The development of antiretroviral (ARV) combinations that prevent the emergence of HIV drug resistance was central to developing successful ARV therapy (ART). The creation of national ART programs in resource limited regions saved millions of lives and provided hope to millions of others. However, the increasing numbers of patients with HIV drug resistance is a major obstacle to continued success. Persons with drug-resistant HIV have fewer treatment options and are at increased risk of morbidity and mortality, particularly in resource-limited regions where ART choices are fewer and virological monitoring infrequent. If levels of drug resistance increase at current rates, >420,000 people are predicted to die as a consequence of HIV drug resistance in the next 5 years and costs of ART delivery will increase by nearly $3 billion.
Comprehensive, accurate, and publicly available HIV drug resistance data are essential for population-based monitoring of ADR and TDR, for guiding salvage ART, and for identifying drug-development needs. However, the variability of viruses that comprise the HIV pandemic, the high mutation rate of HIV, and the many ARVs and ARV combinations make it difficult to quantify ADR and TDR, to optimally interpret HIV genotypic resistance tests, and to identify drug-resistance mutations (DRMs) most relevant to the development of future ARVs.
We maintain a free publicly available, online HIV drug resistance database (HIVDB) for HIV care providers managing patients with HIV drug resistance, for public health officials performing HIV drug resistance surveillance, and for scientists developing new ARV drugs. During the past 15 years, we have published many influential papers on the clinical management, epidemiology, and laboratory science of HIV drug resistance. Our meta-analyses and collaborations have led to much of the knowledge underpinning the global response to combatting HIV drug resistance. Our software tools have also become an increasingly important resource for clinicians, public health organizations and many researchers.
We collect, annotate, and analyze genotypic drug resistance data from ARV-naïve and ARV-experienced patients to identify the HIV DRMs most useful for monitoring ADR and TDR. We systematically aggregate genotype-treatment correlations from many published studies to create a resource that is indispensable to clinical and molecular epidemiologists. We use these data to analyze temporal trends in the DRMs responsible for ADR and TDR in different parts of the world.
Our work is made challenging by the extensive sequence heterogeneity of HIV, including the many different HIV subtypes and recombinant forms. Nonetheless, by collecting well-characterized HIV sequences from nearly 150,000 individuals, we have been able to distinguish naturally occurring HIV mutations from those that are selected by ARV therapy and contribute to HIV drug resistance. We are collaborating with researchers from around the world to track global trends in ADR and TDR and to characterize the genetic mechanisms of drug resistance to several of the most commonly used ARVs for which more data are needed including tenofovir-containing first-line regimens, atazanavir- and lopinavir-containing second-line regimens, and integrase inhibitor-containing regimens.
The presence of HIV-1 genotypic resistance prior to the start of an ARV-treatment regimen is a strong predictor of the success or failure of that regimen. Genotypic resistance testing (GRT) using DNA sequencing is performed routinely in upper-income countries before patients start ARV therapy and in patients who develop virological failure during treatment. In low- and middle income countries (LMICs) it is used sparingly but with increasing frequency. Interpreting HIV-1 genotypic resistance tests is difficult because there are many different DRMs, which occur in complex patterns and which have diverse effects on the ARVs within each drug class.
Since 2000, we have maintained a free publicly available GRT interpretation system that can be accessed via an HTML or an automated web service. The GRT interpretation system has been integrated into the workflows of the World Health Organization’s HIV drug resistance network, the NIH-funded AIDS Clinical Trials Group, and several hundred research and clinical laboratories worldwide. Several thousand clinicians use the program’s HTML interface each month to help choose the most appropriate treatment for their patients.
The HIVDB GRT interpretation system is a rules-based system in which ARV mutation penalties are assigned to different DRMs and to different combinations of DRMs for each of the four most commonly used ARV classes: nucleoside RT inhibitors (NRTIs), nonnucleoside RT inhibitors (NNRTIs), protease inhibitors (PIs), and integrase inhibitors (INIs). The resistance interpretation for an ARV is then determined by adding each of the mutation penalties for that ARV. In addition, to providing a drug resistance interpretation, each report is accompanied by an assessment of sequence quality, an HIV-1 subtype classification, a scoring table, a list of comments for each DRM and a link to a regularly updated document containing published references for each DRM.
The HIVDB system considers three main criteria in developing rules that infer the predicted activity of an ARV based on the DRMs in a virus sequence: (i) the relative frequency of each DRM in ARV-naïve and -experienced patients – Darwinian evidence for the role of a DRM in causing drug resistance; (ii) the contribution of each DRM to reduced in vitro susceptibility; and (iii) the association of each DRM with reduced virological response to an ART regimen containing the ARVs under evaluation. This system is regularly updated with the publication of new drug resistance studies, the development of new HIV treatment practices, and in consultation with a panel of experts in HIV treatment and drug resistance.
Analyses of data in HIVDB often lead to laboratory experiments to fill gaps in the published literature, to evaluate novel sequencing technologies, and to create virus clones that are useful to other researchers.
Because novel patterns of DRMs continue to be identified, we often perform experiments to determine the effects that these DRMs have on in vitro susceptibility. The in vitro susceptibility of novel DRMs informs genotypic resistance test interpretation and provides insight into interactions between ARVs and their molecular targets. We also have an ongoing study to determine whether PI resistance can be caused by mutations outside of the protease gene.
HIV, hepatitis B virus (HBV), and hepatitis C virus (HCV) are the most prevalent deadly chronic viral diseases. Although HIV is a retrovirus, HBV is a double-stranded DNA virus, and HCV is a single-stranded RNA virus, each has a high mutation rate and exists within individuals as a swarm of innumerable related sequence variants. Since 2007, we have been doing next-generation deep sequencing (NGS) of the many virus variants within patients and have published a series of seminal studies on the use of NGS for the study of HIV and HBV drug resistance. As NGS is being increasingly used it has become necessary to determine how the interpretation of NGS will differ from Sanger sequencing for GRT.
The many genetic manifestations of resistance present challenges to researchers who require representative resistant viruses for in vitro mechanistic studies and for developing new inhibitors active against the most clinically drug-resistant viruses. During the past 10 years, we have developed four panels of prototypical recombinant molecular infectious HIV clones resistant to NRTIs, NNRTIs, PIs, and INSTIs and have contributed them to the NIH AIDS Reagent Program to enable drug developers to evaluate novel compounds using representative highly resistant HIV clones.
The increasing prevalence of HIV-1 drug resistance is a threat to HIV-1 treatment and prevention in the low- and middle-income countries hardest hit by the HIV-1 pandemic. An inexpensive point-of-care (POC) genotypic resistance test (GRT) would enable HIV-1 care providers to make informed treatment decisions for patients starting therapy and for patients who develop virological failure (VF) while on therapy. Through the delivery of immediately actionable information upon the detection of VF, a POC GRT would benefit treatment programs by reducing treatment delivery costs and would benefit patients by reducing travel costs and work absenteeism and by increasing the likelihood that treated patients achieve and maintain suppressed virus levels.
The development of a POC HIV-1 GRT will be the first step in addressing the broader challenge of detecting mutations in rapidly evolving epidemic viruses whether they are DRMs or critical vaccine-escape or gain-of-function mutations. The main challenge in developing a POC GRT is the genetic variability at and surrounding each DRM position. We have begun to collaborate with the biotechnology company InSilixa Inc. (Sunnyvale, CA) to develop the required solid-phase fluorogenic probe sets and a multiplex asymmetric RT PCR assay protocol for detecting each of the reported 42 codons at six key DRM positions 65, 103, 106, 181, 184, and 190.
If these preliminary experiments are successful, we will extend our work to detect additional RTI, PI, and INI resistance mutations; to evaluate pre-PCR sample processing modules to separate plasma from blood, concentrate virus, and extract RNA that can be delivered to a fluidic chamber so that the complete GRT process can take place on a single hand-held device within an hour; to evaluate this device on a range of clinical samples from our laboratory and from LMIC clinics; and to assess the possibility of developing a quantitative real-time PCR assay using the same GRT POC platform.
HIVDB has become an increasingly influential knowledge environment for those studying HIV drug resistance. In contrast to most other sequence databases, the data in HIVDB are not confined to passively acquired publicly available data or to data provided by a consortium of research groups required to submit data to a central repository. The majority of data in HIVDB are obtained from GenBank, peer-reviewed publications, and often from the authors of these publications. The continued improvement and long-term sustainability of HIVDB will require novel and transformative approaches such as the creation of consensus data sharing guidelines specific to antiviral drug resistance studies and automated tools to benchmark adherence to data sharing guidelines. This will require collaboration with experts in antiviral drug resistance, journal editors, funding agencies, and stakeholders in public health, academia, and industry.
Several of the tools we have developed or are developing for HIVDB will be useful for piloting databases for drug therapy targets in other viruses of biomedical importance. The development of these databases will require standardized approaches to sequence alignment, post-alignment processing, and quality control. High quality alignments that leverage knowledge from published studies will make it possible to ascertain the prevalence of each amino acid according to virus genotype or subtype and to identify positions under selective immune or drug pressure. Such alignments will be particularly important as an increasing proportion of sequence data is likely be generated by high-throughput, but lower quality, NGS platforms. Finally, as the genetic sequencing of other viruses becomes more common, expert systems for interpreting these sequences will become increasingly necessary to optimize surveillance, clinical research, and clinical practice efforts.