Strengthening the Diagnosis of Drug-Resistant Tuberculosis Using NGS-Based Approaches and Bioinformatics Pipelines for Data Analysis in India

Abstract

In India, drug-resistant tuberculosis (DR-TB) is a major public health issue and a significant challenge to stop TB program. An estimated 27% of new TB cases and 44% of previously treated TB cases are resistant to at least one anti-TB drug. The conventional methods for DR-TB diagnosis are time-consuming and have limitations, leading to delays in treatment initiation and the spread of the disease. Next-generation sequencing (NGS) based approaches have emerged as a promising tool for diagnosing DR-TB, simultaneously offering rapid and accurate detection of resistance mutations in multiple genes. NGS-based approaches generate a large amount of data, which requires efficient and reliable bioinformatics pipelines for data analysis. TBProfiler and Mykrobe are the bioinformatics pipelines that have been created to analyze NGS data for the diagnosis of DR-TB. These pipelines use reference-based and machine-learning approaches to detect resistance mutations and predict drug susceptibility, enabling clinicians to make informed treatment decisions. Implementing NGS-based approaches and bioinformatics pipelines for DR-TB diagnosis can potentially improve patient outcomes by facilitating early detection of drug resistance and guiding personalized treatment regimens. However, the widespread adoption of these approaches in India faces several challenges, including high costs, limited infrastructure, and a lack of trained personnel. Addressing these challenges requires concerted effort to ensure equitable access to and effective implementation of these innovative technologies.

Multi-drug-resistant tuberculosis (MDR-TB) is the leading cause of death and a global health issue. Drug-resistant tuberculosis (DR-TB) poses a significant challenge to TB control and eradication in Asian countries. Rapid detection of DR-TB is critical to improving patient outcomes. According to the World Health Organisation (WHO) global high burden country list 2018, India occupies 5th place [1]. Asian countries contribute to about one-third of the overall MDR burden [2]. Diagnoses of DR-TB using the gold standard culture method face many challenges. Traditional culture-based methods (BACTEC MGIT 960 liquid culture or Lowenstein–Jensen Solid culture) for drug susceptibility testing (DST) requires a long time, often several days to weeks, to detect the threat of possible culture contamination. Moreover, it requires trained technicians. WHO has endorsed various molecular technologies for the detection of DR-TB, such as GenoType MTBDRplus (2008); Xpert MTB/RIF (2010), GenoType MTBDRsl (2016), Loopamp MTBC assay (2016); Xpert MTB/RIF Ultra (2017) Truenat MTB plus and Truenat MTB-RIF Dx (2020), etc. [3]. These techniques have a shorter turnaround time from hours to days, thus delivering faster results. However, these methods carry limitations in detection capacity and are confined to identifying DR-TB with known mutations only. Furthermore, they can detect only a few first-line and second-line drug resistance organisms [4]. In some instances, strains identified as resistant by the proportion method were found to be susceptible to LPA, suggesting that this could be due to silent mutations that failed to illicit any different phenotype [5]. Heteroresistance in TB further complicates the diagnosis and management of DRTB. Incomplete characterization of DR-TB by current culture-based and molecular diagnostic tools results in delayed opposite treatment leading to increased incidences of drug resistance. Thus resulting in additional healthcare costs and increased morbidity besides the unrelenting spread of drug-resistant strains [6]. The limitation, as mentioned earlier, necessitates the induction of next-generation sequencing (NGS) and the development of bioinformatics pipelines for the rapid detection of DR-TB. Besides providing rapid detection, these tools will also enable understanding of transmission patterns that would benefit public health initiatives. Whole-genome sequencing (WGS) may be essential in detecting DR-TB and identifying novel mutations that can lead to resistance. However the analysis of WGS data has issues as there is no standard tool which can perform analysis for all the functions such as resistance detection, identify lineages, phylogenetic analysis and transmission dynamics. Diagnostic performance of WGS and phenotypic DST in DR-TB using tools such as KvarQ, TGS-TB, CPL-TB, CASTB, MTBseq, ReSeqTB-UVP PhyResSE, TB-Profiler, TB Portals and Mykrobe have yielded good results Table 1 [7, 8]. TB-Profiler and Mykrobe are open-resource online bioinformatics pipelines that allow whole genome sequence analysis of Mycobacterium tuberculosis (MTB) to predict drug resistance using the curated tbdb database and lineage (http://tbdb.bu.edu/). TB-Profiler enables variation and deletion analysis for resistance using default Trimmomatic, Burrows-Wheeler Aligner (BWA), and Genome Analysis ToolKit (GATK). It uses fastq files that are easily uploaded and can be analyzed within a few minutes. The analysis requires no expertise [9, 10]. Mykrobe works offline and requires no specialized training for analysis and interpretation [11]. TB-profiler and Mykrobe support both the Illumina and Nanopore sequencers. Both pipelines are open source. MTBseq is used to perform phylogenetic and clustering/network analysis while CPL-TB provides details on spoligotypes and MIRU-VNTR types. Tools such as KvarQ] and PhyResSE provide genotyping analysis but not in the formats of IS6110-RFLP and MIRU-VNTR. However, there is a dearth of Indian platforms which will have data regarding mutations and strain variations from our country. There is an urgent need for real-time monitoring of the incidence of MTB drug resistance in India. NTEP has also sought expansion of genomic sequencing for lineage determination and drug susceptibility testing in NSP 2020–2025. However, data analysis for actual time application and clinical decision-making is lacking. A Customized bioinformatics pipeline for analysis of circulating drug-resistant strains in India is required to detect drug resistance in Indian strains accurately. Recognizing the need, DBT India has initiated a project on data-driven research to eradicate TB, “Dare2eraD TB,” which includes the Indian TB genomic surveillance consortium for the collection and storage of genomic data. (https://dbtindia.gov.in/sites/default/files/Dare2eraD%20TB.pdf). A broad representation of strains from all over the country is being collected to achieve the goal. All the diagnostic laboratories in the NTEP ambit should support this initiative and the data needs to be collated and used for developing customized tool for the country. This initiative will support the larger goal of TB elimination.

Table 1.

Bioinformatics tools for whole genome sequencing data analysis

Sr.	Tools	Pipelines	Input file	Report	Platform	Country
1	PhyResSE	BWA mapping (H37Rv Version 3), GATK. FastQC, BWA, QualiMap, SAMtools	Fastq	Strain lineage and antibiotic resistance	Web-based	Germany
2	CPLP-TB	Spoligotyping and mycobacterial interspersed repetitive units—variable number of tandem repeats (MIRU-VNTR)	NA	Strains	NA	Portugal
3	GMTV	Python and MySQL database	VCF file and a FASTQ	Molecular variation and clinical consequences as well as facilitating epidemiological surveillance of TB and HIV/TB co-infection	Web genome browser, GMTVB, based on JBrowse platform	Russia
5	MIRUVNTRplus	JavaServer, MySQL, variable number of tandem repeats (VNTR)-typing, mycobacterial interspersed repetitive units (MIRU) as genetic markers	MS Excel-file or a CSV-file	Large sequence polymorphisms (LSPs or RDs; regions of difference) or single nucleotide polymorphisms (SNPs)	Web-based	NA
6	ReSeqT	UVP with custom Python programing scripts	Fastq	Strain lineage and drug resistance	NA	California, United States
7	SITVIT2	Spoligotyping and MIRU-VNTR patterns from WGS data (SpolPred, SpoTyping, MIRU-profiler and MIRUReader)	Fastq	Lineage and/or antibiotic resistance prediction	Command line	France
8	tbvar	NA	NA	Variant calling and phylogenetic analysis of TB genomes		United States
9	TB Portals	Web interface, PostgreSQL for the database, Highcharts v6.1.1, Javascript charts, and R v3.2.5	Fastq	Variants, resistance prediction and SNPs	Web-based and command line	United States
10	MTBseq	BWA, SAMtools, PICARD-tools, Genome Analysis Toolkit (GATK)	Fastq	Variant positions annotated with known association to antibiotic resistance and performs a lineage classification based on phylogenetic single nucleotide polymorphisms (SNPs)	Command line
11	Mykrobe	Stampy mapping (H37Rv Version 2); variant calling SAMtools	Fastq	Variety of mapping and variant calling techniques to determine potential resistance mutations	Application based	United States
12	TBProfiler	Snap mapping (selected regions of H37Rv version 3); variant calling SAMtools		Genotyping and drug resistance prediction		United Kingdom
13	CASTB	Velvet de-Novo assembly; BIGSdb; MUMmer mapping; custom scripts		Strains prevalent in China		China
14	TGS-TB	NA	Paired-end fastq.gz	Drug resistance prediction and lineage analysis	Web-based	NA
15	KvarQ	Python modules/packages with C extensions packages with C extensions	Fastq	Variant calling SNPs detection	Command-line tool	South Korea
16	Galaxy/vSNP	SAMtools, BWA-MEN (BAM)	Fasta, Fastq, VCF, CSV	Variants, resistance prediction and SNPs	Web based and command line	Australia
17	SNiPgenie	Phyton	Fastq	Variants, SNPs and INDELs	Command-line tool	NA
18	MycoVarP	NA	Fastq	Variants to the drug-susceptible	Command-line tool	NA
19	SNVPhyl	SNVPhyl is implemented as a Galaxy workflow	Fastq	Strain typing and lineage assignment based on single nucleotide variants (SNVs)	Command-line tool	Canada

Author Contributions

Conceptualization: VKT, JB; Methodology: VKT, NSP, SR, JB; Writing the original draft; VKT, Reviewing and editing: SR, JB.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.