Antimicrobial Susceptibility Patterns among a Large, Nationwide Cohort of Abiotrophia and Granulicatella Clinical Isolates

ABSTRACT

Antimicrobial susceptibility patterns from 599 A. defectiva, G. adiacens, and G. elegans clinical isolates were determined by broth microdilution. We observed significant differences in susceptibility across species, particularly to penicillin and ceftriaxone, and across geographical regions. A. defectiva was the least susceptible species overall to penicillin. All isolates were susceptible to vancomycin and >90% were susceptible to levofloxacin.

INTRODUCTION

The nutritionally variant streptococci (NVS) were first described in 1961 as fastidious Gram-positive cocci that grew as satellite colonies around other bacteria (1). They required supplemental l-cysteine or pyridoxal hydrochloride to promote growth in culture media (2). Over the past 4 decades, the taxonomy of these organisms has continued to evolve. The species that have been identified in human specimens are now classified as Abiotrophia defectiva, Granulicatella adiacens, and Granulicatella elegans based on 16S rRNA sequencing (3). An additional proposed species, “Abiotrophia para-adiacens,” has also been isolated from human specimens, while the remaining valid species, Granulicatella balaenopterae, has not, to date (3, 4).

The NVS are commensal oral and intestinal flora in humans but can also cause significant infections such as endocarditis and septicemia or bacteremia (5). Other reported infections include keratitis, postneurosurgical brain abscess, pancreatic abscess, and infections of bone and joints, including prosthetic joints (5–10). Infections with NVS have high treatment failure rates, including high rates of relapse, complications such as heart failure, embolism, and perivalvular abscess, and high mortality rates (11, 12). Due to their stringent growth requirements and relatively infrequent recovery from clinical specimens, susceptibility testing of these organisms is not routinely performed in clinical microbiology laboratories. Most susceptibility data in the literature, as well as NVS treatment recommendations such as those published in the American Heart Association (AHA) endocarditis guidelines (13), were based on studies with small numbers of isolates. These studies often failed to identify NVS isolates to the species level, and different susceptibility breakpoints were used due to lack of standardized testing methods and interpretive criteria prior to the publication of the Clinical and Laboratory Standards Institute's (CLSI) M45 guidelines for fastidious organisms (14). Not surprisingly, there have been discrepancies among the reported susceptibility patterns of NVS (15–21). These limitations can lead to the use of empirical penicillin and gentamicin without susceptibility testing, which could lead to the unnecessary risk of gentamicin toxicity. In 2006, the CLSI published guidelines for antimicrobial susceptibility testing (AST) of fastidious bacteria, including Abiotrophia and Granulicatella species (14, 22–24). Using these standardized guidelines to interpret AST data from a large collection of Abiotrophia and Granulicatella species isolated across the United States, we identified differences in susceptibility patterns across species, in temporal trends of nonsusceptibility, and in the geographic distribution of penicillin susceptibility. These data may aid in determining appropriate empirical therapy and influence future guidelines for in vitro AST and treatment of NVS infections.

RESULTS AND DISCUSSION

AST was performed on 599 NVS isolates identified as G. adiacens (427 isolates), A. defectiva (152 isolates), or G. elegans (20 isolates). Identification of 73% (436/599) of isolates was performed by Associated Regional and University Pathologists (ARUP) Laboratories, including 70% of G. adiacens (298/427), 78% (119/152) of A. defectiva, and 95% (19/20) of G. elegans isolates. The majority of isolates were recovered from blood (71%), followed by wounds (8.2%), body fluids (6.5%, including unspecified body fluid [2.3%], peritoneal/abdominal fluid [2%], pleural fluid [0.8%], urine [0.7%], cerebrospinal fluid [CSF; 0.5%], and pericardial fluid [0.2%]), tissue (5.7%), joint fluid and bone (5.2%), eye (2.8%), and unspecified sources (0.5%).

Only 14% (21/152) of A. defectiva and 39% (168/426) of G. adiacens isolates were susceptible to penicillin (Table 1). Interestingly, G. elegans was significantly more susceptible to penicillin (80%, 16/20) than G. adiacens and A. defectiva (P < 0.001). Among the penicillin-susceptible G. adiacens isolates, 167 had ceftriaxone susceptibility performed, and we found that only 65% (109) were susceptible. However, the vast majority of penicillin-susceptible A. defectiva isolates (95%) remained susceptible to ceftriaxone, which is similar to what was seen in recent studies (25, 26). Of the penicillin-resistant G. adiacens isolates, 95% (40/42) were also resistant to ceftriaxone, whereas 97% of penicillin-resistant A. defectiva isolates remained susceptible to ceftriaxone (Fig. 1; see also Table S1 in the supplemental material). In contrast to what was reported by Alberti et al., we did encounter ceftriaxone-nonsusceptible A. defectiva, although not to the extent described by Mushtaq et al., who saw 8% nonsusceptibility (25, 26). Overall, the majority of A. defectiva (98%, 148/151) and G. elegans (90%, 18/20) isolates were susceptible to ceftriaxone (Table 1), but G. adiacens had a significantly lower proportion of susceptible isolates (47%, 201/426, P < 0.001). Importantly, there was no significant decrease over the 8 years of data (2007 to 2014) in the rates of susceptibility to penicillin or ceftriaxone for G. adiacens and A. defectiva using the Mann-Kendall test, although G. adiacens showed a nonsignificant trend toward decreased susceptibility to ceftriaxone (P = 0.386) (see Fig. S1 in the supplemental material).

TABLE 1.

Antimicrobial susceptibility of G. adiacens, G. elegans, and A. defectiva

Antimicrobial agent and species (no. of isolates)	MIC (μg/ml)			Interpretive breakpoint (μg/ml)a or % of isolates
	Range	MIC50	MIC90	Susceptible	Intermediate	Resistant
Penicillin				≤0.12	0.25–2	≥4
G. adiacens (426)	≤0.03 to ≥16	0.25	2	39.4b	50.7b	9.9
G. elegans (20)	≤0.03–8	≤0.03	0.25	80c	15c	5
A. defectiva (152)	≤0.03–8	0.5	8	13.8d	63.2d	23d
Ceftriaxone				≤1	2	≥4
G. adiacens (426)	≤0.06 to ≥32	2	≥16	47.2b	24.4	28.4b
G. elegans (20)	≤0.06 to ≥16	0.12	1	90	5	5
A. defectiva (151)	0.12 to ≥16	0.5	1	98d	1.3d	0.7d
Erythromycin				≤0.25	0.5	≥1
G. adiacens (413)	≤0.12 to ≥8	≥8	≥8	44.6	0.2	55.2
G. elegans (19)	≤0.12 to ≥8	≤0.12	≥8	52.6		47.4
A. defectiva (143)	≤0.12 to ≥8	≥8	≥8	39.9	0.7	59.4
Clindamycin				≤0.25	0.5	≥1
G. adiacens (414)	≤0.03 to ≥32	0.06	≥4	77.3	2.9	19.8
G. elegans (19)	≤0.03 to ≥4	0.12	0.5	89.4	5.3	5.3
A. defectiva (143)	≤0.03–4	0.12	0.25	92.3d	5.6	2.1d
Meropenem				≤0.5	1	≥2
G. adiacens (425)	≤0.06 to ≥16	0.12	0.5	91.3	6.4	2.4
G. elegans (20)	≤0.06–1	≤0.06	0.25	95	5
A. defectiva (150)	0.12–2	0.5	1	80.7d	18.7d	0.6
Levofloxacin				≤2	4	≥8
G. adiacens (414)	≤0.25 to ≥16	1	2	91.6	0.7	7.7
G. elegans (19)	≤0.25–1	0.5	1	100
A. defectiva (144)	0.25–4	≤0.25	0.5	99.3d	0.7	0d
Vancomycin				≤1
G. adiacens (423)	≤0.5–1	1	1	100
G. elegans (20)	≤0.5–1	1	1	100
A. defectiva (152)	≤0.5–1	0.5	1	100
Daptomycin
G. adiacens (343)	≤0.25 to ≥8	4	≥8
G. elegans (19)	≤0.25 to ≥8	0.5	2
A. defectiva (119)	≤0.25–4	2	4
Linezolid
G. adiacens (413)	≤0.25–2	1	1
G. elegans (19)	0.5–1	0.5	1
A. defectiva (143)	≤0.25–2	1	1
Gentamicin
G. adiacens (414)	≤0.12 to ≥16	2	4
G. elegans (19)	0.25–2	1	2
A. defectiva (143)	≤0.12 to ≥16	2	4
TMP/SMX
G. adiacens (391)	≤0.12/2.4 to ≥8/152	0.25/4.8	4/76
G. elegans (18)	0.25/4.8–4/76	0.25/4.8	4/76
A. defectiva (128)	≤0.12/2.4 to ≥4/76	0.25/4.8	0.25/4.8
Doxycycline
G. adiacens (410)	≤0.06 to ≥16	≤0.12	8
G. elegans (19)	≤0.12–8	≤0.12	8
A. defectiva (144)	≤0.06–8	≤0.12	2
Rifampin
G. adiacens (387)	≤0.5 to ≥8	≤0.5	≤0.5
G. elegans (18)	≤0.5	≤0.5	≤0.5
A. defectiva (126)	≤0.5–1	≤0.5	≤0.5

^aInterpretive breakpoints are shown in bold for each antibiotic.

^bP value shows significant differences (P < 0.05) between G. adiacens and G. elegans.

^cP value shows significant differences (P < 0.05) between G. elegans and A. defectiva.

^dP value shows significant differences (P < 0.05) between A. defectiva and G. adiacens.

FIG 1.

Scattergram comparisons of ceftriaxone and penicillin MICs for G. adiacens (A), G. elegans (B), and A. defectiva (C). Shading indicates the relative frequency of observations.

The AHA infective endocarditis guidelines recommend different treatment strategies based on penicillin MIC: highly penicillin-susceptible isolates (MIC ≤ 0.12) generally do not require addition of gentamicin; for relatively penicillin-resistant isolates (MIC, >0.12 to <0.5), a combination of penicillin and gentamicin is reasonable; and for penicillin-resistant isolates (MIC ≥ 0.5), combination therapy with penicillin and gentamicin is recommended (13). It is important to note that these categories differ from the CLSI penicillin susceptibility categories (MIC ≤ 0.12, susceptible; MIC = 0.25 to 2, intermediate; MIC ≥ 4, resistant) (23). Because there have been differences in penicillin susceptibility reported in the literature (15–21, 24–26), we evaluated the geographic distribution of penicillin susceptibility across different U.S. Census regions using the AHA infective endocarditis guidelines' penicillin MIC categories (Table 2). We found that the Northeast region had the highest proportion of penicillin-susceptible G. adiacens isolates (57.5%), while the Midwest had the highest percentage of isolates with a penicillin MIC of >0.12 (69.2%). It is interesting that A. defectiva isolates were largely of the penicillin-resistant phenotype (78 to 85%) across the country, without any significant regional bias. We were unable to perform a complete geographical analysis for G. elegans, as none of the isolates were from the Northeast.

TABLE 2.

Geographic distribution of highly penicillin-susceptible (MIC, ≤0.12), relatively penicillin-resistant (MIC, >0.12 to <0.5), and penicillin-resistant (MIC, ≥0.5) isolates (American Heart Association endocarditis treatment categories) of G. adiacens and A. defectiva

U.S. census region	% (no.) of G. adiacens isolates			% (no.) of A. defectiva isolates
	Highly penicillin susceptible	Relatively penicillin resistant	Penicillin resistant	Highly penicillin susceptible	Relatively penicillin resistant	Penicillin resistant
Midwest	30.8 (32)a	24 (25)	45.2 (47)	5.9 (2)	8.8 (3)	85.3 (29)
South	37.6 (41)b	15.6 (17)	46.8 (51)b	12.5 (4)	3.1 (1)	84.4 (27)
Northeast	57.5 (23)	15 (6)	27.5 (11)	21.1 (4)		78.9 (15)
West	41.6 (72)	22.5 (39)	35.8 (62)	16.4 (11)	6 (4)	77.6 (52)

^aP value shows significant differences (P < 0.05) in isolate susceptibility between Midwest and Northeast.

^bP value shows significant differences (P < 0.05) in isolate susceptibility between South and Northeast.

In contrast to the findings of Alberti et al. (25) but similar to the data from Mushtaq et al. (26), we found that not all NVS were susceptible to meropenem. In fact, significantly more A. defectiva isolates were susceptible to ceftriaxone than meropenem (P < 0.001). Meropenem nonsusceptibility among NVS has been described (16, 18, 26), suggesting that this difference could be due to the evaluation of different populations, as the Alberti et al. study was based on isolates from a single county whereas our isolates were referred from hospitals in 42 states. Another possible reason for this difference is that Alberti et al. used lower concentrations of lysed horse blood (LHB; 2.5% versus 5% in this study) and pyridoxal HCl (1 μg/ml versus 10 μg/ml) than those used in our study. The CLSI currently recommends 2.5 to 5% LHB and 10 μg/ml pyridoxal HCl for NVS AST. Although we saw substantially more nonsusceptibility to meropenem than did Alberti et al., most nonsusceptible isolates (84%) were intermediate rather than resistant. Further, our MIC₅₀ values were identical, and our MIC₉₀ values were the same for G. adiacens and differed by only 1 dilution for A. defectiva (Table 1). Our data were essentially identical to the meropenem values reported by Mushtaq et al., suggesting that large and broad isolate collections reveal modal MICs for meropenem that overlap the current susceptible breakpoint. Even larger multicenter studies may be required to fully resolve the normal meropenem MIC distribution for these organisms. As with penicillin and ceftriaxone, there was no significant decrease in meropenem susceptibility over the course of the study, although there was a trend for A. defectiva (P = 0.095).

Erythromycin was the least active of all tested drugs across the NVS group, especially for G. adiacens and A. defectiva, which had MIC₅₀ values of ≥8 μg/ml (25, 26). Only 39 to 53% of NVS species were susceptible to erythromycin, which is consistent with data from Alberti et al. (25). Among the erythromycin-resistant isolates, 63% (144/228) of G. adiacens isolates were susceptible to clindamycin (Table 3). A similar pattern, although with even higher clindamycin susceptibility rates, was also seen in A. defectiva (91%, 76/84 susceptible) and G. elegans (89%, 8/9 susceptible). The rate of clindamycin nonsusceptibility among erythromycin-resistant isolates of G. adiacens was similar to that previously described for NVS and other streptococci in the United States (Table 3) but was considerably lower for A. defectiva and G. elegans. The erythromycin-resistant/clindamycin-susceptible pattern has been associated with macrolide efflux and may suggest this rather than ribosomal methylation as a predominant mechanism of erythromycin resistance among A. defectiva and G. elegans isolates (15, 17). Zheng et al. found that five of eight erythromycin-resistant NVS contained the mef(A) gene, encoding macrolide efflux, whereas the remaining three contained the erm(B) gene, encoding ribosomal methylation leading to the constitutive macrolide-lincosamide-streptogramin B (cMLS_B) phenotype (15). There is currently no evidence for inducible MLS_B among NVS, likely due to the lack of reliable, standardized testing methods for the phenotype in these fastidious organisms. As with the beta-lactams, there was no significant change over time in erythromycin or clindamycin susceptibility among G. adiacens isolates (P = 0.266 and P = 0.711, respectively) or A. defectiva isolates (P = 0.174 and P = 0.803, respectively). Levofloxacin showed very good activity across all three species with >90% susceptibility, although roughly 8% of G. adiacens isolates tested resistant, similar to what was seen in recent studies (25, 26). All isolates were susceptible to vancomycin, also similar to what was seen in other studies (15–18, 25, 26).

TABLE 3.

Clindamycin interpretations among erythromycin-resistant NVS

Clindamycin interpretive categories	% (no.) of NVS species:
	G. adiacens (n = 228)	G. elegans (n = 9)	A. defectiva (n = 84)
Susceptible	63.2 (144)a	88.9 (8)	90.5 (76)
Intermediate	3.5 (8)		7.1 (6)
Resistant	33.3 (76)a	11.1 (1)	2.4 (2)

^aP < 0.0001 for differences with A. defectiva.

We examined MIC distributions for several antibiotics that do not have CLSI interpretive breakpoints (Table 1) and found similar MIC₅₀ and MIC₉₀ values for linezolid, gentamicin, and rifampin across the three species. A. defectiva showed substantially lower MIC₉₀ values for trimethoprim-sulfamethoxazole (TMP-SMX) and doxycycline than either Granulicatella species. We did not find any gentamicin MIC₉₀ greater than 4 in our study. Daptomycin MIC_50/90 values were similar to those seen by Alberti et al. and Mushtaq et al., with G. adiacens and A. defectiva showing MIC₉₀s of 4 μg/ml or greater when sufficiently high concentrations were tested (25, 26), whereas G. elegans tended to have slightly lower daptomycin MICs (Table 1).

Among this large, well-defined collection of NVS isolates, there were significant differences in susceptibility profiles across species, particularly for penicillin and ceftriaxone. This demonstrates that identification to the species level and routine susceptibility testing may help in narrowing and improving empirical therapy of NVS infections. The high penicillin MIC phenotype was significantly more common in A. defectiva across all U.S. Census regions but was also substantial among G. adiacens isolates. Because only 14% of A. defectiva and 39% of G. adiacens isolates have penicillin MICs of ≤0.12 and these two species account for approximately 97% of all NVS isolates, these data suggest that the empirical addition of gentamicin (until penicillin MICs are determined) may be reasonable if penicillin was considered for the treatment of endocarditis. Unfortunately, there is no consensus on treatment of nonendocarditis NVS infections. By expanding our understanding of the activity of a broad selection of drugs against a large collection of isolates, these data may contribute to improved treatment of NVS infections.

MATERIALS AND METHODS

AST was performed on 879 NVS isolates received from across the United States between September 2006 and August 2015. Data from the 599 nonredundant isolates that were identified to the species level as A. defectiva, G. adiacens, and G. elegans were included for analysis. Isolates of the same species from different body sites or those collected from the same site but ≥45 days from the 1st collection date were included. Only the first result from duplicate submissions on the same patient or the first of multiple morphotypes in a culture was included. Identification was performed at ARUP Laboratories by 16S rRNA gene sequencing or matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS; Bruker Biotyper, Billerica, MA) or prior to submission for AST by client laboratories. The identification of 20 client-identified isolates was confirmed at ARUP, and MIC distributions for all isolates identified by clients were analyzed by Fisher's exact and Welch's t tests and were found to be substantially equivalent to those from isolates identified by ARUP. Antimicrobial susceptibility testing was performed using custom-made broth microdilution panels (Trek, Oakwood Village, OH) in cation-adjusted Mueller-Hinton broth supplemented with 5% lysed horse blood and 0.001% pyridoxal HCl (Sigma-Aldrich, St. Louis, MO) according to CLSI M45 guidelines for fastidious organisms (14, 23). Daptomycin wells were supplemented with calcium to a final concentration of 50 μg/ml. MIC values and susceptibility interpretations were determined for penicillin, ceftriaxone, vancomycin, clindamycin, erythromycin, levofloxacin, and meropenem according to CLSI M45 (23), and MICs alone were determined for daptomycin, linezolid, gentamicin, TMP-SMX, doxycycline, and rifampin. Quality control (QC) was performed in accordance with CLSI guidelines using Streptococcus pneumoniae ATCC 49619, Staphylococcus aureus ATCC 29213, and Enterococcus faecalis ATCC 29212 (23, 27), and results were reported only if QC values were in range. MIC_50/90 values were calculated using WHONET software (v. 5.6; World Health Organization [WHO] Collaborating Centre for Surveillance of Antimicrobial Resistance [http://www.whonet.org/software.html]). Statistical analyses were performed using Fisher's exact test to compare differences in susceptibility prevalence across species, among antimicrobial agents, and across geographic regions. The Mann-Kendall test was used to analyze trends for G. adiacens and A. defectiva over the eight full calendar years in the study (2007 to 2014) in susceptibility to penicillin, ceftriaxone, meropenem, clindamycin, erythromycin, and levofloxacin.