Comparison of Lateral Flow Immunochromatography and Phenotypic Assays to PCR for the Detection of Carbapenemase-Producing Gram-Negative Bacteria, a Multicenter Experience in Mexico

The identification of carbapenemase-producing Enterobacterales and Pseudomonas aeruginosa is important for treating and controlling hospital infections. The recommended methods for their identification require a long waiting time, technical training, and expertise. Lateral flow immunoassays such as NG-Test CARBA 5® overcome these needs. We analyzed 84 clinical isolates of carbapenem-resistant Enterobacterales and P. aeruginosa from four different hospitals in a two-year period. Antimicrobial resistance patterns were confirmed with the broth dilution method. Evaluation of KPC, VIM, NDM, IMP, and OXA-48-like enzymes was performed and compared to NG-Test CARBA 5 and phenotypic assays. Enterobacterales represented 69% of isolates and P. aeruginosa represented 31%. Carbapenemase-producing strains were 51 (88%) of Enterobacterales and 23 (88.4%) of P. aeruginosa; 20 (34%) and 23 (88%) were Class B ß-lactamases, respectively. The NG-Test CARBA 5® assay for Enterobacterales showed high sensitivity (98%), specificity (100%), and PPV (100%); however, it did not for P. aeruginosa. The Kappa concordance coefficient was 0.92 for Enterobacterales and 0.52 for P. aeruginosa. NG-Test CARBA 5® is a fast and easy-to-use assay. In Enterobacterales, we found excellent agreement in our comparison with molecular tests. Despite the low agreement in P. aeruginosa, we suggest that this test could be used as a complementary tool.


Introduction
Hospital-acquired infections, predominantly blood-stream infections and pneumonia, are an important, expensive, and alarming problem for health systems worldwide, as well as being the main cost of bed-days lost [1]. Enterobacterales and Pseudomonas aeruginosa are the leading bacteria associated with healthcare infections; furthermore, there has been an alarming increase in multidrug-resistant microorganisms, including resistance to carbapenems [2]; by far the most important mechanism is related to ß-lactamases synthesis, such as carbapenemases [3][4][5].
In a global surveillance study, the prevalence of meropenem non-susceptible Enterobacterales clinical isolates in Latin America was 5.3% vs. 1.4% in the United States of America [6]. Currently, the prevalence of carbapenem resistance in Mexico has been reported from 3 to 12% for Enterobacterales and 35% in P. aeruginosa, showing a significant increase in the last decade [7,8]. There are substantial epidemiological and clinical reasons to identify carbapenemases, such as stewardship programs, reduced resistance propagation, and the use of appropriate treatment, while adequately employing new ßlactamase inhibitors, cefiderocol, or polymyxins. The shortened time-to-result is critical in life-threatening infections such as sepsis [9].
The phenotypic and molecular typification of carbapenemases represents a challenge within clinical microbiology laboratories [10]. The Clinical & Laboratory Standards Institute (CLSI) has published recommendations to make an intentional search for carbapenemases while using phenotypic tests, such as the Carba NP test (colorimetric assay), and modified carbapenem inactivation method (mCIM), with and without EDTA as a quelant to identify metallo ß-lactamases (MBLs), and molecular methods for genes that codify these enzymes [11]. Recently, MALDI-ToF/MS has been evaluated to document carbapenemases, showing faster and more accurate results in comparison to mCIM or Carba NP; however, it is an expensive tool requiring experienced personnel, challenging its worldwide access in routine laboratories, especially in developing countries [12].
Furthermore, some immuno-assays have been widely used in the clinical microbiology field. Lateral flow assay, an immunochromatographic method, has been developed in order to have a fast and straightforward way to identify carbapenemases [13]. NG-Test CARBA-5 ® (NG-Biotech, Guipry, France) was approved by the FDA in 2019, and it can identify the five most prevalent carbapenemases (KPC, NDM, IMP, VIM, and OXA-48 like), giving results in fewer than 15 min without specialized training. Some previous studies showed 100% sensitivity and 99.75% specificity for Enterobacterales using molecular biology as the gold standard [14]. The most recent works showed 100% positive agreement (CI 95%: 97.8-100%) and 95.1% of negative agreement (CI 95%: 90.3-97.6%) compared to the reference method (mCIM and molecular methods) [14,15]. Although it has exemplary diagnostic performance for identifying Enterobacterales, results for P. aeruginosa have unfortunately not been comparable, mainly with IMP producers [16]. According to previous reports of carbapenemases in Mexico, the main enzymes in P. aeruginosa and Enterobacterales belong to MBLs [17]. The aim of this study was to evaluate the performance of NG-Test CARBA 5 ® in clinical isolates in a region where carbapenem resistance due to carbapenemases, such as MBLs, is increasing.

Discussion
Carbapenem-resistant infections secondary to Enterobacterales or P. aeruginosa are of serious clinical worldwide concern, because the antimicrobial clinical treatment options are limited due to harbor resistance mechanisms inherent to many antibiotic classes [18]. Gram-negative bacteria are promiscuous microorganisms capable of sharing information mostly located in plasmids [19]. Epidemiological information has been kept relevant for many years in order to satisfy WHO recommendations regarding to the research of antimicrobial resistance [20]. Unfortunately, the Latin America region is one of the biggest stimulants for ß-lactamase production, with an enormous diversity of different antimicrobial mechanisms [21,22]. Rapid and accurate detection of carbapenemases is critical to diagnose and stop the spread inside the hospital and shorten the time of accurate treatment [23].
Compared to CLSI-endorsed routine methods for carbapenemases detection, NG-Test CARBA 5 ® has advantages as it is a fast and simple test with minimal technical training. This facilitates its use in general hospitals or non-specialized clinical laboratories. The main advantage over colorimetric tests (RAPIDEC CARBA NP ® , ß-CARBA ® , CARBA NP) is its ability to detect enzymes with a weak hydrolytic capacity such as OXA-48-like, which results in multiple false negatives when employing colorimetric methods. On the other hand, an important advantage of NG-Test CARBA-5 ® is the ability to discriminate between the different classes of carbapenemases [15,16,24,25]. There is also a trend of applying it directly into some clinical samples, such as positive blood culture bottles, in order to decrease results time [26][27][28].
Unlike colorimetric tests, the mCIM with and without EDTA has better performance for the detection of carbapenemases [12]. Unfortunately, it requires 24 h after microbiological isolation for its interpretation; as with CARBA NP, the discrepancy between those of the serine class and MBLs is only achieved with the method when modified by EDTA in order to reveal the MBLs. Finally, this method is not approved for use directly from positive blood culture bottles in comparison with NG-Test CARBA 5 ® [11,29]. On the other hand, eCIM is only validated for use with Enterobacterales but not with P. aeruginosa; this is due to substantial discrepancies in the results [11].
Molecular methods based on polymerase chain reaction are usually the ideal methods for the identification of carbapenemases; however, they require technical training, and it is not available in several Latin American laboratories. Furthermore, the cost of homemade PCR, or even commercial kits, in lower-income countries exceeds the purchasing range, making it difficult to access these technologies in routine laboratories [30].
It is important to mention that NG-Test CARBA 5 ® performance has been generally poor with P. aeruginosa [16]. This is relevant in settings such as ours, because close to 50% of clinical strains carbapanemase-producers P. aeruginosa are MBLs, predominantly VIM and IMP; on the other hand, according to some studies in Mexico, the second most frequent P. aeruginosa carbapenemase in Mexico belongs to GES, and this enzyme is not detectable with NG-Test CARBA 5 ® , which is a weakness of the test [17]. In our study, NG-Test CARBA 5 ® was able to identify all VIM strains (13/13). However, in strains harboring IMP genes, NG-Test CARBA 5 ® showed many inconsistencies (5/8 detected). This problem was already mentioned by Proton et al., who found a particular problem with IMP, but also found acceptable agreement with VIM and NDM [16]. We recognized three different types of IMP genes: bla IMP-15 , bla IMP-75 , and bla IMP-62 ; the first has already been described in the NG-Test CARBA 5 ® literature, while the other two have not been previously described for NG-Test CARBA 5 ® [16,36].
Compared to recent publications, this study had the advantage of evaluation employing the immunochromatographic test against multiple phenotypic tests while using molecular techniques as the gold standard [24,37]. Carbapenemases distribution was found to be like previously reported in Mexico, predominantly MBLs and OXA-48-like enzymes [17]. Although it is a weakness that we found few Enterobacterales VIM-producers and few P. aeruginosa NDM-producers, and no Enterobacterales IMP-producers, it is worth mentioning that sensitivity and specificity of phenotypic methods, especially of Carba NP-EDTA and eCIM, exhibited a worse performance in comparison to other publications. An opportunity of study is the sequencing of the carbapenemase encoding-genes, which might elucidate possible new enzymes not reported by this assay or characterize the carbapenemases that were falsely negative by the NG-Test CARBA-5 ® [38,39].
MBLs are an emerging antimicrobial resistance threatening problem among Enterobacterales and P. aeruginosa strains present in nosocomial infections. Latin America has reported several hospital outbreaks related to different types of these enzymes, which is an increasing and perpetual problem [40][41][42]. It should be noted that even though OprD in P. aeruginosa is one of the main resistance mechanisms for carbapenems along with efflux pumps, this test might be used to identify the enzymatic mechanism and to help choose the adequate treatment. For carbapenem-resistant P. aeruginosa, we suggest incorporation of phenotypic assays or genome sequencing in order to elucidate the core mechanism according to local resistance; for example, one of the last CDC reports revealed a prevalence of 87% dysfunction of the porin OprD in P. aeruginosa [30,43].
This work shows the use of this assay in real life because we demonstrated that the NG-Test CARBA 5 ® had excellent agreement (Kappa 0.9) for carbapenem resistance Enterobacterales (CRE); furthermore, in a region with a higher prevalence of MBLs and OXA-48 like carbapenemases, NG-Test CARBA 5 ® could be an acceptable and fast option for routine laboratories. Based on our findings, we propose an algorithm (Figure 2) for CRE detection, applicable in countries with a high prevalence of antimicrobial resistance with an increased burden of ß-lactamases [17]. Algorithm for CRE detection. *We don't recommend further testing in a routine laboratory that does not have PCR assays or genome sequencing; instead, we recommend contacting the central state laboratory. **The mCIM test has a NPV close to 100%; in case of a negative test, the carbapenemase production seems to be unlikely and these strains must be investigated for another carbapenem-resistance mechanism like hyper AMP-c production.
In conclusion, NG-Test CARBA 5 ® is a fast and easy-to-use assay. In Enterobacterales, we found excellent agreement with molecular tests (Kappa 0.9). On the other hand, despite the low agreement in P. aeruginosa (Kapa 0.5), we suggest that this test might be used as a tool for identifying the enzymatic mechanisms with the aid of other assays.

Materials and Methods
Clinical strains. We included all non-duplicated Enterobacterales and P. aeruginosa bacteria showing resistance to at least one carbapenem drug from January 2019 to March 2020. We defined a non-duplicated strain as the first isolated bacterial found in a patient clinical sample independent of isolation sites. There were four participating hospitals: Instituto Nacional de Rehabilitación Luis Guillermo Ibarra Ibarra (INRLGII), Instituto Nacional de Cancerología (INCan), Hospital General Dr. Manuel Gea González (HGDMGG), and Hospital Civil de Guadalajara Fray Antonio Alcalde (HCGFAA). All centers performed the identification and susceptibility tests using the Vitek 2 System (BioMérieux, Marcy 1′Étoile, France).
Minimal inhibitory concentrations. Broth microdilution was employed to corroborate carbapenem resistance for all strains following the CLSI recommendations Figure 2. Algorithm for CRE detection. * We don't recommend further testing in a routine laboratory that does not have PCR assays or genome sequencing; instead, we recommend contacting the central state laboratory. ** The mCIM test has a NPV close to 100%; in case of a negative test, the carbapenemase production seems to be unlikely and these strains must be investigated for another carbapenem-resistance mechanism like hyper AMP-c production.
In conclusion, NG-Test CARBA 5 ® is a fast and easy-to-use assay. In Enterobacterales, we found excellent agreement with molecular tests (Kappa 0.9). On the other hand, despite the low agreement in P. aeruginosa (Kapa 0.5), we suggest that this test might be used as a tool for identifying the enzymatic mechanisms with the aid of other assays.

Materials and Methods
Clinical strains. We included all non-duplicated Enterobacterales and P. aeruginosa bacteria showing resistance to at least one carbapenem drug from January 2019 to March 2020. We defined a non-duplicated strain as the first isolated bacterial found in a patient clinical sample independent of isolation sites. There were four participating hospitals: Instituto Nacional de Rehabilitación Luis Guillermo Ibarra Ibarra (INRLGII), Instituto Nacional de Cancerología (INCan), Hospital General Dr. Manuel Gea González (HGDMGG), and Hospital Civil de Guadalajara Fray Antonio Alcalde (HCGFAA). All centers performed the identification and susceptibility tests using the Vitek 2 System (BioMérieux, Marcy 1 Étoile, France).
Minimal inhibitory concentrations. Broth microdilution was employed to corroborate carbapenem resistance for all strains following the CLSI recommendations [18]. The evaluated antibiotics were: ertapenem (not in P. aeruginosa), meropenem, imipenem, and doripenem. Breakpoints were defined according to the M100 from CLSI [11].
Phenotypic assays for the presence of carbapenemases. All assays were performed in the Clinical Microbiology Laboratory at INR LGII. Phenotypic assays were performed according to CLSI M100 for mCIM, eCIM, CARBA NP, and CARBA NP-EDTA. In the case of Enterobacterales, the eCIM and CARBA NP-EDTA tests were performed to identify MBLs; only the eCIM test was used to identify P. aeruginosa isolates [11].
Carbapanemase related-genes detection. Sequencing. Prior to the sequencing process, re-amplification and labeling using the BigDye ® Direct Cycle Sequencing Kit (Applied Biosystems, Waltham, MA, USA) were performed. This step was done separately for each one of the primers previously described for the amplification; thus, sequencing was done using the forward and the reverse primers separately POP-7 (Applied biosystems, EUA). GenBank was used to compare the sequences obtained with those of reference organisms in this database. The identification was corroborated by performing independent sequencing for each primer.
NG-Test CARBA 5 ® . The test was performed according to manufacturer recommendations. The lecture was executed after 15 min. The validated reaction was taken when a red line appeared in position C. The type of carbapenemase was defined when the red line appeared in the respective letter (K: KPC, O: OXA-48-like, V: VIM, I: IMP, and N for NDM).
Statistical analysis. Measures of central tendency were calculated. The detection capacity was calculated with the Kappa coefficient between PCR and phenotypic assays, as well as with the Kappa coefficient between PCR and NG-Test CARBA 5 ® . Comparable agreements were defined as those closer to 1, statistical significance was defined as p ≤ 0.05. Sensitivity, specificity, negative predictive, and positive predictive values were also calculated for phenotypic assays and NG-Test CARBA 5 ® for Enterobacterales and P. aeruginosa. Software SPSS 22 was used for statistical analysis.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics12010096/s1, Table S1. Oligonucleotide sequence for carbapenemases genes; Table S2. The proportion of genus and specie of the Enterobacterales. Table S3. Complete description of Enterobacterales and P. aeruginosa collected.

Conflicts of Interest:
The authors declare no conflict of interest.