FARMS: A New Algorithm for Variable Selection.

Large datasets including an extensive number of covariates are generated these days in many different situations, for instance, in detailed genetic studies of outbreed human populations or in complex analyses of immune responses to different infections. Aiming at informing clinical interventions or vaccine design, methods for variable selection identifying those variables with the optimal prediction performance for a specific outcome are crucial. However, testing for all potential subsets of variables is not feasible and alternatives to existing methods are needed. Here, we describe a new method to handle such complex datasets, referred to as FARMS, that combines forward and all subsets regression for model selection. We apply FARMS to a host genetic and immunological dataset of over 800 individuals from Lima (Peru) and Durban (South Africa) who were HIV infected and tested for antiviral immune responses. This dataset includes more than 500 explanatory variables: around 400 variables with information on HIV immune reactivity and around 100 individual genetic characteristics. We have implemented FARMS in R statistical language and we showed that FARMS is fast and outcompetes other comparable commonly used approaches, thus providing a new tool for the thorough analysis of complex datasets without the need for massive computational infrastructure.

Clinical value of ultradeep HIV-1 genotyping and tropism testing in late presenters with advanced disease.

OBJECTIVE: This article aims to investigate if the detection of preexisting drug-resistant minority variant (DRMV) and/or X4 HIV-1 variants could improve the efficacy of first-line combined antiretroviral therapy (ART) in late presenters. DESIGN: Post-hoc, combined analysis of two open-label, prospective, randomized clinical trials comparing first-line ART with efavirenz (EFV) vs. ritonavir-boosted protease inhibitor (PI/r)-based regimens in ART-naive, HIV-1-infected patients, with CD4 T-cell counts less than 100 cells/µl and wild-type HIV-1 by bulk sequencing. METHODS: Pre-ART samples were reanalyzed for the presence of DRMVs and X4 HIV-1 using 454 sequencing. Kaplan-Meier curves and Cox regression were used to evaluate the association between X4 HIV and DRMVs and risk of virological failure. RESULTS: From 141 evaluable patients, 57 received EFV, and 84 received PI/r, including first-line ART. Median pre-ART CD4 T-cell counts and HIV-1 RNA levels were 39 cells/µl and 257 424 copies/ml, respectively; 35.5% of patients had X4 HIV variants. Detection of DRMVs leading to an ART-specific cumulative HIVdb score of at least 10 increased the risk of virological failure in patients initiating EFV [log-rank P = 0.048, hazard ratio = 4.3 (95% confidence interval: 0.8, 25.0), P = 0.074], but not in those starting PI/r. Presence of X4 HIV did not affect virological outcomes, but was associated with impaired CD4 T-cell count recovery over 2 years (214 vs. 315 cells/µl with X4 vs. R5 HIV-1 tropism, respectively, P = 0.017). CONCLUSION: Accounting for preexisting DRMVs may improve the outcomes of first-line nonnucleoside reverse transcriptase inhibitor-based ART in late presenters with advanced immune suppression. Presence of X4 HIV-1 at diagnosis predicts impaired immune restoration under ART.

The HLA-C*04: 01/KIR2DS4 gene combination and human leukocyte antigen alleles with high population frequency drive rate of HIV disease progression.

OBJECTIVE: The objective of this study is to identify human leukocyte antigen (HLA) class I and killer-cell immunoglobulin-like receptor (KIR) genotypes associated with different risks for HIV acquisition and HIV disease progression. DESIGN: A cross-sectional study of a cohort of 468 high-risk individuals (246 HIV-positive and 222 HIV-negative) from outpatient clinics in Lima (Perú). METHODS: The cohort was high-resolution HLA and KIR-typed and analysed for potential differences in single-allele frequencies and allele combinations between HIV-positive and HIV-negative individuals and for associations with HIV viral load and CD4 cell counts in infected individuals. RESULTS: HLA class I alleles associated with a lack of viral control had a significantly higher population frequency than relatively protective alleles (P = 0.0093), in line with a rare allele advantage. HLA-A02 : 01 and HLA-C04 : 01 were both associated with high viral loads (P = 0.0313 and 0.0001, respectively) and low CD4 cell counts (P = 0.0008 and 0.0087, respectively). Importantly, the association between HLA-C04 : 01 and poor viral control was not due to its linkage disequilibrium with other HLA alleles. Rather, the coexpression of its putative KIR ligand KIR2DS4f was critically linked to elevated viral loads. CONCLUSION: These results highlight the impact of population allele frequency on viral control and identify a novel association between HLA-C04 : 01 in combination with KIR2DS4f and uncontrolled HIV infection. Our data further support the importance of the interplay of markers of the adaptive and innate immune system in viral control.

Alternative effector-function profiling identifies broad HIV-specific T-cell responses in highly HIV-exposed individuals who remain uninfected.

The characterization of host immune responses to human immunodeficiency virus (HIV) in HIV controllers and individuals with high exposure but seronegativity to HIV (HESN) is needed to guide the development of effective preventive and therapeutic vaccine candidates. However, several technical hurdles severely limit the definition of an effective virus-specific T-cell response. By using a toggle-peptide approach, which takes HIV sequence diversity into account, and a novel, boosted cytokine staining/flow cytometry strategy, we here describe new patterns of T-cell responses to HIV that would be missed by standard assays. Importantly, this approach also allows detection of broad and strong virus-specific T-cell responses in HESN individuals that are characterized by a T-helper type 1 cytokine-like effector profile and produce cytokines that have been associated with potential control of HIV infection, including interleukin 10, interleukin 13, and interleukin 22. These results establish a novel approach to improve the current understanding of HIV-specific T-cell immunity and identify cellular immune responses and individual cytokines as potential markers of relative HIV resistance. As such, the findings also help develop similar strategies for more-comprehensive assessments of host immune responses to other human infections and immune-mediated disorders.

Monotherapy with boosted PIs as an ART simplification strategy in clinical practice.

BACKGROUND: Data on the efficacy of simplifying therapy using darunavir/ritonavir and lopinavir/ritonavir monotherapy in clinical practice remain limited. METHODS: A retrospective single-centre study including patients initiating darunavir/ritonavir or lopinavir/ritonavir monotherapy with a plasma HIV-1 viral load (pVL) <50 copies/mL and at least one subsequent follow-up visit. The primary endpoint was the percentage of patients remaining free of virological failure (VF; defined as a confirmed pVL >50 copies/mL or as any change in the regimen after a single determination with a pVL >50 copies/mL) during the follow-up. We also evaluated the percentage of patients remaining free of treatment failure (TF; defined as VF or the early discontinuation of monotherapy for any reason) and compared the effectiveness of the two regimens. Effectiveness was evaluated using cumulative survival analysis (at Weeks 48 and 96). Factors associated with VF and TF were analysed using Cox regression. RESULTS: A total of 522 patients were included (309 receiving lopinavir/ritonavir and 213 receiving darunavir/ritonavir). The median follow-up was 64.3 (30.5-143.0) weeks. The percentage of patients free of VF and TF was 94% (95% CI 91%-96%) and 79% (95% CI 75%-82%) at 48 weeks, respectively, and 86% (95% CI 81%-89%) and 62% (95% CI 57%-67%) at 96 weeks, respectively. The risk of VF was similar for the two regimens (HR=1.0, 95% CI 0.6-1.8; P=0.962). Lopinavir/ritonavir monotherapy was associated with a 1.5-fold greater risk of TF (95% CI 1.1-2.1; P=0.012) and a 2.3-fold greater risk of discontinuation of therapy due to adverse events (95% CI 1.3-3.9; P=0.003). CONCLUSIONS: The virological efficacy of darunavir/ritonavir and lopinavir/ritonavir monotherapy is high in clinical practice. Treatment discontinuation due to safety issues is more frequent with lopinavir/ritonavir.

Improved prediction of salvage antiretroviral therapy outcomes using ultrasensitive HIV-1 drug resistance testing.

BACKGROUND: The clinical relevance of ultrasensitive human immunodeficiency virus type 1 (HIV-1) genotypic resistance testing in antiretroviral treatment (ART)-experienced individuals remains unknown. METHODS: This was a retrospective, multicentre, cohort study in ART-experienced, HIV-1-infected adults who initiated salvage ART including, at least 1 ritonavir-boosted protease inhibitor, raltegravir or etravirine. Presalvage ART Sanger and 454 sequencing of plasma HIV-1 were used to generate separate genotypic sensitivity scores (GSS) using the HIVdb, ANRS, and REGA algorithms. Virological failure (VF) was defined as 2 consecutive HIV-1 RNA levels ≥200 copies/mL at least 12 weeks after salvage ART initiation, whereas subjects remained on the same ART. The ability of Sanger and 454-GSS to predict VF was assessed by receiver operating characteristic (ROC) curves and survival analyses. RESULTS: The study included 132 evaluable subjects; 28 (21%) developed VF. Using HIVdb, 454 predicted VF better than Sanger sequencing in the ROC curve analysis (area under the curve: 0.69 vs 0.60, Delong test P = .029). Time to VF was shorter for subjects with 454-GSS < 3 vs 454-GSS ≥ 3 (Log-rank P = .003) but not significantly different between Sanger-GSS < 3 and ≥3. Factors independently associated with increased risk of VF in multivariate Cox regression were a 454-GSS < 3 (HR = 4.6, 95 CI, [1.5, 14.0], P = .007), and the number of previous antiretrovirals received (HR = 1.2 per additional drug, 95 CI, [1.1, 1.3], P = .001). Equivalent findings were obtained with the ANRS and REGA algorithms. CONCLUSIONS: Ultrasensitive HIV-1 genotyping improves GSS-based predictions of virological outcomes of salvage ART relative to Sanger sequencing. This may improve the clinical management of ART-experienced subjects living with HIV-1. CLINICAL TRIALS REGISTRATION: NCT01346878.

Stable HIV-1 integrase diversity during initial HIV-1 RNA decay suggests complete blockade of plasma HIV-1 replication by effective raltegravir-containing salvage therapy.

BACKGROUND: There is legitimate concern that minority drug-resistant mutants may be selected during the initial HIV-1 RNA decay phase following antiretroviral therapy initiation, thus undermining efficacy of treatment. The goal of this study was to characterize viral resistance emergence and address viral population evolution during the first phase of viral decay after treatment containing initiation. FINDINGS: 454 sequencing was used to characterize viral genetic diversity and polymorphism composition of the HIV-1 integrase gene during the first two weeks following initiation of raltegravir-containing HAART in four ART-experienced subjects. No low-prevalence Raltegravir (RAL) drug resistance mutations (DRM) were found at baseline. All patients undergoing treatment received a fully active ART according to GSS values (GSS ≥ 3.5). No emergence of DRM after treatment initiation was detected. Longitudinal analysis showed no evidence of any other polymorphic mutation emergence or variation in viral diversity indexes. CONCLUSIONS: This suggests that fully active salvage antiretroviral therapy including raltegravir achieves a complete blockade of HIV-1 replication in plasma. It is unlikely that raltegravir-resistant HIV-1 may be selected in plasma during the early HIV-1 RNA decay after treatment initiation if the administered therapy is active enough.

HIV-1 tropism testing in subjects achieving undetectable HIV-1 RNA: diagnostic accuracy, viral evolution and compartmentalization.

BACKGROUND: Technically, HIV-1 tropism can be evaluated in plasma or peripheral blood mononuclear cells (PBMCs). However, only tropism testing of plasma HIV-1 has been validated as a tool to predict virological response to CCR5 antagonists in clinical trials. The preferable tropism testing strategy in subjects with undetectable HIV-1 viremia, in whom plasma tropism testing is not feasible, remains uncertain. METHODS & RESULTS: We designed a proof-of-concept study including 30 chronically HIV-1-infected individuals who achieved HIV-1 RNA <50 copies/mL during at least 2 years after first-line ART initiation. First, we determined the diagnostic accuracy of 454 and population sequencing of gp120 V3-loops in plasma and PBMCs, as well as of MT-2 assays before ART initiation. The Enhanced Sensitivity Trofile Assay (ESTA) was used as the technical reference standard. 454 sequencing of plasma viruses provided the highest agreement with ESTA. The accuracy of 454 sequencing decreased in PBMCs due to reduced specificity. Population sequencing in plasma and PBMCs was slightly less accurate than plasma 454 sequencing, being less sensitive but more specific. MT-2 assays had low sensitivity but 100% specificity. Then, we used optimized 454 sequence data to investigate viral evolution in PBMCs during viremia suppression and only found evolution of R5 viruses in one subject. No de novo CXCR4-using HIV-1 production was observed over time. Finally, Slatkin-Maddison tests suggested that plasma and cell-associated V3 forms were sometimes compartmentalized. CONCLUSIONS: The absence of tropism shifts during viremia suppression suggests that, when available, testing of stored plasma samples is generally safe and informative, provided that HIV-1 suppression is maintained. Tropism testing in PBMCs may not necessarily produce equivalent biological results to plasma, because the structure of viral populations and the diagnostic performance of tropism assays may sometimes vary between compartments. Thereby, proviral DNA tropism testing should be specifically validated in clinical trials before it can be applied to routine clinical decision-making.

Definition of the viral targets of protective HIV-1-specific T cell responses.

BACKGROUND: The efficacy of the CTL component of a future HIV-1 vaccine will depend on the induction of responses with the most potent antiviral activity and broad HLA class I restriction. However, current HIV vaccine designs are largely based on viral sequence alignments only, not incorporating experimental data on T cell function and specificity. METHODS: Here, 950 untreated HIV-1 clade B or -C infected individuals were tested for responses to sets of 410 overlapping peptides (OLP) spanning the entire HIV-1 proteome. For each OLP, a "protective ratio" (PR) was calculated as the ratio of median viral loads (VL) between OLP non-responders and responders. RESULTS: For both clades, there was a negative relationship between the PR and the entropy of the OLP sequence. There was also a significant additive effect of multiple responses to beneficial OLP. Responses to beneficial OLP were of significantly higher functional avidity than responses to non-beneficial OLP. They also had superior in-vitro antiviral activities and, importantly, were at least as predictive of individuals' viral loads than their HLA class I genotypes. CONCLUSIONS: The data thus identify immunogen sequence candidates for HIV and provide an approach for T cell immunogen design applicable to other viral infections.

Geographic variation in the prevalence of Kaposi sarcoma-associated herpesvirus and risk factors for transmission.

BACKGROUND: The aim of the present study was to estimate the prevalence of Kaposi sarcoma-associated herpesvirus (KSHV) in the female general population, to define geographic variation in and heterosexual transmission of the virus. METHODS: The study included 10,963 women from 9 countries for whom information on sociodemographic characteristics and reproductive, sexual, and smoking behaviors were available. Antibodies against KSHV that encoded lytic antigen K8.1 and latent antigen ORF73 were determined. RESULTS: The range of prevalence of KSHV (defined as detection of any antigen) was 3.81%-46.02%, with significant geographic variation noted. In Nigeria, the prevalence was 46.02%; in Colombia, 13.32%; in Costa Rica, 9.81%; in Argentina, 6.40%; in Ho Chi Minh City, Vietnam, 15.50%; in Hanoi, Vietnam, 11.26%; in Songkla, Thailand, 10%; in Lampang, Thailand, 8.63%; in Korea, 4.93%; and in Spain, 3.65%. The prevalence of KSHV slightly increased with increasing age among subjects in geographic areas where the prevalence of KSHV was high, such as Nigeria and Colombia, and it significantly decreased with increases in the educational level attained by subjects in those areas. KSHV was not statistically associated with age at first sexual intercourse, number of sex partners, number of children, patterns of oral contraceptive use, presence of cervical human papillomavirus DNA, or smoking status. CONCLUSIONS: The study provides comparable estimates of KSHV prevalence in diverse cultural settings across 4 continents and provides evidence that sexual transmission of KSHV is not a major source of infection in the general population.

Genetic variants in apoptosis and immunoregulation-related genes are associated with risk of chronic lymphocytic leukemia.

To identify low-penetrance susceptibility alleles for chronic lymphocytic leukemia (CLL), we performed a case-control study genotyping 768 single-nucleotide polymorphisms (SNP) in 692 cases of CLL and 738 controls. We investigated nonsynonymous SNPs, SNPs with potential functional effect, and tag SNPs in regulatory gene regions in a total of 172 genes involved in cancer biology. After adjustment for multiple testing, we found a strong association between CLL risk and six genetic variants: CCNH (rs2266690, V270A), APAF1 (rs17028658, 3'region), IL16 (rs4505265, first intron), CASP8 (rs1045485, D302H), NOS2A (rs2779251, promoter), and CCR7 (rs3136687, intron 1). We found association with CLL susceptibility and 22 haplotypes in APAF1, IL6, TNFRSF13B, IL16, CASP3, CCR7, LTA/TNF, BAX, BCL2, CXCL12, CASP10/CASP8, CASP1, CCL2, BAK1, and IL1A candidate genes. Finally, we evaluated using public data sets the potential functional effect on gene expression levels of the CLL associated genetic variants detected in regulatory regions. Minor alleles for APAF1 and IL16 were associated with lower mRNA levels; no expression differences were observed for CCR7, whereas NOS2A could not be assessed. This study suggests that common genetic variation in apoptosis- and immunoregulation-related genes is associated with the CLL risk.