An Update on the Streptococcus bovis Group: Classification, Identification, and Disease Associations

Abstract

The Streptococcus bovis group has undergone significant taxonomic changes over the past 2 decades with the advent of new identification methods with higher discriminatory power. Although the current classification system is not yet embraced by all researchers in the field and debate remains over the performance of molecular techniques for identification to the species level within the group, important disease associations for several members of the group have been clarified. Here, we provide a brief overview of the history of the S. bovis group, an outline of the currently accepted classification scheme, a review of associated clinical syndromes, and a summary of the performance and diagnostic accuracy of currently available identification methods.

INTRODUCTION

The history of the Streptococcus bovis group is complicated and confusing due to conflicting classical distinctions based on imperfectly differentiating phenotypic attributes and due to modern disagreements concerning the optimal molecular methods for identification to the species level. Consequently, even the current classification system remains subject to debate and is not accepted by all researchers in the field. In this Minireview, we provide a brief overview of the history of the S. bovis group and description of a current classification system, followed by a discussion of disease associations. We conclude with an overview of currently available diagnostic methods for species- and subspecies-level identification.

A BRIEF TAXONOMIC HISTORY OF THE S. BOVIS GROUP

In her seminal work, Rebecca Lancefield defined group D carbohydrate antigen reactivity among a collection of beta-hemolytic streptococci isolated from cheese (1). Studies of a more diverse group of streptococci over subsequent years revealed that a number of nonhemolytic and alpha-hemolytic isolates from both animal and human sources also demonstrated group D antigen reactivity and shared important characteristics with one another (2). Within this serologic group were bovine and dairy-associated streptococci, some of which were referred to as Streptococcus bovis or S. bovis group, based loosely on shared characteristics with the S. bovis strain that was originally defined by Orla-Jensen in 1919 (3, 4). Streptococcus equinus, Streptococcus faecalis and Streptococcus faecium strains were also within the D serologic group, although the latter two were separable from the other group D streptococci based on their ability to grow in 6.5% NaCl, hydrolyze arginine, and decarboxylate tyrosine, and they were ultimately renamed enterococci (3, 5). In the early 1960s, classification systems were developed that placed S. bovis into a more defined scheme among the group D streptococci (3).

Though the founding members of the group D streptococci were animal in origin, there was an emerging appreciation that human-origin group D isolates were likely important agents of human infection. Facklam emphasized that accurate methods of identification to the species level were essential to understanding the species distribution of the group D streptococci among human infections and their antimicrobial susceptibilities (6). Potentially important disease associations, along with better methods for identification to the species level, stimulated much interest in the group D streptococci, which led to the current classification system.

CURRENT TAXONOMY OF THE S. BOVIS GROUP

Following the taxonomic separation of the enterococci from the group D streptococci, a set of strains referred to as the S. bovis-S. equinus complex remained (7). In older and in some current literature, strains of this complex have been classified on the basis of their ability to ferment mannitol. Those strains that could ferment mannitol were designated biotype I, and those that could not, as biotype II (also referred to as S. bovis variant in some literature). Biotype II was further divided into biochemical subtypes II/1 and II/2 based initially on bile-esculin reaction, acidification of trehalose, and hydrolysis of starch and subsequently on sequence-based analysis (8–11). These biotype classifications retain relevance due to their frequent citation in the literature.

A current taxonomic system that follows recent developments in the literature supplemented by genetic analysis includes seven divisions within the group D streptococci (biotype classifications are shown in parentheses but are not part of official taxonomic nomenclature): S. equinus, Streptococcus infantarius subsp. coli (biotype II/1), S. infantarius subsp. infantarius (biotype II/1), Streptococcus alactolyticus, Streptococcus gallolyticus subsp. gallolyticus (biotype I), S. gallolyticus subsp. pasteurianus (biotype II/2), and S. gallolyticus subsp. macedonicus (7, 12, 13). Though this nomenclature is still subject to debate and has not yet been universally embraced, we will use this system for consistency with other recent attempts to harmonize analysis of the literature (14).

DISEASE ASSOCIATIONS OF THE S. BOVIS GROUP

It had been appreciated since at least the early 1950s that group D streptococci were a cause of infective endocarditis. However, the detailed species associations were not well understood, and some early nonenterococcal S. bovis isolates may have been classified either as enterococci due to their bile insolubility and ability to hydrolyze bile esculin or as Streptococcus viridans due to their penicillin susceptibility (15). As more refined methods to separate the enterococci and nonenterococcal group D organisms were developed, it became clear that the nonenterococcal group D strains were indeed a significant cause of endocarditis.

An association between colorectal carcinoma and group D endocarditis was appreciated at least as early as 1951 (16). In 1974, Hopes and Lerner (17) reported a case series suggesting a relationship between nonenterococcal S. bovis and colorectal carcinoma, as well as other gastrointestinal diseases. Klein et al. (15) studied the relationship between gastrointestinal colonization with S. bovis and a variety of gastrointestinal conditions, including colonic carcinoma, inflammatory bowel disease, peptic-ulcer disease, diverticular disease, and gastrointestinal bleeding of unknown etiology using the methods of Facklam for identification to the species level (6). These authors found S. bovis in fecal cultures from 35 of 63 (56%) patients with colonic carcinoma, representing a significantly greater proportion than in any other group studied. A review of earlier literature suggested that possibly six of nine previous cases that had reported an association between colorectal carcinoma and infective endocarditis involved an S. bovis group organism (15).

In 1989, Ruoff and colleagues (18), using more accurate identification techniques, found striking correlations between S. bovis biotype I (S. gallolyticus subsp. gallolyticus) bacteremias and infective endocarditis (94%) and colonic neoplasm (71% of patients overall and 100% of those who underwent thorough examination). These rates were substantially greater than those associated with the S. bovis biotype II isolates in the study (18% association with endocarditis, and 17% overall association with colonic carcinoma), suggesting the possibility of a specific disease association with S. gallolyticus subsp. gallolyticus. Corredoira et al. (19) studied S. bovis isolates that had been collected over a 16-year period and found that 74% (31 of 42) of S. bovis biotype I (S. gallolyticus subsp. gallolyticus) isolates were associated with endocarditis and 57% (24/42) were associated with colonic tumors. Similar to those seen in Ruoff et al. (18), these rates were significantly greater than the rates associated with S. bovis biotype II isolates (19). Another study of 58 consecutive S. bovis bacteremia isolates found that 9 of 21 S. gallolyticus subsp. gallolyticus isolates were associated with infective endocarditis, and 2 of 21 of these isolates were associated with colonic carcinoma; however, only a limited number of patients in this study underwent colonoscopy, leading to a potential underestimation in the actual association (9).

In total, much evidence has accumulated that supports a specific association of S. gallolyticus subsp. gallolyticus with infective endocarditis and colorectal cancer. A recent thorough meta-analysis (14) found that patients who had S. bovis biotype I (S. gallolyticus subsp. gallolyticus) infection had a strongly increased risk of having colorectal carcinoma (odds ratio, 7.26; 95% confidence interval, 3.94 to 13.36) compared with those who had S. bovis biotype II infections. This meta-analysis also found that S. bovis biotype I was a much more common cause of infective endocarditis than S. bovis biotype II. The authors point out that these strong disease associations underscore the importance of proper identification to the species level and classification of S. bovis isolates by clinical microbiology laboratories and the interpretation by physicians.

Interestingly and importantly, a different pattern of disease association for S. gallolyticus subsp. pasteurianus is implied by recent literature. In 1997, Cohen et al. reported one case of S. bovis biotype II isolated from cerebrospinal fluid from a patient with meningitis and another isolated from a patient with a subdural empyema (20). The authors performed a literature review and found 14 previously reported cases of meningitis attributed to S. bovis. Since the time of the Cohen et al. review (20), at least 16 additional cases of adult and pediatric meningitis attributed to S. bovis biotype II/2 (S. gallolyticus subsp. pasteurianus) have been documented, though the actual number may be much larger, given that, in many cases, the isolates were not identified to the subspecies level (21–31). Notably, most of these cases occurred in patients who lacked an identified colonic neoplasm, and the meningitis occurred without an identified bacteremia in some cases. Interestingly, there was an association with Strongyloides stercoralis infection in 14 cases of S. gallolyticus meningitis reported by van Samkar et al., suggesting an underlying mechanism by which cerebrospinal fluid infection occurred, though it should be noted that not all isolates in this case series were of clearly defined subspecies (31). The emerging picture from this literature suggests that S. gallolyticus subsp. pasteurianus may indeed be a significant (though rare) cause of meningitis in patients without identified colonic neoplasm and that it is a pathogen that may have an underappreciated significance as an agent of systemic disease in infants.

Given a new understanding of the techniques and strategies that are required to separate members of the S. bovis group, it is clear that, in some cases, accurate disease associations have been obscured in older and possibly in some current literature and that a re-examination of isolates in light of current taxonomy is required where possible.

DIAGNOSTIC METHODS FOR THE IDENTIFICATION OF SPECIES AND SUBSPECIES WITHIN THE S. BOVIS GROUP

The specific disease associations within the S. bovis group described above underscore the importance of accurate species- and subspecies-level classification of clinical isolates. However, the complexity of taxonomic changes over the last decade has made it difficult to ensure accurate and complete classification in many cases. Contributing factors have included the absence of a curated sequencing database, the lack of revised nomenclature in culture collection deposits, and the irregularity (or lack) of updates to commercial phenotypic-based identification databases, in part due to the lengthy FDA regulatory approval process to which such changes are subject. Regular taxonomic review and incorporation of nomenclature changes into routine clinical microbiology reporting are now mandated by certain laboratory accrediting agencies, such as the College of American Pathologists, because of the potential effects that such changes may have on antimicrobial choice and/or interpretive criteria (CAP Accreditation Program item MIC.11375) (32). It is therefore the responsibility of the laboratory to understand currently accepted taxonomy and the limitations of contemporary diagnostic platforms in order to provide accurate organism identification that may impact clinical guidance for specific disease syndromes.

The performance of phenotypic, biochemical, matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), and sequencing-based methods for species- and subspecies-level identification within the S. bovis group are discussed below. A summary of S. bovis group members that are listed by the manufacturer to be represented in each commercial test system is provided in Table 1. Importantly, no single test system can provide unequivocal identification, and molecular techniques are often used to complement phenotypic findings.

TABLE 1.

Summary of commercial identification databases and the manufacturer's listed coverage of species and subspecies within the S. bovis group^a

Species or subspecies	Coverage for biochemical system			Coverage for MALDI-TOF MS system
	API 20 strep	Rapid ID 32 strep	Vitek 2 GP ID card	Bruker Biotyperb (no. of representative mass spectral profiles)	Vitek MS 2.0	Vitek MS 3.0c	Saramis
S. alactolyticusd	No	Yes	Yes	Yes (1)	Yesc	Yes	No
S. equinusd	Yes	No	Yes	Yes (2)	Yesc	Yes	Yes
S. gallolyticus subsp. gallolyticusd	Yes	No	Yes	Yes (2)	Yesc	Yes	Yes
S. gallolyticus subsp. macedonicusd	No	No	No	Yes (2)	No	No	Yes
S. gallolyticus subsp. pasteurianusd	Yes	No	Yes	Yes (3)	Yesc	Yes	Yes
S. infantarius subsp. colid	Yes	No	Yes	Yes (6)e	Yesf	Yes	Yes
S. infantarius subsp. infantariusd	Yes	No	Yes	Yes (1)g	Yesf	Yes	Yes
S. bovis	No	Yes	No	No	No	No	No
S. gallolyticus	No	No	No	Yes (3)	No	No	No

^aRefer to the text for a discussion of each assay performance.

^bBruker Biotyper v3.1 research-use-only (RUO) database.

^cClaimed only by Conformité Européenne for in vitro diagnostics (IVD-CE).

^dIndicates true and most current designation in the S. bovis group.

^eLabeled as S. lutetiensis in Bruker Biotyper v3.1 RUO.

^fCleared by the U.S. Food and Drug Administration for in vitro diagnostics (IVD-FDA) and claimed by IVD-CE.

^gLabeled as S. infantarius in Bruker Biotyper v3.1 RUO.

PHENOTYPIC AND BIOCHEMICAL METHODS

Until the early 2000s, Streptococcus species were identified primarily by hemolytic reactions, carbohydrate utilization, and phenotypic tests. Though the taxonomic history has been complicated, the S. bovis group has traditionally been described as Gram-positive cocci in pairs and short chains that were cultured as small gamma- or alpha-hemolytic colonies and that were catalase negative, leucine aminopeptidase positive, pyrrolidonyl arylamidase negative, bile tolerant, esculin hydrolysis positive, salt intolerant, and Lancefield group D positive (33). Further delineation into species and subspecies relied on the testing of β-glucuronidase, α- and β-galactosidase, β-mannosidase, and acid production from starch, glycogen, inulin, and mannitol (33).

Although phenotypic and biochemical tests remain the reference standards for identification within this group, recent studies (9, 22) have demonstrated that these traditional phenotypic and biochemical traits are not universal and can lead to potential misidentifications. In their study of 58 blood isolates, Beck et al. demonstrated that four major phenotypic characteristics that were traditionally used to define the S. bovis group and differentiate them from the enterococci (salt intolerance, bile tolerance, esculin hydrolysis, Lancefield group D typing) were in some instances unreliable (Table 2), emphasizing the significance of the pyrrolidonyl arylamidase test for separating the two groups (9). The variability in Lancefield group D agglutination using modern commercial test kits was further supported by Youn et al. who showed that only 40 of 51 (78%) S. bovis group isolates (as identified by molecular and phenotypic methods) typed as group D, while nine were untypeable and two typed as group A (34). These findings are significant given the wide adoption of Lancefield grouping in clinical laboratories, especially in settings where antigen grouping with commercial kits may be the only discriminatory test performed for Streptococcus spp. depending on organism source and laboratory complexity. Inconsistencies in enzymatic activity and acid production have also been found in a few studies (7, 9, 22).

TABLE 2.

Summary from Beck et al. (9) demonstrating the unreliability of key biochemical reactions that were traditionally used to differentiate the S. bovis group from Enterococcus spp.

Organism	Salt intolerance (%)	Growth and hydrolysis on bile esculin agar (%)	Lancefield group D positive (%)
S. gallolyticus subsp. gallolyticus	72	100	72
S. gallolyticus subsp. pasteurianus	58	83	8
S. infantarius subsp. coli	88	35	59

Several commercial panels are available for Streptococcus identification, but few studies have focused on the accuracy of identification within the S. bovis group specifically. The API 20 strep and Rapid ID 32 strep systems (bioMérieux, Marcy l'Etoile, France) are the most widely referenced panels in S. bovis group literature. Beck et al. found a 97 to 98% correlation with 16S rRNA gene sequencing for identifying S. gallolyticus subsp. gallolyticus, S. gallolyticus subsp. pasteurianus, and S. infantarius subsp. coli (n = 58) (9). In 2009 to 2010, however, both panels underwent reformulation and taxonomic database updates (bioMérieux package inserts), and performance of the updated system with reference molecular testing has not been well studied. In another study, Youn et al. compared the Vitek 2 GP ID card (bioMérieux) to sodA sequencing and found 75% agreement between the methods (34).

Therefore, while commercial phenotypic systems have streamlined workflow by grouping many biochemical tests into a single panel, they do not permit unambiguous identification, because it is difficult to build a flexible and reproducible system that will encompass the different and somewhat limited biochemical traits exhibited by different strains of the same species or subspecies. Care therefore must be taken with final interpretation of the organism identification based on phenotypic and biochemical findings alone.

MALDI-TOF MS

Over the last 5 years, MALDI-TOF MS has transformed the clinical microbiology and infectious diseases fields, enabling significantly faster time to identification at a discriminatory power that is considerably higher than phenotypic and biochemical identification alone and similar to that of 16S rRNA gene sequencing (35). The clinical utility of this technology has been frequently reported; however, only a few studies have evaluated the performance of MALDI-TOF MS for accurate species and subspecies identification within the S. bovis group. By analyzing 52 previously identified S. bovis group isolates, Romero et al. reported that only 27 of 40 (68%) S. gallolyticus isolates were identified by the Biotyper database (Bruker Daltonics, Billerica, MA) to the species level, but none were identified to the subspecies level when a combination of 16S rRNA gene and sodA sequencing was used as the reference standards for identification (36). Identification accuracy was not addressed in this study, but the authors commented that subspecies-level identification at the time of study (2011) was not achievable with MALDI-TOF MS (36). These findings are in striking contrast with those of Hinse et al., who, working with a different system, demonstrated species- and subspecies-level discrimination based on dendrogram analysis of mass spectral profiles using the SARAMIS database (bioMérieux) with the AXIMA confidence MALDI-TOF MS instrument (Shimadzu, Japan), although the authors commented that S. gallolyticus subsp. pasteurianus could not always be separated from S. gallolyticus subsp. gallolyticus (37). These discrepant conclusions suggest that identification accuracy of MALDI-TOF MS is strongly dependent on technical details given the different instruments, spectral databases, and algorithms employed in the two studies (Table 1). To address this issue comprehensively, Youn et al. systematically analyzed 51 previously identified S. bovis group isolates with various protein extraction methods against two different MALDI-TOF MS instruments (Bruker Microflex and bioMérieux Vitek MS) and four different MALDI-TOF MS databases (Bruker Biotyper v3.1, Vitek MS v2.0, Vitek MS v3.0, and SARAMIS) using sodA sequencing and a general BLAST analysis of the NCBI database as the reference standard (34). In this study, the Vitek MS v3.0 system performed best, with 76 to 92% agreement with the results of sodA sequencing, depending on the protein extraction method used (34).

These studies demonstrate that MALDI-TOF MS can provide reasonably accurate species- and subspecies-level identification within the S. bovis group; however, the accuracy and discriminatory power of this technology is highly dependent upon the instrument, methods, and databases employed. Supplementation with biochemical and/or molecular testing may be needed for discrimination in some cases.

GENOMIC ANALYSIS

Over the years, numerous target genes, including sodA, 16S rRNA gene, rpoB, groEL, gyrB, and recN, have been analyzed for their discriminatory potential in providing species- and subspecies-level differentiation within the S. bovis group (13, 34, 38, 39). All such studies face similar challenges in defining reference identifications for the sets of isolates against which new molecular methods are tested. This has proved difficult and controversial in many cases, given the lack of an accepted reference molecular identification method and changing taxonomic definitions. Even though partial sequencing of the 16S rRNA gene is generally considered adequate for most routine clinical bacterial identification, analysis of six S. bovis group type strains showed a percent identity that was too close (97.1% to 99.8%) to differentiate species and subspecies within the group (13). In contrast, other authors have argued on the basis of comparative studies that partial sequencing of the sodA gene (∼480 bp; 82% coverage) may provide better discriminatory power by dividing the S. bovis group into five main clusters (13, 38). More recently, a comprehensive study of 65 Streptococcus type strains found that partial sequencing of the groEL gene (757 bp, ∼47% coverage) may be another useful target (39). Additional testing of groEL sequences on a larger subset of isolates will be useful to understand its utility relative to the other methods mentioned above.

A current major limitation is the lack of curated, well-populated, and updated sequence databases that can provide reliable organism identification. Several contradictory results have been published in the S. bovis group literature, with unreliable 16S rRNA gene and sodA sequences of type strains deposited in GenBank (9, 37). Many anticipate that use of new approaches, such as whole-genome sequencing, may result in another re-evaluation of taxonomic status in the future (40). Keeping up to date with these changes and understanding their clinical relevance will be imperative for clinical microbiologists and infectious disease physicians alike.

ANTIMICROBIAL SUSCEPTIBILITY

Antimicrobial susceptibility data for the S. bovis group have remained relatively stable over the years, with MICs in the susceptible range reported for β-lactams (penicillin, ampicillin, amoxicillin-clavulanate, ceftriaxone, oxacillin, meropenem) and vancomycin; however, variable results have been observed for clindamycin, erythromycin, and levofloxacin (9, 34, 36). Surprisingly, a D test positivity rate of 89% for clindamycin-susceptible, erythromycin-resistant isolates was observed in one study (34), and all S. bovis group isolates in another study exhibited low-level resistance to the aminoglycosides (36). A high percentage of tetracycline resistance reported by Beck et al. (9) may be possibly explained by the recent identification of pSGG1, a novel plasmid carrying genes for tetracycline resistance that has strong sequence similarities to plasmids and chromosomes in several ruminal and gastrointestinal bacteria, thus suggesting a potential for horizontal gene transfer (40). Future genomic studies may highlight other potential mechanisms of resistance.

CONCLUSIONS

Species- and subspecies-level identification within the S. bovis group has become increasingly important as specific disease associations, including gastrointestinal neoplasms, meningitis, and endocarditis, are recognized. Reference identification in the S. bovis group still relies on biochemical and phenotypic-based assays; however, data show that single-gene-based molecular testing (16S rRNA gene and sodA) and MALDI-TOF MS may be useful in providing additional information to help enable species- and subspecies-level discrimination. When possible, both gene targets can be sequenced to increase confidence in identification accuracy; however, genomic sequencing is beyond the capabilities of many clinical microbiology laboratories. Studies published to date suggest that the accuracy of MALDI-TOF MS is highly dependent on the system used and the databases employed. For these systems, the responsibility falls on the laboratory to validate and determine the accuracy of identifications provided by research-use-only (RUO) databases. In time, and with use of appropriately supplemented and updated databases, commercial platforms may be able to identify S. bovis group members at the species- and subspecies-level accurately and consistently. Table 1 illustrates the many differences observed between various commercial platforms, making it difficult to develop a simple and practical identification algorithm for microbiologists. Until further data are generated, many labs may choose to continue to report S. bovis at the group level, with exclusive reporting to the species or subspecies level performed only upon request and with careful consideration of the results obtained across multiple test methods and the clinical context of the patient. A thorough review of culture collection strains is warranted, because many isolates may have been originally misidentified and/or deposited before taxonomic revision. Care should also be taken when reviewing the literature due to changing nomenclature and the application of different methods for species and/or subspecies designation over time. It is envisaged that, once accurate identification models are in place, clinicians may be better equipped to predict potential disease associations and/or discover new organism-disease associations.

Biographies

John P. Dekker received his M.D. and Ph.D. degrees from Harvard Medical School through the NIH Medical Scientist Training Program, with graduate work in the field of ion channel biophysics. He completed Pathology residency and fellowship training in Medical Microbiology at Massachusetts General Hospital. He is board certified through the American Board of Pathology and is a Fellow of the College of American Pathologists. In 2013, he joined the medical staff of the NIH Clinical Center, where, with Dr. Anna Lau, he codirects the Bacteriology, Specimen Processing, Parasitology, and Molecular Epidemiology sections of the Clinical Center's Microbiology Service in the Department of Laboratory Medicine. His translational and basic research interests include proteomics and next-generation sequencing technologies and their applications in diagnostic microbiology. He is also interested in mechanisms of antimicrobial resistance and has served on the FDA Anti-Infective Drugs Advisory Committee (AIDAC).

Anna F. Lau received her Ph.D. from the University of Sydney, Australia, where her research focused on the development of novel diagnostic platforms for the diagnosis of invasive fungal diseases. She completed her CPEP Clinical Microbiology Fellowship training at the NIH Clinical Center and is board certified through the American Board of Medical Microbiology. In 2013, she joined the microbiology faculty in the Department of Laboratory Medicine of the NIH Clinical Center, where, with Dr. John Dekker, she codirects the Bacteriology, Specimen Processing, Parasitology, and Molecular Epidemiology sections. Dr Lau's translational research focuses on the development of new rapid-diagnostic platforms for microbial identification and the detection of resistance mechanisms using molecular-based techniques and mass spectrometry. In 2014, Dr. Lau was recognized with the prestigious Forbes 30 Under 30 award in Science and Healthcare for her development of the NIH MALDI-TOF MS mold database.