Background: HIV positive patients with pulmonary tuberculosis (TB) are thought to be less likely to transmit M.tuberculosis, even accounting for sputum bacillary load, because they have a shorter duration of disease. Whole genome sequencing of M.tuberculosis can identify who transmits to whom, and provides a new method to test this hypothesis.
Methods: In Karonga District Malawi all TB patients are identified and asked about prior contacts with patients with TB. Whole genome sequencing with ~100 fold coverage was carried out on all available cultures from 1996-2010. Transmission networks were constructed using SeqTrack, including links ≤10 mutations, and ensuring coherence of dates. The relative risk of transmitting and causing disease was estimated using ordered logistic regression, using the last 3 years of data only for assessing transmissions. In addition, the probability that a TB case acquired TB from their named contacts with smear-positive TB was calculated, by HIV status.
Results: Overall 1687 high quality sequences were available: 72 % of all culture positive cases in the district over this period. After excluding cases from the last 3 years, 431/1346 patients with pulmonary TB were identified as a source of transmission, resulting in disease in ≥1 (and up to 12) further patients. 181/588 (30.8%) HIV positive patients and 91/318 (28.6%) HIV negative patients transmitted. (8% of HIV positive patients were on antiretrovirals before TB diagnosis.) Using ordered logistic regression there was little difference in transmission from HIV positive and HIV negative patients (odds ratio 1.1, 95%CI 0.81-1.5), and this hardly changed when adjusting for age, sex, lineage, sputum smear status and year.
In the analysis of named contacts, among index case-prior contact pairs, transmission was confirmed from 27/71 (38.0%) HIV positive smear-positive prior contacts and 29/63 (46.0%) HIV negative smear-positive prior contacts (risk ratio 0.83, 0.55-1.2). This reduced slightly after adjusting for closeness of contact (adjusted risk ratio 0.78, 95%CI 0.54-1.1).
In this population about 60% of smear-positive TB patients were HIV positive. If transmission from HIV positive patients was reduced by 22%, then they would still account for 47% of transmission.
Conclusions: Contrary to what is sometimes stated, HIV positive patients appear to play an important role in onward transmission of M.tuberculosis. Where a high proportion of pulmonary TB patients are HIV positive they may be the main source of infection.