Antibioticuse and colorectal cancer: A causal association?

Antibiotics have revolutionized our ability to fight infectious diseases that are major causes of morbidity and mortality. However, the widespread use of these powerful agents has led tounintended consequences that reflect their broad effects onmicrobial community structure. Even short-term antibiotic treatment causes shifts in gut microbiota including, but not limited to, alterations in the abundance of specific taxa and adecrease in overall diversity. Clostridium difficilecolitis and vaginal candidiasis are established examples of conditionsthat originate from gut dysbiosis and opportunistic pathogen colonization induced by short-term antibiotics.Emerging data also suggest that antibiotic-induced perturbationscan persist for years after treatment and contribute to long-term dysregulation of host immune homeostasis [1]. In turn, this can potentially increase susceptibility to chronic disorders with an immune basis, including asthma,inflammatory bowel disease, and obesity.

Recent epidemiological studies have extended the association between antibiotic exposureand chronic disease to risk of colorectal adenoma and cancer (CRC).[2]In parallel, increasingevidence has demonstrateda pivotal role for the interplay between the gut microbiome and lifestyle factors in initiating and promoting CRC.[3] Several species of bacteria have been shownto be potential drivers ofcarcinogenesis through specific biological mechanisms. For example, Fusobacterium nucleatum expresses adhesins, including FadA and Fap2, which bind to tumor cells and directly promote carcinogenesis by activating oncogenic Wnt/β-catenin signaling and dysregulating immune-cell infiltration and antitumor immunity. Enterotoxigenic Bacteroides fragilisforms biofilms in the colonic mucosal membrane, promotinginflammation and tumor development. Similarly, Polyketide synthase-expressing (pks⁺)Escherichia colimay influence carcinogenesis through generation ofthe genotoxin colibactin and subsequent DNA damage.[4,5]In addition to these specific microbes, altered abundance of other microbial members andshifts in metabolic potential,including enriched amino acid and choline metabolism and depleted carbohydrate degradation, have been shown to differ in CRC cases compared with controls.[6]

Thus, the association between the gut microbiome and colorectal carcinogenesis lends critical biological plausibility to thepotential link between antibiotic exposure and development of CRC.The disruption in gut microbiota induced by antibiotics maypromote acquisition or colonization of pro-neoplastic microbiota that influenceCRCpathogenesis.Evidence from animal studies also suggests that antibiotic treatment may impact microbiome-related pathways of CRC. In mousemodels, antibiotics have been shown to decreaselevels of short-chain fatty acidsderived from microbial fermentation.[7]These fatty acids are critical mediators incolorectal carcinogenesisby virtue of their role in regulatinginflammation, immune response,cell proliferation, differentiation, and apoptosis.Moreover, antibiotics can increase intestinal permeability, whichcontributes to bacteria translocation and activation of components of the innate and adaptive immune system, thereby promoting chronic inflammation.[8]Taken together, this evidence supports a potential detrimentalrole for antibiotics on colorectal carcinogenesis. However, data also exist suggesting a potential beneficial effect of antibiotics. A recent study reported that the antibioticmetronidazole could slow the growth of Fusobacterium nucleate-positive tumor samples in patient-derived mouse xenograft models.[9] Therefore, the association between antibiotics and CRC is likely complex, reflecting multiple potential mechanisms of action of these drugs and the biological heterogeneity of the human host.

In this issue of Gut,[10]Zhang et al utilizedthe Clinical Practice Research Datalink (CPRD), a largeelectronic medical record database of general practices in the UK,to examine the association between oral antibiotic use and risk of CRC in a matched case-control study of 28,980 CRC cases and 137,077 controls.They report an association between use of antibiotics, primarily with anti-anaerobic activity, and increased risk of colon cancer, which was limited to the proximal colon. Risk was increased after even minimal antibiotic use. In contrast, antibiotic use was associated with a reduced risk of rectal cancer.In analyses according to antibiotic class, use of penicillins was associated with an increased risk of colon cancer (especiallyof the proximal colon), whereas use of tetracyclines was associated with a reduced risk of rectal cancer.This striking difference in the association with cancers of the proximal colon compared to rectum has not been previously reported.This result is consistent with our growing appreciation of the biogeographical heterogeneity in gut microbial abundance and function as well as the distinct molecular pathways underlying tumorigenesis in different regions of the intestine.

Theinvestigators should be applauded for leveraginga largedataset withprospectively collected, detailedinformation about classes of antibiotics, duration of exposure, and diagnoses of CRC according to anatomic subsite to enablea more comprehensive analysis of the antibiotic-CRC association than has been previously reported.However, the study has several limitations. First, although studies have validated the quality of CPRD data, antibiotic exposure is derived from prescription records which may not reflect actual use.Furthermore, the results on subsite-specific CRC are largely dependent on the accuracy of the coding of the tumor locations, which, to the best of our knowledge, has not been validated. In the UK, the majority of diagnoses of CRC are made in the course of secondary/specialist or hospital-based care rather than in the primary care setting, with the diagnosis subsequently communicated to general practitioners (GPs).These GPs then record and select the appropriate READ codes. Thus, this process may be susceptible to miscommunication or miscoding, which will lead to measurement error.Second, there is potential for reverse causation since these drugs are commonly prescribed for treatment of conditions that either predispose to cancer or symptoms associated with CRC prior to formal diagnosis.Although the authors tried to mitigate this possibility through sensitivity analyses limited to antibiotic use more than 10 years before CRC diagnosis, this remains a key concern for any pharmacoepidemiologic studies. Finally, the relatively modest effect estimates leave open the possibility of confounding by key risk factors for CRC, including diet, physical activity, and family history of CRC,for which data were not available. Thus, causal interpretation of the antibiotic-CRC association still requirescaution.

Further studies are warrantedto better understand the impact of antibiotic exposure on gut microbial composition and function, particularly as related tomechanisms that underliecolorectal carcinogenesis. Prospective, longitudinal cohort studies, ideallywith collection of data on antibiotic exposure and concurrent sampling of the gut microbiota prior to development of colorectal adenomas and CRC,would provide critically important data toadditionally validate these results and better infer causality.

The observations that even short-term or low-dose antibiotic treatment may perturb the gut microbiome and lead to long-term detrimental effects on CRC offer yet anotherrationale for minimizing inappropriate use of broad-spectrum antibiotics.[11]Moreover, these results lend support for the critical role of the gut microbiome in colorectal carcinogenesis, validating the human relevance of the exciting data emerging from experimental model systems. This sets the stage for another revolution associated withmanipulating gut bacteria that can address new challenges beyond the treatment of infections.Perhaps in the not too distant future, microbiome-based interventions may soon be available to prevent or treat chronic diseases such as CRC.

Abbreviations:

CRC

colorectal cancer

CPRD

Clinical Practice Research Datalink