Esther Fearnhill

An Evolutionary Model-Based Algorithm for Accurate Phylogenetic Breakpoint Mapping and Subtype Prediction in HIV-1

Genetically diverse pathogens (such as Human Immunodeficiency virus type 1, HIV-1) are frequently stratified into phylogenetically or immunologically defined subtypes for classification purposes. Computational identification of such subtypes is helpful in surveillance, epidemiological analysis and detection of novel variants, e.g., circulating recombinant forms in HIV-1. A number of conceptually and technically different techniques have been proposed for determining the subtype of a query sequence, but there is not a universally optimal approach. We present a model-based phylogenetic method for automatically subtyping an HIV-1 (or other viral or bacterial) sequence, mapping the location of breakpoints and assigning parental sequences in recombinant strains as well as computing confidence levels for the inferred quantities. Our Subtype Classification Using Evolutionary ALgorithms (SCUEAL) procedure is shown to perform very well in a variety of simulation scenarios, runs in parallel when multiple sequences are being screened, and matches or exceeds the performance of existing approaches on typical empirical cases. We applied SCUEAL to all available polymerase (pol) sequences from two large databases, the Stanford Drug Resistance database and the UK HIV Drug Resistance Database. Comparing with subtypes which had previously been assigned revealed that a minor but substantial (≈5%) fraction of pure subtype sequences may in fact be within- or inter-subtype recombinants. A free implementation of SCUEAL is provided as a module for the HyPhy package and the Datamonkey web server. Our method is especially useful when an accurate automatic classification of an unknown strain is desired, and is positioned to complement and extend faster but less accurate methods. Given the increasingly frequent use of HIV subtype information in studies focusing on the effect of subtype on treatment, clinical outcome, pathogenicity and vaccine design, the importance of accurate, robust and extensible subtyping procedures is clear.

Transmission network parameters estimated from HIV sequences for a nationwide epidemic.

Background. Many studies of sexual behavior have shown that individuals vary greatly in their number of sexual partners over time, but it has proved difficult to obtain parameter estimates relating to the dynamics of human immunodeficiency virus (HIV) transmission except in small-scale contact tracing studies. Recent developments in molecular phylodynamics have provided new routes to obtain these parameter estimates, and current clinical practice provides suitable data for entire infected populations. Methods. A phylodynamic analysis was performed on partial pol gene sequences obtained for routine clinical care from 14 560 individuals, representing approximately 60% of the HIV-positive men who have sex with men (MSM) under care in the United Kingdom. Results. Among individuals linked to others in the data set, 29% are linked to only 1 individual, 41% are linked to 2–10 individuals, and 29% are linked to ≥10 individuals. The right-skewed degree distribution can be approximated by a power law, but the data are best fitted by a Waring distribution for all time depths. For time depths of 5–7 years, the distribution parameter ρ lies within the range that indicates infinite variance. Conclusions. The transmission network among UK MSM is characterized by preferential association such that a randomly distributed intervention would not be expected to stop the epidemic.

Molecular Phylodynamics of the Heterosexual HIV Epidemic in the United Kingdom

The heterosexual risk group has become the largest HIV infected group in the United Kingdom during the last 10 years, but little is known of the network structure and dynamics of viral transmission in this group. The overwhelming majority of UK heterosexual infections are of non-B HIV subtypes, indicating viruses originating among immigrants from sub-Saharan Africa. The high rate of HIV evolution, combined with the availability of a very high density sample of viral sequences from routine clinical care has allowed the phylodynamics of the epidemic to be investigated for the first time. Sequences of the viral protease and partial reverse transcriptase coding regions from 11,071 patients infected with HIV of non-B subtypes were studied. Of these, 2774 were closely linked to at least one other sequence by nucleotide distance. Including the closest sequences from the global HIV database identified 296 individuals that were in UK-based groups of 3 or more individuals. There were a total of 8 UK-based clusters of 10 or more, comprising 143/2774 (5%) individuals, much lower than the figure of 25% obtained earlier for men who have sex with men (MSM). Sample dates were incorporated into relaxed clock phylogenetic analyses to estimate the dates of internal nodes. From the resulting time-resolved phylogenies, the internode lengths, used as estimates of maximum transmission intervals, had a median of 27 months overall, over twice as long as obtained for MSM (14 months), with only 2% of transmissions occurring in the first 6 months after infection. This phylodynamic analysis of non-B subtype HIV sequences representing over 40% of the estimated UK HIV-infected heterosexual population has revealed heterosexual HIV transmission in the UK is clustered, but on average in smaller groups and is transmitted with slower dynamics than among MSM. More effective intervention to restrict the epidemic may therefore be feasible, given effective diagnosis programmes.

Effect of HIV-1 subtype on virologic and immunologic response to starting highly active antiretroviral therapy.

BACKGROUND It has been proposed that subtype-related human immunodeficiency virus type 1 (HIV-1) variability may influence virologic and immunologic responses to highly active antiretroviral therapy (HAART). Studies to date, however, have described treatment outcomes predominantly in persons with subtype B infection or compared subtype B with diverse non-B subtypes grouped together. METHODS With use of data from the linked UK Collaborative Group on HIV Drug Resistance and the UK Collaborative HIV Cohort Study databases, time to viral load undetectability (viral load, <50 copies/mL), time to virologic rebound (viral load, >1000 copies/mL), and increases in the CD4 cell count were compared for a median of 39 months (interquartile range, 23-67 months) in drug-naive patients infected with subtype B (n=1550), subtype C (n=272), subtype A (n=66), circulating recombinant form AG (n=57), or subtype D (n=41) disease who started HAART. RESULTS Overall, 1906 (90%) of 2116 patients achieved viral load undetectability within 12 months after they started HAART, of whom 335 (18%) subsequently experienced virologic rebound. In adjusted analyses, viral load suppression occurred more rapidly in patients infected with subtype C (hazard ratio, 1.16; 95% confidence interval, 1.01-1.33; P=.04) and subtype A (hazard ratio, 1.35; 95% confidence interval, 1.04-1.74; P=.02) relative to subtype B infection. The virologic rebound occurred marginally more rapidly in patients with subtype C infection (hazard ratio, 1.40; 95% confidence interval, 1.00-1.95; P=.05), but the hazard of virologic rebound was similar with other subtypes. Although persons with subtype B infection showed higher baseline CD4 cell counts and maintained the advantage throughout therapy, CD4 cell count recovery occurred at similar rates with all subtypes. CONCLUSIONS Patients infected with prevalent non-B subtypes were as likely to achieve viral load suppression as persons infected with subtype B and showed comparable rates of CD4 cell count recovery. HAART achieves excellent outcomes regardless of the infecting subtype.

Automated analysis of phylogenetic clusters

BackgroundAs sequence data sets used for the investigation of pathogen transmission patterns increase in size, automated tools and standardized methods for cluster analysis have become necessary. We have developed an automated Cluster Picker which identifies monophyletic clades meeting user-input criteria for bootstrap support and maximum genetic distance within large phylogenetic trees. A second tool, the Cluster Matcher, automates the process of linking genetic data to epidemiological or clinical data, and matches clusters between runs of the Cluster Picker.ResultsWe explore the effect of different bootstrap and genetic distance thresholds on clusters identified in a data set of publicly available HIV sequences, and compare these results to those of a previously published tool for cluster identification. To demonstrate their utility, we then use the Cluster Picker and Cluster Matcher together to investigate how clusters in the data set changed over time. We find that clusters containing sequences from more than one UK location at the first time point (multiple origin) were significantly more likely to grow than those representing only a single location.ConclusionsThe Cluster Picker and Cluster Matcher can rapidly process phylogenetic trees containing tens of thousands of sequences. Together these tools will facilitate comparisons of pathogen transmission dynamics between studies and countries.

Identification of accessory mutations associated with high-level resistance in HIV-1 reverse transcriptase

Objective:To identify accessory mutations associated with high levels of resistance to reverse transcriptase inhibitors (RTI). Design:HIV pol sequences from routine resistance tests from over 3000 patients who were treatment experienced and infected with subtype B HIV-1 were analysed. Changes relative to the wild-type amino acid for codons 1–400 of reverse transcriptase were determined in two series: (i) samples showing the accumulation of thymidine analogue mutations (TAMs); and (ii) samples showing the accumulation of non-nucleoside reverse transcriptase inhibitor (NNRTI) resistance mutations. Methods:Accessory mutations were identified as occurring in those codons in which there was a statistically significant association between the distribution of amino acids and the number of TAMs/NNRTI resistance mutations. Positions already established as drug resistance positions by the International AIDS Society – USA Panel were not included in this analysis. Results:Twenty-four positions were identified where accessory mutations were associated with the accumulation of TAMs (20, 35, 39, 43, 60, 83, 98, 101, 122, 135, 179, 196, 203, 208, 218, 223, 228, 284, 322, 356, 359, 360, 371 and 381). Likewise changes at 25 positions (6, 20, 35, 39, 43, 53, 68, 90, 98, 101, 122, 179, 200, 203, 208, 218, 221, 223, 228, 284, 318, 320, 348, 359 and 371) were associated with NNRTI mutations. The accumulation of accessory mutations correlated with increasing numbers of TAMs and NNRTI mutations. Conclusion:The development of high-level resistance to RTI is associated not only with the accumulation of recognized mutations but also with accessory mutations.

Phylogenetic analyses reveal HIV-1 infections between men misclassified as heterosexual transmissions.

Objective:HIV-1 subtype B infections are associated with MSM in the UK. Yet, around 13% of subtype B infections are found in those reporting heterosexual contact as transmission route. Using phylogenetics, we explored possible misclassification of sexual exposure among men diagnosed with HIV in the UK. Design:Viral gene sequences linked to patient-derived information were used to identify phylogenetic transmission chains. Methods:A total of 22 481 HIV-1 subtype B pol gene sequences sampled between 1996 and 2008 were analysed. Dated phylogenies were reconstructed and transmission clusters identified as clades of at least two sequences with a maximum genetic distance of 4.5%, a branch support of at least 95% and spanning 5 years. The characteristics of clusters containing at least one heterosexually acquired infection were analysed. Results:Twenty-nine percent of the linked heterosexuals clustered exclusively with MSM. These were more likely to be men than women. Estimated misclassification of homosexually acquired infections ranged between 1 and 11% of the reported male heterosexuals diagnosed with HIV. Black African heterosexual men were more often phylogenetically linked to MSM than other ethnic group, with an estimated misclassification range between 1 and 21%. Conclusion:Overall, a small proportion of self-reported heterosexual men diagnosed with HIV could have been infected homosexually. However, up to one in five black African heterosexual men chose not to disclose sex with men at HIV diagnosis and preferred to be identified as heterosexual. Phylogenetic analyses can enhance surveillance-based risk information and inform national programmes for monitoring and preventing HIV infections.

Long-Term Probability of Detecting Drug-Resistant HIV in Treatment-Naive Patients Initiating Combination Antiretroviral Therapy

BACKGROUND Robust long-term estimates of the risk of development of drug resistance are needed for human immunodeficiency virus (HIV)-infected patients starting combination antiretroviral therapy (cART) regimens currently used in routine clinical practice. METHODS We followed a large cohort of patients seen in 1 of 11 HIV clinics in the United Kingdom after starting cART with nucleoside reverse-transcriptase inhibitors and either a nonnucleoside reverse-transcriptase inhibitor (NNRTI) or a ritonavir-boosted protease inhibitor (PI/r). Survival analysis was employed to estimate the incidence of virological failure and of detected drug resistance. RESULTS Seven thousand eight hundred ninety-one patients were included; 6448 (82%) started cART with an NNRTI and 1423 (17%) with a PI/r. The cumulative risk of virological failure by 8 years was 28%. The cumulative probabilities of detecting any mutation, > or =1 major nucleoside reverse-transcriptase inhibitor International AIDS Society-United States of America (IAS-USA) mutation, > or =1 major NNRTI IAS-USA mutation (in those starting an NNRTI), and > or =1 major PI IAS-USA mutation (in those starting a PI) were 17%, 14%, 15%, and 7%, respectively, by 8 years. The probability of detecting PI mutations in people who started PI/r-based regimens was lower than that of detecting NNRTI mutations in those starting NNRTI-based regimens (adjusted relative hazard, 0.36; 95% confidence interval, 0.26-0.50; P<.001). The risk of detecting nucleoside resistance did not vary according to whether an NNRTI or a PI/r was used in the regimen (adjusted relative hazard, 1.00; 95% confidence interval, 0.80-1.26; P=.98). CONCLUSIONS In patients who started modern cART in clinical practice in the United Kingdom, virological failure by 8 years was relatively common and was paralleled by an appreciable risk of resistance detection, although the detection rate of class-specific resistance was lower for those who started a PI/r-based regimen.

Integrase inhibitor (INI) genotypic resistance in treatment-naive and raltegravir-experienced patients infected with diverse HIV-1 clades

Objectives The aim of this study was to characterize the prevalence and patterns of genotypic integrase inhibitor (INI) resistance in relation to HIV-1 clade. Methods The cohort comprised 533 INI-naive subjects and 255 raltegravir recipients with viraemia who underwent integrase sequencing in routine care across Europe, including 134/533 (25.1%) and 46/255 (18.0%), respectively, with non-B clades (A, C, D, F, G, CRF01, CRF02, other CRFs, complex). Results No major INI resistance-associated mutations (RAMs) occurred in INI-naive subjects. Among raltegravir recipients with viraemia (median 3523 HIV-1 RNA copies/mL), 113/255 (44.3%) had one or more major INI RAMs, most commonly N155H (45/255, 17.6%), Q148H/R/K + G140S/A (35/255, 13.7%) and Y143R/C/H (12/255, 4.7%). In addition, four (1.6%) raltegravir recipients showed novel mutations at recognized resistance sites (E92A, S147I, N155D, N155Q) and novel mutations at other integrase positions that were statistically associated with raltegravir exposure (K159Q/R, I161L/M/T/V, E170A/G). Comparing subtype B with non-B clades, Q148H/R/K occurred in 42/209 (20.1%) versus 2/46 (4.3%) subjects (P = 0.009) and G140S/A occurred in 36/209 (17.2%) versus 1/46 (2.2%) subjects (P = 0.005). Intermediate- to high-level cross-resistance to twice-daily dolutegravir was predicted in 40/255 (15.7%) subjects, more commonly in subtype B versus non-B clades (39/209, 18.7% versus 1/46, 2.2%; P = 0.003). A glycine (G) to serine (S) substitution at integrase position 140 required one nucleotide change in subtype B and two nucleotide changes in all non-B clades. Conclusions No major INI resistance mutations occurred in INI-naive subjects. Reduced occurrence of Q148H/R/K + G140S/A was seen in non-B clades versus subtype B, and was explained by the higher genetic barrier to the G140S mutation observed in all non-B clades analysed.

The contribution of viral genotype to plasma viral set-point in HIV infection.

Disease progression in HIV-infected individuals varies greatly, and while the environmental and host factors influencing this variation have been widely investigated, the viral contribution to variation in set-point viral load, a predictor of disease progression, is less clear. Previous studies, using transmission-pairs and analysis of phylogenetic signal in small numbers of individuals, have produced a wide range of viral genetic effect estimates. Here we present a novel application of a population-scale method based in quantitative genetics to estimate the viral genetic effect on set-point viral load in the UK subtype B HIV-1 epidemic, based on a very large data set. Analyzing the initial viral load and associated pol sequence, both taken before anti-retroviral therapy, of 8,483 patients, we estimate the proportion of variance in viral load explained by viral genetic effects to be 5.7% (CI 2.8–8.6%). We also estimated the change in viral load over time due to selection on the virus and environmental effects to be a decline of 0.05 log10 copies/mL/year, in contrast to recent studies which suggested a reported small increase in viral load over the last 20 years might be due to evolutionary changes in the virus. Our results suggest that in the UK epidemic, subtype B has a small but significant viral genetic effect on viral load. By allowing the analysis of large sample sizes, we expect our approach to be applicable to the estimation of the genetic contribution to traits in many organisms.