Capillary dilatation because of an injury response (caused by histamine and other mediators of vasodilatation released from damaged tissues) can be captured as a red flare as a result of arteriolar dilatation through a local axon reflex. The axon reflex might be elicited by stimulation of pain nerve fibers (45) or through iontophoresis of acetylcholine, which mediates the release of nitric oxide through the axon reflex (46). A recent study has used heating the skin to 44°C to evoke the flare (LDIflare) and assessed it using a laser Doppler imager to show that it demonstrates C-fibre dysfunction before it can be detected by CASE IV (47).
Nerve biopsy, typically of the sural nerve has been used for many years in the study of peripheral neuropathy, particularly when the etiology is unclear or in patients with diabetes with atypical neuropathies (48). However, this is an invasive procedure with recognized sequelae, which might include postoperative pain at the site of biopsy, sensory deficits in the nerve distribution and allodynia, particularly in diabetic patients (49). A number of morphological parameters including myelinated fibre density, regenerative cluster density, axonal atrophy, and axo-glial dysjunction have been used as morphological end points to determine treatment efficacy (50,51). However, the need to repeat contra-lateral nerve biopsy and questions on the utility and reliability of the methods used to evaluate axo-glial dysjunction and axonal atrophy has posed serious questions regarding its use as an end point in any future trials of human diabetic neuropathy.
An alternative less invasive technique of a 3-mm skin biopsy enables a direct study of small nerve fibres and has been proposed for use in clinical trials (52). Although a number of neuronal markers including neurone-specific enolase and somatostatin have been used to immunostain skin nerves, protein gene product-9.5 has proven to be the best cytoplas-mic axonal marker. To define alterations in the most distal nerves and hence those likely to sustain the earliest damage, 50 ^m formalin-fixed frozen sections have been used to visualize and quantify intraepidermal nerve fibres (IENF) density, as number/length of epidermis in idiopathic sensory neuropathy and in patients with impaired glucose tolerance (53,54). A recent study used a new morphometric modification to assess nerves per epidermal area, which correlated highly (r = 0.945) with the accepted gold standard assessment of nerves per epidermal length (55). This method appears reproducible, diagnostically sensitive, less time-consuming than IENF-counting, and might be adopted in any laboratory familiar with basic immunohistochemical methodology for protein gene product-9.5 staining (55). It has been proposed that the rate of epidermal nerve fibre regeneration before and after intervention could be utilized as an end point in clinical trials (56).
Magnetic resonance imaging has been used to demonstrate that patients with DPN have a lower cross-sectional spinal cord area than healthy controls in the cervical and thoracic regions (57). However, progression or regression of this abnormality has not been evaluated in prospective studies, and therefore, the potential for its use as an end point in clinical trials of human diabetic neuropathy remains to be established.
The cornea is the most densely innervated part of the human body containing AS and unmyelinated C-fibres and derives its innervation from the ophthalmic division of the trigeminal nerve. Corneal confocal microscopy (CCM) permits sequential observations of the corneal subbasal nerve plexus comparable or even superior to that obtained with histopathological examination (58). CCM detects significant alterations in corneal nerve fibre density, branching, and tortuosity in patients with mild diabetic neuropathy and these alterations relate to the severity of somatic neuropathy (59,60) (Fig. 1). Furthermore, in a recent prospective study corneal nerve fibre density has been shown to improve with improved glycemic control (61). Therefore, the ability of CCM to define the extent of nerve damage and repair cross-sectionally but also longitudinally in patients with diabetes appears significant. In particular the noninvasive and hence, reiterative facility of CCM provides a means of expediting drug development programmes of therapies deemed to be beneficial in the treatment of diabetic peripheral neuropathy (Fig. 3).
At present there is no effective treatment for neuropathy therefore, the focus of defining those with neuropathy should be on those who have a significant risk of foot ulceration. The methods used for this purpose must be rapid, simple, and predict ulceration. In a recent study of 2022 diabetic subjects and 175 nondiabetic control subjects, "peripheral polyneuropathy" was diagnosed by assessing VPT at the tip of both great toes using a Rydel-Seiffer 128-Hz tuning fork and a neurothesiometer, abnormal 10-g monofilament test, and the presence of neuropathic symptoms. 5.2% of patients with diabetes had an abnormal vibration test, of whom 66.7% had an abnormal monofilament response and 68.6% had a missing Achilles' tendon reflex. But those with an abnormal tuning fork test had a significantly higher vibration perception threshold (32 +/- 9.8 vs 12.5 +/- 6.4 V), which would be consistent with defining those at risk of ulceration. Thus, it would appear that the simple tuning fork was a useful screening test for those at risk of ulceration rather than diabetic neuropathy as concluded by the authors (62). The predictive value of defining vibration perception threshold was established more than 10 years ago in a large prospective study, which showed that a VPT >25 effectively predicted the risk of foot ulceration in diabetes (63). In a multicenter prospective follow-up study to determine which risk factors in foot screening have a high association with the development of foot ulceration, 248 patients underwent an assessment of the neuropathy symptom score, NDS, VPT, Semmes-Weinstein monofilaments, joint mobility, peak plantar foot pressures, and vascular status at baseline and every 6 months for a mean period of 30 months (range 6-40) (64). Foot ulcers developed in a high proportion (29%) of patients suggesting they were all extremely at high risk. They were more frequently men, had a longer duration of diabetes, nonpal-pable pedal pulses, reduced joint mobility, higher NDS, higher VPT, and an inability to feel a 5.07 monofilament. The NDS alone had the best sensitivity, whereas the combination of the NDS and the inability to feel a 5.07 monofilament reached a sensitivity of 99% (64).
To determine the incidence of, and clinically relevant risk factors for new foot ulceration in a large cohort of patients with diabetes in the community healthcare setting, 9710 patients with diabetes underwent an assessment of the neuropathy symptom score, NDS, sensitivity to the 10 g monofilament, foot deformities, and peripheral pulses at baseline for 2 years in six districts of North-West England (19). The annual incidence of new foot ulcers was 2.2% and the following factors independently predicted new foot ulcer risk: ulcer present at baseline (RR [95% confidence interval]) 5.32 (3.71-7.64), past history of ulcer 3.05 (2.16-4.31), abnormal NDS (>6/10) 2.32 (1.61-3.35), any previous podiatry attendance 2.19 (1.50-3.20), insensitivity to the 10 g monofilament 1.80 (1.36-2.39), reduced pulses 1.80 (1.40-2.32), foot deformities 1.57 (1.22-2.02), abnormal ankle reflexes 1.55 (1.01-2.36), and age 0.99 (0.98-1). Thus, the study recommended that the NDS, 10 g monofilament and palpation of foot pulses be used as screening tools in general practice (19).
Methodological issues of paramount importance for success of any DPN trial are: (1) population selection, (2) end points, (3) study duration, and (4) confounding factors.
Because DPN is a complex disease of multifactorial etiology, a randomly chosen population of patients with this condition are likely to be heterogenous not only with respect to the form and severity of neuropathy but also with respect to the type of diabetes and other risk factors for neuropathy (22). Therefore, to detect a significant drug effect in such a population large numbers of patients and/or impractically long study durations are required. Additionally, DPN evolves from a stage of subclinical alteration with early neurophysiological dysfunction (36) to eventual significant clinical and pathological deficits (43). It is reasonable to assume that any therapeutic intervention is likely to be more beneficial earlier in the course of the disease. For the purpose of a clinical trial, patients with clinical stage 1 or 2a should be targeted. Those with more advanced disease—stage 2b and, especially, stage 3, are less likely to respond to any given therapeutic intervention. Furthermore to limit variability in the assessment of neuropathy because of the involvement of multiple examiners participating in a multicenter trial, it is desirable to establish entry criteria, such as nerve conduction attributes or quantitative sensory tests to optimize homogeneity. To eliminate the possible effects of glycemic control over the duration of the trial it is important to have a run-in period reserved for optimizing glycemic control and eliminating those who are unable to achieve acceptable levels of HbA1C (i.e., 9%). Because patients with type 1 and type 2 diabetes differ considerably in demographic characteristics, comorbidities, concomitant medication, and incidence of neuropathy, proper stratification techniques, and/or study analysis methods are necessary. Regarding neuropathy itself, patients not experiencing steady levels of symptoms and signs for at least 6-12 months should be excluded and the adjustment should be done for the duration of both diabetes and neuropathy. This approach is desirable but might limit recruitment. An effective compound will only succeed if a fair balance is established for study entry criteria that reflect the population with diabetic neuropathy in general.
Ideally, the end points should be clinically relevant, such as the development of clinically manifest neuropathy or progression of existing neuropathy to the point of an insensate foot or foot ulceration. However, because of the natural history of the disease, use of such endpoints makes studies prohibitively long, expensive, and potentially obsolete.
Thus, one has to rely on surrogate end points, which by definition exhibit a causal relationship with the clinical outcome and are laboratory measurements or physical signs which can be measured rapidly with good sensitivity and specificity.
The duration of the trial needed to demonstrate a difference in outcomes between active and placebo groups is directly dependent on the desired statistical power of the study and relative success rates observed in both groups. This success rate will depend on the severity and evolution of change in the chosen end point and the duration of treatment. In patients with mild neuropathy, using outcome measures such as nerve conduction studies (NCS) and QST, it might take 3-5 years to demonstrate a clinically meaningful difference. If clinical signs of neuropathy or composite scores consisting of signs, NCS and QST are being used, it has been postulated that the trial should recruit at least 68 patients per treatment arm and should last 3.7 years (43). If however, one chooses a hard end point like the annual incidence of foot ulceration then approximately 3000 patients are required in 5 years (65).
First, the phenomenon of "regression to the mean" is particularly relevant to trials of symptomatic human diabetic neuropathy. Thus, patients who enroll in a trial might do so when their symptoms are at their peak. Because the natural history of painful diabetic neuropathy is one of spontaneous improvement then the next assessment after being randomized, irrespective of treatment benefit is biased toward an improvement. Second, the placebo effect might play a significant role, especially as this might be higher than the conventionally quoted 20-30% particularly in painful diabetic neuropathy.
Was this article helpful?