Costa et al . : The First National Anti-TB Drug Resistance Survey in Timor-Leste

Constantino Lopes National TB Programme, Ministry of Health, Timor-Leste Debashish Kundu (  debashishkundu@yahoo.com ) WHO: Organisation mondiale de la Sante https://orcid.org/0000-0002-3756-8623 Ismael Da Costa Barreto National Health Laboratory, Ministry of Health Bernardino da Cruz National AIDS Programme, Ministry of Health S Siva Kumar National Institute for Research in Tuberculosis (WHO SNRL), Chennai Sureshbabu Ramalingam National TB Reference Laboratory, Dili Anna S. Dean Global TB Programme, World Health Organization, Geneva Vineet Bhatia World Health Organization Regional O ce for South-East Asia Prabhu Seenivasan National Institute of Research in TB (WHO SNRL), Chennai C Padmapriyadarsini National Institute of Research in TB (WHO SNRL), Chennai Olga Tosas Auguet Global TB Programme, World Health Organization, Geneva


Introduction
Timor-Leste is a low-middle income young island nation with a population of about 1.2 million and an annual population growth rate of 1.8%. The country is in the eastern end of the Indonesian archipelago and is divided into 12 municipalities and one special economic zone (Oecusse). The country has an area of approximately 15,000 km 2 , a large part of which consists of hilly and mountainous terrain which is di cult to access. The population is thus scattered and there are large distances between people's residences and the closest TB diagnostic centre. The country has the second highest estimated tuberculosis (TB) incidence rate and highest estimated mortality rate in the South-East Asia Region 1,2 .
In 2018, WHO estimated a burden of 3.1% (1.1%-6.1%) rifampicin-resistant (RR) TB among new cases, and 15% (7.8% − 23%) among retreatment TB cases for Timor-Leste 3 . Accordingly, it was estimated that 240 cases of RR-TB (a rate of 19/100,000 population) 3 are emerging annually in the country. These presurvey estimates were model-based, and not based on country data due to the lack of a continuous surveillance system for drug-resistant (DR) TB. The number RR-TB cases that were detected and noti ed between 2013-2018 as per the WHO Global TB Reports ranged from 2 to 12. In 2019, 9 were laboratory con rmed RR-TB cases. 52% of pulmonary TB cases in 2019 were bacteriologically con rmed, and only 11% (458/ 4050) of new and relapse cases reported using a WHO-recommended rapid diagnostic -Xpert MTB/RIF as the initial diagnostic test.
The rst national DRS was conducted by the National TB Programme (NTP) and the National TB Reference Laboratory (NTRL) in 2019, with the technical assistance from the World Health Organization (WHO) and the National Institute of Research in Tuberculosis (NIRT), Chennai, India (a WHO Supranational TB Reference Laboratory, SNRL).
The primary objective of the DRS was to determine the prevalence of resistance to rifampicin among new and previously treated sputum smear-positive pulmonary TB cases in Timor-Leste, to inform the planning of the programmatic management of DR-TB (PMDT), including the roll-out of the WHO-recommended alloral shorter regimen for RR-TB 4 , and guide the resource needs. The secondary objectives were to describe the sociodemographic and clinical characteristics of bacteriologically-con rmed pulmonary TB patients; investigate potential risk factors for RR-TB; establish baseline data for surveillance of DR-TB in order to allow the observation of trends over time; and strengthen the routine surveillance of anti-TB drug resistance in Timor-Leste.

Study population
The national cross-sectional study targeted all newly registered (new and previously treated) pulmonary TB patients in the country who were sputum smear-positive by microscopy. Following a one-month pilot to test all study procedures, eligible patients were enrolled into the survey over nine months (from 21st January 2019 to 5th September 2019). Clinically diagnosed TB patients (i.e. those diagnosed without bacteriological con rmation of TB by microscopy), extra-pulmonary TB patients, sputum smear-negative cases, patients who had already received more than seven days of treatment in their current treatment course, and prisoners were excluded from enrolment. Study design and sampling strategy STROBE statement was followed to include check-list items in reporting this cross-sectional study 5  WGS was conducted using Illumina HiSeq X Ten system (Illumina, San Diego, CA, USA), following extraction of genomic DNA from the clinical isolates using the CTAB (cetyl trimethylammonium bromide) method. DNA was puri ed using the Genomic DNA Clean and Concentrator kit and assessed for quality and quantity using Nano Drop and Qubit dsDNA Assay kits (Thermo Fisher Scienti c, Waltham, MA, USA). Libraries were prepared using NEBNext Ultra DNA Library preparation kit (Illumina, San Diego, CA, USA). For the subset of isolates tested by WGS, mutations conferring resistance to isoniazid, ethambutol, streptomycin, pyrazinamide and uoroquinolones were also examined.

Data management
An electronic web-based DRS database (https://tb.ms.gov.tl/tb1) 9 was designed prior to the start of the survey for the purposes of data entry and storage of the data from the DRS. The DRS database was designed to meet the requirements of con dentiality principles and was password protected and backedup daily to a secure server.
The database contained all the data elds from patient questionnaires as well as all NTRL and SNRL laboratory test results. A separate section of the database was designed for the electronic tracking of samples, shipments, and reports of laboratory results to CHCs.

Data analysis
Survey summary statistics, data cross-tabulations and analysis of potential risk factors of RR-TB were conducted using R statistical software (version 3.5.2) 10 . For each anti-TB drug, the prevalence of resistance along with the corresponding 95% con dence intervals (95% CI), was estimated by tting a logistic regression and then taking the inverse logit of the model coe cient and con dence interval.
Resistance to rifampicin was determined by considering both Xpert MTB/RIF and WGS test results, with the latter providing a conclusive result in case of discrepancies. Associations between RR-TB and clinical and demographic traits were explored in univariate logistic regression analyses adjusted by previous anti-TB treatment history. Measures of association for potential predictors were summarised by odds ratios with the corresponding 95% con dence intervals.

ETHICAL CLEARANCE
Ethical clearance was obtained from the Instituto Nacional da Saude (INS), Ministry of Health (MOH).

Results
Clinical and demographic characteristics of the TB population A total of 953 eligible patients were enrolled from 69s CHC across 12 municipalities and one special economic zone over the study period. Seven CHCs did not report any TB patients. The median time taken from sample collection to receipt of the samples at NRTL was 2 days (interquartile range of 1-3 days). Approximately 19.3% (184/953) of the samples took more than 4 days to arrive to NTRL.
Out of 953 enrolled patients, 917 were con rmed as pulmonary TB cases by either Xpert MTB/RIF and/or culture ( Figure 1). More than half of bacteriologically con rmed TB patients were male (57.1%), and 8.0% had been previously treated for TB (Table 1). Only 2.1% of TB cases were children aged <15 years. More TB cases were identi ed from individuals aged 15-34 years (45.9%). A total of 12 (1.3%) patients were coinfected with HIV. Most cases were detected from the capital, Dili, and the Ermera municipality, which together comprised 59.4% of the cases ( Table 1). The age and sex population structure for bacteriologically con rmed pulmonary TB patients enrolled in the study (n=917) is shown in Figure 2.
Sample processing work ow at NTRL and SNRL: Out of 953 patients enrolled into the study based on smear microscopy ndings at CHCs, 913 were con rmed as pulmonary TB patients by Xpert MTB/RIF and a further four were con rmed by culture (n=917). See Figure 1. Out of those con rmed as pulmonary TB by Xpert MTB/RIF, seven were RR-TB and 905 were RS-TB. Rifampicin resistance was indeterminate by Xpert MTB/RIF for one case. Phenotypic DST for this indeterminate specimen, plus four isolates that were con rmed as MTB by culture, showed all ve isolates were RS-TB, totaling 7 RR-TB isolates plus 910 RS-TB isolates by Xpert MTB/RIF and/or culture.
WGS was conducted for validation purposes on 65 isolates (29% of RR-TB (2/7) and 7% of RS-TB (63/910) by Xpert MTB/RIF or pDST). Of these, one isolate was excluded due to inconclusive speciation results, resulting in non-interpretable resistance pro ling (Figure 1). The original test results agreed with the WGS validation results for 62/64 isolates (97%). WGS re-classi ed one RR-TB by Xpert MTB/RIF as susceptible to rifampicin, and one RS-TB by Xpert MTB/RIF as resistant to rifampicin. For the case of the isolate that was con rmed as resistant by WGS, a rare mutation associated with rifampicin resistance (Ser441Gln) was identi ed in the RRDR region. This was an unusual double mutation -where two consecutive nucleotides are replaced by two different nucleotides on the genome-leading to rpoB 441 st codon TCG being replaced by CAG (HGVS nomenclature on rpoB gene sequence: c.1321_1322delinsCA).
No resistance mutations were identi ed in the case of the isolate that was con rmed as susceptible by WGS.

Resistance to anti-TB drugs
The prevalence of rifampicin resistance among new cases was 0.6% (95% CI 0.2 -1.3) and that among previously treated TB cases was 2.7% (95% CI 0.5 -8.2). The prevalence of rifampicin resistance among these groups combined was 0.8% (95% CI 0.3 -1.5). See Table 2. Of the two RR-TB with WGS test results, one also had resistance mutations for isoniazid, ethambutol and streptomycin, and a second one showed no resistance mutations to evaluated drugs (i.e. isoniazid, ethambutol, streptomycin, pyrazinamide and uoroquinolones). Among the 64 isolates with WGS test results, no mutations conferring resistance to pyrazinamide or uoroquinolones were identi ed; mutations conferring resistance to isoniazid, ethambutol and streptomycin were identi ed from seven, two and one isolates respectively. Due to the low testing coverage of M. tuberculosis by WGS (7% [64/917]), resistance estimates and corresponding 95% con dence intervals for drugs other than rifampicin were not calculated in this study.
No factors were associated with RR-TB in univariate logistic regression analyses (Table 3). Multiple regression analysis was not conducted due to the small number of RR-TB cases (n = 7) identi ed.

Discussion
This is the rst national survey of drug resistance among people with TB in Timor-Leste; most TB cases identi ed were male and under 55 years of age. Notably, children <15 years old account for 11% of TB globally, but only represented 2% in Timor-Leste during this 9-month period. The WHO Global TB Report 2020 11 estimated that 8% of noti ed TB cases were children in Timor-Leste, including both clinically diagnosed and bacteriologically con rmed pulmonary and extra-pulmonary TB. This difference may be partially explained by the DRS enrolling only the bacteriologically con rmed pulmonary TB cases. Obtaining bacteriological con rmation of TB in children remains a challenge.
The overall prevalence of RR-TB was 0.8% (95% CI 0.3-1.5%), with 0.6% (95% CI 0.2-1.3%) among new TB patients and 2.7% (95% CI 0.5-8.2%) among previously treated TB patients. This is notably lower than previous modelled estimates prior to the survey. The results of this survey show a lower prevalence than in neighbouring Indonesia (2.4% (1.8-3.3) among new cases and 13% (9-18) among previously treated cases) and in the WHO Southeast Asia Region (2.5% (1.9-3.3) among new cases and 14% (7.7-21) among previously treated cases) 12 . The lower levels of rifampicin resistance in Timor-Leste may relate to the fact that rifampicin was only introduced into the four-month continuation phase of the rst-line treatment regimen from the second quarter of 2015 (in line with the WHO-recommended regimen of 2HRZE+4RH), which was much later than in neighbouring countries. The lower level of rifampicin resistance found in this study in Timor-Leste might also be attributed to a good treatment success rate among TB patients who were enrolled on treatment (88% in new and relapse TB cases in 2018). However, treatment coverage is low (63%) compared to the estimated number of incident TB cases in Timor-Leste, and this also may have led to a lower selection pressure on DR-TB strains. Transmission rates of TB remain high, with the second highest incidence rate (498 cases per 100,000 population) in the WHO South-East Asia Region and among the top ten in the world.
This survey did not nd any signi cant risk factors associated with resistance to rifampicin, likely as a result of the small number of RR-TB cases identi ed (7 cases only).
Although this study did not formally evaluate resistance to drugs other than rifampicin due to the limited number of sequenced isolates, mutations conferring resistance to isoniazid, ethambutol and streptomycin were identi ed, with 7/64 isolates displaying genotypic resistance to isoniazid. No resistance-conferring mutations for pyrazinamide or uoroquinolones were observed among sequenced samples, which may bode well for the management of DR-TB in the country. However, WGS was not systematically performed for all samples.
The WHO End TB Strategy calls for the early diagnosis of TB and universal drug susceptibility testing (DST), highlighting the critical role of laboratories for rapidly and accurately detecting TB and drug resistance. DR-TB is a public health crisis and a major threat to global TB control. WHO recommends the use of Line Probe Assay (LPA) as a rapid diagnostic test for the detection of rifampicin, isoniazid, uoroquinolone and pyrazinamide resistance 12 . A Regional Green Light Committee (r-GLC) mission in 2018 in Timor-Leste recommended investment by the country in LPA 13 , to rapidly detect isoniazid and uroquinolone resistance to inform appropriate patient management. The TB Joint Monitoring Mission (JMM) conducted in Timor-Leste in 2019 also recommended that the NTP consider introducing LPA capacity into the NTRL. By mid-2021, it is anticipated that all bacteriologically-con rmed TB patients will be offered DST for rifampicin and isoniazid, as well as uoroquinolones among cases of RR-TB, thus facilitating timely clinical management of TB patients and appropriate implementation of WHO guidelines on use of all-oral shorter regimen.
This study showed good agreement between Xpert MTB/RIF and WGS results (97%), suggesting that nationwide diagnosis of RR-TB through Xpert MTB/RIF is adequate in this setting. However, WGS also identi ed an unusual double mutation in the RRDR region of the genome which was missed by Xpert MTB/RIF. The occurrence of mutations missed by routine diagnostic tools should be regularly monitored in the region through collaborations with supranational referral laboratories, SNRLs, for quality assurance purposes. One RR-TB case by Xpert MTB/RIF, later found to be susceptible to rifampicin by WGS, may be explained by the analysis of an heteroresistant sample by Xpert MTB/RIF and/or the preferential selection of a susceptible strain upon culturing a sputum sample harbouring a mixed infection, prior to WGS. We also found that of the 40 samples that tested negative to M. tuberculosis by Xpert MTB/RIF at NTRL, 4 were positive by culture. It is possible that the bacterial load may have been too low for detection by Xpert MTB/RIF. Similar ndings have been reported in other studies 14 .
Currently, eight GeneXpert machines are functional in Timor-Leste. An additional two machines of 16 modules have been received for the Covid-19 response, which can also be used by the NTP for TB diagnosis. Linkages with the GeneXpert machines will also be prioritised for intensi ed case nding at all facilities with a high outpatient caseload 15 . During the survey, sputum samples were transported from all the municipalities in the country to the GeneXpert sites and NTRL with minimal delay (median of 2 days) through human carriers who were adequately incentivized for this work. Delay was observed in sample shipment from NTRL, Dili, to the SNRL, Chennai, and the availability of the courier agency to ship biohazard samples, which led to decreased viability and contamination of transported samples. Any discordance between Xpert and LPA tests could be resolved by phenotypic testing, once capacity is established in-country, and appropriate clinical evaluation of the patient For routine patient management and surveillance of drug resistance, it is planned that sputum specimen samples will be routinely transported from all the municipalities to the eight GeneXpert sites, as per the revised NTP guidelines. An e cient specimen referral system is expected to be developed in the next Global Fund cycle with the support of partners, including linking with active TB case nding initiatives led by the NTP and community volunteers The same mechanism could be replicated for routinely transporting samples from the microscopy labs to the GeneXpert sites and could address barriers to accessing TB diagnosis and care from remote, hilly terrains in this island nation.
In conclusion, the survey highlighted the need for strengthening sputum specimen transportation, establishing electronic recording and reporting, achieving NTRL accreditation, and ultimately implementing universal DST in Timor-Leste for both rst-and second-line drugs. The relatively low prevalence of RR-TB in Timor-Leste is an encouraging nding, but gaps remain in obtaining bacteriological con rmation of TB and routine rifampicin testing among bacteriologically con rmed cases. This study showed that the highest burden of TB was in economically productive age groups and predominantly in male gender, as well as reproductive age groups, as in other countries. Thus, the survey highlights the signi cant negative impact of TB on the country's economic growth, and therefore addressing TB and DR-TB in particular, merits sustained investment.

Declarations
Source of Support: The Global Fund grant Con ict of Interest: None declared.
Contributorship: DK developed the concept, designed and performed the experiment, analysed the data, wrote the paper, developed, edited and revised the nal manuscripts; CL performed the experiments, reviewed and revised the paper; SB reviewed the paper; IB reviewed and revised the paper; VB reviewed and revised the paper; BC contributed materials; SK, PC and PP reviewed and analysed the data; OT designed the experiment, analysed the data, edited the manuscript and reviewed the paper; AD designed the experiment, analysed the data, reviewed, edited and revised the nal manuscript. All authors read and approved the nal manuscript.