Background & aims: Colorectal cancer (CRC) development has been associated with increased proportions of Bacteroides fragilis and certain Streptococcus, Fusobacterium, and Peptostreptococcus species in the intestinal microbiota. We investigated associations between bacteremia from specific intestinal microbes and occurrence of CRC.

Methods: We performed a retrospective study after collecting data on 13,096 adult patients (exposed group) in Hong Kong hospitalized with bacteremia (identified by blood culture test) without a previous diagnosis of cancer from January 1, 2006 through December 31, 2015. We collected data on intestinal microbes previously associated with CRC (genera Bacteroides, Clostridium, Filifactor, Fusobacterium, Gemella, Granulicatella, Parvimonas, Peptostreptococcus, Prevotella, Solobacterium, and Streptococcus). Clinical information, including patient demographics, comorbid medical conditions, date of bacteremia, and bacterial species identified, were collected. The incidence of biopsy-proved CRC was compared between the exposed and unexposed (patients without bacteremia matched for age, sex, and comorbidities) groups.

Results: The risk of CRC was increased in patients with bacteremia from B fragilis (hazard ratio [HR] = 3.85, 95% CI = 2.62-5.64, P = 5.5 × 10^-12) or Streptococcus gallolyticus (HR = 5.73, 95% CI = 2.18-15.1, P = 4.1 × 10^-4) compared with the unexposed group. In addition, the risk of CRC was increased in patients with bacteremia from Fusobacterium nucleatum (HR = 6.89, 95% CI = 1.70-27.9, P = .007), Peptostreptococcus species (HR = 3.06, 95% CI = 1.47-6.35, P = .003), Clostridium septicum (HR = 17.1, 95% CI = 1.82-160, P = .013), Clostridium perfringens (HR = 2.29, 95% CI = 1.16-4.52, P = .017), or Gemella morbillorum (HR = 15.2, 95% CI = 1.54-150, P = .020). We observed no increased risk in patients with bacteremia caused by microbes not previously associated with colorectal neoplasms.

Conclusions: In a retrospective analysis of patients hospitalized for bacteremia, we associated later diagnosis of CRC with B fragilis and S gallolyticus and other intestinal microbes. These bacteria might have entered the bloodstream from intestinal dysbiosis and perturbed barrier function. These findings support a model in which specific members of the intestinal microbiota promote colorectal carcinogenesis. Clinicians should evaluate patients with bacteremia from these species for neoplastic lesions in the colorectum.