In a recent review of the literature, Klein and co-workers
(10) noted that 37% of a total of 251 patients with septicemia
caused by S. bovis had evidence of colonic disease, including
carcinoma, polyps, or unidentified colonic masses. Of 34
patients with prospective lower gastrointestinal tract evaluation,
50% were found to exhibit these conditions. These
observations are similar to those of our study in that 39% of
41 patients with clinically significant bacteremia caused by
S. bovis or S. salivarius had colonic neoplasms and 57% of
28 patients who underwent thorough colonic examination
had such lesions. However, more detailed identification of
streptococcal blood isolates to the biotype level revealed an
even more striking association between bacteremia caused
by S. bovis T and colonic neoplasia. Seventy-one percent of
patients with bacteremia due to S. bovis I overall and 100%
(12 of 12) of such patients who underwent thorough colonic
examination had neoplasms detected. In contrast, only 25%
of the remaining 16 patients with bacteremia caused by other
biotypes of S. bovis or by S. salivarius had neoplasms
detected on thorough colonic examination. These data suggest
that identification of S. bovis biotype I strains among
streptococcal isolates causing bacteremia identified patients
at even higher risk of having colonic neoplasia than previously
documented and that all patients with bacteremia
caused by this organism should undergo colonoscopy unless
contraindicated.
The striking association between bacteremia caused by S.
bovis biotype I and both colonic neoplasia (71%) and bacterial
endocarditis (94%), compared with bacteremias caused
by the closely related organisms S. bovis variant and S.
salivarius, suggests the possibility of specific bacterium-host
cell interactions involving S. bovis biotype I organisms.
Such a specific interaction could involve, for example,
selective adherence of this biotype to surface receptors on
neoplastic colonic cells or cardiac endothelium. Further
studies will be needed to address the possibility of such
specific interactions more directly.
Bacteremia due to S. salivarius was less likely to reflect a
clinically significant infection than was bacteremia due to S.
bovis. Nevertheless, of seven patients with clinically significant
bacteremia caused by S. salivarius, three patients had
bacterial endocarditis and one had an associated colonic
neoplasm. Roses et al. (15) reported a case of endocarditis
due to S. salivarius associated with colonic adenocarcinoma,
and Hoecker and co-workers (7) reported six cases of sepsis
due to S. salivarius in children with underlying malignant
diseases. Our data suggest a possible hepatobiliary origin for
many clinically significant infections with S. salivarius and
S. bovis subbiotype II/1, an anatomical correlation not
previously recognized for either of these organisms.
We conclude that accurate differentiation between isolates
of S. bovis and S. salivarius and biotyping of isolates of S.
bovis provide clinically useful information. Problems in
identifying these physiologically similar organisms with presumptive
tests have been discussed previously (16), but our
experiences and those of others (6) have demonstrated the
utility of the API Rapid Strep system for identification of
these organisms; other commercially available systems may
also be of use for this purpose (4). No clear distinction in the
clinical significance of the subbiotypes of S. bovis Il could be
observed from our data. Our findings suggest that identification
of S. bovis to the level of biotype I or Il and
distinction between these organisms and S. salivarius should
be more widely applied.