You are here
WHOLE-GENOME SEQUENCING AND SPATIAL ANALYSIS OF XDR TB TRANSMISSION IN SOUTH AFRICA
Kristin N. Nelson1, Neel R. Gandhi1, Sara C. Auld1, James C. Brust2, Barun Mathema3, Nazir Ismail4, Pravi Moodley5, Angela Campbell1, Salim Allana1, N. Sarita Shah6
1Emory Univ, Atlanta, GA, USA,2Albert Einstein Coll of Med, Bronx, NY, USA,3Columbia Univ, New York, NY, USA,4NICD, Johannesburg, South Africa,5Univ of KwaZulu-Natal, Durban, South Africa,6CDC, Atlanta, GA, USA
Transmission of drug-resistant tuberculosis (TB) is a major threat to TB control, especially in high HIV prevalence settings. We have previously shown that nearly 70% of extensively drug-resistant (XDR) TB cases in KwaZulu-Natal province, South Africa are due to transmission, rather than unsuccessful TB treatment. We identified epidemiologic links through close contacts or hospital admission for 30% of cases, but found no epidemiologic link for the remaining 70%. We hypothesize that genomically-linked cases live or seek healthcare near one another and may have unrecognized epidemiologic links. We compared geospatial distances between homes and health facilities for participants with ≤ 5 single nucleotide polymorphism (SNP) differences to determine geographic proximity.
We enrolled culture-confirmed XDR TB cases in KwaZulu-Natal from 2011-2014. We collected clinical and demographic characteristics, GPS coordinates of homes and XDR TB diagnosis facility, and performed whole genome sequencing (WGS) of TB isolates. We defined a SNP difference of ≤ 5 as genomic evidence of transmission between two cases ('case-pair'). We calculated spatial distances using Haversine's formula and defined geographic proximity as < 20 km. We stratified the analysis by participants' HIV status to determine the association with spatial clustering.
We enrolled 296 participants with WGS results, from all 11 districts in KwaZulu-Natal. Among these, 179 (66%) participants formed 671 case-pairs with ≤ 5 SNPs. The median distance between home residences of case-pairs was 115 km (IQR 63-154) (Figure 1); the median distance between diagnosing facilities was 93 km (IQR 44-131). Only 116 (17%) case-pairs lived or were diagnosed within 20 km of each other. Median distance did not vary significantly when the SNP threshold was reduced to 3 and 1 SNP (p=0.27 for homes, p=0.56 for diagnosing facilities). There was no significant difference in geographic distances for case-pairs based on whether both were HIV-positive, HIV-negative or had discordant HIV status (p=0.87 for homes, p=0.20 for diagnosing facilities).
Although two-thirds of XDR TB participants from KwaZulu-Natal, South Africa were genomically-linked, only 17% lived or were diagnosed at a health facility within 20 km of their case-pair. Further research examining migratory patterns, particularly between rural and urban areas, is needed to determine their role in TB transmission and the spread of drug resistance.