Microbiology of Actinomyces Species Isolated From Patients With Invasive Disease and Contaminated Samples in a Comprehensive Cancer Center

Abstract

Background

Actinomyces are mucous membrane commensals that infrequently cause invasive disease. Our goal was to define Actinomyces species prevalence, the predominant disease site and risk factors for actinomycosis.

Methods

We retrospectively reviewed patients with growth of Actinomyces species from cultures in a single-cancer center from July 2007 to June 2020. Proven invasive actinomycosis was defined as the presence of compatible clinical syndrome and radiographic findings with histopathological confirmation or culture from a normally sterile site. Probable invasive actinomycosis was defined based on the same criteria but without histologic confirmation. Contaminants were defined as culture growth in the absence of clinical or radiological findings consistent with disease. Speciation of Actinomyces was performed by the bioMerieux VITEK 2 anaerobic and coryneform identification card.

Results

Of 235 patients, 179 (76.2%) had malignancy. Among 90 (38.3%) patients with invasive actinomycosis, A odontolyticus was isolated in 32 (35.6%), followed by A meyeri in 20 (22.2%), and A naeslundii in 17 (18.9%). Among 145 (61.7%) colonized patients, A odontolyticus was isolated in 67 (46.2%), followed by A naeslundii in 27 (18.6%). Abdominopelvic infection was the most common site for invasive actinomycosis documented in 54 patients (60.0%) followed by orocervicofacial in 14 (15.6%) and thoracic in 10 (11.1%).

Conclusions

A odontolyticus, A meyeri, and A naeslundi were the most frequently isolated species causing invasive actinomycosis, and A odontolyticus and A nauslendii among colonizers. Abdominopelvic represented the most frequent site for invasive disease. Further studies are needed to investigate the epidemiology of Actinomyces species in this population.

The genus Actinomyces consists of a diverse group of gram-positive, anaerobic, non–acid-fast, rod-shaped bacteria that form branching filaments [1]. They are commensals of human mucous membranes found frequently in the oral cavity [2, 3], but also in the gastrointestinal and female genital tracts [1, 4, 5]. Invasive actinomycosis is classically described as an indolent, slowly progressive infection characterized by abscesses, fistulae, and fibrosis; however, these organisms may also cause more acute infections, such as cutaneous abscesses, or be found in mixed infections, such as intra-abdominal infections, without classical features. Invasive disease originates from the patient's own flora and is precipitated by compromised mucosal or epithelial barriers during dental procedures, surgery, trauma, or radiation [6–11]. Predisposing conditions include poor oral hygiene, malignancy, immunosuppressive conditions, and bisphosphonate-induced osteonecrosis [6, 12, 13]. Actinomycosis is classified into distinct clinical forms according to the anatomical site involved, including orocervicofacial, thoracic, abdominal/pelvic, central nervous system, musculoskeletal, and disseminated disease [1, 11, 14].

The Actinomyces genus contains many species with several implicated in human disease. A israelii was described in 1896 [5]. In 1951, A. naeslundii was implicated in human disease [15]. A. odontolyticus were describe subsequently [16, 17]. With the introduction of new microbiological techniques such as amplified 16S ribosomal DNA restriction analysis, new species have been identified [18, 19] and certain sequencing profiles, such as A naeslundii and A viscosus, have been reassigned within species [3]. The spectrum of the causative species of actinomycosis continues to evolve with new species identified [20]. Invasive actinomycosis is most commonly caused by A israelii [1, 14], with A odontolyticus, A meyeri, A naeslundii, A. viscosus, A gerencseriae, and A graevenitzii described with less frequency.

Despite A israelii having been described as the most common pathogenic species among the genus Actinomyces, we observed other species being isolated at high frequency among our patients. This prompted a clinical investigation on colonization and invasive disease caused by Actinomyces species with the aim to study the epidemiology, clinical manifestations, and the responsible species at our cancer center.

METHODS

We performed a retrospective cohort study of patients at Roswell Park Comprehensive Cancer Center between July 2007 and June 2020 who had 1 or more cultures with growth of Actinomyces species. The study was approved by the institutional review board. Informed consent was waived as the study included a review of electronic medical records and clinical microbiology data. Roswell Park Comprehensive Cancer Center is a 133-bed National Cancer Institute–designated comprehensive care center in Buffalo, New York. Patients who receive care at Roswell Park include those with solid and hematological malignancies and the institution performs and cares for those with stem cell transplant and adoptive cellular therapy. The institution also receives referrals for suspected malignancy and for specialized procedures or care in those with nonmalignant conditions, but these patients are a smaller proportion of our population.

All patients with positive cultures for Actinomyces species and sufficient clinical data were included in the study. Patients were identified through microbiology database review extracted from TheraDoc Clinical Surveillance Software. Patients’ characteristics and clinical data were extracted from patients’ electronic medical records. Data included age, sex, underlying malignancy, inciting factors, surgery involving the gastrointestinal or female genital tract within 3 months from positive culture, radiation therapy at the site of infection anytime in the past, comorbid conditions, smoking history, and white blood cell counts at the time of positive culture. We also obtained data on histopathologic evidence of invasive disease and radiographic findings.

Probable invasive actinomycosis was defined as a compatible clinical syndrome, suggestive radiographic findings, and a positive culture without histopathological confirmation. Proven invasive actinomycosis required the presence of a compatible clinical syndrome, suggestive radiographic findings, positive cultures, and histopathological confirmation of invasive disease, such as demonstration of gram-positive branching filaments consistent with Actinomyces spp. in pus and/or the presence of sulfur granules. Histopathologic confirmation was not required if the positive sample was obtained from an otherwise sterile site (eg, blood, cerebrospinal fluid, intraoperative tissue samples of normally sterile tissue). Colonizers were defined positive samples in the absence of a compatible clinical presentation or radiologic findings. Severe neutropenia was defined as an absolute neutrophil count of ≤0.5 × 10⁹/L at the time of positive culture.

All specimens were cultured under aerobic conditions. Anaerobic cultures were ordered at the discretion of the ordering physician. Identification of Actinomyces at the species level was primarily performed by the clinical laboratory using VITEK 2 automated system anaerobic and coryneform identification card (bioMerieux, Durham, NC). This method reliably identifies A bovis, A israelii, A meyeri, A naeslundii, A neuii, A odontolyticus, and A turicensis species [21]. Other methods included the RapID ANA II System (Remel, Lenexa, KS), API Coryne (bioMerieux), or referral to a reference laboratory when required. This study timeframe preceded the introduction of mass spectrometry for identification of microbes at our institution.

Categorical data were described with frequency counts and the percentage of patients within each category. Continuous data were summarized with descriptive statistics (median, range). Continuous variables were compared with the Student t-test and categorical variables with the chi-squared test with Fisher exact adjustment when necessary. P values < .05 were considered statistically significant.

RESULTS

Between 2007 and 2020, there were a total of 246 positive cultures corresponding to 235 patients. Demographics and clinical characteristics of the 235 patients are presented in Table 1. Median age was 60 years (range, 19–86). Of the 235 patients, 179 (76.2%) had concurrent malignancy, including 152 (64.7%) with solid tumors and 27 (11.5%) with hematological malignancies. Fifty-six (23.8%) were referred for suspected malignancy but clinical investigation ruled out malignancy.

Of 235 patients, 90 (38.3%) met criteria for invasive actinomycosis and 145 (61.77%) were contaminants (Table 1). Of 90 cases of invasive actinomycosis, there were 18 (20.0%) proven and 72 (80.0%) probable cases. Invasive actinomycosis was documented in 73 (81.1%) patients with malignancies and in 17 (30.4%) of noncancer patients (P = NS).

Smoking history was available for 201 patients (85.5%), with 115 (48.9%) of patients being either active smokers or having a history of smoking. Among 90 cases of invasive actinomycosis, inciting factors included previous surgery involving the gastrointestinal or female genital tract in 31 (33.7%) compared to 25 (17.2%) in colonized individuals (P = .002). Prior radiation therapy at the site of infection was documented in 15 (16.7%) of patients with invasive actinomycosis compared with 8 (5.5%) in colonized patients (P = .005). Among 13 cases of orocervicofacial actinomycosis, osteonecrosis of the jaw was present in 4 (30.8%), with 2 from previous radiation and another 2 from bisphosphonate therapy.

The most common sites of invasive actinomycosis (Figure 1) were as follows: 54 (60.0%) abdominopelvic, 14 (15.6%) orocervicofacial, and 10 (11.1%) thoracic.

Of 246 specimens, 78 (31.7%) were cultured under aerobic and anaerobic conditions, whereas 168 (68.3%) were cultured under aerobic conditions only. Among 78 specimens cultured under both conditions, Actinomyces were isolated under aerobic conditions in 37 (47.4%), under anaerobic conditions in 35 (44.9%), and 6 (7.7%) under both conditions.

Ninety-six isolates were obtained from 90 patients with invasive actinomycosis. A odontolyticus was isolated in 34 patients (35.4%), followed by A meyeri in 20 (20.8%), A naeslundii in 20 (20.8%), A turicensis in 10 (10.4%), A israelii in 7 (7.3%), and A neuii in 4 (4.2%) with 1 isolate (1.0%) not further speciated (Figure 2).

Among 150 isolates that were considered colonizers A odontolyticus was isolated in 68 (45.3%), followed by A naeslundii in 28 (18.7%), A meyeri in 24 (16.0%), A turicensis in 15 (10.0%), A israelii in 7 (4.7%), A neuii in 3 (2.0%), A viscosus in 1 (0.7%), and A bovis in 1 (0.7%) with 3 isolates (2.0%) not further speciated (Figure 2). Species association with invasive disease was 33.3% in A odontolyticus (n = 34), 45.5% in A meyeri (n = 20), 41.7% in A naeslundii (n = 20), 40.0% in A turicensis (n = 10), 50.0% in A israelli (n = 7), and 57.1% in A neuii (n = 4) with A viscosus and A bovis not associated with invasive disease (Figure 2). Differences between the groups were not statistically significant. The origin of samples in contaminated specimens is illustrated in Figure 3.

Cultures were polymicrobial in 73 (81.1%) patients with invasive actinomycosis and in 125 (86.2%) contaminated samples (n = NS). Companion microbial growth mainly reflected the local normal flora. Microbial recovery was also affected by growth under anaerobic conditions if performed.

Actinomyces spp. were recovered in blood cultures in 14 patients. In 6 (42.9%), the cultures were considered skin colonizers (ie, contaminated blood cultures) because of a lack of compatible clinical syndrome and radiographic findings. Of 8 (57.1%) patients with bacteremia from Actinomyces, abdominopelvic was the source in 7 (87.5%), dental (orocervicofacial) was the source in 1 (12.5%), and 3 (37.5%) occurred in the setting of severe neutropenia.

DISCUSSION

We reviewed the microbiology of Actinomyces species in patients with and without invasive disease. Our study was driven by the observation that the most commonly isolated species are A odontolyticus, A naeslundii, and A meyeri rather than A israelii, also suggested by older literature. Indeed, A odontolyticus was the dominant species in both cases of invasive actinomycosis (35.4%) and positive cultures without clinical disease (colonization) (45.3%), followed by A naesludii and A meyeri. A israelii was less frequently isolated.

In a study that included 1997 cases of cervicofacial actinomycosis, A israelii and A gerencseriae (formerly identified as A israelii serovar 2) together, were the etiologic agent in 67.7% of cases [3]. The study employed traditional morphological examination and biochemical methods for identification of Actinomyces. However, recent literature from the United States suggests that A odontolyticus and A meyeri are more frequently involved in invasive disease than previously thought. In a retrospective analysis from a large university hospital in the United States, among 36 patients with invasive disease (including 6 patients with malignancy), A meyeri was the most common causative species (33.0%) followed by A israelii in 8% [22]. In a retrospective study in a cancer population that included 46 patients with invasive disease, A odontolyticus was the most frequently isolated species identified in 26% of cases, followed by A israelii in 24% [23]. A study employing parallel sequencing of bacterial 16S rRNA genes among 25 patients with brain abscess documented the presence of Actinomyces species as part of polymicrobial flora in 14 patients (56.0%), with A meyeri identified as the leading species in 12 (85.7%) [24].

In a longitudinal study investigating the age-related oral colonization by Actinomyces species in saliva among 39 healthy infants at various time points A odontolyticus was the most common species at all 5 sampling times followed by A naeslundii [25]. Although these studies vary by methodology and population, they suggest species such as A meyeri and A odontolyticus are more frequently involved in colonization and invasive disease than previously documented.

The discordance between the dominant species in our cohort and older literature could be explained by the more sensitive microbiological techniques currently employed for species identification [21]. Furthermore, geographic variation in species distribution, the potential effect of nutrition and advances in oral hygiene are unexplored, potentially modifying factors. There is no evidence for species-specific preferential sites of colonization that would drive the microbiology of invasive actinomycosis and might have affected our results because our study involved abdominopelvic infections at a higher frequency. Finally, no species appeared to have a propensity for invasive disease, although our study did not investigate virulence (Figure 1).

Constituents of colonizing flora vary in health and disease state. Normal flora is promptly modified by antibiotic pressure and other pathological conditions. Microbiome studies have demonstrated that colonizing phyla vary in health and disease state [26]. Loss of diversity and significant variation were documented in patients with oral squamous cell carcinoma [27] and lung cancer [28], but also through progression of malignancy [29]. Although these studies investigated the colonizing flora at a phylum level, they are indicative of the dynamic interaction between microbial communities and host conditions during disease process.

The most common site of invasive actinomycosis in our cohort was abdominopelvic (60.0%) followed by orofacial (15.6%) and thoracic (11.1%), contrasting with the literature that suggested orofacial as the most common site of invasive actinomycosis [6, 14, 30]. We attributed this difference in disease location to our population's distinct features rather than trends in pathogen's tropism. Our cohort included a disproportionately larger number of patients with infections following surgeries involving the gastrointestinal or gynecological tract, which are established reservoirs for Actinomyces [3, 5]. In addition, isolation of Actinomyces as part of polymicrobial growth was the rule. Because the attribution of Actinomyces to the pathogenesis of infection could not be reliably determined we used a more inclusive definition for invasive disease that regarded any isolation of Actinomyces spp. as a contributing rather than a companion pathogen [1, 6, 11, 22].

Most cases of actinomycosis occurred among patients with solid tumors (70.0%) rather than in those with hematological malignancies (11.1%). This difference is most certainly driven by the risk factors for invasive disease by Actinomyces spp., which include compromise mucosal or epithelial barriers by radiation or chemotherapy and injury associated with surgery [10, 11, 31]. Indeed, surgery involving the gastrointestinal and genital tract and radiation therapy were associated with invasive actinomycosis at a statistically significant level (P = .002 and P = .005, respectively).

Cultures were polymicrobial in 81.1% of samples from patients with invasive actinomycosis and in 86.2% of contaminated samples reflecting the polymicrobial nature of source flora. This finding is consistent with the published literature [3, 6, 11, 22] and highlights the uncertainty in assessing the attribution of Actinomyces isolates to the infectious process and therefore duration of therapy.

Limitations of our study include its retrospective design, single-center study that includes patients receiving cancer care, and lack of confirmatory pathology with a large proportion of patients having probable disease. In addition, we were limited by the spectrum of species identified by the VITEK 2 automated system that detects the most frequently involved pathogenic species but may miss rarer Actinomyces and the potential incorrect speciation based on standard biochemical panels [32].

In conclusion, among patients with invasive actinomycosis, A odontolyticus, A meyeri, and A neaslundii were the most frequently isolated species rather than A israelii. Similarly, among contaminated samples, A odontolyticus and A nauslendii were more commonly identified. Abdominopelvic and orocervicofacial disease were the most frequent sites involved among patients with malignancies reflecting the characteristics of population under investigation. Surgery involving the gastrointestinal and female genital tract, as well as prior radiation therapy at the site of infection was associated with invasive actinomycosis at a statistically significant level. Further studies are necessary to investigate the epidemiology and virulence of Actinomyces species.

Notes

Acknowledgments. M.E.: data collection and curation, drafting and preparation of the original manuscript. B.H.S., J.P.B. and M.E.G.: manuscript review and editing. N.G.A.: manuscript writing, review and editing, conceptualization, and study supervision.

Financial Support. None.

References

Author notes