Genomic epidemiology of tuberculosis in eastern Malaysia: insights for strengthening public health responses

Abstract

Tuberculosis is a leading public health priority in eastern Malaysia. Knowledge of the genomic epidemiology of tuberculosis can help tailor public health interventions. Our aims were to determine tuberculosis genomic epidemiology and characterize resistance mutations in the ethnically diverse city of Kota Kinabalu, Sabah, located at the nexus of Malaysia, Indonesia, Philippines and Brunei. We used an archive of prospectively collected Mycobacterium tuberculosis samples paired with epidemiological data. We collected sputum and demographic data from consecutive consenting outpatients with pulmonary tuberculosis at the largest tuberculosis clinic from 2012 to 2014, and selected samples from tuberculosis inpatients from the tertiary referral centre during 2012-2014 and 2016-2017. Two hundred and eight M. tuberculosis sequences were available for analysis, representing 8 % of cases notified during the study periods. Whole-genome phylogenetic analysis demonstrated that most strains were lineage 1 (195/208, 93.8 %), with the remainder being lineages 2 (8/208, 3.8 %) or 4 (5/208, 2.4 %). Lineages or sub-lineages were not associated with patient ethnicity. The lineage 1 strains were diverse, with sub-lineage 1.2.1 being dominant (192, 98 %). Lineage 1.2.1.3 isolates were geographically most widely distributed. The greatest diversity occurred in a border town sub-district. The time to the most recent common ancestor for the three major lineage 1.2.1 clades was estimated to be the year 1966 (95 % HPD 1948-1976). An association was found between failure of culture conversion by week 8 of treatment and infection with lineage 2 (4/6, 67 %) compared with lineage 1 strains (4/83, 5 %) (P<0.001), supporting evidence of greater virulence of lineage 2 strains. Eleven potential transmission clusters (SNP difference ≤12) were identified; at least five included people living in different sub-districts. Some linked cases spanned the whole 4-year study period. One cluster involved a multidrug-resistant tuberculosis strain matching a drug-susceptible strain from 3 years earlier. Drug resistance mutations were uncommon, but revealed one phenotype-genotype mismatch in a genotypically multidrug-resistant isolate, and rare nonsense mutations within the katG gene in two isolates. Consistent with the regionally mobile population, M. tuberculosis strains in Kota Kinabalu were diverse, although several lineage 1 strains dominated and were locally well established. Transmission clusters - uncommonly identified, likely attributable to incomplete sampling - showed clustering occurring across the community, not confined to households or sub-districts. The findings indicate that public health priorities should include active case finding and early institution of tuberculosis management in mobile populations, while there is a need to upscale effective contact investigation beyond households to include other contacts within social networks.

Keywords: Malaysia; Mycobacterium tuberculosis; whole-genome sequencing.

Fig. 1.

Study diagram. Figure illustrates numbers of samples provided for analysis from patients in periods 1 and 2; numbers of samples obtained from the different patient cohorts (outpatients and inpatients); final numbers available for analysis. NTM, non-tuberculous mycobacteria.

Fig. 2.

Phylogeny of lineage 1 isolates from Kinabalu, Malaysia and 480 publicly available Thai M. tuberculosis genomes [22] (blue tips). Maximum-likelihood tree with bootstrap values above 75 for all major branches created using RAxML at 1000 bootstraps, GTRCAT nucleotide substitution model based on 8321 SNPs. Clade branch colours: 1.2.1.1 – red, 1.2.1.2 – purple, 1.2.1.3 – green, 1.2.1.X – yellow.

Fig. 3.

Map of Kota Kinabalu highlighting the geographical distribution of M. tuberculosis cases according to lineages/sub-lineages. Letters represent the different sub-districts/villages: A, Telipok; B, Sulaman; C, Menggatal; D, Inanam; E, Kolombong; F, Likas; G, Gaya; H, Kinarut; I, Kota Kinabalu; J, Sembulan; K, Bundit Padang; L, Luyang; M, Tanjung Aru; N, Kopungit; O, Lido; Q, Bundusan; R, Taman Penampang; S, Kepayan; T, Petagas; U, Putatan. Inset: geographical location of the eastern part of Kota Kinabalu district (red) within Sabah state, Malaysia.

Fig. 4.

Bayesian timed phylogeny of the dominant sub-lineage 1.2.1 and drug susceptibility profiles. Tree branch ends: blue, inpatients; without colour, outpatients. Field shapes on the outer edge of the tree: red square, streptomycin-resistant; purple circle, rifampicin-resistant; green star, isoniazid-resistant; black square, ethambutol-resistant. Outermost circle shows immigrant status: red circle, immigrant; open circle, Malaysian born; missing circle, unknown. Clades: green, L1.2.1.3; purple, L1.2.1.2; red, L1.2.1.1; white, L1.2.1.X. L1.2.1.1 had an MRCA of 1966 (95 % HPD 1948–1976); L1.2.1.X was rooted in 1910 (95 % HPD 1888–1930).

Fig. 5.

Plausible transmission links among inpatients and outpatients from Kota Kinabalu, Malaysia, 2012–2017. Genomic clusters were identified using transmission threshold (T)=19 and transmission rate=(β), equivalent to a loose singlenucleotide polymorphism (SNP) threshold of 12. Each circle represents an isolate, while the number linking the circles represents the SNP difference. Ethnicity of individuals from whom isolates were obtained is shown as Malaysian, migrant or unknown. All migrants were Filipino, except the patient who supplied sample OP032 from Brunei.