Assessing advances in three decades of clinical antiretroviral therapy on the HIV-1 reservoir

Abstract

BACKGROUNDAntiretroviral therapy (ART) has improved the clinical management of HIV-1 infection. However, little is known about how the latest ART recommendations affect the heterogeneity of the HIV-1 reservoir size.METHODSWe used a complete statistical approach to outline parameters underlying the diversity in HIV-1 reservoir size in a cohort of 892 people with HIV-1 (PWH) on suppressive ART for more than 3 years. Total HIV-1-DNA levels were measured in PBMCs using digital droplet PCR (ddPCR).RESULTSWe classified 179 (20%) participants as being low viral reservoir treated (LoViReT) (<50 HIV-1-DNA copies/106 PBMCs). Twenty variables were collected to explore their association with the LoViReT phenotype using machine learning approaches. LoViReT status was closely associated with higher nadir CD4, lower zenith pre-ART viral load, lower CD4 recovery, shorter time from diagnosis to undetectable viral load, and initiation of treatment with an integrase inhibitor-containing (InSTI-containing) regimen. Initiation of ART with any InSTI was also linked with a shorter time to undetectable viremia. Locally estimated scatterplot smoothing (LOESS) regression revealed a progressive reduction in the size of the HIV-1 reservoir in individuals who started ART after 2007. Similarly, a higher nadir CD4 and a shorter time to undetectable viremia were observed when treatment was initiated after that year.CONCLUSIONOur findings demonstrate that the progressive implementation of earlier, universal treatment at diagnosis and the use of InSTIs affected the size of the HIV-1 reservoir. Our work shows that effective management of infection is the first step toward reducing the reservoir and brings us closer to achieving a cure.FUNDINGNIH; Division of AIDS at the National Institute of Allergy and Infectious Diseases (NIAID), NIH; Merck Sharp & Dohme.

Keywords: AIDS/HIV; Drug therapy; Medical statistics; Molecular pathology; Virology.

Figure 1. Distribution of the HIV-1 reservoir.

Distribution of total HIV-1–DNA for the 892 volunteers screened by ddPCR. LoViReT participants (<50 HIV-1–DNA copies/10⁶ PBMCs) are shown in blue. Open symbols indicate values that fall below the detection limit; in these scenarios, the limit of detection varied according to sample/volume input.

Figure 2. Multivariate analysis to characterize the LoViReT phenotype.

(A) Representation of the top 15 low viral reservoir–associated variables reported by random forest machine learning algorithm analysis. Demographic, clinical, virologic, and immunologic variables are highlighted in gray, green, blue, and orange, respectively, while ART drug families are featured in dark red. Time taking a specific ART family is designated with a “t_” prefix. t_InSTI denotes time taking the integrase strand transfer inhibitors; MeanDecreaseGini is a measure of the importance of a variable based on the Gini Impurity index in the Random Forest algorithm. (B) PCA bidimensional plot showing data for LoViReT individuals (<50 HIV-1–DNA copies/10⁶ PBMCs) highlighted in blue and individuals harboring more than 50 HIV-1–DNA copies/10⁶ PBMCs in light gray. The names of the variables representing each arrow are detailed. (C) OR estimation and CI based on a logistic regression model for LoViReTs (<50 HIV-1–DNA copies/10⁶ PBMCs). CD4 nadir and CD4 recovery factors are displayed according to a magnitude change of 300 cells/μL, in accordance with the estimated mean of the nadir of the participants from the cohort.

Figure 3. Subanalysis of InSTIs.

The effect of initiating ART with InSTI-containing regimens from 2009 on (A) the proviral levels quantified in the study population and (B) over the time to suppression of the viral load. (C) The InSTI-independent effect on the time to suppression was also evaluated considering the various InSTI regimens in the clinical history of our cohort. LoViReT participants (<50 HIV-1–DNA copies/10⁶ PBMCs) are shown in blue. The Mann-Whitney U test was used to analyze the effect of InSTI-containing regimens on HIV-1 proviral levels, while Kruskal-Wallis and Conover post hoc tests were used to compare 2 or 3 groups of participants, respectively. NS, P ≥ 0.05; *P < 0.05 and ***P < 0.001.

Figure 4. Progression of proviral HIV-1–DNA and low viral reservoir–associated variables over time depending on the participants’ ART initiation date.

Distribution of the (A) total HIV-1–DNA levels, (B) nadir CD4, and (C) time from diagnosis to the first viral load suppression data at sampling versus the participants’ ART initiation year. Data on LoViReTs (<50 HIV-1–DNA copies/10⁶ PBMCs) are shown in blue. Open dots refer to values below the detection limit. The biphasic curve area from 2007 is highlighted in pale red. The time course of the approval of several antiretroviral drugs over time is shown above the plot, as are distinct milestones of HIV-1 management (navy blue bold) since the first reported AIDS cases. CCR5 antagonists, fusion inhibitors, InSTIs, nucleoside RT inhibitors, non-nucleoside RT inhibitors, and protease inhibitors are shown in green, magenta, fuchsia, yellow, violet, and orange, respectively, while drug combinations are shown in black. ABC, abacavir; ATV, atazanavir; BIC, bictegravir; CAB, cabotegravir; COBI; cobicistat; DRV, darunavir; DTG, dolutegravir; DOR, doravirine; EFV, efavirenz; EVG, elvitegravir; FTC, emtricitabine; T-20, enfuvirtide; ETR, etravirine; FPV, fosamprenavir; FTR, fostemsavir; HAART, highly active antiretroviral therapy; 3TC, lamivudine; LPV, lopinavir; MVC, maraviroc; NVP, nevirapine; RAL, raltegravir; RPV, rilpivirine; RTV, ritonavir; SQV, saquinavir; TAF, tenofovir alafenamide; TDF, tenofovir disoproxil fumarate; TPV, tipranavir; ZDV, zidovudine.