What are the associations between multimorbidity or individual long-term conditions and colorectal cancer risk?

Problem

Early identification of cancer and specifically colorectal cancer (CRC) is a major international focus in current primary care. CRC survival is improving and early diagnosis has major benefits. Multimorbidity (the presence of ≥2 long-term conditions (LTCs)) is another growing public health problem but the interaction between CRC and multimorbidity is complex and poorly-understood.Understanding the interplay between multimorbidity and CRC outcomes is important to enhance understanding of implications for screening or points for future intervention.This work explores the relationship between CRC and multimorbidity by examining associations between multimorbidity and particular LTCs, and CRC incidence and mortality.

Approach

Prospective population-based study using UK Biobank. Demographic characteristics (age/sex/socioeconomic status), lifestyle factors (BMI/physical activity/smoking/alcohol use), multimorbidity (measured by LTC count, of a possible 42) were recorded at baseline. Outcomes: CRC diagnosis and CRC-specific mortality, determined from linked cancer/mortality registry data.Cox regression models analysed associations between multimorbidity and CRC outcomes, adjusting for demographic/lifestyle factors. F-test variable selection modelling identified LTCs that were potential significant predictors of CRC outcomes. Predictor LTCs were tested for significance in fully adjusted Cox regression models.

Findings

The sample included 500,222 participants, aged 37–73 (mean age 56.5; 54.5% female), recruited between 2006-10. CRC was diagnosed in 3669 (0.73%) participants and 876 (0.18%) died of CRC during follow-up (median follow-up duration 7 years).CRC incidence increased with age; those aged 60-73 had higher CRC incidence (hazard ratio (HR) 5.00, 95% confidence interval (CI) 4.36-5.72) and higher CRC mortality (HR 4.69 95% CI 3.66-6.02) compared to those aged 37-49. Multivariate analysis demonstrated male sex (HR 1.43, CI 1.33-1.54), BMI >40 (HR 1.30, CI 1.02-1.66), previous smoking history (HR 1.21 CI 1.13-1.30), and higher alcohol intake (HR 1.42, CI 1.26-1.61) were significantly associated with CRC incidence. Having ≥2 LTCs did not show a statistically significant association with CRC incidence. CRC mortality was significantly more likely in males (HR 1.62, CI 1.42-1.86), BMI >40 (HR 1.54, CI 1.05-2.27), ex-smokers (HR 1.17, CI 1.03-1.34), high alcohol intake (HR 1.54, CI 1.33-2.03), no physical activity (HR 1.55, CI 1.12-2.13) and multimorbidity, HR of 1.71 (CI 1.38-2.12) in participants with ≥4 LTCs.F-test logistic regression identified hypertension as a potentially significant predictor of CRC incidence and diabetes of CRC mortality. Adjusted HR for CRC incidence in participants with hypertension compared to those without was 1.1 (CI 1.03-1.19); for CRC mortality in those with diabetes adjusted HR = 1.3 (CI 1.08-1.69) compared to those without.

Consequences

This work demonstrates higher mortality from CRC in those with multimorbidity and identifies hypertension as a potential predictor of CRC incidence and diabetes of CRC mortality.These results will guide further research on relationships between CRC risk, other LTCs and multimorbidity that should include exploring causative mechanisms and patient perspectives.