TABLE 4.
Genotype |
N (frequency) |
OR [95% CI] |
||||
---|---|---|---|---|---|---|
GADA positives | GADA negatives | Unconditional | Conditional on HLA-DQB1 | |||
03/03 | 163 (0.14) | 63 (0.05) | 2.60 [1.86–3.65] | 3.20 [2.11–4.85] | 3.89 [2.10–7.18] | 12.24 [2.68–55.86] |
03/07 | 37 (0.03) | 16 (0.01) | 1.81 [0.96–3.40] | 2.22 [1.13–4.36] | 2.60 [1.17–5.78] | 7.80 [1.62–37.47] |
03/04 | 420 (0.35) | 441 (0.36) | 1.00 (reference) | 1.23 [0.90–1.68] | 1.00 (reference) | 3.14 [1.29–7.66] |
03/13 | 31 (0.03) | 31 (0.03) | 0.81 [0.47–1.41] | 1.00 [0.55–1.82] | 1.20 [0.51–2.84] | 3.42 [0.84–14.03] |
04/07 | 64 (0.05) | 63 (0.05) | 0.95 [0.64–1.41] | 1.17 [0.74–1.85] | 0.92 [0.62–1.37] | 2.92 [1.15–7.39] |
04/11 | 22 (0.02) | 8 (0.01) | 2.20 [0.94–5.15] | 2.70 [1.12–6.55] | 1.33 [0.50–3.58] | 2.10 [0.79–5.63] |
01/03 | 57 (0.05) | 53 (0.04) | 1.08 [0.70–1.65] | 1.33 [0.81–2.17] | 1.27 [0.66–2.43] | 1.39 [0.49–3.97] |
03/09 | 20 (0.02) | 11 (0.01) | 1.85 [0.85–4.00] | 2.27 [1.01–5.10] | 1.68 [0.64–4.40] | 1.66 [0.32–8.60] |
04/08 | 22 (0.02) | 48 (0.04) | 0.46 [0.26–0.80] | 0.56 [0.31–1.03] | 0.25 [0.09–0.75] | 1.02 [0.14–7.52] |
Rares | 96 (0.08) | 114 (0.09) | 0.77 [0.56–1.06] | 0.94 [0.63–1.41] | 0.63 [0.36–1.12] | 1.04 [0.53–2.02] |
04/04 | 108 (0.09) | 115 (0.09) | 0.81 [0.59–1.12] | 1.00 (reference) | 0.52 [0.29–0.93] | 1.00 (reference) |
03/08 | 16 (0.01) | 11 (0.01) | 1.47 [0.65–3.34] | 1.81 [0.77–4.25] | 1.23 [0.39–3.90] | 0.87 [0.20–3.71] |
01/04 | 93 (0.08) | 155 (0.13) | 0.53 [0.39–0.72] | 0.66 [0.45–0.97] | 0.39 [0.20–0.76] | 0.86 [0.32–2.31] |
04/13 | 45 (0.04) | 83 (0.07) | 0.40 [0.26–0.60] | 0.49 [0.30–0.79] | 0.42 [0.19–0.94] | 0.74 [0.31–1.75] |
Effect estimates were calculated accounting for the known covariates: sex, duration of disease, and age-at-diagnosis, whereas genotype frequencies were not. Rare genotypes (39 in total) were grouped at 0.01 frequency. HLA-DRB1*0403 and *0407 alleles have opposite effects in type 1 diabetes to the remaining HLA-DBR1*04 alleles; however, they are of such low frequency in the case subjects that an extra subtype was not warranted and so are included with the HLA-DRB1*04 allele. The apparent association of certain HLA-DRB1 genotypes following conditioning on HLA-DQB1 are likely to be attributable to effects outside of the class II region (see main text) and all have very large 95% CIs reflecting uncertainty in the model.