Our study confirmed the low prevalence (0.7%) of diagnosed symptomatic CD at the time of clinical diagnosis but document by screening an increasing prevalence of silent CD during a 5-yr follow-up to reach an overall prevalence of 10%. We suggest that children with T1DM should be screened for CD at the onset of T1DM and annually for a minimum of at least 2 yr. HLA genotypes among T1DM patients developing CD were not different from those among patients with T1DM alone.

Aims/hypothesis High birthweight and increased childhood growth are risk factors for type 1 diabetes. Relative birthweight is associated with HLA genotypes that confer a high risk of diabetes. Our aims were to test whether young children prior to clinical onset of type 1 diabetes have increased: (1) birthweight or birth length standard deviation scores (SDS); (2) height development SDS; or (3) BMI SDS during first 18 months of life and whether these parameters are related to HLA genotypes or mid-parental height (MPH). Methods Birthweight, birth length, weight and height were obtained from 58 type 1 diabetes children and 155 controls matched for HLA or not in the Diabetes Prediction in Skåne study. Results Birth length SDS corrected for MPH was increased in children developing diabetes compared with all (p< 0.048) and with non-HLA-(p<0.050) but not with HLAmatched controls. Children developing diabetes had increased height gain at 0 to 18 months of age (p<0.005). Diabetic children were significantly taller from 6 to 18 months of age when correcting for MPH compared with non-HLA-matched as well as HLA-matched controls, but BMI was not increased. Conclusions/interpretation Birth length SDS was associated with diabetes risk HLA. When corrected for MPH, children developing diabetes were taller at birth than non-HLA-but not taller than HLA-matched controls. Diabetic children had increased MPH-corrected height up to 18 months of age compared with both HLA-and non-HLA-matched controls. High-risk HLA affects prenatal growth, but other factors may explain the increased postnatal linear growth in children developing diabetes.

The mechanisms by which gestational infections, blood incompatibility, birth weight, mother's age and other prenatal or neonatal events increase the risk for type 1 diabetes are not understood. Studies so far have been retrospective, and there is a lack of population-based prospective studies. The possibility of identifying children at type 1 diabetes risk among first-degree relatives has resulted in prospective studies aimed at identifying postnatal events associated with the appearance of autoantibody markers for type 1 diabetes and a possible later onset of diabetes. However, the majority (85%) of new onset type 1 diabetes children do not have a first-degree relative with the disease. Population-based studies are therefore designed to prospectively analyse pregnant mothers and their offspring. One such study is DiPiS (Diabetes Prediction in Skåne), which is examining a total of about 10,000 pregnancies expected every year in the Skåne (Scania) region of Sweden that has 1.1 million inhabitants. Blood samples from all mothers in this region are obtained during pregnancy and at the time of delivery. Cord blood is analysed for HLA high-risk alleles and for autoantibodies against the 65 kD isoform of glutamic acid decarboxylase (GADA), the protein tyrosine phosphatase-related IA-2 antigen (IA-2A) and insulin (IAA) as a measure of prenatal autoimmune exposure. Identifying high-risk children by genetic, autoimmune and gestational risk factors followed by prospective analyses will make it possible to test the hypothesis that gestational events may trigger beta cell autoimmunity as a prerequisite for childhood type 1 diabetes.

The aim of this study was to compare the frequency of human leukocyte antigen (HLA) genotypes in 1-18-year-old patients with type 1 diabetes newly diagnosed in 1986-1987 (n = 430), 1996-2000 (n = 342) and in 2003-2005 (n = 171). We tested the hypothesis that the HLA DQ genotype distribution changes over time. Swedish type 1 diabetes patients and controls were typed for HLA using polymerase chain reaction amplification and allele specific probes for DQ A1* and B1* alleles. The most common type 1 diabetes HLA DQA1*-B1*genotype 0501-0201/0301-0302 was 36% (153/430) in 1986-1987 and 37% (127/342) in 1996-2000, but decreased to 19% (33/171) in 2003-2005 (P \ 0.0001). The 0501-0201/0501-0201 genotype increased from 1% in 1986-1987 to 7% in 1996-2000 (P = 0.0047) and to 5% in 2003-2005 (P > 0.05). This study in 1-18-year-old Swedish type 1 diabetes patients supports the notion that there is a temporal change in HLA risk.

No abstract