathic ulceration during the trial. Another study (Guy et a!. 1988) reported the effects of sorbinil therapy for 12 months in a double-blind controlled study of 39 patients with established neuropathy, 21 of whom received sorbini!. Sophisticated neurophysiological assessments were made ·with particular reference to the upper limb, and included sensory perception thresholds, conventional nerve conduction velocity measurements, somatosensory evoked potentials, and autonomic function tests. This study reported no improvement in any of these measures after treatment with the aldose reductase inhibitor. However, the high incidence of foot ulcers and autonomic symptoms in patients studied indicates that the study group had advanced neuropathic changes, which may not be amenable to any form of treatment. The conclusions of this report were that neurophysiological measures of early neuropathic change were necessary in addition to clinical assessment, and that aldose reductase inhibitors were not effective in established diabetic neuropathy. Certainly the most encouraging study of sorbinil so far published is that of Sima et a!. (1988). This was a trial which involved morphological examination of sural nerve biopsy specimens at baseline and after 12 months of sorbinil treatment. 10 patients on sorbinil and 6 on placebo completed the study with paired biopsy specimens. The analysis included clinical, biochemical and .morphological data and the results show a significant effect

Aldose Reductase Inhibitors in Diabetic Neuropathy

of sorbinil in the stimulation of repair phenomena associated with the typical structural lesions of diabetic neuropathy. The treatment group showed a reduction in nerve sorbitol, demonstrating the effectiveness of aldose reductase inhibition in all but I patient whose serum sorbinil concentration was estimated to be subtherapeutic. Myelinated fibre density (reduced from healthy control values by 50% at baseline) increased by 33% in patients treated with sorbinil, with no change observed in the placebo group. Evidence of repair (regenerating fibres and remyelination) was increased by a factor of nearly 4, and 'subtle and selective' clinical and electrophysiological improvements correlated across groups with morphological phenomena. This is a dramatic report which is very encouraging for the future therapy of diabetic neuropathy, as it suggests that even with advanced structural damage improvement is possible. However, in advanced neuropathy resulting from other causes, significant clinical improvement seldom follows removal of the cause, whereas in early nerve dysfunction due to toxic agents, for example, there is often a progressive and sustained improvement when the insult is removed. This is not the case in myoinositol supplementation and aldose reductase inhibitor administration (Low 1987). Another concern is that the significance of electrophysiological benefit was rather less than that of the morphological benefit, although there was a positive correlation. Repeat biopsies were taken from the ipsilateral limb at a proximal site, and the effect of distal biopsy on proximal parts of the nerve is not known. The numbers are small, but nonetheless the results cannot be ignored. The only other report of paired biopsies after aldose reductase inhibitor administration concerned 6 patients and 3 controls (Dyck et al. 1988), and did not report polyol-associated improvements in morphological phenomena. It will be interesting to see the results of a larger biopsy study with aldose reductase inhibitor treatment if such a trial can be undertaken. There are 3 aldose reductase inhibitors currently undergoing clinical study - tolrestat, ponalrestat, and ONO-2235 - all of which are structurally distinct from sorbinil and each other (fig.

197

4). They appear to be less toxic than the previous aldose reductase inhibitors discussed. Tolrestat has been demonstrated to be a potent aldose reductase inhibitor in animal studies (Simard-Dequesne et al. 1985; Sredy et al. 1986). Preliminary reports of large multicentre studies with strict entry and exclusion criteria have suggested that tolrestat is capable of improving both subjective symptoms and objective measures of nerve function, with increases in motor nerve conduction velocity. Boulton et al. (1986) reported a study of 550 patients from 19 centres randomised to take varying doses of tolrestat or placebo for I year. Clinical response was reported in the group receiving the highest dose of tolrestat. Patients who had improvement in both symptoms and objective parameters (nerve conduction velocities) were considered to be clinical responders, those with no change or a deterioration in either were designated treatment failures, and the remainder partial responders. Significant response was seen at 24 weeks and sustained to the end of the study in 28% of treated patients. Ryder et al. (1986) reported similar results in a study of260 patients from 15 centres who had chronic painful neuropathy. Patients were randomised to receive placebo or high-dose tolrestat for 8 weeks. The improvements in the responders in both these studies were modest, but highly statistically significant. The only significant toxicity associated with tolrestat administration to date is an asymptomatic and reversible elevation in hepatic transaminases (Ryder et al. 1987). Similar efficacy data have recently been presented from a European study (Boulton et al. 1989). One report of the use of ponalrestat over 6 months in a small group of patients has failed to demonstrate significant treatment effects (FIorkowski et al. 1989). Preliminary data have been reported from several studies using ONO-2235 (e.g. Goto et al. 1986; Hotta et al. 1989) which appear to show some benefit, but clearly large double-blind controlled trials are required to clarify such reports. Large multicentre studies are in progress with all of these agents in diabetic peripheral neuropathy and the results of these will be of considerable interest.

198

Drugs 39 (2) 1990

CH3

CH30

I

W ~

N............... COOH

~I

~

~

b

CF3

Sorbinil

Ponalrestat

Tolrestat

ONO·2235

Fig. 4. The structures of the major aldose reductase inhibitors currently undergoing clinical trial.

4. Toxicity of Aldose Reductase Inhibitors The early compounds were associated with considerable side effects. Alrestatin caused a very frequent photosensitive skin rash, and was quickly abandoned. Sorbinil was also associated with the occurrence of fever and skin rash in around 10% of patients, and several more serious hypersensitivity reactions have occurred including StevensJohnson syndrome and toxic epidermal necrolysis (Pitts et al. 1986); however, no significant organ toxicity was reported. The later drugs have been associated with less acute toxicity. Tolrestat caused a reversible elevation in hepatic transaminases in 2% of treated patients, and there was occasional dizziness, thought to be related to postural hypotension (Ryder et al. 1987). There are no indications so far of significant toxicity from the use of ponalrestat or ONO-2235. No data are available on potential long term toxic effects.

5. Discussion The demonstration of clinical benefit in patients with peripheral neuropathy is difficult. As previously discussed, the neuropathies associated with diabetes are heterogeneous, with the major clinical problems being pain and neuropathic foot ulcera-

tion. The latter represents an 'end-stage' of sensory deficit with profound fibre damage, and it would be asking much for any therapy to have a significant beneficial effect. Clinical experience with severe neuropathies of other types would not suggest that removal of the cause at a late stage of denervat ion will lead to major improvements. The former can occur with or without severe structural damage. Many of the studies thus far published have had loose entry criteria and have included patients with heterogeneous clinical problems. There is difficulty in defining 'hard' endpoints for all measures of nerve integrity. Symptoms have usually been undefined, and subjective assessment is hampered by large placebo effects. The usual method of assessment is simply to have a 5-point scale of intensity or to use a continuous visual analogue scale. A greater insight into symptomatic experience can be gained rapidly by the use of simple questionnaires such as the McGill pain questionnaire, in which patients select words from groups describing different aspects of sensory experience. This approach has been advocated by Dyck (1988) and a recent study (Masson et al. 1989) has shown this approach to have some specificity for symptoms in diabetic neuropathy. It has not, however, been employed in trials of aldose reductase inhibition.

Aldose Reductase Inhibitors in Diabetic Neuropathy

Standard electro physiological measures have a relatively low variability in skilled hands so it follows that reports of improvements in, for example, conduction velocities are indeed evidence of real drug effect. However, the translation of such changes to clinical benefit is not straightforward. Conduction velocities only assess the integrity of large myelinated fibres, which are a numerical minority of the whole nerve fibre population. 'Objective' measures of sensory perception thresholds are an attractive concept as they are closer to measuring real functional capability. Unfortunately, they are subject to much greater variability and methodology is far from standardised (Bloom et al. 1984; Dyck et al. 1987; Jamal et al. 1985; Williams et al. 1988). It is possible, therefore, that improvements in these measures could be missed due to a lack of precision in their measurement. Sophisticated methods of the assessment of nerve integrity have been described, and recommendations for complex and precise criteria for the definition of neuropathy have been made (Dyck 198,8). However, the methodology described is expensive, time-consuming and requires considerable equipment and expertise. It is unlikely that such techniques will be practical for the large trials which are necessary for the definitive testing of drug effects. A better approach, and one which is being applied to on-going studies of aldose reductase inhibitors, is to adopt strict entry criteria and to make strenuous efforts to refine and standardise existing conventional assessment techniques, particularly between centres, to ensure that meaningful and consistent data are obtained in multicentre trials. The literature to date consists mainly of studies with small patient numbers and short term treatment with aldose reductase inhibitors. Most of the reported benefits of sorbinil have appeared early and have not been progressive. The early data from studies of to Ires tat suggest that improvement takes longer to become evident, but responses do appear to be sustained. Although larger patient numbers are involved, this is still only preliminary data, and full reports have not yet been published. These data are encouraging as they would appear to show that

199

even patients with advanced disease may indeed have some benefit from these agents. As neuropathic change is usually progressive it seems possible that small improvements, or even lack of progression, may be a worthwhile goal. The only detailed morphological study (Sima et al. 1988) apparently shows clear and considerable benefit from sorbinil treatment, which seems difficult to reconcile with the limited clinical effect reported by even the most optimistic of other authors. The data from animal experiments are primarily related to early functional changes such as reductions in conduction velocity, and reduced axonal transport. These, however, are not only clearly prevented and reversed by administration of an aldose reductase inhibitor, but also by correction of blood glucose. Although it is a therapeutically attractive hypothesis, these changes have not yet proven to be the forerunners of later clinical and structural neuropathy. Indeed, there are some data which suggest that later animal neuropathy is less affected by aldose reductase inhibitors (Turtle 1986), and reversal of the diabetic state by pancreatic transplantation has been reported to produce no clinically significant improvement in human diabetic neuropathy (Solders et al. 1987). Thus, if aldose reductase inhibitors are confirmed to be effective agents in human diabetic neuropathy, their primary role may be that of prophylaxis rather than the treatment of an advanced clinical problem. This in turn implies that although they may be of benefit to those with Type I (insulin-dependent) diabetes, who are in fact a small minority of our patients, they cannot be a panacea for all. The majority of our patients have Type 2 (non-insulin-dependent) diabetes, and some of these present with advanced neuropathic change which may already be irreversible. However, improved health screening to detect asymptomatic diabetic individuals may improve the prognosis in these patients. Studies of prophylaxis are fraught with problems for many reasons. Only a proportion of diabetic patients are likely to develop clinical complications, and there is at present no way of detecting those who are at risk. Therefore, many

Drugs 39 (2) 1990

200

patients would have to undergo long term drug treatment for a problem which may never develop. Although the newer agents undergoing investigation appear to have low toxicity, the effects oflong term administration are not known. Aldose reductase is a widely distributed enzyme in mammalian tissues and it is difficult to see why it should be so if it genuinely has no physiological function in many of the sites in which it is found. However, no such physiological role has so far been demonstrated, so it is unlikely that long term inhibition of this enzyme would have detrimental effects. The published data to date would appear to show that patients with painful neuropathy may derive some symptomatic benefit from the administration of aldose reductase inhibitors. Limited improvements in electrophysiological parameters have occurred even in advanced neuropathy. The early intervention trials at present under way may provide further evidence to confirm the beneficial role of aldose reductase inhibition in diabetic peripheral sensorimotor neuropathy, and the results are eagerly awaited by both the scientific and clinical communities.

6. Conclusions Much frustrating and painstaking work with aldose reductase inhibitors over the last 10 years has failed to produce unequivocal evidence of clinical benefit to patients with diabetic peripheral sensorimotor neuropathy. However, substantial progress has been made with the design of trials and the selection of patients in order to produce further data from large multicentre trials in the future. Studies of prophylaxis must wait for evidence of effect in patients with early changes, and confirmation of the safety oflong term administration. The results from clinical trials of the aldose reductase inhibitors in this difficult area are sufficiently encouraging to allow us to be reasonably optimistic about their future development, and continuing work should clarify their potential role with respect to prophylaxis and treatment.

Acknowledgement Dr Masson and Dr Boulton are supported by the Wellcome Trust.

References Archer AG, Watkins' PJ, Thomas PK, Sharma AK, Payan J. The natural history of acute painful neuropathy in diabetes mellitus. Journal of Neurology, Neurosurgery and Psychiatry 46: 491-499,1983 Bloom S, Till S, Sonksen P, Smith S. Use of a biothesiometer to measure individual vibration thresholds and their variation in 519 non-diabetic subjects. British Medical Journal 288: 17931795, 1984 Boulton AJM, Atiea J, Deleeuw IH. et al. The efficacy and safety of the aldose reductase inhibitor tolrestat in the treatment of chronic sensori motor diabetic neuropathy. Abstract. Diabetologia 32: 469A. 1989 Boulton AJM, Knight G, Drury J. Ward JD. The prevalence of symptomatic. diabetic neuropathy in an insulin-treated population. Diabetes Care 8: 125-128. 1985 Boulton AJM . the Tolrestat in Neuropath y Study Group. Effect of tolrestat. a new aldose reductase inhibitor. on nerve conduction and parasthetic symptoms in diabetic neuropathy. Abstract no. 6 \. Diabetologia 29: 521 A. 1986 Boulton AJM. Ward JD. Diabetic neuropathies and pain. Clinics in Endocrinology and Metabolism 15(4): 917-931 . 1986 Brownlee M. Cerami A. Vlassara H. Advanced glycosylation end products in tissue and the biochemical basis of diabetic complications. New England Journal of Medicine 318: 1315-1321. 1988 Bruyn GW. Garland H. Neuropathies of endocrine origin. In Vinken pj & Bruyn GW (Eds) Handbook of clinical neurology. Vol. 8. pp. 29-71. North Holland Publishing Company. Amsterdam. 1970 Christiensen JE. Varnek L. Gregersen G. The effect of an aldose reductase inhibitor (sorbinil) on diabetic neuropathy and neural function of the retina: a double-blind study. Acta Neurologica Scandinavica 71: 164-167. 1985 Clements RS. Bell DSH. Diagnostic. pathogenic and therapeutic aspects of diabetic neuropathy. Special Topics in Endocrinology and Metabolism 3: 1-43. 1982 Consensus Statement. Report and recommendations of the San Antonio Conference on Diabetic Neuropathy. Diabetes 37: 1000-1004. 1988 Culebras A. Alio J. Herrera J-L. Lopez-Fraile IP. Effect of an aldose reductase inhibitor on diabetic peripheral neuropathy. Archives of Neurology 38: 133-134. 1981 Dyck PJ . Detection. characterisation and staging of polyneuropathy. assessed in diabetics. Muscle and Nerve II: 21-32. 1988 Dyck PJ . Bushek W. Spring EM. Karnes JL. Litchy WJ. et al. Vibratory and cooling detection thresholds compared with other tests in diagnosing and staging diabetic neuropathy. Diabetes Care 10: 432-440. 1987 Dyck PJ. Karnes JL. O'Brien P. Okazaki H. Lais A. et al. The spatial pattern of fibre loss in diabetic neuropathy suggests ischaemia. Annals of Neurology 19: 440-449. 1986 Dyck PJ. Sherman WS. Hallcher LM, Service FJ. O·BrienPC. et al. Human diabetic endoneurial sorbitol. fructose and myoinositol related to sural nerve morphometry. Annals of Neurology 8: 590-596. 1980 Dyck PJ . Zimmerman BR. Vilen TH. Minncrath SR. Karnes JL. et al. Nerve glucose. fructose. sorbitol. myoinositol . and fiber

Aldose Reductase Inhibitors in Diabetic Neuropathy

degeneration and regeneration in diabetic neuropathy. New England Journal of Medicine 319: 542-548, 1988 Fagius J, Brattberg A, Jameson S, Berne C. Limited benefit of treatment of diabetic polyneuropathy with an aldose reductase inhibitor: a 24 week controlled trial. Diabetologia 28: 323-329, 1985 Fagius J, Jameson S. Effect of aldose reductase inhibitor treatment in diabetic polyneuropathy: a clinical and neurophysiological study. Journal of Neurology, Neurosurgery and Psychiatry 44: 991-1001, 1981 Rorkowski CM, Rowe BR, Nightingale S, Barnett AH. A six month randomised placebo-controlled trial of the aldose reductase inhibitor 'statil' in chronic symptomatic peripheral neuropathy. In Ward JD (Ed.) Proceedings of the 1988 IDF Satellite Symposium on Diabetic Neuropathy, John Wiley & Sons Ltd, Chichester, in press, 1989 Flynn TG. The aldehyde reductase family of enzymes. Bulletin of the Canadian Biochemical Society 20: 38-44, 1983 Gabbay KH. The sorbinil pathway and complications of diabetes. New England Journal of Medicine 194: 1085-1087, 1973 Gabbay KH, Merola LO, Field RA. Sorbitol pathway: presence in nerve and cord with substrate accumulation in diabetes. Science 151: 209-210,1966 Goto Y, Shigeta Y, Sakamoto N, Kito S, Matsuoka K. Clinical effect of aldose reductase inhibitor (ONO-2235) on diabetic neuropathy: multi-institutional studies. Abstracts of the International Symposium on the Polyol Pathway and its Role in Diabetic Complications, p. 15, Kashikojima, October 28-30, 1986 Greene DA, Dejesus PV, Winegrad AI. Effects of insulin and dietary myoinositol on impaired motor nerve conduction velocity in acute streptozotocin diabetes. Journal of Clinical Investigation 55: 1326-1336, 1975 Greene DA, Lattimer SA. Action of sorbinil in diabetic peripheral nerve: relationship of polyol (sorbitol) pathway inhibition to a myoinositol mediated defect in sodium-potassium ATPase activity. Diabetes 33: 712-716, 1984 Greene DA, Lattimer SA, Sima AAF. Are disturbances of Sorbitol, Phosphoinositide, and Na+-K+-ATPase regulation involved in the pathogenesis of diabetic neuropathy? Diabetes 37: 688-693, 1988 Gregerson G, Bertelsen B, Harbo H, Larsen E, Anderson JR, et al. Oral supplementation of myoinositol: effects on peripheral nerve function in human diabetes and on the concentration in plasma, erythrocytes, urine, and muscle tissue in human diabetics and normals. Acta Neurologica Scandinavica 67: 164172,1983 Guy RJC, Gilbey SG, Sheehy M, Asselman P, Watkins PJ. Diabetic neuropathy in the upper limb and the effect of twelve months sorbinil treatment. Diabetologia 31: 214-220, 1988 Hale PJ, Nattrass M, Silverman SH, Sennit C, Perkins CM, et al. Peripheral nerve concentrations of glucose, fructose, sorbitol and myoinositol in diabetic and non-diabetic patients. Diabetologia 30: 464-467, 1987 Handelsman DJ, Turtle JR. Clinical trial of an aldose reductase inhibitor in diabetic neuropathy. Diabetes 30: 459-464, 1981 Hotta N, Kakuta H, Koh N, Sakakibara F, Nakamura J, et al. A long-term experimental and clinical study of an aldose reductase inhibitor, ONO-2235 on diabetic neuropathy. In Ward JD (Ed.) Proceedings of the 1988 IDF Satellite Symposium on Diabetic Neuropathy, John Wiley & Sons Ltd, Chichester, in press, 1989 Jamal GA, Hansen S, Weir AI, Ballantyne JP. An improved automated method for the measurement of thermal thresholds. I. Normal subjects. Journal of Neurology, Neurosurgery and Psychiatry 48: 354-360, 1985 Jaspan J, Herold K, Maselli R, Bartkus C. Treatment of severely painful diabctic neuropathy with an aldose reductase inhibitor:

201

relief of pain and improved somatic and autonomic function. Lancet 2: 758-762, 1983 Judzewitsch RG, Jaspan JB, Polonsky KS, Weinberg CR, Halter JB, et al. Aldose reductase inhibition improves nerve conduction velocity in diabetic patients. New England Journal of Medicine 308: 119-125, 1983 Lewin IG, O'Brien lAD, Morgan MH, Corrall RJM. Clinical and neurophysiological studies with the aldose reductase inhibitor, sorbinil, in symptomatic diabetic neuropathy. Diabetologia 26: 445-448, ·1984 Llewelyn JG, Thomas PK, Gilbey SG, Watkins PJ, Muddle JR. Pattern of myelinated fibre loss in the sural nerve in neuropathy related to Type I (insulin-dependent) diabetes. Diabetologia 31: 162-167, 1988 Low PA. Recent advances in the pathogenesis of diabetic neuropathy. Muscle and Nerve 10: 121-128, 1987 Low PA, Schmelzer JD, Ward KK, Yao JK. Experimental chronic hypoxic neuropathy: relevance to diabetic neuropathy. American Journal of Physiology 250: E94-99, 1986 Malik RA, Newrick PG, Sharma AK, Tomlinson DR, Ward JD. Pathogenesis of human diabetic neuropathy. In Ward JD (Ed.) Proceedings of the 1988 IDF Satellite Symposium on Diabetic Neuropathy, John Wiley & Sons Ltd, Chichester, in press, 1989 Masson EA, Church SE, Woodcock AA, Hanley SP, Boulton AJM. Is resistance to ischaemic conduction failure induced by hypoxia? Diabetologia 31: 762-765, 1988 Masson EA, Gem J, Hunt L, Boulton AJM. A novel approach to diagnosis and assessment of symptomatic diabetic neuropathy. Pain 38: 25-28, 1989 Mayer JH, Tomlinson DR. Prevention of defects of axonal transport and nerve conduction velocity by oral administration of myoinositol or an aldose reductase inhibitor in streptozotocindiabetic rats. Diabetologia 25: 433-438, 1983 Mayhew JA, Grillon KRW, Hawthorne IN. Free and lipid inositol, sorbitol, and sugars in sciatic nerve obtained post-mortem from diabetic patients and control subjects. Diabetologia 24: 13-15, 1983 Mulder DW, Lambert EH, Bastron JA, Sprague RG. The neuropathies associated with diabetes mellitus: a clinical and electromyographic study of 103 unselected diabetic patients. Neurology II: 275-284, 1961 Nakuda H. Increased susceptibility to ischaemia in diabetic nerve. Abstract ORA-014-005. Diabetes Research and Clinical Practice 5 (Suppl. I): S57, 1988 Newrick PG, Wilson AJ, Jakubowski J, Boulton AJM, Ward JD. Sural nerve oxygen tension in diabetes. British Medical Journal 293: 1053-1054, 1986 O'Hare JP, Morgan MH, Alden P, Chissel S, O'Brien lAD, et al. Aldose reductase inhibition in diabetic neuropathy: clinical and neurophysiological studies of one year's treatment with sorbinol. Diabetic Medicine 5: 537-542, 1988 Pavy FW. Address on diabetes, Washington International Congress. Medical News (Philadelphia) September 24: 357, 1887 Pirart J. Diabetes mellitus and its degenerative complications: a prospective study of 4,400 patients between 1947 and 1973. Diabetes Care I: 168-188, 252-263, 1978 Pitts NE, Gundersen K, Mehta DJ, Vreeland F, Shaw GL, et al. Clinical experience with sorbinil: an aldose reductase inhibitor. Metabolism 35 (Suppl. I): 96-100, 1986 Ryder S, Sarokhan B, Shand DG, Mullane JF. Human safety profile of tolresta!: an aldose reductase inhibitor. Drug Development Research II: 131-143, 1987 Ryder S, the Tolrestat in Painful Neuropathy Study Group. Objective (nerve conduction) and subjective (painful symptom) improvement following administration of tolrestat, a new aldose reductase inhibitor. Abstract No. 436, Diabetologia 29: 588A, 1986 Salway JG, Finnegan JA, Barnett D, Whitehead L, Karunanyaka

202

A, et aL Effect of myoinositol on peripheral nerve function in diabetes. Lancet 2: 1282-1284, 1978 Sima AAF, Bril V, Nathaniel V, McEwen TA, Brown MB, et aL Regeneration and repair of myelinated fibres in sural nerve biopsy specimcns from patients with diabetic neuropathy treated with sorbiniL New England Journal of Medicine 319: 548-555, 1988 Sima AAF, Lattimer SA, Yagihashi S, Greene DA. Axo-glial dysjunction: a novel struclUrallesion that accounts for poorly reversible slowing of nerve conduction in the spontaneously diabetic bio-breeding rat. Journal of Clinical Investigation 77: 474484. 1986 Simard-Dequesnc N. Greslin E, Dubuc J, Dvornik D. The effect of a new aldose rcduetase inhibitor (tolrestat) in galactosaemic and diabctic rats. Mctabolism 33: 885-892, 1985 Soldcrs G, Gunnarsson R, Pcrsson A, Wilczek H, Tyden G, et aL Eflccts of combincd pancreatic and renal transplantation on diabctic ncuropathy: a two-year follow-up study. Lancet 2: 1232-1235. 1987 Srcdy.l. Millcn J. McCalcb ML Tutwiler GF. Inhibition of pol yo I accumulation in lens, sciatic nerve, and RBC in vitro by the aldosc reductasc inhibitor. tolrestat. Federation Proceedings 45: 902. 1986 Stcwart MA. Shcrman WR, Kurien MM, Moonsammy GI, Wisgcrhof M. Polyol accumulation in nervous tissue of rats with cxpcrimcntal diabctcs and galactosaemia. Journal of Neurochemistry 14: 1057-1066, 1967 Tcrkildscn :'\.B. Christensen NJ. Reversible nervous abnormalitics in juwnile diabetics with recently diagnosed diabetes. Diabctologia 7: ID-117, 1971 Thomas PK. Eliasson SG. Diabetic neuropathy. In Dyck PJ et aL (Eds) Peripheral neuropathy, pp. 956-981, Saunders, Philadelphia. 1975 Timperley WR, Boulton AJM, Davies-Jones GAB. Jarratt JA, Ward JD. Small vessel disease in progressive diabetic neuropathy associated with good metabolic control. Journal of Clinical Pathology 38: 1030-1038, 1985

Drugs 39 (2) 1990

Tomlinson DR, Moriarty RJ, Mayer JH. Prevention and reversal of defective axonal transport and motor nerve conduction velocity in rats with experimental diabetes by treatment with the aldose reductase inhibitor SorbiniL Diabetes 33: 470-476, 1984 Tuck RR, Schmelzer JD, Low PA. Endoneurial blood flow and oxygen tension in the sciatic nerves of rats with experimental diabetic neuropathy. Brain 107: 935-950, 1984 Turtle JR. Aldose reductase in the pathophysiological aspects of diabetic neuropathy. Abstracts of the International Symposium on the Polyol Pathway and its Role in Diabetic Complications, p. IO~ Kashikojima, October 28-30, 1986 Van Heyningen R. Formation of polyols in the lens of the rat with 'sugar' cataract. Nature 184: 194-195, 1959 Ward JD, Barnes CG, Fisher DJ, Jessop JD. Improvement in nerve conduction following treatment in newly diagnosed diabetics. Lancet I: 428-431,1971 Williams G, Gills JS, Aber V, Mather HM. Variability in vibration perception threshold among sites: a potential source of error in biothesiometry. British Medical Journal 296: 233-235, 1988 Winegrad AI, Simmons DA, Martin DB. Pathogenesis of diabetic neuropathy. Correspondence. New England Journal of Medicine 320: 58-59, 1989 Yasuda H, Dyck PJ. Abnormalities of endoneurial microvessels and sural nerve pathology in diabetic neuropathy. Neurology 37: 20-28, 1987 Young RJ, Ewing DJ, Clarke BP. A controlled trial of sorbinil, an aldose reductase inhibitor, in chronic painful diabetic neuropathy. Diabetes 32: 938-942, 1983

Authors' address: Dr A.1.M. Boullon. Senior Lecturer in Medicine and Honorary Consultant Physician, Department of Medicine, Manchester Royal Infirmary, Oxford Road, Manchester M 13 9WL, England.