Hormone replacement therapy in relation to the risk of colorectal cancer in women by BMI: a multicentre study with propensity score matching

Abstract

Epidemiological evidence about hormone replacement therapy and colorectal carcinogenesis by demographic and clinical traits remains unclear. We aimed to assess this postulated association in a large multicentre study and further explore the modification effect by BMI and others. We retrospectively collected records of women diagnosed with colorectal cancer (CRC) at the age of 50years and older during 2014-2017 and their HRT dispensing prior to CRC diagnosis in three tertiary hospitals in China. CRC cases were matched with controls at a ratio of 1:3 using nearest neighbour propensity scores matching to better control for the remaining imbalance between groups, which generated a total of 824 cases with 2472 controls. Our study confirmed the inversed association between colorectal cancer risk and hormone replacement therapy (OR, 0.62; 95% CI, 0.54-0.75), which was more prominent among women having multiple HRT dispenses (OR, 0.60; 95% CI, 0.52-0.76). Furthermore, significant associations were consistently observed for the short-term (OR, 0.69; 95% CI, 0.57-0.88), middle-term (OR, 0.51; 95% CI, 0.41-0.66), and long-term HRT users (OR, 0.70; 95% CI, 0.43-0.90). Estrogen-related regimen reduced CRC risk more than progestogen-only. We, for the first time, found that the modifying effect of BMI on HRT use and CRC risk was in different ways when BMI was categorized by a medium level of 27. Our findings mainly suggest that there might be a different mechanism for the reversed association between HRT and colorectal tumorigenesis by BMI level, providing thoughts on clinical treatment of CRC.