Our retrospective single-centre study of 754 children with diabetes in a highly consanguineous population found that 95% had a clinical diagnosis of type 1 diabetes. To our knowledge, this is the first such study from a region with a high consanguinity rate of up to 44%. The other subtypes of diabetes were much less common: 1.9% had neonatal diabetes, 1.5% had clinically defined MODY, 1.1% had type 2 diabetes and 0.9% had syndromic diabetes. In a study from Turkey (20% consanguinity), 84% of participants were classified as having type 1 diabetes and 5.7% as having type 2 diabetes [8]. A more recent Turkish study focused on children with monogenic diabetes and found a prevalence of 3.1%, which was similar to that in a cohort from the UK (lower rate of consanguinity) [20]. Our study found the prevalence of monogenic diabetes to be 4.2% among all participants followed at the single centre. However, because of the lack of standard antibody testing in those with diabetes, a proportion of individuals with MODY subtypes (such as HNF1A/HNF4A-MODY) may be misdiagnosed as having type 1 diabetes. Therefore, the actual percentage of individuals with MODY in this population may be higher, which could lead to a greater overall prevalence of monogenic diabetes. The higher prevalence may be attributed to the presence of more recessive forms of diabetes due to the high percentage of consanguinity.

The rate of consanguinity in the study region of Sulaimani has not been formally documented, but a study investigating congenital heart disease in the same region reported 41% consanguinity among the study population [12]. In a 2010 study evaluating consanguinity in the region of Baghdad, Iraq, the rate was 44% [13]. There was a significant association between consanguinity and sociodemographic characteristics, such as differences in rates among urban and rural populations. Furthermore, in a study from north-west Iran, which is geographically close to Kurdistan, Iraq, the rate of consanguinity was 39.1% [14]. These data can be used to make an approximation of the possible rate of consanguinity in the region of Sulaimani.

The consanguinity rate among participants followed at our centre was 36.5%. This is close to the rate among the majority group of participants—those with type 1 diabetes (35%). Although previous studies have shown that the prevalence of type 1 diabetes is not influenced by consanguinity, there is a higher risk of development of type 1 diabetes if there is a history of diabetes in first cousin parents [24]. We confirmed that there was no statistically significant association between consanguinity and type 1 diabetes.

It would be expected that a child from a first cousin marriage would have 6% of the genome covered in ROH. Probands with positive consanguinity who underwent genetic testing (all from reported first cousin marriages) had 3.8–13.5% ROH (Table 1). The participants with no reported consanguinity had 0.0–2.0% ROH. However, because only coding regions were analysed by WES, there may be additional ROH in non-coding regions of the genome.

A genetic diagnosis was identified in 83% of participants with neonatal diabetes who were available for testing by WES (i.e. 10/12). Among these, 80% were homozygous for pathogenic variants causing the disease. There were only two participants with transient neonatal diabetes, with variants in ABCC8 and PTF1A (VUS). In a study comparing genetic causes of neonatal diabetes among consanguineous and non-consanguineous populations, the most common cause of neonatal diabetes among participants born to consanguineous parents was recessive EIF2AK3 gene variations causing Wolcott–Rallison syndrome, whereas in non-consanguineous populations pathogenic variants in the KCNJ11 and ABCC8 genes accounted for the majority of cases (46%). These genes (KCNJ11 and ABCC8) accounted for only 12% of cases in the consanguineous group [10]. Moreover, there was a much higher incidence of recessive forms of neonatal diabetes in consanguineous regions, which we also found in our study.

The overall spectrum of monogenic diabetes genes found in our study population was different from what is found in non-consanguineous populations [10, 25]. We observed homozygous, causal variants in genes such as PTF1A, GLIS3, INSR and SLC29A3, which are uncommon in non-consanguineous populations. Consanguineous populations may differ in their genetic burden because of founder effects and the frequency of heterozygotes in potentially pathogenic genes. This is apparent in regard to the number of PTF1A variants in our study population. In comparison, the study from Turkey had a predominance of recessive variants in the WFS1 and SLC19A3 genes [20]. Therefore, it can be concluded that each consanguineous population is unique, which can allow specific insights into the genetics of conditions such as monogenic diabetes [26].

Most participants with syndromic diabetes in our cohort had relatively distinct phenotypic features suggestive of a monogenic condition, for example the participant with the INSR pathogenic variant who had features typical of Rabson–Mendenhall syndrome. At 63 days, she underwent a bilateral oophorectomy owing to the presence of bilateral cysts and the suspicion of a juvenile granulosa cell tumour. This led to hypergonadotropic hypogonadism with absent pubertal development. In our participant, surgery was carried out very early; however, individuals with INSR defects can present peripubertally with features resembling polycystic ovary syndrome or adrenache and, if genetic testing is carried out in a timely manner, invasive measures can be avoided.

In the participant with the SLC29A3 variant causing H-syndrome, the genetic diagnosis was crucial for correct management. Initial evaluation of this participant was performed because of their short stature, hepatosplenomegaly and camptodactyly, which led to examination for mucopolysaccharidosis, revealing mildly decreased levels of alpha-iduronidase. The diagnosis of mucopolysaccharidosis type I was confirmed and the participant was on expensive enzyme therapy (with no effect) until genetic diagnosis.

One participant who was suspected of having syndromic diabetes had a pathogenic variant in the GNPTG gene, confirming a diagnosis of mucolipidosis gamma. With regard to his diabetes, we did not find a causal variant. He is currently being treated with metformin, so it is uncertain if his diabetes could be clinically classified as type 2 diabetes or if there is an impact from an unrecognised gene variant.

This raises an interesting point about the diagnosis and classification of individuals with syndromic diabetes, especially in consanguineous regions. It can be argued that the usual set criteria for syndromic diabetes can be misleading in some cases [20]. In addition, in consanguineous families, individuals may have a single gene condition causing the extra-pancreatic phenotype and concurrently develop type 1/type 2 diabetes. Another possibility is the presence of multiple causative homozygous variants causing two conditions, including monogenic diabetes.

Our results showed that consanguinity was significantly associated with syndromic diabetes (p=0.0023) but not with other diabetes subtypes. Therefore, genetic testing in individuals with a suspicion of syndromic diabetes from consanguineous regions is crucial. However, setting criteria for genetic testing in such individuals is restricted by factors such as limited antibody testing. Our testing criteria (see Methods) yielded a high percentage of positive results. We observed that the presence of short stature and hepatosplenomegaly were crucial in finding monogenic diabetes variants in certain participants. A recent study also found that the presence of specific non-autoimmune extra-pancreatic features (deafness, anaemia and developmental delay) markedly improved the identification of autosomal recessive monogenic diabetes [20]. Consanguinity of parents was a helpful identifying factor as well [20]. Taken together, we suggest testing all individuals matching our selection criteria, with special emphasis on the presence of short stature, hepatosplenomegaly, deafness, anaemia and/or developmental delay.

Enabling genetic testing of people with diabetes in consanguineous populations is important in improving diagnostic criteria for monogenic diabetes [20]. In addition, studies in consanguineous populations have led to the ongoing discovery of novel genes and pathophysiological pathways [26]. One pathogenic variant among our cohort was in the ZNF808 gene (Table 1), which was identified very recently as causing neonatal diabetes in consanguineous families [27].

Our study provides a new rare insight into the influence of consanguinity on diabetes subtypes in Kurdistan, Iraq, and on the spectrum of genes that are causative of monogenic diabetes (specifically neonatal and syndromic diabetes). The main limitations of our research are the lack of a computerised system for collecting and maintaining patient data, leading to possible transcription errors and missing data for some participants. Furthermore, a lack of records on family history of diabetes and a lack of or limited access to certain laboratory tests such as routine antibody testing and evaluation of C-peptide levels were limiting factors in calculating genetic risk scores. Subtypes were mostly defined clinically by the attending physicians. Genetic testing in those with a clinical suspicion of MODY was not carried out because of a lack of informed consent and available blood samples for genetic testing in many. Furthermore, we believe that MODY prevalence was underestimated because of the reasons mentioned above, the lack of a comprehensive family history of diabetes and the lack of preventive check-ups to identify cases of hyperglycaemia.

Our single-centre study provides a unique insight into the prevalence and genetic causes of neonatal and syndromic diabetes in a highly consanguineous population. Our data confirm that, even in such populations, type 1 diabetes is the prevailing paediatric diabetes subtype. It was found that syndromic diabetes is strongly associated with consanguinity. The causative gene in monogenic diabetes was successfully elucidated in 83% of participants with neonatal diabetes and 57% of participants with syndromic diabetes. Homozygous variants made up 80% of all pathogenic variants identified. The spectrum of causative genes (PTF1A, GLIS3, WFS1, INSR, SLC29A3, ZNF808, ABCC8, INS) is markedly different from the monogenic diabetes genes seen in non-consanguineous cohorts, and also different from those seen in other consanguineous populations. In addition, we observed that phenotypic features such as short stature and hepatosplenomegaly may be important diagnostic criteria for syndromic diabetes in consanguineous populations, in whom diagnosis can be complicated due to the presence of concomitant conditions.