Impact of Whole-Genome Sequencing of Mycobacterium Tuberculosis On Treatment Outcomes for MDR-TB/ XDR-TB: A Systematic Review Protocol

The emergence of drug-resistant tuberculosis (DR-TB) is a persistent threat to public health. The detection of DR-TB requires culture-based drug susceptibility testing (DST) or rapid molecular assays for targeted genes. The recent advances in Whole-genome sequencing (WGS) technology have offered a new capacity to identify resistance-conferring mutations in Mycobacterium tuberculosis (MTB). This study reviews and quanties the emerging evidence on the association between genomic markers of drug resistance in MTB identied by WGS and treatment outcomes for DR-TB.


Introduction
Despite the global efforts, drug-resistant tuberculosis (DR-TB) remains a major obstacle in tuberculosis (TB) eradication programs. Globally in 2019, the World Health Organization (WHO) reported around 465,000 new cases and approximately 182 000 deaths due to multi-drug resistant TB (MDR-TB)/ rifampicin-resistant TB (RR-TB) [1]. As per WHO, 17.7% of prior treated cases and 3.3% of new patients had RR/MDR-TB, and successful treatment outcomes were accounted for only 57% of cases with RR/MDR-TB [1]. Mycobacterium tuberculosis (MTB) isolates resistant to rst-line drugs isoniazid and rifampicin are de ned as MDR-TB, and extensively drug-resistant TB (XDR-TB) are referred to as MDR isolates, along with resistance to a uoroquinolone and another Group A drug (e.g., linezolid or bedaquiline) [2][3][4]. The detection of DR-TB using culture-based drug susceptibility testing (DST) can be challenging. It remains time-consuming, expensive, and requires a high-capacity laboratory, as well as methodological standardization of testing for second-line antimycobacterial agents [5,6]. In 2019, only 71% of RR/MDR-TB cases were tested for second-line drug uoroquinolones, which could be due to inaccessibility of test facilities or culture of slow-growing MTB for phenotypic DST [1].
Currently available genotypic assays can rapidly provide common drug resistance which is advantageous over time-consuming phenotypic DST; however, these diagnostic methods can only detect speci c genetic mutations already known to confer drug resistance [7,8]. Whole-genome sequencing (WGS) has become a reliable tool to overcome the limitations of probe-based diagnosis methods and provide a better insight into emerging clinically relevant mutations in the MTB genome [9,10]. WGS is relatively rapid and has the capacity to identify a broad range of genetic polymorphisms, which can guide us to new treatment strategies and improve case management for DR-TB [8,11]. The WGS -based applications are expanding to TB diagnostics and recent studies demonstrated the applicability of WGS in the clinical management of tuberculosis [10,12,13]. Also, previous systematic reviews of Mycobacteriology have emphasized the application of WGS from a different perspective. The use of WGS for TB transmission and outbreak investigations was reviewed by Hatherell HA [14] and Van Der Werf MJ et.al. [15], whereas, Papaventsis D et.al [8] and Nieto Ramirez LM et.al. [16] addressed the diagnostic accuracy of WGS for detecting the mutations conferring drug resistance. This systematic review will bridge the knowledge gaps on the e cacy of MTB whole-genome sequencing on patients' treatment outcomes.
The standardized treatment approach for individual MDR-TB cases does not always lead to successful treatment outcomes. As several patients with a history of TB treatment develop MDR/XDR TB, baseline drug resistance has been recognized as a major factor that impacts the treatment e cacy [17]. Therefore, in this present review, we aimed to assess the applicability of WGS to detect genomic markers of drug resistance in M. tuberculosis and explore their association with treatment outcomes.

Objectives:
To evaluate the e cacy of whole-genome sequencing as a predictor of treatment outcomes for multidrug-resistant/extensively drug-resistance tuberculosis compared to phenotypic drug-susceptibility testing.

Methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) statements were used to develop this systematic review protocol [18] (Additional le 1). This present protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) database (registration ID: CRD42020197099).

Eligibility criteria
We will include non-Randomized clinical trials (NRCT), Randomized clinical trials (RCT), and observational studies, such as case-control, cohort, case series describing different applications of WGS to genotypic resistance testing for TB and detection of MDR-TB/ XDR-TB as well as treatment outcomes of the patients. The exclusion and inclusion criteria of this study are de ned using PICOS (Population, Intervention, Comparator, Outcome, Study design) framework [19] (Table 1). Five electronic databases will be used to conduct a literature search i.e. Scopus, Cochrane Library, Web of Science, NCBI PubMed (MEDLINE), and CINAHL (Ebsco). The structured search strategy will be employed based on the keywords related to "drug resistance tuberculosis", 'tuberculosis', "treatment outcome" and "whole genome sequencing". Also, the site-speci c biomedical terms like Medical Subject Headings (MeSH), Emtree will be included and the Boolean operators will be used to combine all of the search terms [20] (Additional le 2). The search data will be reviewed and validated by a specialist health science librarian. The results will be limited to the English language and original research articles from the year 2000 onwards will be included. Letters to the editor, review articles and comments will not be included in this study. There will not be any geographical lter as the review intends to include global data.

Selection of studies
The search results retrieved from the electronic databases will be imported in Zotero and duplicate studies will be removed. For the title-abstract screening phase, the retrieved studies will be exported to a Microsoft Excel spreadsheet (Microsoft, 2019). Two reviewers will independently assess the abstracts using eligibility criteria. Studies identi ed as relevant by at least one of the reviewers will be included for full-text assessment. The full text will be retrieved and assessed by two reviewers independently. Any disagreement between reviewers' interpretation will be initially discussed between them, and, if not resolved, will be clari ed by a third reviewer. The nal decision of including/ excluding a study will be con rmed by all reviewers and appropriately documented in the PRISMA ow chart [21].
Data extraction: Pertinent data will be extracted separately by the primary reviewers from each included article. A standardized data extraction spreadsheet will be developed by the reviewers after a pilot assessment. The key details to be extracted will include the following:

Risk of bias and quality assessment
The quality assessment of included studies will be performed using the Cochrane Risk of Bias 2 (RoB 2) tool for RCT [22], Cochrane Risk of Bias in Non-randomised Studies of Interventions (ROBINS-I) tool for NRCT [23] Newcastle-Ottawa Scale (NOS) for quasi-experimental and observational studies [24] or based on the study designs. Two reviewers individually will determine the risk of bias and based on the study design, studies will be classi ed as having a 'low,' 'high,' or 'unclear' risk of bias. A third reviewer will address any discrepancies between the primary reviewers.

Data synthesis and analysis
The study characteristics and outcome data will be collated into tables and summarised narratively. The I 2 statistics will be used to assess the heterogeneity of the included studies and if the data are su ciently homogenous, a meta-analysis will be performed. Random or xed-effect meta-analysis will be carried out to estimate the individual study effect and produce a pooled estimate summarizing data on the predictors of all distinct treatment outcomes [25,26]. Standardized mean differences will be used to assess continuous outcomes and odds ratio (OR) and Risk ratio (RR) will be used to estimate the effect sizes of categorical outcomes, with a 95% con dence interval (CI) [27]. The degree of heterogeneity among the studies will be represented visually using a forest plot [28] and the value of the I 2 statistics and the Egger's test and visual representation of the funnel plot will be used to assess publication bias [29,30]. Narrative data synthesis will be conducted for all the included studies if performing meta-analysis is not possible. We will adhere to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach to assess the data quality presented in the systematic review [31,32].
Reporting of the systematic review and meta-analysis-PRISMA 2020 guidelines will be used to report this systematic review [21].

Subgroup analysis
A subgroup analysis will be conducted, in cases of a signi cant heterogenicity observed in pooled estimates, considering the following grouping characteristics [27,33] i. Type of drug-resistant strains of MTB: MDR-TB, XDR-TB.
ii. Participants demographic and risk factors: age, HIV status.
iii. Study characteristics: geographical area and study design.
iv. Types of sequencing platforms and data analysis tools.

Discussion
Despite recent advances in the application of WGS to the prediction of drug resistance in MTB, the association between detection of resistance or susceptibility from genome sequence data and the treatment outcomes in the case of DR-TB/ MDR-TB/ XDR-TB remains untested. This systematic review will interrogate the existing data on the feasibility and added value of WGS in improving treatment outcomes in TB patients. The rapid detection of drug-resistance conferring mutations and selection of appropriate drug regimens is likely to improve the cure rates while minimizing adverse events and treatment costs for TB patients. Hence, the outcome of this systematic review will be bene cial for policymaking and will guide clinical laboratory practice to improve drug resistance diagnostic capacity and treatment outcomes. While the genomic sequencing methods and bioinformatics analyses employed in studies included in this systematic review will not be speci cally scrutinised, we acknowledge the importance of wet and dry laboratory techniques and analytic pipelines in the implementation of WGSguided selection of therapy for DR-TB [34]. We will pay particular attention to the interpretation of drugresistance mutations, their impact on phenotypic DST results, and the ranking of their signi cance and will recommend a minimum data required for a standardised assessment of genetic mutation-clinical outcome associations in order to avoid biases. In addition, this review will shed light on important challenges in the translation of genomic research into TB control and personalised therapeutic interventions and identify relevant research gaps for further studies. Availability of data and materials: Not applicable

List Of Abbreviations
Competing interests: The authors declare that they have no competing interests.

Funding: None
Authors' contributions: The study was conceptualized by VS, KC, DH, CL and supervised by KC, VS, VSD, and BTV. The search strategy was designed by DH, VSD, and CL, and the literature review was performed by DH and CL. The manuscript has been written by DH and reviewed and improved by VS, KC, VSD, CL, and BTV. All authors critically revised the current protocol and approved the nal version of this manuscript.