Detection of Frequent Superinfection among Kenyan Women Using Ultra-deep Pyrosequencing
Keshet Ronen*1, C McCoy1, F Matsen1, F Bushman2, S Grunberg2, D Boyd1, S McClelland3, W Jaoko4, K Mandaliya5, and J Overbaugh1
1Fred Hutchinson Cancer Res Ctr, Seattle, WA, US; 2Univ of Pennsylvania, Philadelphia, US; 3Univ of Washington, Seattle, US; 4Univ of Nairobi, Kenya; and 5Coast Provincial Gen Hosp, Mombasa, Kenya
Background: Superinfection, the re-infection of an HIV-infected individual, has been reported in several cohorts at various frequencies. Such events are of interest since they allow us to explore whether infection provides any protection from re-infection, but studies to date have been relatively small. Larger-scale screens are required, but existing techniques for the identification of superinfection are not amenable to high-throughput screening.
Methods: An analysis pipeline was developed to amplify and analyze ~500 bp amplicons in gag, pol, and env from plasma RNA. Amplicons were sequenced on the Roche 454 GS-FLX sequencing platform. The AmpliconNoise software package was used to correct sequencing error. Cleaned sequences were aligned and phylogeny inferred using BEAST. Posterior probabilities of monophyly, confirmed by visual inspection of maximum clade credibility trees, were used to identify cases of multiple infection. The pipeline was used to study a female sex worker (FSW) cohort in Mombasa, Kenya. Samples from within 6 months of infection, over 2 years after infection, and 1 to 3 more intervening time points were analyzed. Between 100 and 2000 sequences were obtained per genomic region per time point for a total of ~380,000 sequences.
Results: Previous screens of 56 women in this cohort using single-copy PCR and Sanger sequencing identified 12 cases of superinfection. We validated the 454 sequencing-based pipeline with samples from 6 individuals screened in the prior studies, of whom 2 were superinfected. Both previously identified cases of superinfection were detected as superinfected using the new method, and single infection confirmed in the rest. Following pipeline validation we have screened a further 54 women in the cohort and identified 8 new cases of superinfection, totaling 20 cases of superinfection among 110 women screened to date.
Conclusions: Using a combination of deep sequencing and single genome sequencing, we have identified 20 cases of superinfection over 542 person-years of follow up among 110 women in this FSW cohort, corresponding to an incidence of 3.69% per person-year of follow up. Screening of an additional 39 women is ongoing to enable formal comparison of the incidence of initial infection and superinfection in the cohort and identification of cases of superinfection for studies of correlates of immune protection from HIV-1 infection.