Association between antibiotic usage and colorectal cancer development: a territory-wide study

Abstract

Introduction: Recent studies suggested that antibiotics may modulate colorectal cancer (CRC) risk due to gut dysbiosis. We aimed to investigate the specific and temporal effects of different antibiotics on CRC development in older subjects. Methods: This is a retrospective cohort study using a territory-wide electronic healthcare database in Hong Kong including patients aged ≥60 years who had undergone colonoscopy between 2005 and 2013. Exclusion criteria included inflammatory bowel disease, prior colectomy, prior CRC and CRC detected within 6 months of index colonoscopy. The primary outcomes were CRC diagnosed >6 months after index colonoscopy. The adjusted hazard ratio (aHR) of CRC with antibiotics (defined as any antibiotic use up to 5 years before index colonoscopy) was derived by multivariable Cox proportional hazards model with adjustment for other covariates (including patient’s demographics, history of colonic polyps/polypectomy, concurrent drug usage [aspirin, NSAIDs, COX-2 inhibitors and statins] and endoscopy centre’s performance [polypectomy rate and colonoscopy volume]). Eleven classes of antibiotics were included: penicillins, cephalosporins, macrolides, carbapenems, quinolones, tetracyclines, aminoglycosides, nitroimidazoles, glycopeptides, sulpha/trimethoprim, and others (clindamycin, nitrofurantoin, linezolid, rifampicin, rifaximin, and daptomycin). Stratified analysis was performed according to cancer location and nature of antibiotics. Results: Among 97 162 eligible patients (52.3% male), 1026 (1.1%) patients developed CRC after colonoscopy (proximal: 171 [16.7%], distal: 254 [24.8%], and rectal: 601 [58.6%]). The median age of cancer diagnosis was 79.1 years (interquartile range=72.9-85.0). There were 58 704 (60.4%) antibiotic users. Antibiotic use was associated with lower rectal cancer risk (adjusted hazard ratio [aHR]=0.64; 95% confidence interval [CI]=0.54-0.76), but higher proximal colon cancer risk (aHR=1.63; 95% CI=1.15-2.32). Th effect was neutral on distal colon cancer development (aHR=0.99; 95% CI=0.76-1.30). These effects varied according to anti-anaerobic/anti-aerobic activity, broad-/narrow-spectrum, and route of administration of antibiotics. Notably, penicillins were associated with a lower CRC risk (aHR=0.83; 95% CI=0.73-0.96), while aminoglycosides were associated with a higher risk (aHR=1.53; 95% CI=1.05-2.25). Conclusion: Antibiotic use was associated with divergent effects on CRC development in older patients, with respect to cancer location, class of antibiotics, and route of administration.