Searching for Monogenic Diabetes in a High-risk Autoimmune Diabetes Cohort: Needles in a Paperclip Stack.

Classical type 1 diabetes (T1D) is caused by the autoimmune destruction of pancreatic ß cells, leading to the immediate requirement of exogenous insulin to regulate blood glucose. Although the most common cause of childhood diabetes is T1D, some children who present with diabetes (1.2%) (1) actually have monogenic diabetes, either syndromic diabetes or early-onset maturity onset diabetes of the young (MODY). Pathogenic rare variants in genes traditionally associated with MODY may cause onset of diabetes in childhood. There are at least 14 known “MODY genes,” and mutations in GCK (OMIM 125851), HNF1A (OMIM 600496), and HNF4A (OMIM 600281) collectively account for most cases (2). Sulfonylureas are often prescribed as first-line treatment, followed by glucagon-like peptide 1 receptor agonists and insulin (3). Unfortunately, inappropriate use or dosing of insulin may lead to episodes of potentially life-threatening hypoglycemia and even predispose to weight gain (3). Thus, early and correct identification of a monogenic cause, when one is present, should improve disease management over the long term.

Contemporary diagnostic algorithms for MODY (3) recommend DNA sequencing among patients who test negative for autoantibodies (aAbs). Next-generation sequencing (NGS) multigene panels are used frequently, but exome sequencing may be justified in some atypical cases. The variant interpretation step will then seek highly penetrant likely pathogenic and pathogenic variants, and only likely pathogenic and pathogenic variants should be classed as “mutations.” Nevertheless, it will be awhile before NGS tests are sufficiently cost-effective to be ordered during the initial diagnostic workup of suspected pediatric T1D. The analogy of needles in a haystack is often applied to the search for causative genetic mutations within individuals and populations, but the genetic architecture of T1D is largely composed of common low-penetrance variants such as those in the HLA cluster (perhaps symbolized by paperclips, in an admitted extension of the metaphor) rather than highly penetrant rare variants (the needles). Thus, monogenic causes of diabetes are traditionally sought among patients who have been preselected for those who have a high prior probability, which are typically those with an affected first-degree relative such as a parent or sibling.

In this issue of the Journal, Marchand and colleagues (4) set out to test whether preselection for biological markers indicative of a low autoimmune T1D risk would positively enrich for previously unsuspected cases of monogenic diabetes, thereby enhancing the diagnostic efficiency of exome sequencing. To do so, they investigated cases collected within the Type 1 Diabetes Genetic Consortium, which was set up to interrogate the genetic architecture of T1D risk. Because the genetic risk of T1D is most strongly associated with the HLA class II locus, Marchand et al selected for families wherein all those affected were both aAb-negative and did not carry HLA haplotypes conferring strong autoimmune risk for T1D. Fifty-two families met these criteria (~2% of the entire cohort), of which DNA from 46 probands was sequenced for rare coding variants. As expected, in this subgroup they found a high frequency (>40%) of protein-altering variants in known MODY genes, the most common of which were in HNF1A. However, recessive variants in WFS1 were a close second in number. WFS1 is considered primarily to be a syndromic diabetes gene rather than a MODY gene.

Homozygosity or compound heterozygosity for pathogenic mutations in WFS1 is known to cause Wolfram syndrome (OMIM 222300), sometimes referred to as diabetes insipidus, diabetes mellitus, optic atrophy, and deafness. This is a rare disorder with a neurodegenerative component (5). Because individuals known to have syndromic diabetes were deemed ineligible for inclusion, cases of Wolfram syndrome would have been excluded from the Type 1 Diabetes Genetic Consortium, if they had had features other than diabetes in either sibling at the time of enrollment. Interestingly, the prevalence of patients with recessive WFS1 variants was comparable to HNF1A-MODY, and much higher than expected based on the estimated prevalence of Wolfram syndrome in the general population (5). This prevalence suggests that many of these cases may truly be nonsyndromic in nature; put another way, the patients identified by Marchand and colleagues may have childhood-onset diabetes as the only manifestation of their biallelic WFS1 variants. Recessive WFS1 variants have also been found by these researchers in an aAb-negative Chinese cohort (6), who also have a low genetic risk of T1D based on ancestral haplotypes. The ascertainment of WFS1 variants within a cohort that excluded extrapancreatic manifestations in family members at time of recruitment would presumably identify more “diabetes-only” WFS variants than would cohorts preselected for families in which multiple members have more than 1 feature of Wolfram syndrome. Because the phenotype is age-dependent, more detailed functional studies of any putative “diabetes-only” WFS1 variants will need to be paired with clinical follow-up of patients with these variants to test the possibility that these WFS1 variants might result in onset of diabetes in childhood without additional features of Wolfram syndrome later in life (7, 8). Functional studies might also reveal quantitatively milder phenotypic effects from partial loss-of-function variants, and would hopefully elucidate the mechanism(s) whereby these variants lead to childhood diabetes (reviewed systematically by de Heredia et al (8)).

Marchand and colleagues provide an excellent example of how highly penetrant rare variants contribute to what is seen by many as common disease. Although this study was not designed with the intent of changing existing diagnostic screening criteria for monogenic diabetes, the authors provide proof-of-concept that inclusion of multiple biomarkers indicative of a low risk of classical T1D can improve the efficiency of NGS in identifying monogenic diabetes. Furthermore, their findings show that refinement of case selection criteria can aid in the discovery of novel rare variants associated with diabetes. More interestingly, we think, such curation will likely flag variants in genes typically associated with syndromic diabetes in children who do not necessarily manifest the full clinical spectrum. Certainly, the work of Marchand and colleagues can help guide clinical practice and disease management in regions where aAb testing and NGS are more readily available.