Hauke Walter

Diversity and complexity of HIV-1 drug resistance: a bioinformatics approach to predicting phenotype from genotype.

Drug resistance testing has been shown to be beneficial for clinical management of HIV type 1 infected patients. Whereas phenotypic assays directly measure drug resistance, the commonly used genotypic assays provide only indirect evidence of drug resistance, the major challenge being the interpretation of the sequence information. We analyzed the significance of sequence variations in the protease and reverse transcriptase genes for drug resistance and derived models that predict phenotypic resistance from genotypes. For 14 antiretroviral drugs, both genotypic and phenotypic resistance data from 471 clinical isolates were analyzed with a machine learning approach. Information profiles were obtained that quantify the statistical significance of each sequence position for drug resistance. For the different drugs, patterns of varying complexity were observed, including between one and nine sequence positions with substantial information content. Based on these information profiles, decision tree classifiers were generated to identify genotypic patterns characteristic of resistance or susceptibility to the different drugs. We obtained concise and easily interpretable models to predict drug resistance from sequence information. The prediction quality of the models was assessed in leave-one-out experiments in terms of the prediction error. We found prediction errors of 9.6–15.5% for all drugs except for zalcitabine, didanosine, and stavudine, with prediction errors between 25.4% and 32.0%. A prediction service is freely available at http://cartan.gmd.de/geno2pheno.html.

Geno2pheno: estimating phenotypic drug resistance from HIV-1 genotypes

Therapeutic success of anti-HIV therapies is limited by the development of drug resistant viruses. These genetic variants display complex mutational patterns in their pol gene, which codes for protease and reverse transcriptase, the molecular targets of current antiretroviral therapy. Genotypic resistance testing depends on the ability to interpret such sequence data, whereas phenotypic resistance testing directly measures relative in vitro susceptibility to a drug. From a set of 650 matched genotype-phenotype pairs we construct regression models for the prediction of phenotypic drug resistance from genotypes. Since the range of resistance factors varies considerably between different drugs, two scoring functions are derived from different sets of predicted phenotypes. Firstly, we compare predicted values to those of samples derived from 178 treatment-naive patients and report the relative deviance. Secondly, estimation of the probability density of 2000 predicted phenotypes gives rise to an intrinsic definition of a susceptible and a resistant subpopulation. Thus, for a predicted phenotype, we calculate the probability of membership in the resistant subpopulation. Both scores provide standardized measures of resistance that can be calculated from the genotype and are comparable between drugs. The geno2pheno system makes these genotype interpretations available via the Internet (http://www.genafor.org/).

European guidelines on the clinical management of HIV-1 tropism testing

Viral tropism is the ability of viruses to enter and infect specific host cells and is based on the ability of viruses to bind to receptors on those cells. Testing for HIV tropism is recommended before prescribing a chemokine receptor blocker. In most European countries, HIV tropism is identified with tropism phenotype testing. New data support genotype analysis of the HIV third hypervariable loop (V3) for the identification of tropism. The European Consensus Group on clinical management of tropism testing was established to make recommendations to clinicians and clinical virologists. The panel recommends HIV-tropism testing for the following groups: drug-naive patients in whom toxic effects are anticipated or for whom few treatment options are available; patients who have poor tolerability to or toxic effects from current treatment or who have CNS pathology; and patients for whom therapy has failed and a change in treatment is considered. In general, an enhanced sensitivity Trofile assay and V3 population genotyping are the recommended methods. Genotypic methods are anticipated to be used more frequently in the clinical setting because of their greater accessibility, lower cost, and faster turnaround time than other methods. For the interpretation of V3 loop genotyping, clinically validated systems should be used when possible. Laboratories doing HIV tropism tests should have adequate quality assurance measures. Similarly, close collaboration between HIV clinicians and virologists is needed to ensure adequate diagnostic and treatment decisions.

Detection of minor populations of drug-resistant HIV-1 in acute seroconverters.

Objective:The transmission of drug-resistant HIV-1 is a major health concern. To date, most clinical studies have relied on sequencing techniques for genotypic analyses which do not allow quantification of minority viral populations below 25%. As minor populations of drug-resistant HIV-1 could impact the efficiency of antiretroviral therapy, this study was performed to determine the prevalence of minor populations of drug-resistant HIV-1 in acute seroconverters. Design and methods:Forty-nine acute seroconverters from two clinical centers in Germany were included in the study. Individuals were identified between June 1999 and March 2003, and none had received antiretroviral therapy prior to sampling. Minor populations of drug-resistant variants were detected by quantitative real-time polymerase chain reaction using allele-discriminating oligonucleotides for three key resistance mutations: L90M (protease), K103N and M184V (reverse transcriptase). The approximate discriminative power was between 0.01 and 0.2%. Results:Drug-resistant variants were detected in 10 of 49 patients (20.4%). The L90M mutation was found in one of 49 (2%), the K103N mutation in five of 49 (10.2%) and the M184V mutation in six of 49 (12.2%) patients, respectively. In five of the 10 individuals with detectable drug-resistant virus (50%), the detected population represented a minor viral quasi-species (< 25% of viruses) and was not detected by direct sequencing. Conclusions:The prevalence of minor populations of drug-resistant HIV-1 in acute seroconverters can be frequently detected and may impact the success of antiretroviral therapy.

Frequency of genotypic and phenotypic drug-resistant HIV-1 among therapy-naive patients of the German Seroconverter Study.

Summary: Genotypic and phenotypic resistance of viral reverse transcriptase (RT) and protease (PR) was determined for 64 therapy‐naive, HIV‐1‐infected seroconverters of the German Seroconverter Study coordinated by the Robert Koch‐Institut, Berlin. The date of seroconversion of patients and the laboratory, clinical, and therapeutic follow‐up data were documented. Samples were collected between 1996 and 1999. Phenotypic resistant HIV‐1 were found in 8 (13%) seroconverters; in most cases resistance was weak and mainly directed against RT inhibitors (4 nucleoside reverse transcriptase inhibitors [NRTIs], 2 nonnucleoside reverse transcriptase inhibitors [NNRTIs], 1 combination NRTI/NNRTI). Only one infection with a weak PR inhibitor resistance was identified. Transmission of multidrug‐resistant HIV‐1 has not yet been observed. Frequently at least one or more amino acid mutations associated with antiretroviral drug resistance were detected by genotypic analysis. The mean number of resistance‐associated mutations in the RT of the transmitted virus has increased significantly since 1996. Studies have shown the improved benefit of initial antiretroviral therapy if based on genotypic resistance data. In view of the considerably high level of transmission of resistant HIV‐1 in Germany, which is also seen in other studies in Europe and the United States, we suggest determining the genotypic resistance pattern before starting therapy of newly HIV‐1‐infected patients.

Computational methods for the design of effective therapies against drug resistant HIV strains

The development of drug resistance is a major obstacle to successful treatment of HIV infection. The extraordinary replication dynamics of HIV facilitates its escape from selective pressure exerted by the human immune system and by combination drug therapy. We have developed several computational methods whose combined use can support the design of optimal antiretroviral therapies based on viral genomic data.

Tenofovir Resistance and Resensitization

ABSTRACT Human immunodeficiency viruses in 321 samples from tenofovir-naïve patients were retrospectively evaluated for resistance to this nucleotide analogue. All virus strains with insertions between amino acids 67 and 70 of the reverse transcriptase (n = 6) were highly resistant. Virus strains with the Q151M mutation were divided into susceptible (n = 12) and highly resistant (n = 8) viruses. This difference was due to the absence or presence of the K65R mutation, which was confirmed by site-directed mutagenesis. Viral clones with various combinations of the mutations M41L, K70R, L210W, and T215F or T215Y were analyzed for cross-resistance induced by thymidine analogue mutations (TAMs). The levels of increased resistance induced by single, double, and triple mutations at the indicated positions could be ranked as follows: for mutants with single mutations, mutations at positions 41 > 215 > 70; for mutants with double mutations, mutations at positions 41 and 215 > 70 and 215 = 210 and 215 > 41 and 70; for mutants with triple mutations, mutations at positions 41, 210, and 215 > 41, 70, and 215. Viral clones with M184V or M184I exhibited slightly increased susceptibilities to tenofovir (0.7-fold). Almost all clones with TAM-induced resistance were resensitized when M184V was present (P < 0.001). Among the viruses in the clinical samples, the rate of tenofovir resistance significantly increased with the number of TAMs both in the samples with 184M and in those with 184V (P = 0.005 and P = 0.003, respectively). A resensitizing effect of M184V was confirmed for all samples exhibiting at least one TAM (P = 0.03). However, accumulation of at least two TAMs resulted in more than 2.0-fold reduced susceptibility to tenofovir, irrespective of the presence of M184V. Decision tree building, a classical machine learning technique, was used to generate models for the interpretation of mutations with respect to tenofovir resistance. The application of previously proposed cutoffs for a reduced response to therapy and treatment failure demonstrated the central roles of positions 215 and 65 for 1.5- and 4.0-fold reduced susceptibilities, respectively. Thus, clinically relevant resistance may be conferred by the accumulation of TAMs, and the resensitizing effect of M184V should be considered only minor.

Simple algorithm derived from a geno-/phenotypic database to predict HIV-1 protease inhibitor resistance

BackgroundResistance against protease inhibitors (PI) can either be analysed genotypically or phenotypically. However, the interpretation of genotypic data is difficult, particularly for PI, because of the unknown contributions of several mutations to resistance and cross-resistance. ObjectiveDevelopment of an algorithm to predict PI phenotype from genotypic data. MethodsRecombinant viruses containing patient-derived protease genes were analysed for sensitivity to indinavir, saquinavir, ritonavir and nelfinavir. Drug resistance-associated mutations were determined by direct sequencing. geno- and phenotypic data were compared for 119 samples from 97 HIV-1 infected patients. ResultsSamples with one or two mutations in the gene for the protease were phenotypically sensitive in 74.3%, whereas 83.6% of samples with five or more mutations were resistant against all PI tested. Some mutations (36I, 63P, 71V/T, 77I) were frequent both in sensitive and resistant samples, whereas others (24I, 30N, 46I/L, 48V, 54V, 82A/F/T/S, 84V, 90M) were predominantly present in resistant samples. Therefore, the presence or absence of a single drug resistance-associated mutation predicted phenotypic PI resistance with high sensitivity (96.5–100%) but low specificity (13.3–57.4%). A more specific algorithm was obtained by taking into account the total number of drug resistance-associated mutations in the gene for the protease and restricting these to certain key positions for the PI. The algorithm was subsequently validated by analysis of 72 independent samples. ConclusionWith an optimized algorithm, phenotypic PI resistance can be predicted by viral genotype with good sensitivity (89.1–93.0%) and specificity (82.6–93.3%). The reliability and relevance of this algorithm should be further evaluated in clinical practice.

Geno2pheno: interpreting genotypic HIV drug resistance tests

This intelligent system uses information encoded in the HIV genomic sequence to predict the viruss resistance or susceptibility to drugs. To make predictions, geno2pheno employs decision tree classifiers and support vector machines.

Low Level of Cross-Resistance to Amprenavir (141W94) in Samples from Patients Pretreated with Other Protease Inhibitors

ABSTRACT The therapeutic success of an antiretroviral salvage regimen containing protease inhibitors (PI) is limited by PI-resistant viral strains exhibiting various degrees of resistance and cross-resistance. To evaluate the extent of cross-resistance to the new PI amprenavir, 155 samples from 132 human immunodeficiency virus type 1-infected patients were analyzed for viral genotype by direct sequencing of the protease gene. Concomitantly, drug sensitivity to indinavir, saquinavir, ritonavir, nelfinavir, and amprenavir was analyzed by a recombinant virus assay. A total of 111 patients had been pretreated with 1-4 PI, but all were naive to amprenavir. A total of 105 samples (67.7%) were sensitive to amprenavir; 25 samples (16.1%) were intermediately resistant, and another 25 samples were highly resistant (4- to 8-fold- and >8-fold-reduced sensitivity, respectively). The mutations 46I/L, 54L/V, 84V, and 90M showed the strongest association with amprenavir resistance (P < 0.0001). The scoring system using 84V and/or any two of a number of mutations (10I/R/V/F, 46I/L, 54L/V, and 90M) predicted amprenavir resistance with a sensitivity of 86.0% and a specificity of 81.0% within the analyzed group of samples. Of 62 samples with resistance against 4 PI, 23 (37.1%) were still sensitive to amprenavir. In comparison, only 2 of 23 samples (8.7%) from nelfinavir-naive patients with resistance against indinavir, saquinavir, and ritonavir were still sensitive to nelfinavir. Amprenavir thus appears to be an interesting alternative for PI salvage therapy.