The use of rapid diagnostic tests (RDTs) for blood cultures has become standard of care in the United States to inform early antimicrobial optimization. The relative ability of genotypic and phenotypic approaches to identify beta-lactam susceptibility in Escherichia coli, Klebsiella spp., and Proteus mirabilis was evaluated, using incidence rates of resistance mechanisms to third-generation cephalosporins, aztreonam, and piperacillin-tazobactam seen across U.S. census regions.
KEYWORDS: ESBL, KPC, NDM, ceftazidime, ceftriaxone, imipenem, meropenem, piperacillin-tazobactam
ABSTRACT
The use of rapid diagnostic tests (RDTs) for blood cultures has become standard of care in the United States to inform early antimicrobial optimization. The relative ability of genotypic and phenotypic approaches to identify beta-lactam susceptibility in Escherichia coli, Klebsiella spp., and Proteus mirabilis was evaluated, using incidence rates of resistance mechanisms to third-generation cephalosporins, aztreonam, and piperacillin-tazobactam seen across U.S. census regions. Overall, the presence of CTX-M, KPC, and/or NDM genes was 81% (range, 57 to 87%) sensitive for the prediction of ceftriaxone, ceftazidime, and aztreonam resistance and 73% (range, 25 to 90%) sensitive for the detection of piperacillin-tazobactam resistance. The sensitivity of KPC or NDM to predict imipenem or meropenem resistance was 94.3% overall, and for meropenem ranged from 70 to 100% across U.S. census regions. Institutions that use genotypic RDTs to inform therapeutic de-escalation decisions should be aware of the incidence-base performance across U.S. geographies and in different patient populations, where resistance rates may vary.
INTRODUCTION
The use of rapid diagnostic tests (RDTs) for blood cultures to inform early antimicrobial optimization has become the new standard of care in the United States. Several studies have concluded that bloodstream infection (BSI) mortality is directly linked to inadvertent use of inactive empirical antimicrobial therapy (1–3). Further, the value of rapid de-escalation for patients who are treated with overly broad empirical therapy is becoming increasingly apparent, including reduction in hospital length of stay, hospital costs and adverse antimicrobial effects, such as toxicity, infection with multidrug-resistant pathogens, and Clostridioides difficile infection (4, 5).
Recently, data from the MERINO trial brought into question the validity of one of the most commonly prescribed empirical and definitive therapy options for Gram-negative BSI, piperacillin-tazobactam (TZP) (6). MERINO was an international noninferiority open-label randomized clinical trial to evaluate the primary endpoint of 30-day all-cause mortality associated with TZP versus meropenem as definitive therapy for the treatment of BSI caused by ceftriaxone (CRO)-nonsusceptible (intermediate or resistant) Escherichia coli or Klebsiella pneumoniae BSIs. The study found that TZP was associated with an all-cause 30-day mortality of 12.3% (23/187), compared to 3.7% (7/191) for patients treated with meropenem. The absolute risk difference was 8.6% (one-sided 97.5% confidence interval, −∞ to 14.5%; P = 0.90 for noninferiority). Understanding the ramifications of this study to current practices is challenging (7). For instance, the dosing regimen of TZP in the study (4.5 g every 6 h) do not reflect those used by all institutions. Similarly, the epidemiology of E. coli and Klebsiella spp. may be different in the regions where the MERINO trial was conducted (Australia, Asia, and southern Europe) than in other regions. Nonetheless, the outcomes of the MERINO study came as a surprise to many, particularly since the patient population studied was on the lower end of the acuity spectrum, and many agree that TZP should no longer be used as definitive therapy for isolates of E. coli or K. pneumoniae that are not susceptible third-generation cephalosporins (7).
Knowledge of whether a patient’s infecting organism is not susceptible to ceftriaxone (or other empirically used beta-lactams) as soon as possible is therefore needed to inform escalation to a carbapenem or, conversely, de-escalation from a carbapenem if this was used empirically. At present, in the United States, two RDTs cleared by the U.S. Food and Drug Administration (FDA) that might be used to determine an isolate’s susceptibility to ceftriaxone directly from positive blood cultures. These are a molecular test (Verigene GN; Luminex), which detects the presence of CTX-M, KPC, NDM, IMP, VIM, and Oxa-48-like beta-lactamases, and a phenotypic test (Accelerate PhenoTest BC kit; Accelerate Diagnostics, Inc., Tucson AZ), which specifically determines ceftriaxone, ceftazidime, aztreonam, TZP, and meropenem MICs, among those for other antimicrobial agents. The relative ability of a genotypic approach to identify ceftriaxone-, ceftazidime-, aztreonam-, and TZP-nonsusceptible phenotypes has not been evaluated across the varied epidemiology of E. coli and Klebsiella spp. seen in the United States. The intention of the present study was to evaluate the relative rates of ceftriaxone-, ceftazidime-, aztreonam-, and TZP-nonsusceptible E. coli, K. pneumoniae, Klebsiella oxytoca, and Proteus mirabilis in the United States that might be predicted using a genotype evaluation. Prediction of imipenem and meropenem resistance was also evaluated by the genotype approach.
MATERIALS AND METHODS
Raw data were obtained from JMI Laboratories for their 2012 surveillance study of extended spectrum beta-lactamase (ESBL), AmpC, and carbapenemase enzyme prevalence across the United States (8). This survey evaluated 5,739 E. coli (n = 2,767), Klebsiella spp. (n = 2,289), and Proteus mirabilis (n = 683) isolates collected from 72 U.S. hospitals in the 2012 calendar year. One isolate per patient was included, and the isolates were recovered from blood, respiratory tract, intraabdominal, skin and skin structure, and urinary tract sources. Isolates that met Clinical and Laboratory Standards Institute (CLSI) screening criteria for the presence of an ESBL (i.e., MIC > 1 μg/ml to one or more of ceftriaxone, ceftazidime, or aztreonam) as determined by CLSI reference broth microdilution (9) were tested for the presence of ESBL (CTX-M, SHV, and TEM), AmpC (ACC, ACT/MIR, CMYI/MOX, CMYII, DHA, and/or FOX), KPC and NDM beta-lactamases using a Check-MDR CT101 kit (Check-Points, Wageningen, Netherlands). In total, 747 isolates met these criteria and comprise the data set for the present evaluation.
Analytical performance characteristics (sensitivity) of a genotype approach that used detection of CTX-M, KPC, and NDM were evaluated. Because some have endorsed use of ceftriaxone, ceftazidime, or aztreonam for isolates of Klebsiella or E. coli that test negative for CTX-M or carbapenemase (10), isolates that did not harbor these genes but were not susceptible to these antimicrobials were classified as very major errors (VME). Isolates that were positive for CTX-M, NDM, and/or KPC were considered to be in categorical agreement with a nonsusceptibility phenotype to these antimicrobials. Major errors (i.e., false resistance) were not calculated, since the isolates were only tested for presence of beta-lactamases if reduced susceptibility to ceftriaxone, ceftazidime, and/or aztreonam was documented (i.e., was biased against this measurement). Meropenem and imipenem phenotypic results were compared to NDM and KPC results.
Performance was then extrapolated across different rates of resistance, ranging from 5 to 50%. This was done by calculating the number of resistant isolates that would be predicted to be susceptible in a hypothetical sampling of 100 isolates at each prevalence rate. Upper and lower confidence limits of a 95% exact binomial confidence interval were calculated. Average national rates and observed rates by census region were utilized for these analyses.
RESULTS
The composite performance results are shown in Table 1 for ceftriaxone, ceftazidime, aztreonam, and TZP results and in Table 2 for imipenem and meropenem. Overall, 81% of isolates that were ceftriaxone, ceftazidime, or aztreonam resistant were positive for one or more of CTX-M, KPC, or NDM, indicating a 19% VME rate, if a negative result for these genes was used to predict susceptibility to these antimicrobials (Table 1). We also evaluated the frequency of isolates that were negative for all of these genes but resistant to TZP; 27% of the isolates met these criteria. Including isolates with intermediate MICs to these antimicrobials did not significantly reduce the performance of the genotypic prediction method for the beta-lactams evaluated (Table 1). Among the non-CTX-M resistance mechanisms that accounted for resistance to these antimicrobials, SHV ESBLs were the most common, followed by AmpC. Between 12 and 27% of the isolates resistant to ceftriaxone (27%), ceftazidime (12%), and/or aztreonam (25%) and negative for CTX-M were also negative for SHV, TEM, and AmpC (Table 1). For TZP-resistant isolates, 34.2% of CTX-M/KPC/NDM-negative isolates harbored SHV, and 53.4% were negative for all beta-lactamases tested.
TABLE 1.
Utility of a genotypic result of CTX-M/KPC/NDM to predict ceftriaxone, ceftazidime, aztreonam, and piperacillin-tazobactam resultsa
| Phenotype and drug treatment |
No. negative for CTX-M, KPC, and NDM | No. positive for CTX-M, KPC, and/or NDM | % VME | If negative for CTX-M, KPC, and NDM, presence of other BLA, n (%) |
|||
|---|---|---|---|---|---|---|---|
| AmpC | TEM | SHV | Neg | ||||
| Resistant | |||||||
| CRO | 127 | 556 | 18.6 | 43 (33.9) | 1 (0.8) | 49 (38.6) | 34 (26.8) |
| CAZ | 92 | 404 | 18.5 | 34 (37.0) | 1 (1.1) | 46 (50.0) | 11 (12.0) |
| TZP | 73 | 202 | 26.5 | 7 (9.6) | 2 (2.7) | 25 (34.2) | 39 (53.4) |
| ATM | 114 | 501 | 18.5 | 34 (29.8) | 1 (0.9) | 50 (43.9) | 29 (25.4) |
| Nonsusceptible | |||||||
| CRO | 145 | 557 | 20.7 | 49 (33.8) | 1 (0.6) | 53 (36.6) | 42 (29.0) |
| CAZ | 109 | 436 | 20.0 | 39 (35.7) | 1 (0.9) | 49 (45.0) | 20 (18.3) |
| TZP | 93 | 250 | 27.1 | 30 (32.3) | 3 (0.3) | 16 (17.2) | 44 (47.3) |
| ATM | 139 | 533 | 20.7 | 40 (28.8) | 1 (0.7) | 51 (36.7) | 47 (33.8) |
CRO, ceftriaxone, CAZ, ceftazidime, TZP, piperacillin-tazobactam, ATM, aztreonam, BLA, beta-lactamase.
TABLE 2.
Utility of a genotypic result of KPC/NDM to predict imipenem and/or meropenem resistance
| Drug treatmenta | No. negative for KPC and NDM | No. positive for KPC or NDM | % VME | If negative for KPC and NDM, presence of other BLAs, no. (%)b
|
|||
|---|---|---|---|---|---|---|---|
| AmpC | TEM | SHV | CTX-M | ||||
| Imipenem, R | 3 | 150 | 2.0 | 0 (0) | 0 (0) | 0 (0) | 2 (66.7) |
| Meropenem, R | 8 | 145 | 5.2 | 0 (0) | 0 (0) | 0 (0) | 4 (50.0) |
| Imipenem, NS | 19 | 152 | 11.1 | 4 (44.0) | 0 (0) | 0 (0) | 11 (57.9) |
| Meropenem, NS | 13 | 152 | 7.9 | 0 (0) | 0 (0) | 1 (7.7) | 9 (69.2) |
| Carbapenem, R (either) | 9 | 149 | 5.7 | 0 (0) | 0 (0) | 0 (0) | 6 (66.7) |
| Carbapenem, NS (either) | 25 | 153 | 14.4 | 4 (16.0) | 0 (0) | 1 (4.0) | 16 (64.0) |
R, resistant; NS, nonsusceptible.
BLA, beta-lactamase.
The specific MICs to ceftriaxone, ceftazidime, and aztreonam were evaluated for the isolates that were not susceptible to these antimicrobials (Fig. 1). Nearly all isolates positive for CTX-M, KPC, and/or NDM had a ceftriaxone MIC of >8 μg/ml, and only 59.3% of resistant isolates negative for CTX-M, KPC, and NDM had a ceftriaxone MIC of >8 μg/ml. In comparison, 92.7% of isolates positive for CTX-M, KPC, and NDM had a ceftazidime MIC of ≥16 μg/ml, as did 84.4% of isolates that were negative for these genes but resistant to ceftazidime (Fig. 1). Among the isolates positive for CTX-M, KPC and/or NDM, 24 (4.3%) isolates were susceptible to aztreonam, 121 (21.7%) were susceptible to ceftazidime, and 0 (0%) were susceptible to ceftriaxone.
FIG 1.
MIC distribution for isolates with CTX-M, KPC, and/or NDM (A, ceftriaxone; B, ceftazidime) and KPC and/or NDM (C, imipenem; D, meropenem).
Use of KPC or NDM to predict resistance to imipenem and meropenem are shown in Table 2. Overall, 153 isolates were positive for KPC or NDM, and none were positive for both genes. The sensitivity to detect imipenem resistance/nonsusceptibility MICs was 98.0%/88.9%, and meropenem resistance/nonsusceptibility MICs was 94.8%/92.1%. The ability of the genotype to detect resistance/nonsusceptibility MICs to either imipenem or meropenem was 94.3%/86.0%. MIC distributions for KPC/NDM-negative isolates are shown in Fig. 1. In general, meropenem MICs were higher for KPC/NDM-negative isolates than were imipenem MICs, and no KPC/NDM-negative isolates had an imipenem or meropenem MIC of >8 μg/ml. CTX-M was the only beta-lactamase detected in KPC/NDM-negative, imipenem- or meropenem-resistant isolates (Table 2). Some isolates that were not nsusceptible to imipenem harbored AmpC genes alone.
Performance of genotypic method across different ceftriaxone, piperacillin-tazobactam, and meropenem nonsusceptibility rates.
The number of VME associated with using the genotypic method to predict ceftriaxone, TZP, and meropenem susceptibility was evaluated across different prevalences of resistance to these antimicrobials (Table 3). These data demonstrate the number of VME a laboratory may expect per 100 isolates tested. The numbers of VME were 1 to 10.3 for ceftriaxone, 1.3 to 13.3 for TZP, and 0.4 to 4 for meropenem across a resistance rate of 5 to 50%, respectively. FDA guidance for susceptibility testing allows for 0 to 1 VME across these prevalence rates (11).
TABLE 3.
Modeled performance of CTX-M/KPC/NDM to predict nonsusceptible E. coli, Klebsiella species, and Proteus mirabilis across different prevalences of resistancea
| Resistance (%) | No. of VMEb | Ceftriaxone |
TZP |
Meropenem |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| VME (n) | LCL | UCL | VME (n) | LCL | UCL | VME (n) | LCL | UCL | ||
| 5 | 0 | 1.03 | 0.23 | 1.19 | 1.33 | 1.28 | 1.37 | 0.39 | 0.23 | 0.65 |
| 10 | 0 | 2.07 | 0.47 | 2.38 | 2.65 | 2.57 | 2.74 | 0.79 | 0.47 | 1.30 |
| 15 | 0 | 3.10 | 0.70 | 3.57 | 3.98 | 3.85 | 4.11 | 1.18 | 0.70 | 1.95 |
| 20 | 0 | 4.13 | 0.93 | 4.76 | 5.31 | 5.14 | 5.48 | 1.58 | 0.93 | 2.60 |
| 25 | 0 | 5.16 | 1.17 | 5.95 | 6.64 | 6.42 | 6.85 | 1.97 | 1.17 | 3.25 |
| 30 | 0 | 6.20 | 1.40 | 7.14 | 7.96 | 7.70 | 8.22 | 2.36 | 1.40 | 3.90 |
| 35 | 0 | 7.23 | 1.63 | 8.33 | 9.29 | 8.99 | 9.59 | 2.76 | 1.63 | 4.55 |
| 40 | 0 | 8.26 | 1.86 | 9.52 | 10.62 | 10.27 | 10.96 | 3.15 | 1.86 | 5.20 |
| 45 | 0 | 9.29 | 2.10 | 10.71 | 11.95 | 11.56 | 12.33 | 3.55 | 2.10 | 5.86 |
| 50 | 1 | 10.33 | 2.33 | 11.90 | 13.27 | 12.84 | 13.71 | 3.94 | 2.33 | 6.51 |
LCL, lower confidence limit; UCL, upper confidence limit; TZP, piperacillin-tazobactam; VME, very major error (total number expected per 100 isolates tested).
The number of VME accepted by the FDA.
The performance of the genotypic method across geographic areas was evaluated, since beta-lactamase gene prevalence varies geographically across the United States. Among the ceftriaxone-nonsusceptible isolates tested, 57.1 to 87.8% harbored CTX-M, KPC, and/or NDM, with the lowest proportion in the New England region and the highest in the Mid-Atlantic region (Table 4). For meropenem, sensitivity of detection of KPC/NDM to predict meropenem nonsusceptibility ranged from 69.5% (West South Central) to 100% (South Atlantic and East South Central). Use of CTX-M, KPC, and/or NDM to predict TZP resistance ranged from 25% (West North Central) to 90.4% (East North Central).
TABLE 4.
Predicted performance of genotypic approach across U.S. geographic regionsa
| Census region | Ceftriaxone |
Meropenem |
TZP |
|||
|---|---|---|---|---|---|---|
| NS (n) | % VME | NS (n) | % VME | NS (n) | % VME | |
| New England | 42 | 42.86 | 4 | 25.00 | 23 | 60.87 |
| Mid-Atlantic | 255 | 12.16 | 116 | 3.45 | 166 | 9.64 |
| East North Central | 80 | 25.00 | 13 | 7.69 | 31 | 41.94 |
| West North Central | 27 | 33.33 | 0 | NA | 8 | 75.00 |
| South Atlantic | 38 | 28.95 | 5 | 0.00 | 21 | 47.62 |
| East South Central | 51 | 27.45 | 1 | 0.00 | 12 | 66.67 |
| West South Central | 131 | 18.32 | 23 | 30.43 | 56 | 23.21 |
| Mountain | 23 | 26.09 | 2 | 0.00 | 7 | 57.14 |
| Pacific | 55 | 21.82 | 1 | 0.00 | 19 | 47.37 |
| Overall | 702 | 20.66 | 165 | 7.27 | 343 | 27.11 |
The genotypic approach includes detection of CTX-M, KPC, and/or NDM for ceftriaxone and TZP, KPC, or NDM for meropenem. NS, nonsusceptible; NA, not applicable.
DISCUSSION
Resistance to ceftriaxone, ceftazidime, aztreonam, and carbapenems in the Enterobacteriaceae is complex but driven primarily by the activity of extended-spectrum beta-lactamases (CTX-M, SHV, or TEM types), AmpC beta-lactamases, and/or carbapenemases, either alone or in combination with permeability deficiencies (12, 13). In 2010, the CLSI reduced the MIC and disk diffusion clinical breakpoints for the third-generation cephalosporins and monobactam to more accurately predict clinical outcomes (14). This change eliminated the need to perform phenotypic ESBL testing, which was previously recommended for isolates of E. coli and Klebsiella spp. with elevated (>1 μg/ml) MICs to ceftriaxone, cefotaxime, ceftazidime, or aztreonam or isolates of P. mirabilis with elevated (>1 μg/ml) MICs to cefpodoxime, ceftazidime, or cefotaxime. Importantly, the phenotypic ESBL test recommended by the CLSI does not detect AmpC resistance mechanisms and may yield false-negative results if an isolate expresses both an ESBL and AmpC (14).
Results from the MERINO trial have reinvigorated discussions regarding the need for ESBL testing for E. coli and Klebsiella spp. isolated from blood by demonstrating that meropenem is associated with improved outcomes compared to TZP for the definitive therapy for CRO-nonsusceptible bloodstream infections, many of which are due to ESBL expression. Importantly, the primary endpoint of 30-day all-cause mortality did not correlate to TZP MIC, which is in contrast to other antimicrobial agents where MIC correlates better than resistance mechanism with outcome (15). However, the MERINO trial itself, which was conducted outside the United States, reinforced the limitations of both phenotypic and genotypic ESBL detection methods to predict ceftriaxone nonsusceptibility results. Of the isolates evaluated in the study, 14% did not generate a positive phenotypic ESBL test. Furthermore, among 293 isolates available for whole-genome sequencing, 14.7% did not harbor a CTX-M, TEM, or SHV ESBL, and only 71.3% harbored a CTX-M gene, which is the only ESBL detected by FDA-cleared genotypic tests performed on positive blood cultures. Many isolates in the MERINO study harbored OXA-1, which is associated with elevated TZP MICs (6), but this gene has not been detected by any FDA-cleared test.
The prevalence of ESBLs in the current U.S. landscape is not well defined. The most recent data on the prevalence of beta-lactamases in the United States is from surveillance studies conducted by JMI laboratories in 2012, 2014, and 2016 (8, 16, 17). Evaluation of the 2012 surveillance data, in the present study, demonstrated that between 57 and 88% of ceftriaxone-nonsusceptible isolates harbor CTX-M, KPC, and/or NDM genes (Table 4); in other words, the absence of these genes would yield a false-susceptible interpretation for 12 to 43% of ceftriaxone-nonsusceptible isolates (8). Performance of a genotypic method to predict meropenem nonsusceptibility results with the use of KPC and/or NDM was more reliable, yielding a 92% sensitivity overall for resistance (8% VME). However, this, too, was dependent on geography, and ranged from 70% detection to 100% detection (Table 4). The value of these tests to predict TZP nonsusceptibility results was similarly influenced by epidemiology; in the Mid-Atlantic region, where KPC is a prevalent mechanism of resistance, 90.4% of TZP-resistant isolates were positive for CTX-M or KPC, whereas in other regions the sensitivity was lower (Table 4).
While the intended use of genotype methods is not to extrapolate resistance markers to phenotypic susceptibility results, this is often done in clinical practice, and the absence of beta-lactamase gene may be used to prompt antimicrobial de-escalation (10). According to FDA guidance, the acceptable number of VME is based on prevalence and ranges from 0 to 1 for phenotypic antimicrobial susceptibility tests (11). The acceptable error rate for VME using genotypic approaches has not been defined. Genotype tests on market are highly sensitive for detection of CTX-M: clinical trial summary data demonstrated 50/52 (96%) in prospective samples and 151/153 (99%) of all samples that were positive for CTX-M were detected. Hence, the primary limitations of these tests are not the ability to detect targeted genes but rather CRO nonsusceptibility phenotypes driven by other resistance mechanisms. In this data set, the most common mechanisms among CTX-M-negative, ceftriaxone-nonsusceptible isolates were ESBL SHV variants (36.6%) and AmpC (33.8%, Table 1). However, a nearly equal number of isolates were negative with regard to all of the queried beta-lactamases (29.0%), which again reinforces the complex nature of resistance to ceftriaxone and other beta-lactams in Enterobacteriaceae. It should be noted 19/42 isolates negative for CTX-M, SHV, TEM, AmpC, KPC, and NDM enzymes had ceftriaxone MICs >8 μg/ml (not shown). Most isolates positive for CTX-M tested with ceftriaxone MICs of >8 μg/ml (98%, Fig. 1). In contrast, a significant proportion (40%) of CTX-M-negative isolates also had MICs of 2 to 8 μg/ml. Ceftazidime MICs were more evenly distributed between CTX-M-positive and -negative isolates (Fig. 1), which is not surprising since CTX-M does not hydrolyze ceftazidime. In contrast, the presence of NDM or KPC was much more likely to yield an MIC of >8 μg/ml for meropenem or imipenem (Fig. 1).
The number of VME per 100 patients tested was calculated across a variety of resistance rates (Table 3). For institutions with very low ceftriaxone resistance rates (5%), the number of VME predicted by using the genotypic method is low: only 1 patient per 100 tested overall. However, at this low resistance rate, use of ceftriaxone empirically would only be incorrect in 5 of 100 patients, limiting the value of a molecular test. In contrast, at a resistance rate of 50% to ceftriaxone, 10 patients per 100 tested would be incorrectly identified as susceptible to ceftriaxone (Table 3). While the clinical threshold for allowable risk will differ by the patient population, clinicians will rarely tolerate an error rate above 5 to 10%, which occurred at 25% resistance to ceftriaxone, 20% resistance to TZP, and >50% resistance to meropenem (Table 3). In general, institutions in the United States do not witness these high resistance rates. However, the epidemiology of CTX-M varies across the United States. In New England, the prevalence of ceftriaxone-nonsusceptible isolates was 9.7% overall in 2014. In this population, 4 patients in every 100 would be expected to have a false-negative prediction for ceftriaxone susceptibility based on the absence of CTX-M, KPC, and NDM (not shown). In contrast, in the South Atlantic, where the prevalence of ceftriaxone not susceptibility was 20%, 6 patients per 100 would have a false-susceptible prediction by the absence of CTX-M, KPC, and NDM (Table 4 [calculations not shown]).
It is important to note that prevalence of resistance varies not only across census regions but also significantly within each region, depending on the institution. For example, in 2012 at a large tertiary care institution in Los Angeles, the prevalences of ceftriaxone-nonsusceptible E. coli and Klebsiella recovered from blood were 22 and 17%, respectively, values far above the Pacific region prevalence of 9.1% (R. Humphries, unpublished data). Furthermore, when evaluating E. coli isolated from nonurine cultures, ceftriaxone nonsusceptibility rates were 17% among isolates recovered from outpatients, 26% among isolates recovered from hospitalized patients outside the intensive care unit (ICU), and 36% among isolates recovered from patients in the ICU. Most institutions do not stratify antibiograms by specimen type nor by hospital unit, and these differences in the prevalence of CRO-nonsusceptible E. coli may not be appreciated. Importantly, this represents a difference in VME of 4 for outpatients to a VME of 7 for ICU patients, where the risk of a false-susceptible prediction is less tolerated. Institutions that choose to use genotypic RDTs to inform therapeutic de-escalation decisions should be aware of this incidence-base performance across patient populations and routinely review both the prevalence of resistant isolates at their institution across different patient populations and the correlation between gene detection and final MICs.
Limitations of this study include the source of data, which includes isolates from both blood (17.3% of all isolates) and other anatomical sources (8). Similarly, the data used here were from 2012, 6 years ago. However, the spectrum of beta-lactamases in the United States remains largely the same, as documented by Castanheira and coworkers in 2014 and 2016 (16, 18). Data from 2014 demonstrated no major changes in the blaCTX-M-carrying isolates overall and a decrease in isolates carrying blaKPC from 16.5 to 10.9%, mostly due to the decrease in isolates harboring blaKPC in hospitals in the Mid-Atlantic and South Atlantic regions (16). Similarly, this data set only queried KPC and NDM as carbapenemases, and IMP, VIM, and OXA-48 have since been reported in the United States. However, at the time of this study, IMP, VIM, and OXA-48 were not detected among U.S. isolates (M. Castanheira, unpublished data).
While the MERINO trial evaluated TZP versus meropenem for definitive therapy, it is clear that the first 24 h of antimicrobial therapy are critical to patient outcome (19, 20). This raises the question of whether TZP should remain the mainstay empirical therapy, particularly in regions of high ceftriaxone resistance or for patients at high risk for infections caused by CRO-nonsusceptible isolates. The alternative, however, is troubling: empirical use of a carbapenem for uncomplicated patients with BSIs caused by Gram-negative organisms, which undoubtedly increase the selective pressure for carbapenem resistance in the United States, furthering the carbapenem resistance epidemic. It should be noted that TZP prediction was poor by the genotype method, a finding also shown by others (21). Regardless, it is clear that antimicrobial resistance, and the optimal therapy for the treatment of these infections, is complex. Knowledge of the isolate’s MIC as soon as possible may significantly aid in the management of these complicated cases.
ACKNOWLEDGMENT
We are employees of Accelerate Diagnostics, Inc.
REFERENCES
- 1.Ortega M, Marco F, Soriano A, Almela M, Martinez JA, Munoz A, Mensa J. 2009. Analysis of 4758 Escherichia coli bacteraemia episodes: predictive factors for isolation of an antibiotic-resistant strain and their impact on the outcome. J Antimicrob Chemother 63:568–574. doi: 10.1093/jac/dkn514. [DOI] [PubMed] [Google Scholar]
- 2.Peralta G, Sánchez MB, Garrido JC, De Benito I, Cano ME, Martínez-Martínez L, Roiz MP. 2007. Impact of antibiotic resistance and of adequate empirical antibiotic treatment in the prognosis of patients with Escherichia coli bacteraemia. J Antimicrob Chemother 60:855–863. doi: 10.1093/jac/dkm279. [DOI] [PubMed] [Google Scholar]
- 3.Retamar P, Portillo MM, Lopez-Prieto MD, Rodriguez-Lopez F, de Cueto M, Garcia MV, Gomez MJ, Del Arco A, Munoz A, Sanchez-Porto A, Torres-Tortosa M, Martin-Aspas A, Arroyo A, Garcia-Figueras C, Acosta F, Corzo JE, Leon-Ruiz L, Escobar-Lara T, Rodriguez-Bano J, Group S. 2012. Impact of inadequate empirical therapy on the mortality of patients with bloodstream infections: a propensity score-based analysis. Antimicrob Agents Chemother 56:472–478. doi: 10.1128/AAC.00462-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Seddon MM, Bookstaver PB, Justo JA, Kohn J, Rac H, Haggard E, Mediwala KN, Dash S, Al-Hasan MN. 2018. Role of early de-escalation of antimicrobial therapy on risk of Clostridioides difficile infection following Enterobacteriaceae bloodstream infections. Clin Infect Dis doi: 10.1093/cid/ciy863. [DOI] [PubMed] [Google Scholar]
- 5.Schuts EC, Hulscher M, Mouton JW, Verduin CM, Stuart J, Overdiek H, van der Linden PD, Natsch S, Hertogh C, Wolfs TFW, Schouten JA, Kullberg BJ, Prins JM. 2016. Current evidence on hospital antimicrobial stewardship objectives: a systematic review and meta-analysis. Lancet Infect Dis 16:847–856. doi: 10.1016/S1473-3099(16)00065-7. [DOI] [PubMed] [Google Scholar]
- 6.Harris PNA, Tambyah PA, Lye DC, Mo Y, Lee TH, Yilmaz M, Alenazi TH, Arabi Y, Falcone M, Bassetti M, Righi E, Rogers BA, Kanj S, Bhally H, Iredell J, Mendelson M, Boyles TH, Looke D, Miyakis S, Walls G, Al Khamis M, Zikri A, Crowe A, Ingram P, Daneman N, Griffin P, Athan E, Lorenc P, Baker P, Roberts L, Beatson SA, Peleg AY, Harris-Brown T, Paterson DL, Investigators MT, the Australasian Society for Infectious Disease Clinical Research. 2018. Effect of piperacillin-tazobactam versus meropenem on 30-day mortality for patients with Escherichia coli or Klebsiella pneumoniae bloodstream infection and ceftriaxone resistance: a randomized clinical trial. JAMA 320:984–994. doi: 10.1001/jama.2018.12163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hayden MK, Won SY. 2018. Carbapenem-sparing therapy for extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella pneumoniae bloodstream infection: the search continues. JAMA 320:979–981. doi: 10.1001/jama.2018.12565. [DOI] [PubMed] [Google Scholar]
- 8.Castanheira M, Farrell SE, Krause KM, Jones RN, Sader HS. 2014. Contemporary diversity of beta-lactamases among Enterobacteriaceae in the nine U.S. census regions and ceftazidime-avibactam activity tested against isolates producing the most prevalent beta-lactamase groups. Antimicrob Agents Chemother 58:833–838. doi: 10.1128/AAC.01896-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.CLSI. 2018. Performance standards for antimicrobial susceptibility testing, M100 approved standard, 28th ed Clinical and Laboratory Standards Institute, Wayne, PA. [Google Scholar]
- 10.Pogue JM, Heil EL, Lephart P, Johnson JK, Mynatt RP, Salimnia H, Claeys KC. 2018. An antibiotic stewardship program blueprint for optimizing Verigene BC-GN within an Institution: a tale of two cities. Antimicrob Agents Chemother 62:e02538-17. doi: 10.1128/AAC.02538-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.FDA. 2018. Class II special controls guidance document: antimicrobial susceptibility test (AST) systems. U.S. Food and Drug Administration Center for Devices and Radiological Health, Bethesda, MD: https://www.fda.gov/RegulatoryInformation/Guidances/ucm080564.htm. Accessed November 7 2018. [Google Scholar]
- 12.Iredell J, Brown J, Tagg K. 2016. Antibiotic resistance in Enterobacteriaceae: mechanisms and clinical implications. BMJ 352:h6420. doi: 10.1136/bmj.h6420. [DOI] [PubMed] [Google Scholar]
- 13.Ruppe E, Woerther PL, Barbier F. 2015. Mechanisms of antimicrobial resistance in Gram-negative bacilli. Ann Intensive Care 5:61. doi: 10.1186/s13613-015-0061-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dudley MN, Ambrose PG, Bhavnani SM, Craig WA, Ferraro MJ, Jones RN, CLSI. 2013. Background and rationale for revised clinical and laboratory standards institute interpretive criteria (Breakpoints) for Enterobacteriaceae and Pseudomonas aeruginosa: I. Cephalosporins and Aztreonam. Clin Infect Dis 56:1301–1309. doi: 10.1093/cid/cit017. [DOI] [PubMed] [Google Scholar]
- 15.Kahlmeter G. 2008. Breakpoints for intravenously used cephalosporins in Enterobacteriaceae–EUCAST and CLSI breakpoints. Clin Microbiol Infect 14(Suppl 1):169–174. doi: 10.1111/j.1469-0691.2007.01856.x. [DOI] [PubMed] [Google Scholar]
- 16.Castanheira M, Mendes RE, Jones RN, Sader HS. 2016. Changes in the frequencies of beta-lactamase genes among Enterobacteriaceae isolates in U.S. hospitals, 2012 to 2014: activity of ceftazidime-avibactam tested against beta-lactamase-producing isolates. Antimicrob Agents Chemother 60:4770–4777. doi: 10.1128/AAC.00540-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Castanheira M, Deshpande LM, Mendes RE, Canton R, Sader HS, Jones RN. 2019. Variations in the occurrence of resistance phenotypes and carbapenemase genes among Enterobacteriaceae isolates in 20 years of the SENTRY Antimicrobial Surveillance Program. Open Forum Infect Dis 6:S23–S33. doi: 10.1093/ofid/ofy347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mendes RE, Jones RN, Woosley LN, Cattoir V, Castanheira M. 2019. Application of next-generation sequencing for characterization of surveillance and clinical trial isolates: analysis of the distribution of beta-lactamase resistance genes and lineage background in the United States. Open Forum Infect Dis 6:S69–S78. doi: 10.1093/ofid/ofz004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kumar A, Ellis P, Arabi Y, Roberts D, Light B, Parrillo JE, Dodek P, Wood G, Kumar A, Simon D, Peters C, Ahsan M, Chateau D, Cooperative Antimicrobial Therapy Of Septic Shock DRG. 2009. Initiation of inappropriate antimicrobial therapy results in a fivefold reduction of survival in human septic shock. Chest 136:1237–1248. doi: 10.1378/chest.09-0087. [DOI] [PubMed] [Google Scholar]
- 20.Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, Suppes R, Feinstein D, Zanotti S, Taiberg L, Gurka D, Kumar A, Cheang M. 2006. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 34:1589–1596. doi: 10.1097/01.CCM.0000217961.75225.E9. [DOI] [PubMed] [Google Scholar]
- 21.Evans SR, Hujer AM, Jiang H, Hujer KM, Hall T, Marzan C, Jacobs MR, Sampath R, Ecker DJ, Manca C, Chavda K, Zhang P, Fernandez H, Chen L, Mediavilla JR, Hill CB, Perez F, Caliendo AM, Fowler VG Jr, Chambers HF, Kreiswirth BN, Bonomo RA, Antibacterial Resistance Leadership Group. 2016. Rapid molecular diagnostics, antibiotic treatment decisions, and developing approaches to inform empiric therapy: PRIMERS I and II. Clin Infect Dis 62:181–189. doi: 10.1093/cid/civ837. [DOI] [PMC free article] [PubMed] [Google Scholar]