Treatment-induced neuropathy of diabetes: an energy crisis?

This scientific commentary refers to ‘Treatment-induced neuropathy of diabetes: an acute, iatrogenic complication of diabetes’ by Gibbons and Freeman (10.1093/brain/awu307).

Peripheral neuropathy, together with retinopathy and nephropathy, constitute a triad of major complications of chronic hyperglycaemia in diabetes. In addition to the common or garden predominantly sensory small fibre neuropathy, many other types of diabetic neuropathy have been described (Dyck et al., 1993, 1999; Tesfaye et al., 2010). These include autonomic neuropathy, mononeuropathy, mononeuropathy multiplex, radiculopathy, and radiculoplexus neuropathy. Less well known is ‘insulin neuritis’, which develops shortly after the imposition of strict glycaemic control and is characterized by a length-dependent painful neuropathy and autonomic failure. First described by Caravati (1933), the condition is considered rare and its mechanisms are unknown. In a definitive report in this issue of Brain, Gibbons and Freeman provide the first lucid description of this disorder—now known as treatment-induced neuropathy of diabetes (TIND)—in a large group of subjects, and define both the probability of developing TIND and the main risk factors (Gibbons and Freeman, 2014).

By undertaking a retrospective review of all individuals referred to a tertiary care diabetic neuropathy clinic over 5 years, Gibbons and Freeman were able to show that >10% of patients who had a decrease in glycosylated haemoglobin A1C (HbA1c) >2% (104/954) developed TIND, as defined by an acute increase in neuropathic pain or autonomic symptoms or signs. Such symptoms occurred far less frequently when the reduction in HbA1c was smaller, suggesting a causal relationship between the risk of TIND and the size of the HbA1c reduction.

The study has a number of particular strengths. Gibbons and Freeman evaluated all eligible subjects and provided a uniform diagnosis of TIND, defined as >3-point increase on an 11-point Likert scale, and/or autonomic dysfunction developing over 2 weeks and of sufficient severity to cause subjects to seek medical attention, within 8 weeks of a large improvement in glycaemic control (a decrease in HbA1c of ≥2 percentage points over 3 months). As the data were collected in a major centre devoted to diabetes care and research (Joslin Diabetes Centre) and evaluated there or at the Beth Israel Deaconess Medical Centre Neuropathy Clinic, with systematized evaluation and follow-up of subjects, the review was unusually thorough. The authors undertook detailed structured neurological examinations, and obtained logs of glucose control, pain scores, and autonomic symptoms and other microvascular complications. These were measured every 3–6 months for the duration of follow-up. There was a strong statistical correlation between the magnitude of decrease in HbA1c and both the severity of neuropathic pain and degree of autonomic failure (both parasympathetic dysfunction and sympathetic adrenergic function as measured by orthostatic hypotension on tilt table testing and beat-to-beat blood pressure responses to the Valsalva manoeuvre). Of note is that subjects with TIND also had a higher risk of developing retinopathy (P < 0.001) and microalbuminuria (P < 0.001).

The study also has certain limitations. Although it is the largest systematic study of TIND, it is nevertheless retrospective. The calculated risks should be interpreted with caution as the denominator is not all patients with the HbA1c drop seen at the diabetes centre, but only those referred for evaluation of neuropathy. The presented risk calculations would therefore likely overestimate absolute risk. It would have been helpful for the reader had the authors used their unique dataset to address the role of hypoglycaemia in TIND. The data make a strong case for involvement of a rapid fall in blood glucose from very high levels (as reflected by very high A1c) in the genesis of TIND. Subjects with lower A1c (and milder average hyperglycaemia) do not seem to be at the same risk, but are presumably at much greater risk of hypoglycaemia. This would suggest that a large fall in blood glucose is the critical factor in TIND, rather than hypoglycaemia per se. It is not clear if there is a threshold of blood glucose that must be maintained to avoid TIND. Moreover, as discussed by the authors, it is a rapid improvement in glycaemic control, rather than insulin itself, that is responsible for TIND, as the syndrome can also occur with oral hypoglycaemic agents and even strict fasting.

Two observations, the rapid fall from very high levels of HbA1c and the selective involvement of small, predominantly unmyelinated somatic and autonomic C fibres, raise intriguing questions about the disease pathogenesis. The abrupt and concomitant involvement of pain and autonomic fibres, both in large part unmyelinated ‘C’ fibres, evokes a common mechanism of injury. Peripheral nerves in animal models of diabetic neuropathy due to chronic severe hyperglycaemia (which model the current situation of severe chronic hyperglycaemia in humans) have reduced nerve blood flow and oxygenation (Tuck et al., 1984). Severe hyperglycaemia can cause endoneurial oedema and increased intercapillary distance, resulting in hypoxia (Tuck et al., 1984). While changes in blood flow in rat nerves are reversible, nerve ischaemia in human diabetes is irreversible (Dyck et al., 1986). Such an endoneurial microenvironment, with reduced blood flow and hypoxia, would create a disproportionate reliance on anaerobic metabolism for its energy supply. Notably, the subjects most susceptible to TIND had severe hyperglycaemia, a condition that could cause endoneurial oedema (Tuck et al., 1984). As peripheral nerve has poor blood flow autoregulation compared to brain (Low and Tuck, 1984), it would be unusually susceptible to milieu changes. Arguably, a microenvironment characterized by hypoxia, ischaemia and perhaps endoneurial oedema could, if subject to an abrupt reduction in endoneurial glucose, result in an energy crisis. Unmyelinated fibres, which have a much greater surface area to size ratio than myelinated fibres, would be unduly susceptible to a sudden reduction in glucose or ATP, even when hypoglycaemia is avoided.

Beyond identifying the susceptible fibre population, the mechanism(s) of pain and autonomic failure remain unknown. There have been few pathological or physiological studies of nerves affected by TIND. The conclusion that pain is related to the type of fibre and rate of degeneration (Dyck et al., 1976) would suggest that degeneration of both somatic and autonomic unmyelinated fibres is responsible for the autonomic failure and pain in TIND. The study by Llewelyn et al. (1988) on a sural nerve biopsy in a subject with TIND, in which they found many regenerating clusters of nerve fibres, suggests an additional role for fibre regeneration and spontaneous firing as a mechanism for painfulness.

So what is the take-home message? The most important points to note are that TIND is relatively common, and that patients with severe chronic hyperglycaemia are at particularly high risk of developing the syndrome in association with vigorous treatment of their hyperglycaemia. Moreover, TIND is a serious complication, causing severe pain and autonomic failure, which might be irreversible. Damage extends beyond nerves; it also worsens retinopathy and nephropathy. What advice can be gleaned regarding the management of severely hyperglycaemic subjects? The obvious suggestion is to reduce A1c more gradually. The authors sensibly suggest that the goal should be to reduce A1c by <2% over 3 months. This is also good advice in that a graduated plan might avoid worsening retinopathy and nephropathy. Important as the current study is, this is still a discovery cohort. A prospective study comparing two different rates of A1c reduction would be of great interest. Some clues as to likely outcomes of such a study might be obtained from a comparison of two different time periods. Tight glycaemic control gained significant momentum following publication of the DCCT study on type I diabetes (Anonymous, 1998); however, the beneficial effects have not been replicated in type II diabetes, with concerns raised instead over worsening ischaemic complications. It might be interesting to compare the current data set with data from an earlier time point to assess the extent to which outcomes depend upon the rate of HbA1c reduction. How much protection is offered by a reduction over 6 months versus 3 months, for example? The results of such studies would help to inform the treatment of patients with chronic severe hyperglycaemia with a view to reducing the risk of treatment-induced neuropathy.

References