Although early-onset diabetes is a highly heterogeneous group of disorders, in a subgroup of these patients, such as eDia3, genetic factors might play a significant role in the pathophysiology, and the disease may be less influenced by environmental risk factors. Proper classification of these patients is a major challenge to clinicians. A previous study defined the multigenerational form of diabetes mellitus as “familial diabetes of adulthood” (FDA) (30) and revealed significant clinical differences between FDA and T2D. In this study, we performed an extreme case–control study with patients with eDia3 as cases and those with eDia0 as controls. Although the age of diabetes diagnosis was under 40 years, statistically significant differences in diabetes onset age, duration, BMI and metabolic biomarkers were found between the two groups. The results suggested a hypothesis of different pathogenetic backgrounds between the two subgroups.
Current guidelines identify candidates for performing MODY genetic testing, including age at diagnosis typically before 25 years, noninsulin dependence, and family history of diabetes with at least two generations (31). However, a study selected 1,564 probands and reported that using stringent inclusion criteria would miss 70% of cases of monogenic diabetes (32). Meanwhile, with the rise of obesity in young and middle-aged individuals in recent years, remarkable overlaps of characteristics were observed between MODY and T1D/T2D patients. Therefore, this study detected MODY in patients with eDia3 and did not restrict the weight of the patients. Four out of 10 MODY patients in this study had diabetes diagnosed older than 30 years. Potential clinical biomarkers were investigated to help prioritize the strategy of selecting diabetes patients for genetic testing (33). Although our study found that eDia3 is significantly different from eDia0 in clinical characteristics, it was also found that these clinical indicators could not be used as a precise biomarker for known MODY screening.
WES was carried out in the 89 patients with eDia3 and the findings demonstrated that variants of genes related to MODY1-14 were not mainly causes for patients with eDia3 in China. The genetic confirmed MODY was detected in 11.2% patients with eDia3 (10/89) and only in 1.13% early-onset diabetes patients (10/884). These results are comparable to a Korea study which found a prevalence of 12.8% in the four relatively common MODY genes (HNF1A, HNF4A, HNF1B or GCK) among the 109 diabetes patients with onset age ≤ 30 years and a BMI ≤ 30 kg/m2 (34). Similarly, an UK study demonstrated that the mutation pick-up rate of MODY genes (HNF1A, HNF4A, HNF1B or GCK) in South Asian participants was 12.6%, lower than White European group (25.2%)(35). Conversely, a recent Chinese study selected 42 clinically diagnosed MODY aged ≤ 18 years and identified 24 patients (57.1%) had mutations in the known MODY genes(36).This study demonstrated that mutations of genes related to MODY1-14 were not the main cause of eDia3 in Chinese patients, which indicated that the pathogenic background of eDia3 needs further investigation in the future.
Except the 10 variants that likely pathogenic to MODY, we also found 11 rare, non-silent variants in 24 patients, classified as likely-benign or uncertain significance. Of which, the variants of PAX4 were identified in 16 patients, with PAX4 Arg192His variant (chr7:127253550, rs2233580) in 8, PAX4 Arg192Ser variant (chr7:127253551, rs3824004) in 5, and PAX4 Arg31Gln variant (chr7:127255483, rs115887120) in 3. PAX4 is a transcription factor that paly an crucial role in beta cell development, differentiation and survival(37). It had been suggested that the mutations of PAX4 gene were positively and ethnic-specifically associated with the risk of T2D in Asian population(38). Genome wide association studies in Chinese populations identified Pax4 arg192his (rs2233580) as a T2DM susceptibility locus(39). A Korean study found that the combination of PAX4 Arg192His and PAX4 Arg192Ser could be considered as a strong risk factor for T2D, and having two copies of PAX4 Arg192His variant was related to a 7.0 years earlier onset of diabetes(40). Other studies also provided evidence that missense variant rs2233580 (p.Arg192His) in PAX4 gene was significantly associated with T2D, which is related to the reduction of C-peptide and the age of diagnosis in T2D patients (41) Combined with the occurrence of Pax4 arg192his (rs2233580) genotype in eDia3 patients in this study, it is also confirmed that it may be a high-risk genetic factor for eDia3 in China(39).
Previous study suggested HNF1A-MODY (53%), GCK-MODY (32%) were most common subtypes of MODY(42). However, the etiology of the MODY in our study demonstrated that variants of KLF11 genes were more frequently involved. A Chinese research also identified the prevalence of HNF1A-MODY and GCK-MODY were only 9% and 1% in patients with suspected MODY(43). The prevalence of rare subtypes of MODY was relatively high in patients with eDia3 in our study. The cause of the variation in the frequencies of mutations between our data and previous reports remains unclear. The different genetic background might be an important reason for the phenomena. Our findings indicated that the pathogenic background of hyperglycemia had not been elucidated in vast majority of patients with eDia3, especially expanding age and BMI standards, which require further and broader attempts and get deeper insight into the molecular causes in the future investigation.
There are some limitations of our study. First, it was a hospital-based study including patients with relatively high HbA1c and increased prevalence of diabetic vascular complications. Therefore, patients with mild asymptomatic hyperglycemia could not be selected in our study, which may influence the frequency of detected gene mutations. Second, the number of participants was low, and WES test of the eDia0 cohort was not carried out in this study, so it is unclear whether there were genetically diagnosed MODY patients in the control group. We may need to increase the sample amount in later studies, and carried out WES test in eDia0 cohort. Third, some relatives of the genetic confirmed MODY patients could not be connected to perform the genetic testing. In some cases, due to the unavailable information of all family members related to the probands, we could not perform a segregation analysis of some rare potentially pathogenic variants identified in our study.
In summary, eDia3 patients had different clinical characteristics from age-matched T2D patients. Known MODY genes were not common causes of clinically suspected MODY, and KLF11 gene mutations were more frequently identified in these patients in China. Hence, more comprehensive studies are needed.