Sex Differences in Age of Diagnosis, HLA Genotype, and Autoantibody Profile in Children With Type 1 Diabetes

In populations with a high incidence of type 1 diabetes (T1D), more men than women are diagnosed after puberty (1,2). Most children with T1D have one or multiple islet cell autoantibodies prior to diagnosis, such as antibodies against GAD antibody (GADA), insulin autoantibody (IAA), insulinoma-associated protein 2 (IA-2A), and variations of the zinc transporter 8 autoantibody (ZnT8A) (3,4). Studies have shown that in children at high risk of T1D, IAA often occurs first and is more common in very young children with T1D (1,5,6). Some studies have found GADA to be more common in girls (1,7), and IAA, IA-2A, and ZnT8A more common in boys (1,6). A longitudinal study following individuals at risk for T1D showed that girls with multiple autoantibodies had a decreased risk of progression to T1D with increasing age compared with boys (8). Thus, several studies indicate that sex and age may be important in the link between autoantibodies and T1D, but no study has examined potential sex differences across the entire pediatric age range for the four major autoantibodies in children with T1D.

The following haplotypes in the HLA area carry the highest genetic risk for T1D: DQA1*03-B1*0302 (DQ8) and DQA1*0501-B1*0201 (DQ2) (9). A recent study showed that DQ2 is more common in boys and DQ8 more common in girls with T1D (1), while other studies report no sex differences in HLA types (10,11). Thus, the relation between sex and HLA risk in children with T1D remains unclear.

The aim of this study was to examine the relation between sex and age at T1D diagnosis, autoantibodies, and HLA risk in a population-based, nationwide sample of 3,645 Swedish children with newly diagnosed T1D.

The current study used data from the nationwide Better Diabetes Diagnosis (BDD) study, which has been ongoing in Sweden since 2005 (12). Between May 2005 and December 2010, 4,601 children in Sweden were diagnosed with some type of diabetes, of which 3,977 (87%) were included in BDD. Until December 2010, all BDD participants were analyzed for GADA, IA-2A, ZnT8RA, ZnT8WA, ZnT8QA, and IAA. Thereafter, participants were only analyzed for IAA or zinc if GAD and IA-2A were negative (12); thus, we only included participants who were enrolled in BDD prior to 2011. In the current study, data on 3,645 children with a confirmed T1D diagnosis according to the American Diabetes Association criteria (13) were analyzed. Figure 1 shows the selection of participants.

Autoantibodies were analyzed using radioligand binding assays using methods that have been described in detail elsewhere (12). The cutoff levels used in the current study are found in Supplementary Methods 1. HLA typing was performed on dried blood spots with PCR using a DELFIA hybridization assay (PerkinElmer, Boston, MA) (12). Sequence-specific oligonucleotide probes of HLA-DQB1 were used to define the presence of HLA-DQB1*02, 03:02, 03:01, 06:02, 06:03, and 06:04 alleles and another set of HLA-DQA1 probes to define DQA1*02:01, 03, and 05 alleles. The HLA genotypes were classified into different risk groups (high, moderate, and low risk) using normative data from the general population (14); for more information, see the Supplementary Methods 2. Ethical approval for BDD was granted from the Regional Ethics Board at Karolinska Institute in Stockholm, Sweden (Dnr 2009/1684-32, Dnr 2006/1082-32, Dnr 04-826/1, and Dnr 2009/1684-32). Patients and caregivers provided written informed consent/assent.

Statistical analyses were performed with SPSS 25 and R Studio 1.1.447 software. To correct for multiple comparisons, we used an α level of 0.01 as an indicator of statistical significance. For analyses in separate age-groups, where statistical power was lower, we used an α level of 0.05. As boys have been shown to be older at T1D diagnosis, we controlled for age of diagnosis in all models where age of onset was not the dependent variable. Sex differences were examined using the following statistical models: independent-samples t tests (age at diagnosis), one-sample binomial tests (proportion boys/girls in the separate age groups), logistic regression (positivity for different autoantibodies), and ordinal regression (HLA risk). The following age-groups were used: 0.5–2 years, 3–5 years, 6–8 years, 9–11 years, 12–14 years, and 15–17 years.

Table 1 presents demographic and clinical characteristics of the sample. Girls were significantly younger at diagnosis (9.53 vs. 10.23 years; t_3,513 = 4.81, P < 0.001). Boys were more common in the following age-groups: 3–5 years (55.0% boys; P = 0.017), 12–14 years (62.0% boys; P < 0.001), and 15–17 years (64.0% boys; P < 0.001).

Girls were more likely than boys to be autoantibody-positive (94.7% vs. 92.0%; age-adjusted odds ratio [OR] 1.529 [95% CI 1.164, 2.008], P = 0.002). In age-specific analyses (see Fig. 2A), sex differences emerged in 3- to 5-year-olds (χ² = 6.346, P = 0.012), 12- to –14-year-olds (χ² = 6.129, P = 0.013), and 15–17-year-olds (χ² = 5.469, P = 0.019). Age at diagnosis was not a significant predictor of autoantibody positivity/negativity (P = 0.189).

Girls were positive for a larger number of autoantibodies than boys (2.36 vs. 2.21; age-adjusted OR 1.273 [95% CI 1.131, 1.432], P < 0.001). When those with no autoantibodies were excluded, girls were still positive for a larger number of autoantibodies than boys (age-adjusted OR 1.201 [95% CI 1.062, 1.359], P = 0.003). Age-specific differences showed significant sex differences in the number of autoantibodies in 3- to- 5-year-olds (P = 0.031) and 6- to 8-year-olds (P = 0.004) (see Fig. 2B).

Girls were more likely than boys to be positive for GADA (64.9% vs. 49.0%; age-adjusted OR 2.041 [95% CI 1.783, 2.342], P < 0.001), and boys were marginally more likely to be positive for IAA (33.8% vs. 32.3%; age-adjusted OR 1.197 [1.034, 1.387], P = 0.016). Sex differences were significant for GADA across all age-groups (P < 0.001) (see Supplementary Fig. 1A), while differences for IAA were only significant among 15- to 17-year-olds (P = 0.005) (see Supplementary Fig. 1B). Older age at diagnosis was associated with a higher risk of GADA (sex-adjusted OR 1.068 [95% CI 1.051, 1.084], P < 0.001) and ZnT8A (sex-adjusted OR 1.043 [1.027, 1.059], P < 0.001) and a lower risk of IAA (sex-adjusted OR 0.854 [0.840, 0.869], P < 0.001). Age at diagnosis was not associated with IA-2A (P = 0.959). The association between age at diagnosis and the presence/absence of different antibodies was similar between sexes.

Overall, sex was not associated with HLA risk (P = 0.795), but in the separate age-groups, sex differences were present among 3- to 5-year-olds (OR 0.647 [95% CI 0.427, 0.979], P = 0.040) and 6- to 8-year-olds (OR 1.442 [1.017, 2.042], P = 0.040), with boys being more likely to have high HLA risk among 3- to 5-year-olds and girls among 6- to 8-year-olds. Older age at diagnosis was associated with a higher probability of being classified as having a low HLA risk (sex-adjusted OR 1.058 [95% CI 1.033, 1.084], P < 0.001). Sex or age at diagnosis did not interact with HLA risk in relation to positivity/negativity for any of the autoantibodies or being positive for at least one autoantibody.

The current study identified several sex differences in children with newly diagnosed T1D and adds to the current literature by showing that some of these differences are age dependent. Girls were younger at T1D diagnosis and more frequently positive for GADA, which is in line with other studies (1,7,15). For the first time, we show a female preponderance in GADA irrespective of age. Girls were also more likely to be autoantibody-positive and positive for multiple autoantibodies, with the latter being in line with some (16–18), but not all, prior studies (1). Boys in the age of 15–18 years were more often positive for IAA, which has been demonstrated before (19). Differences for the number of autoantibodies were found in the younger age-groups (3–8 years), while being autoantibody-positive was more common in 3- to 5-year-old girls and in girls >12. More boys than girls were diagnosed with T1D, but significant sex differences were only found in 3- to 5-year-olds and in those >12.

Proportions of boys/girls with T1D in this study closely match a comparable Finnish study, the only prior study examining sex differences in pediatric T1D with a similar sample size (1). Our findings confirm that the male predominance in T1D is most pronounced during adolescence (1), but we also found a greater proportion of boys in the 3- to 5-year-old group. Interestingly, the age-groups with a male predominance coincided with the age-groups where boys were more likely to be negative for all autoantibodies. This may reflect that puberty, with increased insulin demand, onsets earlier in girls than boys, with a potential equalization of risks between sexes and the number of individuals lacking autoantibodies during this age period. Increased risk of T1D during periods of distinct growth has been shown by others (20). For HLA, we only found sex differences in the younger age-groups.

In summary, this study identified sex differences for age at diagnosis and autoantibodies in children with newly diagnosed T1D. Overall, our findings imply that the disease mechanisms leading to T1D may influence the immune system differently in girls and boys.

Funding. BDD is funded by Barndiabetesfonden (the Swedish Child Diabetes Foundation).

The study sponsors were not involved in the design of the study, the collection, analysis, and interpretation of data, writing the report, or the decision to submit the report for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.T. wrote the first draft of the manuscript. J.T. and M.C. performed statistical analyses. J.T., M.C., E.H., Q.B., and A.C. were responsible for the study concept. M.C. designed the statistical analyses. G.F., H.E.L., J.L., U.S., C.M., M.P., and A.C. were responsible for acquisition of data. All authors discussed and interpreted the results. All authors edited and reviewed the manuscript and approved the final version. A.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.