The index patient is homozygous for the c.130G>A variant of the INS gene, which most likely fits with the diagnosis monogenic diabetes. To our knowledge, the c.130G>A INS variant presented here has not been described before in a homozygous state. Four patients have been discovered in large-population studies who were heterozygous for the reported variant.3 12 21 22

Earlier reported patient cases, who were heterozygous for the c.130G>A variant of the INS gene, included a woman diagnosed with gestational diabetes12 and a women diagnosed with mild diabetes who was also positive for anti-GAD and anti-IA2 (consistent with auto-immune diabetes).3 22 It remains questionable if the INS gene variant was directly causative for the diabetes onset in these patients. Interestingly, the aforementioned female patient,3 22 as well as another female patient described by Flannick et al,22 were both diagnosed with diabetes outside the neonatal period at the age of 17–29 years, respectively. While both female patients were diagnosed with the same INS gene variant and a mild form of diabetes, one could control her blood glucose levels well with diet alone, while the other needed metformin treatment.23 The fourth known patient was diagnosed with MODY10, as a result of the INS gene variant.21 These cases highlight variability in diabetes presentation among individuals with the INS gene variant, including differences in age of onset, diabetes type and treatment requirements. However, it remains uncertain whether the INS gene variant is directly causative in the onset of diabetes in these patients. On the other hand, one could argue that all these cases, including the patient presented here, show that the c.130G>A variant of the INS gene is consistent with delayed onset diabetes, which is in line with a recent review covering all known INS gene variants and their respective association with diabetes onset.5

The onset of diabetes in our patient could potentially be explained by either a misfolded (pro)insulin with reduced biosynthesis causing pancreatic β-cell dysfunction or an (pro)insulin variant that has less potency to the IR.6–10 To investigate this, we first evaluated if the patient synthesizes a proinsulin-like or insulin-like protein. Second, we evaluated if the interaction and/or potency of this (pro)insulin variant with the IR was affected.

We used several assays to detect the synthesis of insulin and C-peptide proteins. The COBAS 8000 assay did not detect insulin. However, it uses a monoclonal antibody, which potentially binds the epitope containing the amino acid substitution p.Gly44Arg, hence blocking its binding. The RIA assay did detect insulin. However, it uses a polyclonal antibody and is known to cross-react with proinsulin. Therefore, we selected another insulin assay (Lumipulse) that uses a monoclonal antibody with negligible cross reactivity with proinsulin and the C-peptide. The combination of these three assays confirms that an insulin-like protein is synthesized and C-Peptide values are quantifiable. These results suggest that the c.130G>A variant of the INS gene does not affect insulin-synthesis from proinsulin.

Next, we reconstructed the insulin variant in silico and evaluated through 3D modeling whether the interaction with the IR was disturbed. The INS variant (c.130G>A) resulted in an amino acid substitution at position 20 of the mature insulin B-chain (GlyB20→Arg). We concluded that the GlyB20→Arg substitution (structurally) interferes with IR interaction. Moreover, our functional studies showed that the INS variant had reduced potency in binding to the IR and downstream signaling. It remains questionable whether this reduced potency explains the onset of diabetes in this patient, but the bioactivity of this insulin variant may have been sufficient to prevent the onset of diabetes during childhood and adolescence. Nonetheless, INS mutations may result in diabetes due to reduced insulin bioactivity and less severe insulin deficiency,9 24 which is in line with our observations here.

The exact pathophysiological mechanism of the c.130G>A variant remains elusive. We found no empirical evidence that p.Gly44Arg results in a misfolded insulin protein. Nonetheless, it is apparent that INS gene variations introduce misfolded (pro)insulins, as a consequence of endoplasmic reticulum stress, and are a major cause of pancreatic β-cell death resulting in NDM25 and MODY.5 26 Recent research has highlighted a novel mechanism underlying NDM in which, next to pancreatic β-cell death, specific INS mutation results in dedifferentiation of pancreatic β-cells where they lose their specialized function and revert to a more immature or progenitor-like state.27 Interestingly, there are 33 known clinical INS gene variants that all have a different clinical presentation.5 Notably, the β-turn of the insulin B-chain consists of four amino acids (B20–B23) that are all known clinical INS gene variation sites that cause MODY10.5 The fact that B23 amino acid substitutions can lead to either neonatal onset (ValB23) or delayed onset (AspB23) indicates that the specific side chain chemistry plays a role in influencing folding efficiency. SerB24, known as insulin Los Angeles,28 is associated with familial hyperinsulinemia where SerB24-proinsulin may fold correctly in the endoplasmic reticulum and thus processes into mature insulin.29 Given that GlyB20→Arg alters the ϕ dihedral angle, it is plausible that the insulin variant described here is potentially prone to proinsulin misfolding but left unharmed for subsequent processing into insulin. Our findings suggest that the pathological mechanism resulting in a mild late-onset of diabetes is most likely different from the classical NDM patients, which are homozygous for an INS variant.

It is also important to note that INS gene variants not linked to impaired protein folding can still be associated with adult-onset diabetes. Classical insulinopathies like insulin Wakayama30 and Chicago31 (ValA3→Leu and PheB25→Leu, respectively) are causes of familial hyperinsulinemia and give rise to insulin variants known to hinder IR binding.29 Given that each INS gene variant manifests through distinct disease mechanisms and clinical presentations, the genotype-to-phenotype translation is far from straightforward. Although various studies have provided findings that align with our observations, it is plausible that the variant described here may induce a combination of two mechanisms: reduced IR binding affinity and proinsulin misfolding, ultimately leading to beta-cell death or dedifferentiation. This dual mechanism could explain the complex clinical outcome associated with this particular variant.

Interestingly, the patient’s brother is homozygous for the same INS gene variant and has a similar late-onset of (mild) diabetes. Their parents had no history of diabetes, were not consanguineous and likely are heterozygous for this INS variant. One may suggest that a carrier of c.130G>A exhibits normal glucose tolerance, since one (normal) insulin allele could be enough to maintain normal blood sugar levels.

In summary, the index patient is homozygous for an INS gene variant. The insulin variant shows reduced IR affinity and bioactivity, and supposedly combined with pancreatic β-cell apoptosis or dedifferentiation leads to, in later life, a relative deficit leading to adult-onset monogenic diabetes.