Summary. HLA type, time of year of diagnosis, and age at diagnosis were studied in 52 new cases of Type 1 diabetes in Seattle, Washington. Diagnosis was found to be seasonal in diabetic patients positive for DR3 (p < 0.005), with the expected marked reduction in new cases during the summer months. This seasonality was not age-related (p>0.13). Cases who were DR3-negative did not show significant seasonality of diagnosis (p> 0.5). However, when age at diagnosis was adjusted for, a seasonal effect was found in the DR3-negative group (p< 0.006), with older cases favouring a spring onset and younger cases favouring an autumn onset. Thus, DR3-positive cases showed a seasonal diagnosis pattern that did not depend on age, while DR3-negative cases showed an age-dependent seasonal pattern. These differences may reflect the predominance of different aetiological mechanisms in these two genetic groups.Key words: Seasonality, Type 1 diabetes, HLA, aetiology.The onset of Type 1 (insulin-dependent) diabetes follows a seasonal pattern, with increased incidence of new cases in the autumn and late winter, followed by a nadir during the summer [1][2][3][4][5][6]. This pattern appears to be inverted in the southern hemisphere [7] and may not be present in children who develop diabetes in their pre-school years [5,8].Certain HLA antigens are now known to be associated with an increased risk of Type 1 diabetes [9,10]. However, the primary mechanisms of B cell damage may differ in patients with different HLA phenotypes [11,12]. In this study, we have investigated whether certain HLA specificities mark subgroups of Type 1 diabetic patients with different patterns of diagnosis, taking into account a possible relationship between season and age at diagnosis.
Subjects and methodsWe studied 54 unrelated cases of newly diagnosed Type 1 diabetes, with a diagnosis between March 1976 and March 1981, who volunteered for a prospective study [13]. All patients were between 5 and 20years of age at diagnosis; the median age at diagnosis was 10.5 years. Forty-eight patients were typed for HLA-DR and 31 were typed for HLA-D. The four patients typed for HLA-D, but not DR, were all HLA-D3-negative and were assumed to be DR3-negative * Present address: City Hospital, Nottingham NG5 1PD, UK also. This assumption was felt to be reasonable since among 18 D3-negative patients who were also DR typed, only one was DR3-positive. Further, no set of reclassifications for those four patients and subsequent repeat analysis would materially alter our conclusions. Serological typing for DR was performed by standard methods [14,15]. Typing for HLA-D was performed using HLA-D homozygous typing cells as previously described [16].Statistical analysis of this data presents special problems, since time of year should be treated as a circular variable (December 31 being adjacent to January 1). For testing for seasonality, versus a null hypothesis of uniformity, the non-parametric test developed by Hodges and Ajne was applied [17]. The seasonal patterns for DR3-positive ve...