Myasthenia gravis (MG) is a B-cell-driven, T-cell-dependent autoimmune disease (AID), which results from the transmission disorder in the post-synaptic neuromuscular junction caused by the production of antibodies for acetylcholine receptors (AChRs) (1, 2) or other proteins, such as muscle-specific kinase (MuSK) (3). Low-density lipoprotein receptor-4 (LRP-4) and agrin were also considered indicative of MG. The main clinical manifestations of MG include muscle weakness and abnormal fatigability which seriously affect the quality of patients' daily life. Generally, MG can be divided into different clinical subtypes. According to the different distributions of muscle weakness, MG can be divided into ocular MG (OMG) and generalized MG (GMG) (4). Grouped from the age of onset, MG patients also can be divided into early-onset (EOMG) and late-onset (LOMG) forms by the age of 50 years (5). It has been confirmed that the incidence of MG in the elderly of both sexes is increasing (5). Incredibly, very late-onset MG patients (onset age: 65 years or older) are more prone to severity (6). Thymoma-associated myasthenia gravis (TAMG) was considered a different subset from EOMG/LOMG or OMG/GMG clinical classification because of its specific characteristics in pathogenesis and clinical features (7). The traditional therapies of MG are acetylcholinesterase inhibitors (AChEIs), immunomodulatory therapies, thymectomy, and so on (8).

Type 2 diabetes mellitus (T2DM) is a common chronic disease, characterized by elevated blood glucose and attenuated insulin action. The incidence of T2DM has been rapidly increasing over recent decades, which has become a significant public health problem in the general population of China (9). Although the traditional view is that the correlation between MG and DM might be related to the adverse effect of corticosteroid and aggressive therapy (10), a 35-year retrospective study in 2007 raised doubts that it remained unclear whether prednisolone administration directly induced DM or if it merely accelerated disease onset (11). We found that quite a lot of MG patients had T2DM prior to the onset of MG. Traditional views cannot explain the phenomenon we found in the clinical work and may even obscure the true relationship between T2DM and MG.

In the present study, we propose a preliminary hypothesis: T2DM may be associated with MG. A matched case–control study was designed to systematically test the hypothesis. The chronological order of the onset of T2DM and MG was also considered. Furthermore, the correlation between T2DM and MG, potential clinical features, and possible molecular mechanisms were elaborated. This discovery will provide innovative ideas for MG diagnosis and treatment.

Our study showed that the top eight comorbidities of MG with prevalence were hypertension (HTN) (34.04%), thymoma (27.66%), T2DM (23.40%), coronary heart disease (18.44%), pulmonary infection (16.31%), cerebral infarction (16.31%), pneumonia (7.09%), and cataract (7.09%). Most comorbidities, except for thymoma, were related to geriatric chronic disease or infection. The correlation between thymoma and MG is already acknowledged, and the other top two comorbidities (HTN and T2DM) were focused on. The prevalence of HTN and T2DM in the MG inpatients and the general population was separately collected and analyzed. In terms of the prevalence of HTN, there was no statistical difference between controls and MG patients (33.77% vs. 34.56%, P > 0.05), while the T2DM prevalence of MG patients was higher than that of controls (24.26% vs. 9.54%, P < 0.001) (Figure 2).

Research has confirmed that a high proportion of abnormalities in glycolipid metabolism has been present in MG patients before treatment with glucocorticoids (16). According to the International Diabetes Federation (IDF) Diabetes Atlas (eighth edition 2017) and the latest Chinese diabetes epidemiological survey, the prevalence of diabetes (20–79 years old) in China was 10.9% (17), which was far lower than that of MG hospitalized patients in this study (24.26%). The probability of MG patients complicated with diabetes was higher than that of the general population. The risk of MG in association with T2DM exposure was explored through an EMR-based, 1:5 matched case–control study. This study illustrated distinct age- and sex-specific associations between T2DM and MG.

In this study, the results of four datasets showed that patients with T2DM had a higher risk of MG than those without T2DM. The sources of the control population of the four datasets were different. Datasets A and B represented the comparison of MG patients and the general population settled in the area. The findings can be extrapolated to the whole population more reliably by selecting controls from the local population without MG. The result of dataset A could only reveal a statistical association between T2DM exposure and MG but not a causal association. For chronic diseases, the time interval between exposure and illness still needs to be considered to determine the cause and effect of this association. In the MG patients of dataset B, the onset of diabetes preceded MG, which is consistent with the sequence of a causal association. The results of dataset B further proved that T2DM might be one of the causative factors of MG. Datasets C and D represented the comparison of MG inpatients and other inpatients, which made better homology and comparability. The results of dataset C demonstrated that MG is associated with diabetes in the female hospitalized population. The results of dataset D further extended the significance of this study. The controls of dataset D were other AID inpatients. The result of dataset D illustrated that MG, but not any other autoimmune disease, was strongly associated with diabetes.

The female population of four different datasets was further stratified by age. The results suggested that T2DM remained significantly associated with the risk of MG in any dataset. Age may be the deciding determinant among female patients as well. As it is well known, the significant asymmetry of AIDs between sexes is thought to be related to sex hormones, sex chromosomes, gut microbiota, and so on (18). There may be differences in etiology according to sex and aging that affect the progression of MG in diabetic patients. Research on related molecular mechanisms is urgently needed.

The mean onset age of diabetic MG patients was significantly higher than that of non-diabetic MG patients, illustrating that the clinical types of diabetic MG patients were mostly LOMG (Table 5). It has been reported that diabetes was more prevalent in LOMG when compared to EOMG (19, 20). There are no differences in different autoantibodies in MG with type 2 diabetes, compared with patients without T2DM. The AChR antibody still takes the overwhelming majority when tested in Chinese patients (21). The lack of positive results may be related to data deficiencies.

A verdict of causation can be safely made only on a sufficient totality of evidence. Several positive criteria support a judgment of causality, including (22, 23) (1) strength; (2) consistency; (3) specificity; (4) temporality; (5) biological gradient; (6) plausibility; (7) coherence; (8) experiment; and (9) analogy. In our study, the risk for MG in T2DM patients was 1.91 to 4.56 times higher than that of groups without T2DM. After identifying the temporal sequence of DM and MG, the results were still positive. The same answer has been reached by different populations with different circumstances. This study should be viewed in the context of its limitations. First, dose–response relationship analyses about the severity of illness and more clinical subgroup analyses, except for age and sex, were not attempted. Second, there may be other unknown confounding factors or mediating variables related to both MG and T2DM, obscuring or exaggerating the relationship between MG and T2DM.

This new finding has led to our further thinking. The concept of diabetic MG may need to be paid more attention to. It may have unique clinical electrophysiology and also clinical and immunological phenotypes distinct from previous subgroup classifications. Previous studies could not explain the relationship between T2DM and the subsequent onset of MG. A protein named advanced glycation end products (AGEs), which are greatly augmented in the early stage of diabetes, might play an important role in this process. As the main ligand of the receptor for AGEs (RAGE) (24), AGEs stimulate inflammatory signal amplification, produce several cytokines and growth factors (25), enhance the proliferation response of CD4(+)CD28(-) T cells (26), and further act in AIDs (27, 28). RAGE mainly acts in pro-inflammatory cellular responses and contributes to the failure of self-tolerance mechanisms (29). MG patients with strong RAGE expression in the thymus tissue had higher levels of AChR-Ab, especially at the disease onset of TAMG (30). In experimental autoimmune MG (EAMG) animal models, increased expression of RAGE and its ligand S100B enhanced T-cell pro-inflammatory responses, affected the Th1/Th2/Th17/Treg cell equilibrium, and upregulated AChR-specific T-cell proliferation (31). S100B-RAGE interaction promotes the production of antibodies against AchRs from the spleen of EAMG (32). The clinical importance of AGEs-RAGE signaling in diabetic MG pathophysiology should be emphasized.

Recently, our group has confirmed that DM aggravated the aberrant humoral immunity in MG patients by promoting differentiation and activation of circulating follicular helper T (Tfh) cells (33). In vitro, hyperglycemia increased the expression of inducible costimulatory (ICOS) in Tfh cells via the mechanistic target of the rapamycin (mTOR) signaling pathway. This process promoted the differentiation of plasmablasts and IgG secretion. Monitoring humoral immune indicators may be beneficial to patients. The possible immunologic link between MG and T2DM will provide a new strategy for the prevention and treatment of MG.

Overall, this study has shown a significant association between T2DM and MG, suggesting T2DM may be a risk factor for the onset of MG. Basic experiments to explore relevant molecular mechanisms are needed, but there is no denying that the relationship between T2DM and MG has come to our attention. We hypothesize that a diabetic MG subtype may exist in LOMG patients. Further personalized diagnosis, treatment, and management of diabetic MG patients should be on the agenda.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Research Ethics Committee of the First Affiliated Hospital of Shandong First Medical University (No: S030). The patients/participants provided their written informed consent to participate in this study.

R-SD, Y-DL, and FT contributed to the conception and design of this study. X-LL, HL, C-CW, and BY contributed to data acquisition. Y-DL and X-LL organized the database. PZ, C-LY, and TD contributed to the literature search. Y-DL and Y-FL contributed to data analysis and interpretation and verified the underlying data. Y-DL, FT, YL, and R-SD contributed to drafting the manuscript and figures. All authors contributed to the manuscript revision and read and approved the submitted version of the manuscript.