The insulin autoimmune syndrome (IAS) as a cause of hypoglycaemia: an update on the pathophysiology, biochemical investigations and diagnosis

Introduction

Autoimmunity in general is beneficial; however, impairment of the immune system in recognising “self-antigens”, e.g. immunosenescence can trigger autoaggression which in some cases progresses to autoimmune disease [1], [2], [3]. Autoimmune diseases in some cases occur when the very low titre of endogenous antibodies rise causing transformation from benign autoimmunity to pathogenic autoimmunity. This progression is determined by both genetic influences and environmental triggers such as spontaneous genetic instability, acquired genetic mutations or antigenic modifications [3].

The very first case of insulin autoimmune syndrome (IAS) was described in the Tohoku Journal of Experimental Medicine in 1972 [4] by Yukimasa Hirata in a 47-year-old Japanese man with severe spontaneous hypoglycaemia. IAS is now the third most common cause of hypoglycaemia in Japan after insulinoma and extrapancreatic neoplasms. The first case in a Caucasian was independently reported in a 40-year-old Norwegian with hyperglycaemia and profound reactive hypoglycaemia after meals by Følling and Norman, and appeared in Diabetes in 1972 [5]. The very different clinical manifestations in these two very initial cases of IAS are noteworthy. Since then, some 70 cases of IAS in Caucasians with a wider range of clinical presentations have been fully described from almost all western European countries, North and South America and an abstract of one case in 2010 from the UK [6]. Newly found cases of IAS with already well-documented aetiologies and manifestations but lacking new or additional information may be considered unsuitable for further publication by most academic reviewers.

IAS is also known as Hirata’s disease or insulin autoimmune hypoglycaemia (IAH). It is essentially a rare autoimmune disorder caused by the presence of variable and significant increase in the titre and affinity/avidity of endogenous antibodies which bind insulin/proinsulin and/or insulin receptor (insulin mimetic). Antibodies of the IAS have a unique double-whammy impact namely a pathological role in causing hypoglycaemia as well as the potential to interfere in some immunoassays of pancreatic hormone measurements. IAS has been shown to clinically and biochemically mimic other causes of hypoglycaemia such as insulinoma, insulin administration (self-harm or malicious). Appreciating this dual clinical-analytical impact has made its consideration in the differential diagnosis of non-diabetic, non-critically ill patients with indeterminate cause of hypoglycaemia judicious and prudent.

In this short review, the pathophysiology of IAS will be briefly described and the heterogeneity of endogenous antibodies and their clinical and analytical repercussions in mimicking other causes of hypoglycaemia will be summarised. This would help clinical and laboratory practitioners to plan appropriate biochemical investigations, perform and correctly interpret relevant immunoassay results. Occasionally, a paraprotein in some patients may also exhibit an ability to bind insulin, thus causing hypoglycaemia by a mechanism similar to IAS [7], [8], [9].

Brief note on the pathophysiology of hypoglycaemia of IAS

IAS is a rare cause of hypoglycaemia with higher incidence in genetically predisposed individuals such as those with HLA-DR4 [10], [11], [12]. There are two major triggers, namely infection/reinfection and drugs particularly in patients with a history of other autoimmune disorders or autoimmune polyendocrine syndromes [10], [11], [12], [13], [14], [15]. A range of common viral infections/reinfections such as coxsackie B influenza, mumps and rubella are known to cause islet cell antibodies and diabetes type 1. However, these viruses have also been shown to increase the concentration of endogenous insulin antibodies in 1986 [16] and hepatitis C virus was added to the list in 2002 [17]. A higher concentration of endogenous antibodies in asymptomatic non-diabetic individuals is considered a risk factor for potential overt disease [16], [18].

Until 2003, drugs containing sulfur/sulfydryl groups (e.g. thiols with -SH) such as methimazole, captopril, D-penicillamine, hydralazine, glutathione, methionine, mercaptans, clopidogrel, aurothioglucose, imipenem, penicillin G and diltiazem [19], [20], [21] were deemed to have the potential to trigger IAS. An additional drug namely α-lipoic acid (ALA) which also contains sulfur (cyclic disulfide) was also reported to cause IAS in Japanese [19]; this was quickly followed by numerous reports in patients all of whom were also Japanese [21], [22], [23], [24], [25], [26]. The first case in a Caucasian (Italian) appeared in Sicily in 2011 [27]; however, within 22 months, six more cases were also identified by the same medical centre in Sicily [28]. Lipoic-acid is sold over-the-counter (OTC) and on the Internet as a universal antioxidant drug/free radical scavenger because of its unique solubility in both water and fat; it is also prescribed for conditions such as peripheral polyneuropathy [26]. It appears that the incidence of drug-induced IAS can occur in significant number in genetically predisposed Caucasians too. Clinical manifestations of IAS varied, irrespective of the underlying cause, from mild, intermittent and transient to spontaneous, protracted and very severe, necessitating in extreme cases plasmapheresis for glycaemic control [11], [29], [30], [31], [32], [33].

A suggested mechanism of action of endogenous antibodies which bind insulin and proinsulin(s) is that the initial pancreatic response to a rise in blood sugar would be “ineffective” in lowering blood glucose, because insulin/proinsulin(s) bind to these endogenous antibodies causing postprandial hyperglycaemia to persist for longer. This in turn results in prolonged pancreatic secretion of insulin and C-peptide in equimolar amounts as well as ~3% of proinsulin which has 1/10th of the insulin bioactivity, until the endogenous antibodies binding capacity is exceeded. Only then the unbound/free active insulin rises and reduces blood glucose, and at a concentration of ~4.5 mmol/L (i.e. the glycaemic threshold), pancreatic secretion decreases/ceases [31].

In patients with IAS, the t_1/2 of insulin/proinsulin(s) increase manyfold, from minutes to potentially hour(s) [29], [31], [34], [35], [36], [37], but that of C-peptide which may not bind to these antibodies remains unaffected at ~25 min [31]. Insulin/proinsulin(s) bound to these endogenous antibodies in blood represent a “reservoir”; its subsequent dissociation from these antibodies would be dependent on the intrinsic dissociation rate constant (i.e. K₋₁) [38], [39]. A significant K₋₁ would continuously release free/unbound bioactive insulin/proinsulin(s) enough to cause hypoglycaemia of varying severity usually within 2–6 h or longer. Hyperinsulinaemia also inhibits counter-regulatory mechanisms of glucose homeostasis (suppresses glycogenolysis, gluconeogenesis and ketogenesis). The severity, duration and swing from hyper- to hypoglycaemia and the intensity and duration of the hypoglycaemic phase itself have been attributed to (a) the endogenous antibody characteristics, namely its intrinsic association and dissociation rate constants to insulin/proinsulin(s), (b) its titre/concentration and (c) its capacity, i.e. antibodies’ valency. More than one class of endogenous antibodies may also coexist in the same patient, e.g. high affinity/low capacity and high capacity/low affinity [31], [40]. The presence of such endogenous binding antibodies has distorted glucose metabolism [31], [35] with phases of hyper- and hypoglycaemia, which led some to refer to these swings as diabetes and hypoglycaemia [41], [42]. Three classes of immunoglobulin antibodies namely IgG, IgA and IgM are so far been identified with the most common being IgG [7], [19], [29], [31], [33], [43].

Another form of endogenous antibodies namely insulin receptor antibodies was the sole cause of hypoglycaemia in some cases. However, both insulin binding and insulin receptor antibodies have occurred in the same patient, sequentially or simultaneously [43], [44], [45], [46], [47], [48], [49]. A suggested mechanism is an idiotypic immune response (see reference [49] for more details).

Immunoassays of pancreatic hormones and potential interference from endogenous antibodies of IAS

Screening tests for hypoglycaemic drugs (sulfonylureas and biguanides), non-islet cell tumour-induced hypoglycaemia, e.g. testing for IGF(s), β-hydroxybutyrate, pituitary and adrenocortical functions will not be considered being outside the scope of this short review and are well-described elsewhere [50], [51] and in text books of endocrinology.

Current pancreatic immunoassays are almost exclusively commercial, optimised “with justification” to measure these moieties in serum essentially free from endogenous insulin and/or proinsulin antibodies. Proinsulin is proteolytically processed within the pancreas by several endopeptidases to generate and predominantly secrete insulin and C-peptide in equimolar amounts. Some 3% of intact proinsulin is also secreted together with four partially processed split proinsulin intermediates. These split conversions are biologically more potent than proinsulin itself, which has 1/10th of the bioactivity of insulin [52], [53], [54], present in higher concentration in serum than proinsulin [52], [53], [54] and similarly retain both insulin and C-peptide components (see reference [54] for more details).

The inbuilt “insulin component” in “proinsulin and the four proinsulin split intermediates” collectively referred to as proinsulin(s) in this review have the potential to cross-react (albeit at dissimilar rates), measured and may be “improperly” reported as “free” insulin in some immunoassays. The “C-peptide component” may similarly cross-react, measured and also may be “incorrectly” reported as “free” C-peptide by some immunoassays [54], [55], [56], [57]. The presence of insulin, proinsulin(s) binding and/or receptor antibodies has shown to increase the concentration of one or more of these moieties thus causing unpredictable inaccuracy in immunoassay results [53], [54], [55], [56], [57]. For this reason, testing for endogenous insulin antibodies is considered prudent or even necessary in non-diabetic, non-critically ill patients with indeterminate cause of hypoglycaemia. Tests for “functional” insulin/proinsulin binding antibodies are initially carried out (being relatively more common and simple to perform) and if negative, tests for receptor antibodies are then considered if all forms of IAS are to be excluded. Multiple blood samples during phases of hypoglycaemia as well as few hours after a meal or an extended OGTT (~5 h) were also useful tools in diagnosis. Pancreatic hormones which bind to endogenous antibodies progressively rise with time and remain elevated for long while others which do not bind declined with their normal t_1/2 [31], [32], a distorted pattern consistent with the presence of endogenous antibodies.

Initial tests for antibodies were aimed at establishing or excluding the presence of “excessive functional binding” of labelled insulin/proinsulin in patient’s serum vs. normal sera free from endogenous antibodies. Initial and critical screening tests used were able to detect excessive “functional” binding irrespective of antibodies or paraprotein class such as IgM, IgG and IgA [7], [9]. Pragmatic, albeit not the only tests used were polyethylene glycol (PEG 6000 at 12.5% final concentration) or ammonium sulfate (50% final concentration) precipitation of immunoglobulins [34]. The finding of a higher amount of labelled insulin and/or proinsulin in the precipitate in patient’s serum compared with those found in normal sera (controls) was indicative of “endogenous functional binding” consistent with IAS. The limitation of such a simple test has been generally recognised because the amount of labelled insulin bindings to antibodies could vary from one sample to another taken from the same patient, being dependent on the number of “endogenously unoccupied” binding sites on the antibodies in the sample used (potentially being reduced postprandially but increase after prolonged fast). An alternative test used was to repeat the analyses of insulin, proinsulin and C-peptide before and after immunoglobulin/antibody precipitation in a patient’s serum and controls which allow assessment of the ratio of bound/free moieties. Currently, some laboratories routinely report both “free insulin” following PEG precipitation as well as total insulin. An additional test used was the Scatchard plot to identify the nature of the binding’s macromolecule, its class and binding characteristics. Some kit tests detect the physical presence of one specific class of insulin antibodies such as the common IgG and may be considered informative only in positive cases; however, if the outcome is negative, searching for other insulin/proinsulin antibodies of different class such as IgM or IgA as well as other binding macromolecules (e.g. paraprotein) was prudent and diagnostically necessary [7], [8]. Furthermore, kits which detect the immunological presence of a particular class of antibodies with a high degree of specificity may not detect others as well as proinsulin(s) antibodies if present. Investigating the presence of receptor antibodies was also considered in cases of unexplained hypoglycaemia in which exclusion of all forms of IAS per se was pursued and necessary.

On the incongruity of IASs caused by endogenous insulin antibodies

Immunogenicity of a substance is generally influenced by numerous factors such as its molecular size, chemical complexity/heterogeneity and individual genotype. Proinsulin has 86 amino acids with a MW of 9390 Da and is considered more immunogenic [58] than insulin (51 amino acids) with MW of 5808 Da. No specific autoantibody against low MW C-peptide in IAS has been documented (this connecting-peptide with “unknown” function has 31 amino acids and MW of 3021 Da). The t_1/2 of proinsulin and C-peptide are similar at ~25 min and 6-fold greater than the insulin t_1/2 of ~4 min.

Endogenous antibodies are known to bind insulin and/or proinsulin(s) in patients with IAS. However, binding of C-peptide to endogenous antibodies in IAS though unreported remained a possibility, e.g. C-peptide was known to bind to antibodies in some insulin-dependent diabetic patients [58]. Currently, the assumption is that C-peptide does not bind to specific C-peptide antibodies in patients with IAS and its t_1/2 remains unaltered. Elevated concentrations of C-peptide may be attributed to cross-reactivities with the C-peptide in proinsulin(s) which may be immunologically recognised, measured and “inappositely” reported as “free C-peptide” in some immunoassays [54], [55], [56], [57]. The above complexities and the heterogeneity of C-peptide measurements by immunoassays have therefore been a source of confusion, dependent on the nature of endogenous antibodies and immunoassay’s characteristics. The following are examples (without being prescriptive) of the reported adverse clinical and analytical manifestations of endogenous antibodies in patients with IAS.

Comments and conclusions

It may be important to comment further on some aspects of the endogenous antibodies which underpin the aetiology of IAS to help elucidate its highly variable biochemical and clinical manifestations, i.e. onset, spontaneousness, duration and severity of hypoglycaemia. As mentioned before, some viruses can increase insulin antibodies. An important feature of the immune response is that it remembers if it has seen such a pathogen (or a closely related antigen) before, reacts to a second and further exposure in a manner different than after a primary exposure. The reaction to a second exposure, e.g. reinfection(s) or sustained exposure reflects the immune memory response and “affinity maturation”. In such case, antibodies production are of the IgG, IgA, IgD or IgE class rather than the IgM, commonly produced after an initial exposure, i.e. class switching [80]. Furthermore, the immune response to a second and further exposure can be fast and vigorous with progressive increase in the titre of high affinity/avidity endogenous antibodies which in some cases can double every few hours [81], [82]. It is therefore reasonable to expect that the titre, capacity and affinity/avidity of antibody production in response to pathogens and/or antigenic modifications to be variable and unpredictable [2], [3], [16], [17], [18].

Of the many features of antibodies of IAS, probably the most crucial is their affinities/avidities. The binding between an antigen (Ag) and antibody (Ab) is not a reaction but a reversible interaction. The rate law equations (zero, first and second rate kinetic reactions) for this reversible interaction are describable by the forward/association and backward/dissociation rate constants (i.e. K₁ and K₋₁, respectively) and are applicable to any Ag-Ab interaction, both in vivo and in vitro and follow a reversible bimolecular thermodynamic principle (for further details, see [38], [39], [49], [81], [82]). The association binding reaction (K₁) is usually dominant and fast; the reverse dissociation rate reaction (K₋₁) determines the length of time the antigen remains bound to their antibodies and is influenced by the number and types of non-covalent bonds that are formed between the antigenic-determinant, i.e. epitope(s) on the antigen and the corresponding antigen-combining site on the paratope(s) on the antibody [81], [82]. The association reaction ranges over several orders of magnitude; the reverse dissociation rate reaction (K₋₁) has similarly wide range [81], [82]). The equilibrium constant (K_eq) is the ratio between these reactions (reactants) at equilibrium and the greater the strength of the bonds, the higher its K_eq which is typically between 10⁸ and 10¹⁰ (a 100-fold difference). A dominant K₁ is necessary in order to develop an endogenous insulin/proinsulin(s) pool in the first place to cause IAS. The titre and capacity of antibodies would define the size of this pool of bound insulin and/or proinsulin(s) and K₋₁ would determine the rate of insulin/proinsulin(s) released into the blood, affecting the onset, duration and severity of hypoglycaemia [31], [40]. The clinical course of most autoimmune diseases may not be straightforward, and tend to wax and wane by obscure and poorly understood mechanism(s). Of interest is that such a phenomenon has also been demonstrated in patients with IAS as shown by the epidemiological study of Uchigata et al. on 197 Japanese patients in whom many had spontaneous remission without any “positive” treatment and in others after the discontinuation of drugs containing sulfur. In few patients however, recurrent attacks have subsequently ensued without exposure to any drug prior to their recurrent hypoglycaemic episodes [83].

Unlike pancreatic hormones which have t_1/2 of minutes extended to hours in IAS, antibodies per se (irrespective of class or subclass) have a much longer t_1/2 of 3 weeks or more thus allowing belated detection. Testing for “insulin antibodies” is now included among the first-line tests in the recently published clinical practice guidelines by the American Endocrine Society in 2009 [84] and reiterated in a short note in 2012 [85]. Consideration and searching for various forms of functional binding antibodies a priori is important because of the vulnerabilities of many immunoassays to produce results which if interpreted in isolation could be highly misleading. Antibodies which underpin IAS have the ability to bind insulin and/or proinsulin(s) and the potential to distort not only their own measurements [54], [55], [56], [57] but also that of C-peptide immunoassays. This may also help to explain the wide range of concentrations reported in the literature [31], [34], [54], [55], [56], [57], [81], [85], [86], [87]. The specificities of immunoassays of insulin, proinsulin and C-peptide are known to vary from one reagent antibody to another leading to poor “between-methods” interchangeability [54], [55], [56], [57], [88], [89], [90]. However, despite these inaccuracies, immunoassay results remained on the whole useful and informative when interpreted in the context of the prevailing antibodies and with appropriately/timely taken samples during fasting and after meals and/or an extended OGTT [31], [32]. Direct testing for all forms of endogenous insulin/proinsulin antibodies remains however, the sine qua non practices in establishing or excluding the diagnosis of IAS in non-diabetic, non-critically ill individuals [85].

In conclusion, new informative cases of IAS in Caucasians continued to appear in the world literature [27], [28], [69], [70], [75], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99]. The very wide range of their clinical presentations and biochemical findings re-emphasised that IAS is a highly heterogeneous syndrome. Furthermore, it confirms that IAS per se has no specific clinical features of its own but has the potential to mimic other causes of hypoglycaemia clinically and biochemically. In this paper, the term IAS was confined to hypoglycaemia in non-diabetic, non-acutely ill patients in whom thorough testing for insulin/proinsulin binding and receptor antibodies was essential for confirming diagnosis and rational interpretation of pancreatic hormone data. Potential misdiagnosis is therefore likely to occur if these two interdependent entities namely the presence and nature of endogenous antibodies and immunoassays cross-reactivities are not considered and taken into account. Finally, the term IAS must not be confused with the effect of endogenous insulin antibodies in patients with brittle diabetes triggered by exogenous administration of insulin or insulin analogues causing poor glycaemic control and hypoglycaemia by a mechanism similar to IAS [100], [101], [102], [103].