Background: Next-generation sequencing has been transformative to molecular epidemiology of viruses, though technical complexities have slowed adoption in clinical settings. We developed a quantitative genotyping method optimised for a large-scale HIV phylogenetic study linked to the HPTN 071 (PopART) cluster-randomised trial of antiretroviral treatment as prevention in Zambia.
Methods: To date, 4319 HIV-positive patients (ART-naive or >1 year since last ART) from 9 health care facilities have consented to viral sequencing using residual blood from CD4 testing collected at recruitment. Nucleic acid extracted from 0.5 ml plasma from 292 male and 357 female HIV-infected participants was used to produce sequencing libraries without virus-specific PCR. Oligonucleotide-baits, designed to capture the full HIV epidemic diversity enriched libraries for sequencing on MiSeq (Illumina) and portable MinION (Oxford Nanopore) sequencers. Optimizations focused on the simplest, most cost-effective means of maximizing numbers of unique viral RNA templates, including samples with low viral load, while preserving quantitative information and minimizing oversampling of short-RNA fragments. Merged paired-end reads were assembled using SHIVER, depleted of PCR-duplicates and contaminants, then submitted for high-resolution transmission-mapping (phyloscanner) and drug resistance profiling (HIVdb, Stanford). Quantitative sequencing controls were used to estimate viral load. Clinical viral loads were obtained for a random subset of samples (n=126) for cross validation.
Results: Whole genomes were obtained (with minimum depth of 5x) for 80% of all samples and for 97% of samples with clinical viral load >1,000 copies per ml (fig A). Sequence-based viral load correlated with clinical viral load (R2 = 88% and n=126, fig B). By minimizing PCR-amplification of short fragments, a median of 49% of inserts were longer than 350 b.p providing sufficient phylogenetic resolution to assess intrahost diversity, identify transmission pairs and potentially estimate recency of infection. The total processing time from RNA extraction to sequencing can be completed in 48 hours per batch of 90 samples.
Conclusions:


[Quantitative sequencing of HIV]

Our novel laboratory and informatics pipeline provides robust viral genetic, viral load, and minority variant information. Processing times, cost and capabilities for handling low viral load samples are highly-competitive compared to routine viral load or polymerase drug resistance testing and suitable for clinical use.