Rheumatoid arthritis and cancer risk in the Million Women Study

Abstract

Background: Most previous studies of rheumatoid arthritis (RA) and cancer risk have lacked information on potential confounding factors. We investigated RA-associated cancer risks in a large cohort of women in the UK, taking account of shared risk factors.

Methods: In 1996-2001, women aged 50-64, who were invited for routine breast screening at 66 National Health Service (NHS) screening centres in England and Scotland, were also invited to take part in the Million Women Study. Participants provided information on sociodemographic, lifestyle and health-related factors, including RA, and were followed up for cancers and deaths. Cox regression yielded RA-associated hazard ratios (HRs) of 20 cancers, adjusted for 10 characteristics including smoking status and adiposity.

Results: Around 1.3 million women (half of those invited) were recruited into the study. In minimally adjusted analyses, RA was associated with the risk of 13 of the 20 cancers. After additional adjustment for lifestyle factors, many of these associations were attenuated but there remained robust evidence of RA-associated increases in the risk of lung (HR 1.21, 95% confidence interval 1.15-1.26), lymphoid (1.25, 1.18-1.33), myeloid (1.12, 1.01-1.25), cervical (1.39, 1.11-1.75) and oropharyngeal (1.40, 1.21-1.61) cancers, and decreases in the risk of endometrial (0.84, 0.77-0.91) and colorectal (0.82, 0.77-0.87) cancers.

Conclusions: After taking account of shared risk factors, RA is positively associated with lung and certain blood and infection-related cancers, and inversely associated with colorectal cancer. These findings are consistent with existing hypotheses around immune response, susceptibility to infections, and chronic inflammation. The inverse association observed for endometrial cancer merits further investigation.

Keywords: Epidemiology; cancer; rheumatoid arthritis; smoking.

Figure 1.

Hazard ratio of cancer in individuals with rheumatoid arthritis (RA) compared with others in the Million Women Study. Numbers and forest plots show hazard ratios (HR) and 95% confidence intervals. Minimal adjustment includes year of recruitment; year of birth; 10 regions of residence; age. Full adjustment includes year of recruitment; year of birth; 10 regions of residence; age; social deprivation; education; joint alcohol consumption and smoking status; body mass index; strenuous exercise; age at menarche; age at first birth and parity; duration of oral contraceptive use; menopausal status and age; use of menopausal hormone therapy. Number of asterisks (*) indicates level of statistical significance (*<0.05; **<0.005; ***<0.0005; ****<0.00005; *****<0.000005)

Figure 2.

Hazard ratio of haematological malignancies by histological subtypes in individuals with rheumatoid arthritis (RA) compared with others in the Million Women Study. Numbers and forest plots show hazard ratios (HR) and 95% confidence intervals. Minimal adjustment includes year of recruitment; year of birth; 10 regions of residence; age. Full adjustment includes year of recruitment; year of birth; 10 regions of residence; age; social deprivation; education; joint alcohol consumption and smoking status; body mass index; strenuous exercise; age at menarche; age at first birth and parity; duration of oral contraceptive use; menopausal status and age; use of menopausal hormone therapy. Number of asterisks (*) indicates level of statistical significance (*<0.05; **<0.005; ***<0.0005; ****<0.00005; *****<0.000005). CLL/SLL, chronic lymphocytic leukaemia (CLL) and small lymphocytic lymphoma; MPD/MDD, myeloproliferative or myelodysplastic disorders