Association of gyrA and rrs gene mutations detected by MTBDRsl V1 on Mycobacterium tuberculosis strains of diverse genetic background from India

There is limited data on the use of Genotype MTBDRslVersion 1 (MTBDRsl V1) as an initial rapid screening test to rule out XDR-TB and most importantly its performance in various genotypes of Mycobacterium tuberculosis is scarcely studied. A total of 359 MDR-TB isolates were tested for gene mutations representing second line drug resistance, using the MTBDRsl_V.1 and the results were compared with phenotypic method (Bactec MGIT-960 system) for second-line drug (SLD) susceptibility testing. Genetic lineages of all these isolates were also determined using spoligotyping and SITVIT2 WEB database. The MTBDRsl V1 detected mutations in the gyrA, rrs, and emb genes in 108 (30%), 2 (0.5%) and 129 (35.9%) isolates, respectively. Remaining 120 (33.4%) had no second line drug (SLD) resistance. In 17 (4.7%) isolates mutations were detected in both gyrA and rrs genes. Its concordance with MGIT-960 culture drug susceptibility testing (DST) was 97% and 94.1%, 93.5%, 60.5% and 50% for the detection of XDR-TB, pre-XDR, Ethambutol, and Aminoglycosides/Cyclopeptides resistance. The Beijing lineage was predominant (46%) between both the pre-XDR/XDR-TB isolates. We conclude that MTBDRsl is useful for rapid detection of SLD resistance. Also in pre-XDR and XDR-TB isolates the frequency of relevant genetic mutations was significantly higher in the Beijing strains.

The recent estimates of World Health Organization (WHO) show that more than 10.4 million people had incidental tuberculosis (TB) and 1.7 million died of this disease in 2016 alone. The 2016 data also showed that India is the most vulnerable country to TB with 2.8 million new cases (26.9% of global cases) in the year 2016 1 . The scenario was not much different in 2017. Approximately 600,000 Rifampicin resistant (RIF R ) cases were reported, of whom 490,000 had multi-drug resistant TB (MDR-TB). The drug resistance TB has become a major challenge to the success of TB control programmes around the globe 2 . Not only diagnosis, even the treatment of MDR-TB is difficult, requiring much longer duration of treatment with very costly and comparatively more toxic second-line drugs (SLDs) such as fluoroquinolones (FQ) and aminoglycosides/cyclic peptides (AG/CP) 3 . Moreover, emergence of resistance even to the SLDs is being reported from all parts of the world. This form of TB is known as extensively drug-resistant TB (XDR-TB), which is a more devastating condition with very poor treatment success 4 . In 2016, an estimated 6.2% patients with MDR-TB were found to have XDR-TB. India reported first case of XDR-TB 5 in 2007 and by 2015 more than 117 countries were already struggling with this condition 1,5 .
Therefore, it is essential that all suspected cases of MDR-TB must be investigated for susceptibility to second drugs also, in a timely manner to control the spread of spread of XDR-TB. This has become essential in order to optimize the treatment regimen at the earliest. However, the conventional methods of drug susceptibility testing (DST) for SLDs are more complex due to non-standardized methods and protocols, leading to poor reproducibility and reliability 6 . Also, the conventional culture-based methods are labour intensive and require longer turn around time (TAT) with undesirable treatment outcome and wider window to disease transmission 7,8 . In 2013, a Line probe assay (LPA), also known as Genotype MTBDRsl (MTBDRsl) was developed by Hain's Life Science GmbH, Germany and was approved by the WHO also for clinical use. This molecular test rapidly detects genotypic resistance to FQ, AG/CP and Ethambutol (EMB) within 48-72 h and makes it possible to diagnose pre-XDR-TB and XDR-TB at the earliest 9,10 . Therefore, the use of MTBDRsl has been recommended as a rapid and initial diagnostic test to rule out FQ and AG/CP resistance in all MDR-TB patients in order to initiate effective treatment at the earliest 11 . Recently, WHO recommended the use of MTBDRsl test in place of conventional phenotypic culture-based DST methods especially in high TB burden countries 12 . India, a high TB burden country is keen to implement these guidelines but usefulness of MTBDRsl on MDR TB isolates has not yet been evaluated from India.
The genotypic characterization of MTB isolates becomes essential to understand the clonal expansion of the lineages, their transmission dynamics and association with drug resistance 13 . There are several studies which have shown association of different lineages of MTB with variable pathogenicity and vulnerability to drug resistance [14][15][16][17] . For genotyping of the MTB, spoligotyping is a widely used technique which detects the presence or absence of 43 spacer sequences in clustered regularly interspersed short palindromic repeat (CRISPR) region of MTB 14 . Though several studies are published on association of various lineages of MTB with first-line drug resistance, there is not enough literature on the association of MTB lineages with resistance to second-line drugs 14 .
Thus, the main aim of the present study was to evaluate the usefulness of MTBDRsl assay in comparison with phenotypic line drug resistance testing using the Bactec MGIT-960 system in a programmatic mode. We also aimed to find if there was any association between the second line drug resistance pattern and genetic lineages of MTB isolates.

Materials and Methods
Setting. This retrospective study was conducted in the TB research laboratory, which is a certified routine diagnostic laboratory in the Division of Clinical Microbiology & Molecular Medicine, All India Institute of Medical Sciences, New Delhi, India. All the routine mycobacterial isolates are stored and maintained in laboratory repository after characterization. From this repository, 359 MDR-TB isolates that were stored during 2011-2015 were used in the present study. Patient's clinical details were retrieved as published earlier 14,18 (Table 1). These isolations were made as a part of previous study, which was approved by Institutional ethics committee of the All India Institute of Medical Sciences, New Delhi (reference number IESC/T-39/04.01.2013). All in-vitro methods were performed in accordance with the standard guidelines and following the manufacturer's instructions. The mycobacterial culture and DST were performed using the MGIT-960 system and identification of Mycobacterium tuberculosis (MTB) and Non-tuberculous Mycobacteria (NTM) was done by well established in-house multiplex polymerase chain reaction (PCR) 19 .
Genotype MTBDRsl V1. The MTBDRsl test is based on the DNA strip technology having three steps: DNA extraction, multiplex PCR amplification, and reverse hybridization. All steps were performed as per manufacturer's instructions 10,23 . The DNA obtained from the standard MTB-H37RV strain (as positive control) and one negative control was also tested in each batch in order to check the cross-contamination during hybridization assay and other quality parameters. The test was considered as valid; only when the hybridization bands were obtained on MTB complex control (TUB), conjugate controls (CC) and the amplification controls (AC) along with the targeted gene loci controls. For convenience MTBDRsl V1 will be referred as MTBDRsl only hereafter.
Genotyping of MDR-TB isolates by spoligotyping. DNA extraction for spoligotyping. DNA extraction from MTB cultures grown on L-J medium was performed using chloroform iso-amyl alcohol (CI) method as mentioned previously 24 .
Spoligotyping. Spoligotyping was performed using the commercially available kit (Ocimum Biosolutions, Hyderabad, India) by amplification, hybridization and finally detection of hybridizing DNA 25 . In brief, the PCR amplified products were hybridized on a membrane and images were detected with electro-chemi-luminescence (ECL) detection kit (GE Healthcare, Life Sciences, UK) on X-ray films (Kodak, Rochester, NY) 14 . The MTB H37Rv and M. bovis-BCG strains were included as a quality control in every batch of test. The hybridization patterns obtained in the binary format were transformed to an octal code for assessment with the spoligotype patterns using SITVIT2 database, which is an updated version of SITVIT_WEB database 26 . A shared type was defined as a spoligotype pattern common to at least two isolates, and clades were assigned according to the signatures as published earlier 14 .
Statistical analysis. Results of MTBDRsl were analysed and compared with second-line Bactec MGIT-960 DST, which was considered as the gold standard. Data was statistically analysed to calculate the agreement between MTBDRsl and SL-MGIT DST using OpenEpi 3.01. Moreover, Fisher's exact test was performed using STATA 11.1 software to observe significance of the association amongst the second line drug resistant mutation patterns among the different genotypes.
Cluster analysis. The cluster analysis revealed that our isolates belonged to overall 47 clusters with the size of 2-65 isolates in each cluster. The highest clustering (31 clusters) was found among the MDR-TB isolates but sensitive to second-line drugs. Amongst the pre-XDR-TB isolates highest number of clusters were observed in the CAS lineage having 5 clusters (2-18 isolates in each cluster) followed by Beijing with 2 clusters of 2-46 isolates. However, in the XDR-TB isolates only one cluster was found with 8 isolates all belonging to the Beijing lineage.
Isolates showing unique patterns were also found in a single cluster with two isolates in the pre-XDR TB isolates. Interestingly, most of the X, T, orphan and unique strains were un-clustered (Table 5). Minimum Spanning Tree (MST) analysis done by using the MIRU-VNTRplus software, revealed various SITs amongst different regions showing predominant SITs and evolutionary relationship of the lineages and their SITs. MST connects each genotype based on the degree of changes required to go from one allele to another.   The length of the branches denotes the distance between any two patterns whereas the intricacy of the lines indicates the number of spacers between the two patterns. The thicker lines represent 1 change while thinner ones indicate 2 or 3. The size of the circle is comparative to the total number of MTB isolates in this study. The colour of the circles represents the phylogenetic lineage to which the specific pattern belongs. Beijing patterns are circled in red and yellow indicates CAS strains. EAI strains are in dark green colour while EAI strains are in dark green colour. The Clustering and MST of MDR-TB isolates (but susceptible to second line drugs) is shown in Fig. (1). The clustering details for pre-XDR strains are given in Fig. (2). Out of 111 isolates, a total of 88 (79.3%) were grouped into 13 clusters, whereas 23 (20.7%) were non-clustered isolates of which 8 (34.8%.) were unique non-clustered isolates. In case of XDR-TB strains these details are shown in Fig. (3). Out of 17 isolates, 8 (47.1%) isolates could be grouped only in 1 cluster, whereas 9(52.9%) were non-clustered isolates of which 1 (11.1%.) was unique non-clustered isolate.  Table 5. Distribution and clustering pattern of MDR-TB isolates. n = 2) and SIT 1166 (T1, n = 1) were found. In the XDR-TB isolates, SIT 464 (Ural, n = 1) was the only new SIT found to be evolved ( Table 4).
Discussion. The first decade of this century witnessed the emergence of XDR-TB strains for which management is extremely difficult, and this imposed serious concerns for the health care systems around the world 1,5 .
To develop tests to diagnose XDR-MTB strains at the earliest possible time have become an urgent need. Various research organizations are working untiringly to develop such tests and devices. A German based company developed the first version of MTBDRsl for rapid screening of FQ and AG/CP resistance in the MDR-TB isolates. After preliminary evaluations, in 2013, the WHO recommended use of this test to rule out the XDR-TB 27 . However, despite India being the high TB burden country, its usefulness of this test has not been evaluated on a large number of isolates. In the present work, we evaluated the efficiency of MTBDRsl V.1 in comparison to the second-line DST using MGIT-960 system, which still remain the gold standard for second line drug susceptibility testing. We also analysed for the first time, the association of various lineages of MTB with genetic mutational patterns. Even though it is a molecular test, which is prone to several procedural errors, we found that the test protocol is highly standardized giving no invalid results, which means that all (100%) tests were valid. Taking the phenotypic MGIT-960 system as the gold standard, we found very high (97%) concordance of MTBDRsl for detecting the second drug susceptible isolates and for FQ resistance detection (93.5%). However, more improvised version will be required to detect AG/CP resistance, where its performance was not found very high (50%). Similar findings regarding detection of AG/CP resistance have also been previously published from Spain and China reporting sensitivity of 56% and 67% respectively 28,29 . Nevertheless, the test was found extremely good (94.1%) for XDR-TB detection in our study as well as by other studies published from Serbia and Netherlands both showing 100% sensitivity 3,30 . Similar to our results the lower detection rate for EMB resistance of 56.2% was reported from China. These authors performed this test on MDR-TB isolates and emphasized identification of other mutations for detection of EMB resistance for improvement of the test 31 . Hence the identification of novel mutations outside the QRDR gene region of gyrA and gyrB and rrs gene is urgently needed. The new version of the assay (i.e. MTBDRsl V.2.0) has been recently developed by the company for improvement to overall performance of MTBDRsl V.1 and in particular to its sensitivity for detection of kanamycin resistance. However, at the time of study this version was not available in India 32,33 . While correlating the association of gyrA and rrs gene mutations in various lineages 34 we found that even though most of our isolates were from the Northern-Western part of India (270; 75.2%) where prevalence of Figure 3. Minimum spanning tree of XDR-TB isolates using spoligotyping method as mentioned above. Each circle represents a genotype. Each circle represents a genotype. The distance between circles represents how closely related are different genotypes to each other.
SCIeNtIfIC REPORTS | (2018) 8:9295 | DOI:10.1038/s41598-018-27299-z CAS strain is predominant yet the association of Beijing isolates was statistically significantly high (p-0.0006) with gyrA gene mutations but insignificant (p-0.079) with rrs gene mutations. Beijing strains are considerably prevalent in South East Asia and North-Eastern region of India, where its is posing a major concern due to its high prevalence among the MDR-TB patients 19,31,35,36 . However, the recent studies from India show that this strain is spreading fast to other parts of India and neighbouring countries 14 , which can be warning signal to the TB control programme managers. Within the Beijing genotype SIT1 was most predominant share type amongst the pre-XDR and XDR-TB isolates in our study. SIT1 was also reported predominant among XDR-TB isolates from Africa (34%) and Russia (9.5%) 37,38 . We for the first time reported 18 new SITs and 5 Orphans in the Indian MDR-TB isolates, though these types have previously been reported from other countries, like SIT67 (1.4%) from United States and Mexico, SIT243 (1.1%) from Zambia, Vietnam and Italy, SIT 794 (1.1%) from Bangladesh, Pakistan and United States [39][40][41][42][43] . The two newly evolving SITs in the Beijing genotype [SIT 1168 (0.8%) and SIT 190 (0.5%)], were found only in the MDR and pre-XDR TB isolates. These SITs have been reported from Unites States, Thailand, Japan, Vietnam and China, but never from India 41,43 . The Ural SIT 464 was the only strain evolved among the XDR-TB isolate which have never been reported from India but only from United States of America 39 . Evolution of new share types among the MDR, pre-XDR and XDR-TB isolates from India indicates these strains got transmitted to India through migration of population from such geographical regions in recent years 14 . We also report association of Beijing genotypes with very high frequency of 2 gyrA gene mutations-ΔWT3-D94G and ΔWT2-A90V (Table 4).
In conclusion, even though our study had some limitations such as not being able to monitor the progression of disease and treatment outcome of patients due to the use of archival MDR culture isolates, our study signifies that MTBDRsl V1 is a good diagnostic tool for the detection of pre-XDR and XDR-TB. We propose that MTBDRsl V1 should be used on all MDR-TB isolates in place of phenotypic culture DST methods, till its second version is made available, in the programmatic mode. This strategy is more pertinent for countries and regions where pre-XDR and XDR-TB prevalence is high. We also conclude that more prospective genotyping studies along with next-generation sequencing methods be implemented in order to ensure the understanding of the vulnerability of some genotypes to the drug resistance development and the molecular mechanisms leading to the emergence of pre-XDR and XDR-TB strains.