Prevalence of drug-resistant tuberculosis assessed by next-generation sequencing : an 18-month nationwide study in Lebanon

Tuberculosis (TB) is the first killer infectious disease, with 10 million new cases estimated worldwide in 2017. TB drug resistance and its diagnosis are particularly problematic. Only 25% of the 450,000 incident multidrug resistant (MDR) TB patients estimated over the same year were diagnosed and treated as such.Although Lebanon is a low-TB burden country, significant challenges exist in terms of disease control. Lebanon is the country hosting the largest refugee population proportionally to its national population worldwide, with 1.5 million Syrian refugees as a consequence of the war in Syria, in addition to large populations of Palestinian refugees and migrant workers. Such populations are particularly vulnerable to risks of TB and emergence of drug resistance. The last national survey on the prevalence of drug resistant TB was done 15 years ago, well before the start of the Syrian crisis in 2011. Even most recent reported rates of MDR TB largely relied on estimates. Second-line drug susceptibility testing (DST) and individualized extensively drug-resistant (XDR) TB treatments were not available in the country.In order to gain a more comprehensive view of the TB situation, we set up the first nationwide study combining phenotypic and extensive molecular testing to determine the prevalence and extent of TB drug resistance in the country. A total of 417 patients were included, corresponding to all confirmed TB cases reported to the national tuberculosis program between June, 2016 and November, 2017. Lowenstein-Jensen and/or MGIT culturing, and molecular testing using GeneXpert MTB/RIF and/or Anyplex MTB/NTM Real-time were used in Lebanon for diagnostic confirmation and DST. In Lille, we evaluated, for the first time on a nationally representative sample, a new deep sequencing assay called Deeplex-MycTB, for extensive drug resistance prediction and genotyping of patient isolates. MIRU-VNTR typing was used in combination for defining molecular clusters, potentially suggestive of endemically circulating or epidemically transmitted TB strains.For the first time in the country, out of the 354 culture positive TB cases with available DST, 3 XDR cases, resistant to at least rifampicin (RIF), isoniazid (INH), kanamycin (KAN)/amikacin (AMI) and levofloxacin (LFX) were detected, in addition to 5 MDR (resistant to at least RIF, INH) cases and one RIF mono-resistant case. Among the remaining cases, 3.4% (12/354) had resistance to INH and streptomycin (SM), 3.4% (12/354) mono-resistance to INH, 0.3% (1/354) mono-resistance to ethambutol (EMB), 8.5% (30/354) mono-resistance to streptomycin (SM), while 81.9% (290/354) were susceptible to all 4 first line drugs. While none of MDR and XDR TB cases were found in molecular clusters, a large cluster comprising 36 other patients was identified, suggestive of a highly endemic or actively transmitted drug susceptible strain.A total of 4184 out of 4407 (94.9%) possible phenotypes could be predicted by Deeplex-MycTB for 339/348 (97.4%) analyzable samples, of which 1282/1380 (92.9%) matched the available phenotypic results. Based on detectable resistance determinants, INH, RIF, EMB and SM resistance was concordantly predicted with 90.3%, 100%, 100%, 52.8% sensitivity, respectively, and susceptibility with 99.6%, 100%, 99.4%, 99.6% specificity, respectively. While predicted first and second-line drug resistance matched almost completely the available phenotypic profiles of the 8 MDR and XDR cases, mutations were additionally detected in all of these 8 cases in targets predicting supplementary pyrazinamide and/or ethionamide resistance, not phenotypically tested. Moreover, resistance to fluoroquinolones was also predicted in 34/339 (10%) non-MDR cases, not subjected to LFX DST. Finally, the use of advanced molecular testing allowed us to identify the first 12 (3.4%) zoonotic TB cases identified in the country [...]