Introduction
Overweight and obesity are major risk factors for type 2 diabetes [
1], and the association between excessive weight and insulin resistance is well known. Several mediating pathways have been proposed, including ectopic lipid accumulation and lipotoxicity, and the release of proinflammatory cytokines from visceral fat tissue [
2].
Type 1 diabetes has been viewed as a non-obese form of diabetes, but this was challenged by the accelerator hypothesis [
3], which proposes that obesity accelerates disease onset and, further, that insulin resistance is a common underlying feature of all types of diabetes [
3]. Insulin resistance has also been postulated to be an independent risk factor for type 1 diabetes [
4,
5]. Adiposity could potentially affect risk via beta cell autoimmunity; adipokines, which are secreted from excessive fat tissue, have been shown to be involved in various immune-mediating processes [
6]. Subsequent prospective studies have reported a twofold increased risk of type 1 diabetes in obese children [
7] and obese adults [
8], while others find no association [
9]. An association is further supported by the coincident increases in childhood obesity and type 1 diabetes incidence [
10,
11].
Latent autoimmune diabetes in adults (LADA) is an autoimmune form of diabetes with features of type 2 diabetes, including adult onset and insulin resistance [
12]. Autoimmunity is typically less pronounced than in type 1 diabetes, which implies that insulin resistance, increasing the beta cell demand, may play a key role in the promotion of LADA. In line with this, data from cross-sectional studies [
13‐
19] suggest that individuals with LADA tend to have higher BMI than those with type 1 diabetes, but lower than those with type 2 diabetes. Interestingly, the clinical phenotype of LADA is known to vary by degree of autoimmunity, with less autoimmune individuals being more type 2-like. Hence, the role of overweight in the development of LADA may depend on the severity of the underlying autoimmune process. Only one prospective study based on 11 years follow-up of the Norwegian Nord-Trøndelag Health Study (HUNT Study) estimated the risk of LADA in relation to overweight/obesity [
20]. This study was based on only 81 individuals; hence, the influence of excessive weight on more or less autoimmune forms of LADA could not be explored, and confounding control was limited. Other aspects that remain to be addressed include interaction between overweight and family history of diabetes (FHD), which is a strong indicator of type 2 diabetes risk [
21], whether the shape of association is linear or not, and the preventive potential of overweight in the aetiology of LADA.
Our aim was to describe the association between overweight and obesity and LADA compared with type 2 diabetes, taking into account degree of autoimmunity and potential interaction with FHD. We used updated prospective data from the HUNT Study, including 22 years of follow-up, and newly collected data from a Swedish case–control study with incident cases.
Discussion
Our findings using data from two large population-based studies indicate that overweight and obesity are associated with an increased risk of LADA and that the risk is highest in individuals with a combination of overweight and FHD. The association with obesity seemed strongest in LADA with low GADA, but was apparent also in LADA with higher GADA levels. The results indicate that LADA in 31–56% of individuals could be attributed to overweight/obesity, compared with 70–82% of all those with type 2 diabetes.
These findings fit with those of previous cross-sectional studies, which indicated that individuals with LADA tend to be obese but leaner than those with type 2 diabetes [
13‐
19] and with previous reports of LADA being characterised by insulin resistance, but to a lesser extent than type 2 diabetes [
12]. One previous study found that a majority of individuals with LADA have a lean phenotype [
37]. One explanation of this somewhat conflicting result may be the use of a different age criterion (>25 years), as younger age at onset tends to be associated with a more type-1-like phenotype [
38]. In contrast, the large multicentre ADOPT study found that participants with LADA and type 2 diabetes were equally overweight/obese [
39]. In this study, however, GADA was measured in individuals with prevalent diabetes without insulin treatment within the first 3 years of diagnosis. As such, these individuals with LADA were likely to have a more type-2-like phenotype. These findings highlight the heterogeneous nature of LADA and the need for a unified definition.
BMI was positively associated with insulin resistance in LADA, suggesting that this is an underlying pathway. In contrast, there was nothing to suggest that excessive weight would influence autoimmunity per se; there was an inverse association between BMI and GADA level similar to a previous report [
25]. Reports of type 1 diabetes in children are in keeping with our data; obesity has been associated with insulin resistance [
40], but not with autoimmunity, irrespective of number and type of diabetes antibodies in the study participants [
41]. Our findings fit with the accelerator hypothesis [
3], which proposes that insulin resistance plays a role in promoting autoimmune diabetes by increasing the insulin demand—this may accelerate disease onset in individuals with an ongoing autoimmune process. In the case of mild autoimmunity, one can hypothesise that factors related to insulin resistance are more important for progression to overt diabetes. This could explain why we found a stronger association between high BMI and less autoimmune LADA and also why the phenotype of the obese individuals with LADA compared with those with normal weight, in line with previous reports [
13,
16‐
18,
25], was more type-2-like, with higher C-peptide levels, better beta cell function and a higher level of insulin resistance. There was a tendency towards a U-shaped relationship between LADA and BMI. If not occurring by chance, it may reflect the weight loss often seen in individuals with type 1 diabetes prior to diagnosis as a consequence of insufficient insulin production.
The association between BMI and LADA was stronger in the prospective data from HUNT, where BMI was assessed several years prior to diagnosis, than in the Swedish case–control data, where BMI was assessed at time of diagnosis. It is possible that the baseline measurements in HUNT reflect a more aetiologically relevant exposure window. Self-reported weight in the case–control study may also have contributed to dilution of associations. On the other hand, the association between type 2 diabetes and BMI was stronger in ESTRID. Another explanation may be that the LADA populations differ in either genetic or unmeasured phenotypical factors.
We confirm the strong association previously reported of overweight and obesity with type 2 diabetes [
1]. In addition, we confirm that the combination of overweight and FHD dramatically increases the risk of type 2 diabetes [
21] and show, for the first time, that the risk of LADA increases substantially in individuals with FHD and overweight, although the effect is not as pronounced as for type 2 diabetes. Unfortunately, the numbers did not allow us to explore interaction with BMI separately in individuals with a family history of type 1 diabetes vs those with a family history of type 2 diabetes. We have previously shown that LADA is associated with a family history of type 2 diabetes, but even more so with a family history of type 1 diabetes [
42], which is in line with genetic studies showing a strong link between LADA and HLA genotypes associated with type 1 diabetes [
12]. Together with the findings of present study, this supports the idea that LADA is a hybrid form of diabetes promoted by genes associated with autoimmunity and lifestyle factors inducing insulin resistance.
Strengths and limitations
The strengths of this study include the large number of incident cases, detailed information on potential confounders and the use of two well-defined population-based studies. The specificity of the GADA assessment was high, but it is possible that some participants with type 2 diabetes were misclassified as LADA, i.e. were false positives. This may contribute to an association with BMI, especially for LADA
low. It has also been suggested that individuals with LADA and low GADA are actually false positives [
43]. However, we found that these individuals differ from those with type 2 diabetes in several clinical characteristics. Also, previous studies in the HUNT Study indicate a real impact of even low and transient levels of GADA, e.g. individuals with low GADA display lower fasting C-peptide levels than individuals with type 2 diabetes [
44]. Still, the importance of GADA positivity for disease progression in very obese individuals with low GADA levels is unclear. The sensitivity of the method and the use of only one autoantibody imply that some individuals with LADA were classified as GADA negative, i.e. as having type 2 diabetes. Importantly, GADA is by far the most common autoantibody in LADA, present in ~90% of all individuals [
45]. In the HUNT Study, some individuals had GADA measured several years after diagnosis. Because GADA can disappear after prolonged disease duration [
44], it is possible that some individuals with LADA therefore appeared GADA negative and were classified as having type 2 diabetes. Notably, GADA tends to be more stable in LADA than in type 1 diabetes [
12]. Although there is no unified definition of LADA, the present report is consistent with currently used criteria [
12], with the exception of C-peptide which was used in ESTRID as an indicator of remaining insulin production and can be considered a more objective measure compared with the frequently used insulin criterion, i.e. lack of insulin treatment 6–12 months after diagnosis [
12]. Estimation of PAR is based on the assumptions of causality and the absence of measurement errors and bias and should hence be interpreted with caution. As for generalisability, it should be noted that PAR is based on the estimated effect size as well as prevalence of overweight in the population and is, as such, population specific. The study is based on populations in Scandinavia, where the incidence of autoimmune diabetes is high, and the results may be less generalisable to areas with lower incidence. Last, assessment on insulin resistance was based on HOMA and even though HOMA has been validated against the hyperinsulinaemic–euglycemic clamp with good correlation [
46], it is still a crude method.
In conclusion, under the assumption of causality, excessive weight is a strong contributor to development of LADA and maintaining a healthy weight should be a priority, especially in the presence of FHD or autoimmunity. As expected, obese individuals with LADA had a more type-2-like phenotype, but overweight/obesity was also associated with more autoimmune LADA. These findings support the hypothesis that even in the presence of autoimmunity, factors linked to insulin resistance such as excessive weight could promote the onset of diabetes.
Acknowledgements
We thank all participants in HUNT, ESTRID, ANDIS and ANDiU as well as administrative personnel, nurses and research team members from all the studies.
The preliminary results from this work were presented as an abstract at the 53rd EASD Annual Meeting 2017, Lisbon, Portugal.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.