Validation of the FluoroType MTBDR Assay for Detection of Rifampin and Isoniazid Resistance in Mycobacterium tuberculosis Complex Isolates

ABSTRACT For Mycobacterium tuberculosis complex (MTBC), the rapid and accurate diagnosis of drug resistance is crucial to ensure early initiation of appropriate therapy. Recently, a new molecular diagnostic test, the FluoroType MTBDR, aimed at detecting rifampin and isoniazid resistance has become available. This study aimed to evaluate the FluoroType MTBDR in comparison to phenotypic drug susceptibility testing (DST) using M. tuberculosis complex isolates. MTBC isolates underwent phenotypic DST and were tested using the FluoroType MTBDR and Genotype MTBDRplus. Sanger sequencing of the key regions of rpoB, katG, inhA, and aphC was performed for isolates with discordant phenotypic and molecular results. Furthermore, isolates with specific wild-type bands missing in the Genotype MTBDRplus, indicating the presence of a mutation, were investigated by Sanger sequencing. Specificity and sensitivity, defined as the proportions of isolates correctly determined as susceptible and resistant by the FluoroType MTBDR compared to phenotypic DST, were calculated. A total of 180 culture isolates were included; phenotypic DST showed 85 isolates susceptible to isoniazid and rifampin, 7 with isoniazid monoresistance, 7 with rifampin monoresistance, and 81 with multidrug resistance. The specificity of the FluoroType MTBDR was 100% (95% confidence interval [CI], 96.0 to 100%) for both rifampin and isoniazid. The sensitivity was 91.7% (95% CI, 83.6 to 96.6%) for isoniazid and 98.9% (95% CI, 93.8 to 100.0%) for rifampin. The FluoroType MTBDR has a high sensitivity and specificity for the detection of rifampin and isoniazid resistance when using culture isolates.

R apid and accurate laboratory diagnoses of Mycobacterium tuberculosis drug susceptibility and resistance are crucial to ensure early initiation of appropriate therapy, to adequately manage disease, and to control further transmission. Conventional drug susceptibility testing (DST) relies on culture-based methods with results available only after several weeks. Much shorter turnaround times can be achieved with molecular diagnostics performed either directly on sputum samples or on positive cultures (1)(2)(3).
The currently available molecular methods include Xpert MTB/RIF, line probe assays (LPAs), Sanger target sequencing, and next-generation sequencing (4)(5)(6)(7). LPAs are recommended by the World Health Organization (WHO) as rapid diagnostic tests for detection of drug resistance (8). The technology combines PCR with subsequent reverse hybridization. The specifically bound amplicons are made visible in a colorimetric detection reaction resulting in banding patterns suggestive of the presence or absence of mutations in the target genes. The most widely used LPA, the GenoType MTBDRplus, detects the most prevalent mutations in the rpoB, katG, inhA genes conferring resistance to rifampin (RMP) and isoniazid (INH). The Genotype MTBDRplus is designed to specifically detect four rpoB mutations (in codons 516, 526, and 531), two katG mutations (in codon 315), and four inhA mutations in the regulatory region. Other mutations within the amplified region of the target genes are indicated by the absence of a wild-type band without the simultaneous presence of a mutation band.
Recently a new assay, the FluoroType MTBDR, which is directed at the same target genes but uses a different technology, has been developed. This assay combines a linear-after-the-exponential PCR (LATE-PCR) (9) with special probes using lights-on/ lights-off detection technology (10). The readouts of the FluoroType MTBDR are melting curves. The shapes correspond to wild types or to the presence of specific mutations. The FluoroType MTBDR is an open system that identifies characterized mutations via a learning software interpreting the melting curves.
Practically, the FluoroType MTBDR has several advantages over the Genotype MTBDRplus: (i) less hands-on time, (ii) more-rapid results, (iii) decreased risk of DNA contamination, and (iv) automatic interpretation with the possibility to import results directly into a laboratory information system. However, the sensitivity and specificity of this new method for detecting RMP and INH resistance have not been investigated under routine conditions. This study aimed to compare the results of phenotypic DST, GenoType MTBDRplus, and FluoroType MTBDR using cultures positive for M. tuberculosis complex (MTBC).

MATERIALS AND METHODS
Mycobacterial cultures were performed at the National Mycobacterial Reference Laboratory in Borstel, Germany, between May 2016 and October 2017 using mycobacterial growth indicator tubes (MGIT) (Becton Dickinson, USA) and Loewenstein-Jensen and Stonebrink culture slants (Artelt-Enclit, Germany). All cultures were identified to species level using the GenoType MTBC version 1.0 (Hain Lifescience, Nehren, Germany) according to the manufacturer's instructions. The MGIT 960 isoniazidrifampin-ethambutol (IRE) kit (Becton Dickinson) was used according to the manufacturer's instructions. The critical concentrations of rifampin and isoniazid were 1 g/ml and 0.1 g/ml, respectively (11).
For DNA extraction, an aliquot of 500 l was obtained from the bottom of a positive MGIT or two loops (10 l) of bacterial growth were collected from Loewenstein-Jensen or Stonebrink culture slants and added to 500 l of 0.9% NaCl solution. Samples were centrifuged at 3,000 ϫ g for 15 min. Supernatants were removed, and DNA was extracted using the Fluorolyse kit (Hain Lifescience, Nehren, Germany) according to the manufacturer's instructions. In brief, 100 l of lysis buffer and 2 l of internal control were added to the remaining pellets. The mixed samples were incubated at 95°C for 5 min, centrifuged for 1 min, and resuspended in 100 l neutralization buffer. Following centrifugation at 10,000 ϫ g for 5 min, supernatants were transferred into a 0.5-ml cryovial. DNA was stored at Ϫ20°C. For each batch, one negative control using 100 l of lysis buffer, 100 l of neutralization buffer, and 2 l of internal control was included.
DNA was analyzed with the Genotype MTBDRplus version 2 and FluoroType MTBDR assays (Hain Lifescience, Nehren, Germany) as per the manufacturer's instructions. In brief, for the FluoroType MTBDR, 20 l of DNA solution was added to 6 l master mix component A and 14 l master mix component B and transferred into 96-well plates. The plates were then sealed, centrifuged, and analyzed via Fluoro-Cycler 96. For each plate, one positive and one negative control were included. For the Genotype MTBDRplus version 2, 5 l of DNA solution was added to 45 l of master mix containing 10 l master mix component A and 35 l master mix component B. Twenty microliters of PCR products was added to 20 l denaturation reagent in a hybridization well. Hybridization reagents were prepared, preheated, and loaded onto the GT-Blot48 (Hain Lifescience, Nehren, Germany). After an initial hybridization step, strips were added and hybridization performed as per the standard protocol. For each batch, one negative control (45 l master mix, 5 l DNA-free water) was tested. Interpretation of the results was performed according to the manufacturer's instructions.
The package insert of the FluoroType MTBDR identifies the following rpoB, katG, and inhA mutations in amino acid codons (e.g., S531L) and regulatory regions (e.g. Molecular results were interpreted without knowledge of phenotypic DST results. Sanger sequencing in the key regions of rpoB, katG, inhA, and aphC (12) was performed for isolates with discordant results between either the two molecular methods or the molecular and phenotypic methods. If the Genotype MTBDRplus showed a missing wild-type band but not a mutation band, Sanger sequencing in the respective key region was performed to confirm the FluoroType MTBDR result. Sanger sequencing was performed using an ABI 3130xl genetic analyzer (Applied Biosystems) and an ABI BigDye Terminator cycle sequencing kit (version 3.1) according to the manufacturer's instructions.
Molecular results were coded as wild type, individual mutation, mutations in a region of the gene, or indeterminate. The molecular results were interpreted as susceptible, resistant, or indeterminate. Indeterminate results in the Genotype MTBDRplus were defined as detection of MTBC DNA without the presence of a gene locus control. All samples yielding an indeterminate result were retested using a new aliquot of DNA. A mixed culture (heteroresistance) was defined as both wild-type and mutant DNA being present as identified by wild-type and mutant bands (Genotype MTBDRplus) and/or by respective double peaks in the DNA sequence.
To determine the accuracy of the FluoroType MTBDR assay, the results were compared with phenotypic DST results. Sensitivity and specificity were calculated for RMP and INH separately. Sensitivity was defined as the proportion of isolates correctly determined as resistant by the FluoroType MTBDR compared to the phenotypic DST results. Specificity was defined as the proportion of isolates correctly determined as susceptible by the FluoroType MTBDR compared to the phenotypic DST results. The McNemar test on paired proportions was used to compare the sensitivities of the FluoroType MTBDR and the Genotype MTBDRplus for INH. All statistical analysis were performed using Stata 14.0.
Information was extracted from the laboratory information system anonymously. The ethics committee of the University of Luebeck approved the study. Individual consent was not sought as no additional patient information was obtained.

RESULTS
A total of 180 cultures were included in the study. The majority were identified as M. tuberculosis (n ϭ 173), followed by Mycobacterium africanum (n ϭ 5), Mycobacterium bovis subsp. bovis (n ϭ 1), and M. bovis BCG (n ϭ 1 , and the specificity was 100.0% (90/90; 95% CI, 96.0 to 100.0%). The most frequent rpoB mutation detected by Fluo-roType MTBDR was S531L (n ϭ 56). A total of 7 rpoB mutations remained unidentified. Mutations were identified by Sanger rpoB sequencing as H526L (n ϭ 3), S531L (n ϭ 1), Q513K (n ϭ 1), D516Y (n ϭ 1), and, in one mixed culture, H526D and S531L. A total of four mixed culture were identified in the study. One of those cultures was a mixture of an rpoB wild-type isolate and an isolate with a L533P mutation. The FluoroType MTBDR detected the L533P mutations, while the Genotype MTBDRplus did not. The FluoroType MTBDR identified three isolates as having H526P mutations, while Sanger sequencing revealed H526L mutations. The FluoroType MTBDR falsely detected an H526R mutation in a mixed culture with one isolate harboring an H526D mutation and the other a S531L mutation. Of the seven cultures falsely tested as INH susceptible by the FluoroType MTBDR, four had katG and/or inhA mutations detected by the Genotype MTBDRplus. This resulted in a higher, but not significantly, sensitivity of the MTBDRplus (96.0%; 95% CI, 89.9 to 99.3%) for INH than of the FluoroType MTBDR (91.7%; 95% CI, 83.6 to 96.6%) (P ϭ 0.12).

DISCUSSION
This study shows high sensitivity and specificity of the FluoroType MTBDR for the detection of RMP and INH resistance using M. tuberculosis complex isolates. The proportion of indeterminate results was small. Repeat DNA extraction resolved 5 of 9 indeterminate results, suggesting that insufficient or low-quality DNA might have been the problem. Two of the isolates that tested repeatedly indeterminate were mixed cultures, which might explain the result.
For RMP, the sensitivity and specificity of FluoroType MTBDR were comparable to those reported for established molecular diagnostics such as the Genotype MTBDRplus and the Xpert MTB/RIF (2,3,13). While the presence or absence of an rpoB mutation was correctly detected in all isolates, the FluoroType MTBDR failed to identify the exact mutation in 10% (9/87) of cases. Given that silent rpoB mutations are rare (14-16), one might argue that detecting the presence of an rpoB mutation itself is sufficient. In fact,  When wild-type bands were missing in the Genotype MTBDRplus assay, the base pair region is indicated (e.g., 526 -529 the most widely used molecular diagnostic test, the Xpert MTB/RIF, indicates only the presence of RMP resistance without information about the associated rpoB mutation. The sensitivity of the FluoroType MTBDR for detecting katG and/or inhA mutations was lower than that of the Genotype MTBDRplus. Two of the isolates falsely reported as INH susceptible by the FluoroType MTBDR had a rare combination of mutations, an S315T1 katG mutation and a C-17T mutation in the inhA promoter region, and one isolate had a T-8A mutation in the inhA promoter region. Theoretically the FluoroType MTBDR has the potential for continuous improvement when new mutations and their respective melting curve properties are detected and fed back into the analytic software. Whether and how frequently improvement through software updates will occur remain to be seen. WHO recommends that commercial molecular LPAs such as the Genotype MTBDRplus may be used for a positive sputum smear specimen or a cultured isolate of M. tuberculosis complex to detect resistance to RMP and INH (8). The Genotype MTBDRplus offers benefits compared to phenotypic DST when used on cultured isolates. First, results are available within 48 h of the culture becoming positive, in contrast to the 7 to 14 days to results with culture-based DST (17). Second, these assays are not affected by the presence of contamination with bacteria, fungi, or nontuberculosis mycobacteria, as are liquid culture-based DST systems. The Genotype MTBDRplus and FluoroType MTBDR detect mutations in the same three target genes. However, there are some distinct advantages of the FluoroType MTBDR. Except for DNA extraction, it provides a fully automated, closed system with a capacity of testing 95 samples in one batch. This in turn decreases the risk of DNA contamination, which is less problematic when using DNA extracted from cultured isolates but may be a significant problem when testing primary samples. Full automation also means less hands-on time and automated interpretation of results, reducing the risk of analytic errors. Results can be directly imported into a laboratory information system, thus decreasing transcription errors.
This study has several strengths and limitations. It was conducted in a national tuberculosis reference laboratory with standardized procedures for phenotypic and molecular DST and long-standing expertise with LPA and Sanger sequencing. Interpretation of molecular results was performed blinded to the phenotypic DST results. Half of the isolates included in the study were resistant to both RMP and INH. However, the majority of isolates harbored the common S531L rpoB and S315T1 katG mutations, possibly skewing the sensitivity estimate upwards.
In summary, the FluoroType MTBDR assay testing culture isolates of M. tuberculosis complex has a high sensitivity for detecting RMP resistance but a lesser sensitivity for detection of INH resistance. It offers an alternative to the currently WHO-endorsed Genotype MTBDRplus, especially for high-throughput laboratories. However, validation studies from other settings are needed to ensure that isolates with a variety of mutations and combinations of mutations are investigated. Furthermore, studies in low-and middle-income countries with high burdens of multidrug-resistant tuberculosis (MDR-TB) should be performed to assess the performance of the new test in those settings.