Comparison of Three Biochemical Tests for Rapid Detection of Extended-Spectrum-β-Lactamase-Producing Enterobacteriaceae

Abstract

Enterobacterial isolates producing clavulanic-inhibited extended-spectrum β-lactamases (ESBLs) are increasingly spreading in the community and are often responsible for nosocomial infections. Rapid biochemical tests have been developed recently for their detection. Three tests, namely, the Rapid ESBL NDP test, the β-Lacta test, and the Rapid ESBL Screen, have been evaluated with a collection of 108 well-characterized strains, including wild-type strains, strains producing ESBLs, overexpressed cephalosporinases, and carbapenemases. The ESBL NDP test and the Rapid ESBL Screen (a copy of the ESBL NDP test) are aimed at detecting ESBL producers, while the β-Lacta test is aimed at detecting not only ESBL producers but also cephalosporinase and carbapenemase producers. The sensitivity and specificity for detecting ESBL producers (n = 60) were 95% and 100% for the Rapid ESBL NDP test, 80% and 87% (after 30 min) and 92% and 83% (after 2 h) for the Rapid ESBL Screen, and 88% and 71% for the β-Lacta test, respectively. Varied and time-consuming detection (up to 2 h) of ESBLs by the Rapid ESBL Screen and concomitant and varied detection of producers of AmpC and several types of carbapenemases correspond to significant shortcomings of using the Rapid Screen ESBL and β-Lacta tests, respectively.

INTRODUCTION

Acquired resistance to broad-spectrum cephalosporins in Enterobacteriaceae is mainly due to the production of clavulanic acid-inhibited extended-spectrum β-lactamases (ESBLs), which have extensively disseminated worldwide (1). Along with carbapenemase producers, ESBL producers represent the most important resistance trait in Enterobacteriaceae in 2015. The European Antimicrobial Resistance Surveillance System Network (EARS-Net), which includes 30 European countries, reported in 2013 the prevalence rates of nonsusceptibility to broad-spectrum cephalosporins among invasive enterobacterial isolates (2). The proportion of Escherichia coli strains resistant to broad-spectrum cephalosporins ranged from 5% to 39.6%, depending on the country. In Klebsiella pneumoniae, the percentage of resistance to broad-spectrum cephalosporins showed a significant increase from 22.8% in 2012 to 30% in 2013, encompassing 85 to 100% of the ESBL producers. In the United States, the percentages of health care-associated infections caused by broad-spectrum cephalosporin-resistant Enterobacteriaceae have been estimated to be 14% and 23% for E. coli and Klebsiella spp., respectively (http://www.cdc.gov/drugresistance/threat-report-2013). Those ESBL-producing Enterobacteriaceae are identified either as a source of hospital- or community-acquired infections (3).

The rapid detection of ESBL producers is therefore crucial in order to prevent their dissemination and to guide the treatments of infected patients. Several phenotypic techniques are based on the inhibition of ESBL activity by clavulanic acid or tazobactam. Those techniques require a preliminary growth step of 24 to 48 h (4). Molecular detection of ESBL-coding genes is interesting but remains costly, requires expertise, and does not detect all genes encoding enzymes exhibiting ESBL activity (4–6). Other techniques, such as matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) (7), are being developed, but they do require additional material and a significant degree of expertise.

Recently, two rapid diagnostic tests were developed to identify enterobacterial strains that are resistant to broad-spectrum cephalosporins. The Rapid ESBL NDP test (8), which is based on the detection of hydrolysis of the cefotaxime β-lactam ring revealing the production of a broad-spectrum β-lactamase, coupled with a tube containing tazobactam, signaling likely ESBL production. Another test is the β-Lacta test, based on the cleavage of a chromogenic cephalosporin, HMRZ-86 (9, 10) (Bio-Rad, Marnes-la-Coquette, France). It is claimed that this test detects any activity that may lead to the hydrolysis of broad-spectrum cephalosporins, such as ESBLs, overproduction of cephalosporinases, and carbapenemases of the K. pneumoniae carbapenemase (KPC) and metallo-β-lactamase types, without distinction between those mechanisms of resistance. Finally, a commercial adaptation of the Rapid ESBL NDP test (not validated by the inventors of the original Rapid ESBL NDP test) named Rapid ESBL Screen kit 98022 (Rosco-Diagnostica A/S, Taastrup, Denmark) has been developed.

Several authors have reported varied results using those three biochemical tests but using different collections of strains for which resistance mechanisms were in some cases not even identified. The aim of this study was to compare those three tests using the exact same collection of strains possessing well-characterized resistance mechanisms.

MATERIALS AND METHODS

Strain collection.

A total of 108 clinical enterobacterial isolates were included in this study. They were from origins worldwide and had been recovered from different types of clinical specimens. This collection included isolates producing the main ESBL types (CTX-M, TEM, SHV, PER, and GES) (n = 60), and also isolates showing resistance to broad-spectrum cephalosporins due to other mechanisms, such as plasmid-encoded AmpC-type β-lactamases or overproduced chromosomal AmpC. Some strains were also tested as negative controls, including strains producing β-lactamases with a narrow spectrum activity and wild-type strains (n = 24). Some carbapenemase producers (n = 10) were also included in the study. All isolates had previously been characterized at the molecular level to identify the different mechanisms responsible for resistance to β-lactam antibiotics. In addition, they were all clonally unrelated, according to pulsed-field gel electrophoresis (data not shown). They were cultured onto Mueller-Hinton agar plates and incubated for 16 to 24 h at 37°C. The tests were interpreted blindly by three persons from our lab.

Rapid ESBL NDP test.

The Rapid ESBL NDP test was performed as described previously (8). A single full 10-μl calibrated loop of the bacterial colonies studied was suspended in the lysis buffer (B-PERII, bacterial protein extraction reagent; Pierce/Thermo Scientific, Villebon-sur-Yvette, France) and disposed in three 1.5-ml different Eppendorf tubes (A, B, and C). Ten microliters of a concentrated tazobactam solution (40 μg/ml) was added to tube C. Next, 100 μl of the revealing solution containing pH indicator (phenol red) was added to tube A, and 100 μl of the same solution supplemented with cefotaxime (6 μg/ml) was added to tubes B and C. The three tubes were incubated at 37°C for 20 min. The results were considered negative when all tubes were red and thus interpreted as containing non-ESBL strains. When tube B was yellow/orange and both tubes A and C were red, the test result was considered positive (ESBL-producing isolate). When tube A turned to yellow/red, the test result was considered noninterpretable, regardless of any color change for tubes B and C.

β-Lacta test.

The β-Lacta test (Bio-Rad, Marnes la Coquette, France) was performed according to the manufacturer's instructions. A single 1-μl loop of the studied bacterial colonies was put into a microtube with a drop of reagent R1 and a drop of reagent R2. The microtubes were left at ambient temperature, and reading of the results was performed visually within 15 min. No change in color was considered a negative result (no hydrolysis of HMRZ-86), a color change to purple-red was considered a positive result, and a color change to orange was considered a noninterpretable result.

Rapid ESBL Screen kit 98022.

The Rapid ESBL Screen kit 98022 was performed according to the manufacturer's indications (Rosco Diagnostica, Axonlab AG, Baden, Switzerland). Several 1-μl loops of each strain were added to 150 μl of lysis buffer (B-PERII), incubated at room temperature for 30 min, and then 50 μl of this suspension was diluted in a tube with 100 μl of a 0.9% sodium chloride solution. A tablet of cefotaxime plus indicator was added to the tube. The same process was repeated using the tablet with cefotaxime plus tazobactam plus indicator. The tubes were incubated at 37°C from 20 min until 2 h and results interpreted as follows: (i) if cefotaxime plus indicator turned yellow and cefotaxime plus tazobactam plus indicator remained red, the test was considered positive for ESBL; (ii) if cefotaxime plus indicator turned yellow and cefotaxime plus tazobactam plus indicator also turned yellow, the test was considered negative for ESBL production but likely positive for another type of β-lactamase; (iii) if both tubes remained red, the test was considered negative for ESBL production.

Sensitivities and specificities were calculated for each test. A noninterpretable result was included as a negative result.

RESULTS

Evaluation of the Rapid ESBL NDP test.

By using the Rapid ESBL NDP test with all CTX-M producers, all tubes turned from red to yellow in the presence of cefotaxime and remained red in the presence of cefotaxime and tazobactam; therefore, the sensitivity of the test was 100% for detecting ESBL production with that collection of ESBL producers (Table 1). The global sensitivity of the test for detecting ESBL production was 95% (57/60). As expected, the test remained negative with all wild-type strains and for all strains expressing β-lactamases with a narrow spectrum of activity (Table 2).

TABLE 1.

Detection of extended-spectrum β-lactamase-producing isolates using different tests^a

^aShaded cells highlight the nonexpected results.

^b+, color change from red to yellow/orange for the Rapid ESBL NDP test; −, no color change. An overall positive Rapid ESBL NDP test corresponds to a positive result for cefotaxime (CTX) hydrolysis and a negative result when tazobactam (TZB) is added.

^c+, hydrolysis of HMRZ-86 for the β-Lacta test; −, no color change. NI, not interpretable (color change to orange). The β-Lacta test is aimed to detect the activity of all broad-spectrum cephalosporinases.

TABLE 2.

Detection of non-ESBL-producing isolates using different tests

^a+, color change from red to yellow/orange; −, no color change; H, hydrolysis of HMRZ-86; NH, no hydrolysis of HMRZ-86; NI, noninterpretable (color change to orange). An overall positive Rapid ESBL NDP test result corresponds to a positive result for CTX hydrolysis and a negative result when tazobactam (TZB) is added. The β-Lacta test is aimed to detect the activity of all broad-spectrum cephalosporinases. Shaded cells highlight the nonexpected results.

β-Lacta test.

Using the β-Lacta test, all except two CTX-M producers (CTX-M-15-producing Proteus mirabilis and CTX-M-37-producing E. coli) and two non ESBL producers (TEM-12-positive E. coli and TEM-24-positive E. coli) were found to be positive. In addition, a yellow-to-orange color change was observed for three isolates, thus corresponding to noninterpretable results. Therefore, the sensitivity for detecting CTX-M producers was 91% (Table 1). The sensitivity for detecting non-CTX-M ESBL producers was lower (84%), and the overall sensitivity for any type of ESBL was 88%. Noticeably, no color change was observed when wild-type and narrow-spectrum β-lactamase-producing isolates were tested (Table 2).

The β-Lacta test is also aimed at detecting cephalosporinase overproducers and carbapenemases of the KPC type together with metallo-β-lactamases. Here, only four out of 14 AmpC overproducers were detected, and only four out of eight carbapenemase producers (with OXA-48 producers supposed to be not detected) (Table 2). Therefore, the sensitivity of detection was low in both of those cases.

Rapid ESBL Screen kit 98022.

The test results for the Rapid ESBL Screen kit 98022 were interpreted after 30 min of incubation and until 2 h, according to the manufacturer's instructions. After 30 min, no color change was observed for 6 CTX-M and 7 non-CTX-M ESBL producers, respectively (Tables 1 and 2), leading to sensitivities of 83% and 72%, respectively, and a global sensitivity for ESBL detection of 80%. All wild-type and narrow-spectrum β-lactamase-producing strains gave a negative result (Table 2). However, when tubes were incubated up to 2 h, the sensitivities for detecting CTX-M and non-CTX-M ESBL producers reached 94% and 88%, respectively (92% global sensitivity for detecting ESBL), but the specificity was decreased (three wild-type or narrow-spectrum β-lactamase-producing strains gave a false-positive result) (Tables 1 and 2). The specificities of the test after 30 min and 2 h were 87% and 83%, respectively (Table 3).

TABLE 3.

Diagnostic parameters of the different tests

Diagnostic test parameter	Performance (%) by test
	Rapid ESBL NDP	β-Lacta	Rapid ESBL Screen kit
			30 min	2 h
Sensitivity for CTX-M-type ESBL	100	91.4	82.8	94.3
Sensitivity for non-CTX-M-type ESBL	88	84	72	88.0
Global sensitivity for ESBL	95.0	88.0	80	91.7
Global specificity	100	70.8	87	83

DISCUSSION

This study showed that the sensitivity of detection of the three tests was good for all CTX-M producers, although the Rapid ESBL NDP test was the only test able to flag 100% of the CTX-M producers. The global sensitivity of any kind of ESBL producer varied from 80% for the Rapid ESBL Screen test (reading after 30 min) to 95% for the Rapid ESBL NDP test. A partial lack of detection of ESBL-producing isolates that were not CTX-M producers explains some discrepancies published regarding the potencies of the tests (8, 9, 11–13). The Rapid ESBL NDP and Rapid ESBL screen kit 98022 tests are not designed to detect plasmid-mediated and chromosomal overproducers of AmpC. The β-Lacta test, which is designed to detect those AmpC overproducers, failed to detect those strains in 10/14 of the cases. A similar failure of detection was reported by Morosini et al. (13). The overall specificity of the β-Lacta test for detecting ESBL producers was much lower than that of the Rapid ESBL NDP test, since it also partially detects AmpC producers and several types of carbapenemase producers. Indeed, the β-Lacta test is also aimed at detecting the production of carbapenemases of the KPC and metalloenzyme types. The data obtained through this comparative study shows that KPC producers were well detected, but metallo-β-lactamase producers were not. Incomplete detection of AmpC or carbapenemase producers might therefore be a source of confusion when using the β-Lacta test.

From a technical point of view, it must be highlighted that the best results for the detection of ESBL producers using the Rapid ESBL Screen test were obtained after 2 h of incubation. This incubation time does not make the test as rapid as the ESBL NDP test or the β-Lacta test, for which results are obtained within 20 min. In addition, prolonged incubation with the Rapid ESBL Screen test generated false-positive results. It is likely that the weak-positive results obtained several times after the first reading time (30 min) may be associated with weak dissolution of the tablets. Weak-positive results were sometimes difficult to interpret at the first time point (30 min) using the Rapid ESBL Screen test, probably because of the poor dissolution of the tablet. Regarding the β-Lacta test, interpreting any color change as a positive result might also increase sensitivity but, conversely, specificity might be affected.

Noticeably, both the β-Lacta and the Rapid ESBL Screen kit 98022 tests do not include an internal control well that would be free of antibiotic. We believe that this is a major shortcoming in a comparison with the Rapid ESBL NDP test, since such a control allows a better appreciation of the color change (especially for weak-positive strains) and allows the detection of possible false-positive results due to nonspecific reactions (11).

As opposed to the Rapid ESBL NDP test, which does not misidentify a KPC producer as an ESBL producer, the β-Lacta test cannot differentiate between KPC producers and ESBL producers. This feature may be important in countries with high prevalence rates of KPC producers, such as the United States, Canada, Colombia, Italy, and Israel.

In conclusion, and even though all tests evaluated here overall performed well for detecting ESBL producers, greater performance was obtained with the Rapid ESBL NDP test. One main disadvantage of the β-Lacta test is a lack of specificity with AmpC and carbapenemase producers. The Rapid ESBL Screen test, which is actually a copy of the Rapid ESBL NDP test, has much poorer performance. Also, it requires an additional delay (2 h versus 20 min) for reading the results that may be considered significant for patient management and antibiotic stewardship. Finally, we believe that the use of rapid biochemical tests for detecting ESBL producers from clinical sites (11, 14) will be an alternative to molecular techniques, since they are easy to implement, affordable, and may detect any kind of ESBL.

Funding Statement

The stay of J.F. at the University of Fribourg was funded by a grant from the Sociedad Española de Enfermedades Infecciosas y Microbiologia Clínica (SEIMC).