Association of Streptococcus bovis Bacteremia With Colonic Neoplasia and Extracolonic Malignancy FREE

An association between Streptococcus bovis bacteremia and gastrointestinal disease, particularly colonic neoplasia, has been well documented in the literature. McCoy and Mason¹ published a case report of "enterococcal" endocarditis associated with a carcinoma of the cecum in 1951. It is likely that the organism in this case was truly S bovis.² Since that time, a connection between S bovis septicemia and colonic neoplasia has been confirmed by several other case reports as well as reviews of case series and a case-control study. On the basis of these data, several authors have recommended that all patients with S bovis bacteremia undergo complete evaluation of the colon.^3- 12

The reason for this association has not been elucidated. Among the theories are that colonic neoplasia specifically allows for the overgrowth or translocation of this organism³ and that S bovis is in fact causative of the neoplasia itself.^3,13 One study demonstrated a significantly higher rate of fecal carriage of S bovis in patients with colorectal cancer as opposed to controls.³ Subsequent studies, however, have failed to confirm this finding.^14- 16 In a laboratory investigation, antigens from the cell wall of S bovis promoted neoplastic progression in chemically induced tumors in a rat model.¹³

In contrast to the literature regarding an association with colonic neoplasia, less is written about associations of S bovis with other gastrointestinal disease or with other cancers. Some authors have noted that a subset of patients with S bovis bacteremia have cirrhosis or liver dysfunction.^{2,6,8- 10,12,17- 19} One group has postulated a triad of S bovis bacteremia, liver disease, and colonic pathology whereby the liver disease may account for the increased fecal carriage, entry to the portal venous system, or passage from the portal to systemic circulation of S bovis.⁹

The purpose of this study was to review the experience with S bovis bacteremia in our community with respect to associated gastrointestinal pathology, the use of gastrointestinal endoscopy, and associated infections, as well as associated gastrointestinal and extraintestinal malignancies.

Cases of S bovis bacteremia were obtained from a review of all available computerized bacteriology records at 2 hospitals (Yale–New Haven Hospital and the Hospital of Saint Raphael, New Haven, Conn) for the period between January 1986 and January 1999. Hospital records, including pathologic reports, were reviewed for all patients identified as having S bovis bacteremia. Clinical parameters including age, sex, the presence of anemia, the presence of signs or symptoms of gastrointestinal pathology, the use of upper or lower gastrointestinal endoscopy, the presence of infections at extravascular sites at the time of documented bacteremia, and other associated medical conditions such as cancer were recorded. All work was governed by human investigation committee protocols 10491 (Yale–New Haven Hospital) and 1123 (Hospital of Saint Raphael).

The 2-sided t test was used to examine for differences among continuous variables, and the 2-sided Fisher exact test was used to examine for differences among categorical variables.

Literature review was conducted by MEDLINE searches using combinations of the term "Streptococcus bovis" with "colon," "intestine," "gastrointestinal," "liver," "cancer," "neoplasia," and "malignancy." Abstracts and/or full text of all articles written in English were reviewed. Particular attention was paid to series of more than 10 patients. The references of the reviewed articles were also used to identify pertinent literature.

There were 45 patients with S bovis bacteremia documented at 2 hospitals in 1 city during a 12-year period. Of these, 4 were children (newborn, 6 weeks, 6 weeks, and 3 years), and 41 were adults. The median age of the patients was 74 years (range, newborn to 90 years). Thirteen of the patients were female and 32 were male. The clinical characteristics of the patients with S bovis bacteremia are presented in Table 1. There were no significant differences in the demographic and clinical characteristics of the patients seen at the 2 institutions except that the mean age was lower at the tertiary referral hospital (63.0 vs 79.7 years; P = .01), reflecting the fact that all of the pediatric patients were seen at this hospital; the mean age of the adult patients was not statistically different (71.3 vs 79.9 years; P = .10).

Eighteen patients underwent upper or lower gastrointestinal endoscopy after the diagnosis of S bovis bacteremia (1 upper, 17 lower). One colonoscopy was performed 2 years after the diagnosis of S bovis bacteremia; the other examinations were performed within 1 month. In all but 1 of the patients examined, gastrointestinal pathology was diagnosed. The most common pathology identified was colonic polyps (14 patients, 13 with adenomatous polyps and 1 with a hyperplastic polyp). One patient had diverticulosis in addition to an adenomatous polyp. One patient had ischemic colitis. The patient who underwent upper endoscopy had an invasive duodenal adenocarcinoma.

Gastrointestinal endoscopy was not performed subsequent to a diagnosis of S bovis bacteremia in 27 patients (23 adults). Interestingly, 8 patients with signs and symptoms of gastrointestinal pathology or anemia (7 of whom were adults) did not undergo gastrointestinal endoscopy. Three of these patients, all adults, had anemia only. Six of the 7 adults with signs and symptoms of gastrointestinal pathology or anemia in whom endoscopy was not performed had either a known malignancy or liver failure. The additional adult, who had a history of bright red blood per rectum and had not previously had colonoscopy, was 85 years old and had diabetes mellitus, coronary artery disease, severe congestive heart failure, and chronic renal insufficiency.

In total, of the 23 adults who did not undergo endoscopy subsequent to a diagnosis of S bovis bacteremia, 19 were being treated for a known malignancy (11 patients), had liver failure (5), had an extravascular infection thought to account for the S bovis bacteremia (6), or had a combination of these factors (2 with malignancy and infection, 1 with malignancy and liver failure, and 1 with liver failure and infection). The details of the concurrent malignancies are listed in Table 2. Four additional adults who did not undergo endoscopy had no signs or symptoms of gastrointestinal pathology or anemia. One of these patients died of S bovis endocarditis; the autopsy revealed adenomatous polyps in the colon.

There were 23 clinically significant infections associated with the diagnosis of S bovis bacteremia as seen in Table 3. The most common infection was endocarditis, which was present in 12 patients. Of the 12 patients with endocarditis, 6 had prosthetic valves and 6 had native valves affected. One case of S bovis bacteremia was associated with septic shock, and another occurred in the setting of a neutropenic fever in a patient treated for endometrial cancer.

In 10 patients, S bovis bacteremia was felt to be associated with an extravascular site of infection. Cholangitis was present in 3 of these patients, 1 of whom had a biliary stent secondary to an obstructing pancreatic adenocarcinoma. Two cases of S bovis bacteremia were associated with gastroenteritis in infants. Streptococcus bovis bacteremia was also associated with 2 cases of chorioamnionitis. In one case, the mother developed S bovis bacteremia; in another case, it was the infant. One patient with S bovis endocarditis had a septic arthritis documented with a joint aspirate that was cultured for S bovis. Streptococcus bovis bacteremia was diagnosed in a patient with liver failure hospitalized for pneumonia. Streptococcus bovis bacteremia was also diagnosed in a patient hospitalized for an intra-abdominal abscess after resection of a pancreatic adenocarcinoma.

Neoplasia was associated with S bovis bacteremia in 26 (58%) of the 45 patients. The types of neoplasia present in patients with S bovis bacteremia are listed in Table 2. Colorectal adenomas or adenocarcinomas were diagnosed in 16 patients with S bovis bacteremia, and an additional patient had a large ulcerating ascending colon mass, but colonoscopic biopsies revealed only necrotic material. Colonic adenomas were diagnosed within 1 month of S bovis bacteremia in 13 patients, 12 by colonoscopy and 1 by autopsy. An additional patient underwent colonoscopy to reveal an adenomatous polyp 2 years after S bovis bacteremia was diagnosed.

Malignancy was associated with S bovis bacteremia in 13 patients (29%), excluding the patient with the undiagnosed colon mass. Three patients had colorectal adenocarcinomas, 1 found on colonoscopy after the diagnosis of S bovis bacteremia and 2 other recurrent lesions, 1 locally recurrent, and 1 with liver and lung metastases. One patient had a duodenal adenocarcinoma, and 1 had gallbladder adenocarcinoma. Three patients had pancreatic adenocarcinoma, 2 having undergone resection. One developed S bovis bacteremia in the setting of an intra-abdominal abscess after resection of pancreatic cancer. The patient with the unresected pancreatic cancer who developed S bovis bacteremia did so in the setting of cholangitis. Three patients developed S bovis bacteremia with solid-organ malignancies outside the gastrointestinal system. One patient had advanced lung cancer, 1 had ovarian cancer, and 1 had endometrial cancer. The patient with endometrial cancer developed S bovis bacteremia in the setting of a neutropenic fever during adjuvant therapy. Two patients developed S bovis bacteremia in the setting of a non–solid-organ malignancy; one had chronic myelogenous leukemia with blast crisis, and the other had chronic lymphocytic leukemia with end-stage liver disease.

It is now well established that S bovis bacteremia is associated with gastrointestinal disease, particularly colonic neoplasia. Although the nature of this relationship remains unknown, it is accepted that clinicians should screen patients for occult colon cancer whenever possible. The association between S bovis bacteremia and liver dysfunction, particularly cirrhosis, has been described,^9,17 although this relationship is not as well validated and clinical implications of this association are not readily apparent. An association between S bovis bacteremia and extraintestinal malignancy has not been well established. This study, however, indicates that malignancy may be a common feature of S bovis bacteremia, and clinicians should be alert to cancer both inside and outside the gastrointestinal system.

The first description of colorectal carcinoma occurring in association with S bovis bacteremia is often attributed to a 1951 case report.¹ In this report, a woman who had recently undergone resection of a sigmoid adenocarcinoma developed endocarditis with an organism identified as Streptococcus fecalis. S bovis was often mislabeled as an enterococcus, such as Enterococcus fecalis (formerly S fecalis), before identification of this organism based on standardized biochemical and physiological methods was described.²⁰ The sensitivity of the organism to penicillin in this case report is much more typical of S bovis than E fecalis.^2,21

Several subsequent case reports also noted associations between S bovis bacteremia and colorectal cancer. Beginning in the early 1970s, larger series of patients with S bovis bacteremia were reported. Table 4 presents all series of patients (n>10) with S bovis bacteremia who were examined for associations with gastrointestinal or liver pathology and described in the English-language literature. In these series, colonic neoplasia was present in 6% to 71% of patients.

Notably, Klein et al²⁵ prospectively studied 29 patients with S bovis bacteremia. They recommended gastrointestinal evaluation for all their patients, although this was completed in only 15 patients owing to patient refusal, inability to tolerate examinations, or death. Colonic neoplasia was found in 12 patients (11 of whom underwent complete gastrointestinal evaluation), and colorectal cancer was found in 2 patients (both of whom underwent the recommended evaluation).²⁵

Two studies have compared the incidence of gastrointestinal pathology in patients with endocarditis from S bovis with that in patients with endocarditis due to other microorganisms. In the study by Leport et al,⁷ despite similar demographic characteristics and similar use of colonic studies among groups, polyps, carcinoma in situ, and invasive colon cancer were significantly more frequent in the cases of S bovis endocarditis (12, 3, and 6 of 34, respectively) vs endocarditis from other organisms (3, 0, and 1 of 43).⁷ Similarly, in the study by Pergola et al,²⁶ gastrointestinal lesions were found in 22 of 40 patients with S bovis endocarditis compared with 7 of 166 patients with endocarditis from other organisms. The gastrointestinal lesions were predominantly colonic polyps (14 patients) or cancers (4 patients) in the patients with S bovis endocarditis.²⁶

Hoen et al¹⁰ performed a case-control study comparing patients with S bovis endocarditis who underwent colonoscopy with sex- and age-matched unaffected patients who had undergone this procedure. Colonic adenomatous polyps were present in twice as many cases as controls (15 of 32 vs 15 of 64), and colorectal cancer was present approximately 3 times as often (3 of 32 vs 2 of 64).

Notably, a sizable proportion of the colon pathology found in patients with S bovis bacteremia in the previous studies was otherwise occult. Thus, several authors have advocated complete evaluation of the colon for all patients found to have S bovis bacteremia.^3- 12 Nevertheless, colon evaluation has not been uniformly performed in patients with S bovis bacteremia even in the more recent studies, including our own.^{8- 10,17- 18,27} The reasons for this are difficult to discern in retrospective studies. Comorbidities, advanced age, or an infection outside the gastrointestinal tract thought to account for the bloodstream infection may be reasons not to perform colonoscopy. Because it cannot be otherwise determined which patients with S bovis bacteremia will harbor malignancies, all adult patients who can tolerate evaluation of the colon and who might possibly benefit from identification of an occult neoplasm should be studied after an episode of S bovis bacteremia has been documented.

As seen in Table 4, the reported prevalence of liver dysfunction in patients with S bovis bacteremia varies between 0% and 52%, probably reflecting differences in definitions of liver dysfunction. In the series reporting the largest percentage of patients with liver disease, this was defined by abnormalities of liver function tests or radiographic imaging.⁹ However, some variability among studies may be explained by variation in patient populations served by different hospitals. A small study from Spain of 20 patients with S bovis bacteremia describes 9 patients with cirrhosis defined by more stringent criteria, with an additional patient also having a history of alcohol abuse and abnormalities on liver function testing.¹⁷ A significant proportion of the patients with liver dysfunction in both of these studies also had colonic pathology.^9,17 Whereas colonic neoplasia associated with Streptococcus bovis bacteremia can often be occult,²⁵ it is unclear if clinically significant liver disease was found incidentally. Thus, the clinical ramifications of the association between Streptococcus bovis bacteremia and liver disease are not clear.

In our study, from 17 colonoscopies, 13 adenomas and 1 (possibly 2) carcinomas were identified. Although the incidence of colon cancer was not as high as in other series, there was a striking association of S bovis bacteremia with malignancy irrespective of site, with 29% of patients harboring a cancer. Patients in this series had malignancies in the colon, duodenum, gallbladder, pancreas, ovary, uterus, lung, and hematopoietic system. Previous case reports have noted the occurrence of S bovis bacteremia in patients with pancreatic cancer,²⁸ squamous cell carcinoma of the mouth,²⁹ endometrial cancer,³⁰ melanoma metastatic to the gastrointestinal tract,³¹ lymphosarcoma,³² and Kaposi sarcoma.³³ Similarly, in larger series of patients with S bovis bacteremia, isolated patients with esophageal carcinoma,^8,25 gastric carcinoma,^10,12,25,34 gastric lymphoma,²⁵ and pancreatic carcinoma^17- 18 were described.

A recent series examining antibiotic resistance in S bovis noted 31% of patients with S bovis bacteremia to be harboring neoplasia (58% in our study). Among these, 23% had a malignancy (29% in our study). About half of the malignancies originated from the gastrointestinal system.³⁵ Taken together with our results, these data lead us to believe that physicians caring for patients with S bovis need to be alert to the possibility of malignancy, and searching for extracolonic malignancies may be warranted.

In conclusion, our results substantiate the previously reported associations of S bovis bacteremia with colonic neoplasia and liver disease. Furthermore, our results bring to light the underutilization of colonoscopy in patients with S bovis bacteremia despite the well described association with occult colonic malignancies. This study also suggests that extracolonic malignancy is a common feature of a large proportion of patients with S bovis bacteremia. Further studies are needed to determine if S bovis bacteremia is an independent predictor of malignancy and to determine the pathophysiologic features of this relationship.

Accepted for publication December 17, 2003.

Correspondence: Ronald R. Salem, MD, Yale University School of Medicine, Section of Surgical Oncology, TMP 203, 333 Cedar St, New Haven, CT 06520 (ronald.salem@yale.edu).