Gene–environment interaction between HLA-SE (none, single and double copies) and heavy smoking in all rheumatoid arthritis (RA) groups and in groups stratified by serological status in the Nurses’ Health Studies
| HLA-SE | Pack-years smoking | All RA | Seropositive RA | Seronegative RA | ||||
| Cases/controls | OR* (95% CI) | OR† (95% CI) | Cases/controls | OR‡ (95% CI) | Cases/controls | OR§ (95% CI) | ||
| 0 | ⩽10 | 119/161 | 1.00 (reference) | 1.00 (reference) | 64/161 | 1.00 (reference) | 55/161 | 1.00 (reference) |
| 1 | ⩽10 | 86/94 | 1.26 (0.85 to 1.88) | 1.33 (0.88 to 2.00) | 52/94 | 1.36 (0.87 to 2.15) | 34/94 | 1.08 (0.65 to 1.79) |
| 2 | ⩽10 | 32/19 | 2.61 (1.37 to 4.97) | 2.96 (1.51 to 5.80) | 25/19 | 3.56 (1.81 to 7.02) | 7/19 | 1.21 (0.47 to 3.08) |
| 0 | >10 | 86/103 | 1.13 (0.75 to 1.69) | 1.07 (0.70 to 1.62) | 44/103 | 1.08 (0.68 to 1.73) | 42/103 | 1.16 (0.72 to 1.89) |
| 1 | >10 | 78/46 | 2.26 (1.42 to 3.62) | 2.24 (1.38 to 3.62) | 57/46 | 3.14 (1.91 to 5.17) | 21/46 | 1.25 (0.67 to 2.32) |
| 2 | >10 | 24/6 | 5.43 (2.14 to 13.77) | 6.60 (2.49 to 17.46) | 18/6 | 7.47 (2.77 to 20.11) | 6/6 | 2.97 (0.90 to 9.79) |
| Additive: | ||||||||
| AP1 (95% CI) | 0.39 (0.05 to 0.72) | 0.38 (0.03 to 0.73) | 0.54 (0.26 to 0.82) | 0.01 (−0.75 to 0.77) | ||||
| p For interaction | 0.02 | 0.04 | <0.001 | 0.98 | ||||
| RERI | 0.88 (−0.11 to 1.86) | 0.84 (−0.16 to 1.85) | 1.70 (0.29 to 3.11) | 0.01 (−0.94 to 0.97) | ||||
| Synergy index | 3.26 (0.52 to 20.25) | 3.15 (0.48 to 20.54) | 4.81 (0.71 to 32.54) | 1.06 (0.02 to 54.35) | ||||
| AP2 (95% CI) | 0.50 (−0.03 to 1.02) | 0.54 (0.05 to 1.04) | 0.51 (−0.03 to 1.06) | 0.54 (−0.12 to 1.20) | ||||
| p For interaction | 0.06 | 0.03 | 0.07 | 0.11 | ||||
| RERI | 2.70 (−2.34 to 7.73) | 3.57 (−2.75 to 9.90) | 3.83 (−3.59 to 11.24) | 1.60 (−2.00 to 5.20) | ||||
| Synergy index | 2.56 (0.63 to 10.40) | 2.77 (0.69 to 11.06) | 2.45 (0.62 to 9.72) | 5.32 (0.11 to 252.71) | ||||
| Multiplicative: | ||||||||
| p For interaction§ | 0.13 | 0.20 | 0.07 | 0.52 | ||||
*Conditional logistic regression. †Conditional logistic regression, adjusting for parity, breast feeding and menstrual irregularity, and age at menarche, menopausal status and postmenopausal hormone use at diagnosis. ‡Unconditional logistic regression adjusting for year of birth, parity, breast feeding and menstrual irregularity, and age at menarche, menopausal status and postmenopausal hormone use at blood draw and at diagnosis. §p Value calculated from HLA-SE×pack-years interaction term.
AP, attributable proportion due to interaction, an index of additive interactions between binary smoking variable and binary HLA-SE genotype; RERI, relative excess risk due to interaction.