Testing for type 1 diabetes autoantibodies in gestational diabetes mellitus (GDM): is it clinically useful?

Gestational Diabetes Mellitus (GDM) is the most common metabolic disorder in pregnancy, with prevalence between 2 and 17% depending on the genetic background of the studied population [1‐3]. GDM is defined as carbohydrate intolerance diagnosed in the second or third trimester of pregnancy that was not clearly overt diabetes prior to gestation [4]. According to this definition, conditions leading to beta cell deficiency during pregnancy may reveal as GDM, triggered by the impairment in insulin action that physiologically appears during pregnancy, aimed at favouring fetal growth. Normally the β-cell pool adapts to physiological needs and increased functional demands [5]. However, if this state of insulin resistance (IR) is not compensated by an increase in beta-cell insulin secretion, it determines the appearance of GDM and provides a higher risk to develop type 2 diabetes (T2D) [6]. Epidemiological data shows that in a subgroup of women, estimated to be between 0 and 10% [7] of all GDM cases, carbohydrate intolerance is associated with the presence of autoimmunity against β-cells. In these women there is a higher risk of progression to type 1 diabetes (T1D) and/or Latent Autoimmune Diabetes of Adulthood (LADA) after pregnancy [2, 7‐11]. In rare occasions, autoimmune diabetes makes it first appearance in pregnancy as diabetic ketoacidosis (DKA) [12, 13]. When DKA is encountered in pregnancy the possibility of unrecognized pre-existing diabetes (mostly autoimmune) should be strongly considered. Pregnancy itself is a condition that predisposes to ketoacidosis, for example through nausea and vomiting in the first trimester, or insulin-resistance and increased lipolysis in the second and third trimesters [14].

Islet-cell autoantibodies, the markers of beta-cell autoimmunity, are present in sera from women with GDM with variable frequency. The prevalence of diabetes-related autoimmunity in pregnancy is extremely variable depending on the type of the autoantibody under study, the method for detection, and the population under observation.

Many studies have assessed the prevalence of diabetes-related autoantibodies in women with GDM, searching for ICA (islet cell autoantibody), IAA (insulin autoantibody), GADA (glutamic acid decarboxylase autoantibody), IA-2A (tyrosine phosphatase-like islet antigen autoantibody) and, recently, ZnT8-A (Zinc trasporter 8 autoantibody).

In general, titres for all autoantibodies are lower in GDM patients than in cases of newly diagnosed T1D [6, 15‐23] or in first-degree relatives of patients with T1D [24, 25]. These AABs’ titres are similar to those observed in LADA patients, and are considered indicative of a slow-developing autoimmune process in women with GDM that are positive for diabetes-related autoimmune markers [7, 26].

With regards to the prevalence of individual AABs, ICA studies showed a variable prevalence between 1 and 35% [7]. Nonetheless, because of technical (standardization) and methodological (test variability) issues, ICA are now less often measured [27, 28].

As for the other beta-cell autoantibodies, reports on the differences in autoantibodies frequencies and titres between GDM and control women have been conflicting, especially for GADA and IA2-A. The results of studies on GADA in GDM patients and controls vary widely; the overall frequencies of GADA range between 0 and 10.8% [7, 9, 19‐22, 24, 29‐43] (Fig. 1), with some studies showing higher frequency in GDM [9, 21, 22, 30, 39], and other no differences [19, 33‐35, 43, 44]. Also GADA titres have been reported higher in GDM women in some studies [9, 21, 22, 30, 31], with others showing no difference [33‐37]. As for the frequency of IA-2A positivity, this ranges from 0 to 6.2% (Fig. 1) [20, 21, 30, 38] with some papers [21, 30, 39] reporting a higher frequency in GDM patients than in normal controls (up to 26% in only one study [35], while others found no difference [9, 19]. Such contrasting results may be due to different genetic backgrounds within populations, since it is common knowledge that ethnicity plays an important role in determining beta-cell autoimmunity. Finally, a low frequency of IAA (Fig. 1) in GDM patients is reported in the literature [8, 15, 16, 19, 40], and only few studies found a different prevalence of IAA positivity in GDM patients than in the control women [9, 39], also confirming that IAA are more typical in younger ages [45].

The last beta-cell autoantibody to be discovered, ZnT8-A, was shown in a study by Rudland et al. to be the most represented in their GDM cohort, with a prevalence of 4.8%, also associating with higher basal blood glucose values compared to the other AABs [41]. In a more recent study Dereke and collaborators found that 6.8% of GDM women were autoantibody-positive, and 3.2% resulted ZnT8-A positive (Fig.1). Of the autoantibody-positive women with GDM who developed postpartum T1D, that were approximately 19%, all were positive for GADA. It seems therefore that ZnT8A does not have additional predictive values over GADA in the postpartum development of autoimmune diabetes [42].

Finally, very few studies have been conducted on populations with a high genetic risk for autoimmune diabetes, and they have shown a different prevalence of autoantibodies positivity in GDM between studies. For example, in a Finnish population, Jarvela et al. [9] observed a prevalence of autoimmunity of 16.7% in women with GDM vs. 2.8% in controls. Murgia et al. [39] reported in Sardinia a prevalence of 38.8% for at least one AAB vs 7.1% in control women; Cossu and co-workers [43] reported instead an overall prevalence of AABs of 6.4% in Sardinian women, a prevalence that was not different between GDM (5.6%) and control (8.3%) women (Fig. 1). This variability seems to be due to sample selection, screening criteria, different autoimmune measurement tools, all that could lead to conflicting results.

The development of an autoimmune or a non-autoimmune form of GDM depends on a pregnant woman’s genetic susceptibility. GDM may facilitate the identification of women at risk of developing diabetes later in life, and the presence of autoimmunity in GDM helps to estimate the risk of future type 1 or type 2 diabetes [46]. In particular, the risk of future T1D rises with increasing number of positive autoantibodies.

In this respect, it has been estimated that the risk of developing type 1diabetes two years post-partum is 17% in the presence of a single autoantibody, increases up to 61% in the presence of 2 autoantibodies, and to 84% when 3 autoantibodies are present [30]. In this study, among all antibodies tested, GADA had the highest accuracy in predicting autoimmune diabetes (sensitivity 63%) compared to ICA (48%) or IA2 (34%), but, overall, the presence of single GAD autoantibodies appeared to have limited predictive power, as described also for patients with LADA [26]. Löbner and collaborators reported that 96% of GDM women with at least one positive antibody (GADA or IA2-A) develop T1D within 8 years after delivery [47]. A Finnish population study [9] reports that 10% of women with a previous GDM develop type 1 diabetes within 6 years; the risk for type 1 diabetes correlates with an age under 30 years, with the need for insulin therapy during pregnancy, and with a positivity for at least 1 antibody. Another study reports an onset of T1D after 1 year of 50% in AABs-positive GDM women, and of 21% of IFG or IGT [10]. In a recent work from our group, women followed for 2 years after pregnancy had a 2.65 greater risk of having impaired glucose regulation when T1D-related autoantibodies where present, suggesting a continuing damage to beta-cells by the autoimmune process [43].

Overall, understanding the etiopathogenesis (i.e. autoimmune or non-autoimmune) of GDM is useful to evaluate the future risk of glycaemic alterations. The real question is “when it is useful to search for an autoimmune-diabetes related diagnosis?”