ABSTRACT
Daptomycin has become a mainstay therapy for the treatment of serious vancomycin-resistant Enterococcus faecium infections. However, concern exists that current testing methods do not accurately predict the clinical success of daptomycin therapy. We evaluated a collection of 40 isolates of E. faecium across three centers by reference broth microdilution (BMD), and two gradient strips, to determine the precision of daptomycin MICs by these methods and the correlation of daptomycin MIC testing with mutations in the liaFSR system, one of the primary daptomycin resistance mechanisms among the enterococci. Daptomycin MICs spanned 3-log2 dilutions by BMD for 60.0% of isolates, 17.5% spanned 4 dilutions, 2.5% spanned 5 dilutions, and 20.0% spanned 6 or more dilutions. Fifteen isolates had MICs interpreted as susceptible by some tests and nonsusceptible by others. Neither BMD nor gradient diffusion tests could reliably differentiate isolates with or without mutations in liaFSR, resulting in a 59.8% very major error rate compared to determination of genotype by BMD, 63.5% by Etest, and 68.5% by MIC test strip. Imprecision in daptomycin MIC determination for E. faecium make establishment of a revised breakpoint challenging. Clinicians should be aware of this testing variability when making treatment decisions for patients with serious infections caused by this organism.
INTRODUCTION
The treatment of serious infections caused by Enterococcus spp. is difficult due to the intrinsic resistance of these organisms to many antibiotic classes. Traditionally, the synergistic combination of ampicillin/penicillin or vancomycin with an aminoglycoside has been a reliable regimen for severe infections, including endocarditis. However, E. faecium resistant to these agents has become widespread, leaving very few treatment options (1). Daptomycin is a cyclic lipopeptide antibiotic with in vitro bactericidal activity against E. faecium, including vancomycin-resistant isolates. While daptomycin does not have a clinical indication for the treatment of vancomycin-resistant enterococci, in vitro susceptibility of Enterococcus spp. is defined by an MIC of ≤4 μg/ml by the Clinical and Laboratory Standards Institute (CLSI) (2). The term daptomycin “nonsusceptible” is used for isolates with MICs of >4 μg/ml, since an MIC breakpoint that defines resistance has not been established to date due to the relatively rare occurrence of such isolates and lack of corresponding clinical outcome data at the time the susceptible breakpoint was defined (3). However, since the breakpoint was established, isolates of Enterococcus spp. with daptomycin MICs of >4 μg/ml have been increasingly reported from both patients with prior daptomycin exposure (4–7), as well as those without (8). Most importantly, cases of treatment failures have also been reported in patients infected with daptomycin-susceptible isolates with MICs in the higher end of the susceptible range (2 to 4 μg/ml) (9, 10). In vitro studies suggest that such isolates are daptomycin tolerant (11, 12), and recent clinical data supported that patients infected with isolates with MICs of 3 to 4 μg/ml (Etest) are likely to fail daptomycin therapy (9). In particular, the presence of substitutions in the LiaFSR three-component system, predicted to regulate cell membrane stress response in enterococci, is associated with reduced daptomycin susceptibility and tolerance as well as clinical failures of therapy (9, 11), although other resistance mechanisms have also been described. Combined, these data provoked the CLSI to reevaluate daptomycin breakpoints for the enterococci; the discussions for this are ongoing. One primary challenge to lowering the Enterococcus daptomycin-susceptible breakpoint is the fact that the modal daptomycin MIC for wild-type isolates of E. faecium is 2 to 4 μg/ml (3). As such, lowering the susceptible breakpoint may bisect this wild-type MIC distribution and render phenotypic tests, including the reference broth microdilution (BMD), incapable of differentiating daptomycin-susceptible from daptomycin-nonsusceptible isolates. In this study, we evaluated the reproducibility of daptomycin MICs for E. faecium isolates by three antimicrobial susceptibility testing methods, i.e., broth microdilution, Etest, and MTS, and the association of mutations in the liaFSR system with MIC testing results.
RESULTS
Broth microdilution testing.Forty isolates of E. faecium recovered from the blood of patients at multiple institutions across the United States were evaluated. Ten isolates harbored multiple mutations conferring reduced daptomycin susceptibility (liaFSR or other) and were recovered from patients with daptomycin treatment failures (modal MIC, >32 μg/ml, referred to as “frankly resistant” isolates). The remaining 30 isolates were susceptible by the current CLSI breakpoint to daptomycin upon initial testing: 16 of these isolates harbored mutations in the liaFSR system, including 10 isolates recovered from patients with daptomycin treatment failures, and 14 did not have mutation in the liaFSR system. These isolates are outlined in Table S1 in the supplemental material. The distribution of MICs obtained by CLSI reference BMD representing three brands of cation-adjusted Mueller-Hinton broth (CA-MHB) and three laboratories are shown in Table 1. Daptomycin MICs spanned the expected 3-log2 dilution range of reference BMD (i.e., were in essential agreement) for only 60.0% of isolates. An additional 17.5% spanned 4 dilutions (n = 7), 2.5% spanned five dilutions (n = 1), 12.5% spanned six dilutions (n = 5), and 7.5% spanned seven dilutions (n = 3). Fifteen isolates had MIC distributions that spanned the current CLSI susceptible and nonsusceptible categories, resulting in the isolates being interpreted as susceptible by some tests and nonsusceptible by others (Table 1).
Distribution of daptomycin BMD MICs, by isolate evaluated, across three laboratories and three brands of CA-MHBa
No large differences in MICs were noted between CA-MHB from Difco and BBL, although Oxoid CA-MHB resulted in MICs one dilution higher on average, near the 4-μg/ml breakpoint (Fig. 1). This was most apparent when evaluating the proportion of isolates with MICs in the 2 to 8 μg/ml range (Fig. 1). Similarly, when each laboratory was evaluated, laboratory A demonstrated an MIC skew toward lower MICs compared to laboratories B and C (Fig. 2). Colony counts were within expected ranges for all laboratories.
BMD MIC distribution, by testing media. White bars represent BBL, gray bars represent Difco, and black bars represent Oxoid brands of CA-MHB.
BMD MIC distribution, by testing site. White bars represent site A, gray bars represent site B, and black bars represent site C.
Interpretation of BMD MICs by current and alternative MIC breakpoints.For the isolates with initially susceptible MICs, BMD did not differentiate those with liaFSR mutation from those without (Table 1), even when the concentration of 1.5 μg/ml daptomycin was evaluated (not shown). The modal MIC for both populations was 4.0 μg/ml (Table 2). Interestingly, susceptible isolates with liaFSR mutations more often displayed MICs ≤ 0.5 μg/ml (20.1% of tests) than did isolates without this mutation (6.3% of tests), a difference which reached statistical significance (P < 0.01, t test).
BMD results for 40 isolates of E. faecium, as determined by BMD across three testing laboratories
The current breakpoint of ≤4.0 μg/ml resulted in 16.7% (n = 15/90) of the MICs on frankly resistant isolates to be interpreted as susceptible (very major errors [VME]). If the presence of a mutation to liaFSR was used to categorize susceptible isolates as “resistant,” based on data that demonstrate such isolates are daptomycin tolerant (11, 12), 86.8% (n = 125/144) of the tests for the isolates with liaFSR mutation were susceptible. Combining these two organism groups (i.e., frankly resistant and those with a mutation to liaFSR) resulted in an overall 59.8% VME rate (Table 3). In contrast, 4.8% of the tests yielded MICs interpreted as nonsusceptible for susceptible isolates without mutation to these genes (major errors [ME]). The use of alternative breakpoints did not allow for differentiation between these two groups. For example, while a breakpoint of 1.5 μg/ml resulted in only 3.3% VME for isolates with resistance, this came at the cost of 74.0% ME for susceptible isolates with no mutation to liaFSR (not shown). A breakpoint of ≤2 μg/ml would result in 32.1% VME and a breakpoint of ≤1 μg/ml in 15.8% VME. The ME rates for these breakpoints were 15.1 and 73.8%, respectively (Table 3).
Assessment of alternative MIC breakpoints for daptomycin
At present, an intermediate breakpoint does not exist for daptomycin. However, applying a breakpoint of ≤2/4/≥8 μg/ml (susceptible/intermediate/resistant [S/I/R]) would result in 32.1% VME and 4.8% ME, and a breakpoint of ≤1/2–4/≥8 μg/ml (S/I/R) would result in 15.8% VME and 4.8% ME. The results were not impacted by brand of CA-MHB or testing laboratory (data not shown).
Gradient diffusion tests.Similar to the observations for BMD, neither gradient strip adequately differentiated the different isolates by resistance mechanism. The MIC modes were 2 μg/ml (susceptible isolates without mutation), 2 μg/ml (susceptible isolates with mutation), and 8 to 12 μg/ml (frankly resistant isolates). No breakpoint was both sensitive and specific to differentiate the genotype-predicted susceptible isolates from genotype-predicted resistant isolates (Table 4).
Gradient diffusion strip results, by isolate group
The Mueller-Hinton agar (MHA) brand was evaluated for the two laboratories that performed testing using multiple brands. Differences in the mode were observed, with BD MHA yielding an MIC mode one dilution higher for both Etest (Fig. 3) and MTS (data not shown) than MICs obtained on Remel (Lenexa, KS) or Hardy (Santa Maria, CA) MHA, although the differences in the MIC were not significant (P > 0.5) (Fig. 3).
MIC distribution of E. faecium isolates, as determined by Etest, on three brands of MHA.
DISCUSSION
An organism is defined as wild type by the presumed absence of detectable acquired resistance mechanism to the agent in question, a feature that is typically measured phenotypically, by reference BMD. It is common for this wild-type MIC distribution to span 3- to 5-log2 dilutions due to the inherent variability of both the organisms and the test method. This characteristic of phenotypic susceptibility testing becomes a challenge when pharmacokinetic/pharmacodynamic modeling predicts that a breakpoint splits the population, resulting in the wild type being both susceptible and resistant to the agent in question. The present study demonstrates the dilemma of this scenario, i.e., susceptibility results that are not reproducible, even by the reference BMD method. While the precedence is to allow pharmacokinetic/pharmacodynamic predictions to trump genotype testing, in the case of daptomycin and E. faecium, clinical outcome data suggest that genotype may in fact predict treatment success (or failure) for some patients (9, 11). We were unable to define any breakpoint that would successfully and reproducibly differentiate isolates that are genotypically “resistant” from those that were susceptible. Similarly, we were unable to identify a cutoff by Etest or MTS, despite prior data suggesting that these tests are superior to BMD for testing enterococci (11). Of particular concern in the present study was the difficulty in obtaining reproducible MICs for the majority of isolates in the study, including those considered “frankly resistant,” i.e., those with high modal MICs and recovered from patients with daptomycin treatment failures. However, 16.7% of the replicate BMD MICs were ≤4 μg/ml, which is considered susceptible by the current CLSI breakpoint (2). Of note, isolate 31 was particularly problematic, having all MICs ≤4 μg/ml (Table 1).
Although changes in liaFSR are highly associated with daptomycin nonsusceptibility, an important caveat is that there are other genetic pathways associated with daptomycin resistance that were not included in the analysis of susceptible isolates (13), which may have resulted in our underestimating the VME rate. In addition, it cannot be overemphasized that the current standard is to not use resistance mechanisms to predict treatment success or failure, but rather clinical outcomes and pharmacokinetic/pharmacodynamics modeling, based on the MIC. However, in this instance, MICs for the isolates investigated were neither accurate nor precise (Table 1), suggesting an alternative testing strategy is needed. This strategy is yet to be defined, but it may include genotypic approaches or alternative phenotypic testing methods, such as altered media, testing conditions, or a combination of methods. Along these lines, we noted a difference in daptomycin MICs, as measured by both MTS and Etest, based on the brand of MHA. Anecdotally, both testing laboratories that utilized Remel and Hardy MHA commented on poor growth of the enterococci evaluated in this study on these media, which may have contributed to the lower MICs observed compared to MHA purchased from BD. However, we did not systematically evaluate this effect. A second limitation to the present study is that we did not evaluate susceptibility by other methods used commonly by clinical laboratories, such as the automated susceptibility test systems, although we do not anticipate that performance would be better on these systems.
The CLSI is evaluating revision to the enterococcal daptomycin breakpoint, with many indications that this should be lowered. However, the findings of the present study present a significant challenge to establishing a new daptomycin breakpoint for Enterococcus spp. Minimizing the incidence of VMEs, compared to genotypic resistance data, by selecting a breakpoint of ≤1 μg/ml, significantly reduces the utility of daptomycin for the enterococci, since roughly 75% of isolates have daptomycin MICs of >1 μg/ml (3). Introduction of an intermediate breakpoint of 2 to 4 μg/ml, at which higher doses and/or combination therapy must be used, is one option. However, synergy between daptomycin and ampicillin, ceftaroline, or possibly other beta-lactams may only be observed in vitro for those isolates with changes in liaFSR (14), and clinical data are currently lacking. In addition, treatment failures even at maximal doses of 12 mg/kg have been documented for patients with isolates that tested susceptible initially but that harbored resistance mutations (10). The opposite to this scenario is also possible, i.e., successful treatment of isolates that harbor these resistance mutations, particularly in infections with low inocula. Further work is needed to elucidate the best testing method for daptomycin and the enterococci and in order to determine the best approach to testing (if any) for antimicrobial agents where the breakpoint bisects the wild-type population.
MATERIALS AND METHODS
Bacterial isolates.Forty isolates of E. faecium were evaluated in this study, isolated from five centers across the United States (see Table S1 in the supplemental material). These were selected by daptomycin MIC, determined by BMD at or near the time of isolation from the patient, to encompass 30 with “low” daptomycin MICs (≤4 μg/ml) and 10 with frank resistance to daptomycin (MIC > 32 μg/ml on initial testing). The latter group was comprised of isolates from patients who experienced daptomycin treatment failures, as described in the references listed in Table S1 in the supplemental material. Isolates in the susceptible category were subjected to sequencing of the liaFSR region, as described previously (14). Isolates were stored frozen at −70°C, in brucella broth plus 10% glycerol (Hardy) and were shipped to testing sites, coded. Upon receipt, the isolates were subcultured twice from frozen stock on sheep blood agar prior to testing.
Daptomycin susceptibility testing.BMD was performed using frozen-form BMD panels manufactured by Thermo Fisher (Lenexa, KS). Each panel was prepared using CA-MHB supplemented with 50 mg/liter CaCl2 from three different manufacturers (Difco and BBL, BD, Sparks, MD; and Oxoid, Thermo Fisher, Lenexa, KS). Daptomycin concentrations were 128 to 0.12 μg/ml, in doubling dilutions. In addition, 1.5 and 3.0 μg/ml of daptomycin were included in the panels to match the gradient diffusion strip MICs within the high end of the susceptibility range. MIC testing was performed according to CLSI standards at each site.
Gradient diffusion testing was performed at each site using strips from two manufacturers (Etest, bioMérieux, Marcy l'Etoile, France; MTS, Liofilchem, Italy). MHA plates from three manufacturers (BD, Remel, and Hardy) were evaluated by two laboratories. BD MHA alone was used at the third laboratory. Gradient diffusion strip testing was performed according to the respective manufacturer's instructions.
The quality control strains, Staphylococcus aureus ATCC 29213 and Enterococcus faecalis ATCC 29212, were included in all testing.
Study design and data analysis.E. faecium isolates were randomized, coded, and shipped to three testing laboratories. MIC tests, encompassing nine variables for two laboratories (BMD with three brands of CA-MHB, Etest with three brands of MHA, and MTS with three brands of MHA) and five variables for one laboratory (BMD with three brands of CA-MHB and Etest and MTS with one brand of MHA) was performed in parallel, at each site, using the same 0.5 McFarland standard preparation. Colony counts were performed at each site to confirm the concentration of bacteria in the inocula for each test. The numbers of replicate MICs that were categorized as susceptible, with the current CLSI breakpoint of ≤4 μg/ml, and other theoretical breakpoints were evaluated. In addition, the ability of the breakpoints to differentiate isolates with liaFSR mutations and no liaFSR mutations was assessed. In the latter scenario, the presence of an liaFSR mutation was used to define “resistance,” and both major errors and very major errors were calculated. In these calculations, the numbers of replicate MICs measured for an isolate without or with liaFSR mutations were used as the denominator.
FOOTNOTES
- Received 13 April 2018.
- Returned for modification 25 May 2018.
- Accepted 6 June 2018.
- Accepted manuscript posted online 25 June 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.00745-18.
REFERENCES
- Copyright © 2018 American Society for Microbiology.
