Evidence that low-level viremias during effective highly active antiretroviral therapy result from two processes: expression of archival virus and replication of virus

Abstract

Episodes of low-level viremia (LLV), with plasma human immunodeficiency virus type 1 (HIV-1) RNA levels ranging from 50 to 400 copies (c)/ml, occur commonly during highly active antiretroviral therapy (HAART). LLV has been associated with virologic failure of HAART in some studies, while in others LLV did not appear to affect the clinical outcome. To understand the processes leading to LLV, genetic analyses were used to determine whether plasma virions emanated from archived or from newly evolved viral genomes. Episodes of LLV (plasma HIV-1 RNA, 50 to 379 [median, 77] c/ml) were detected in 21/37 (57%) HIV-1-infected children with median plasma HIV-1 RNA levels of <50 c/ml during 79 patient years of HAART. Viral sequences were derived by direct sequencing of PCR products from 21 plasma specimens diluted to end point. In phylogenetic analysis, LLV viral sequences grouped with virus from early in the course of infection in 8/11 subjects. Six specimens had multiple identical viral sequences, suggesting origin from clonally expanded infected cells. LLV plasma virus evolved over time, indicating viral replication, in 3/11 subjects. Two of these had frequent LLV, including the selection of drug-resistant mutants. In summary, plasma virus from episodes of LLV during effective HAART appeared to originate from two distinct processes, (i) clonal outgrowth from long-lived HIV-1-infected cells, presumably following activation and proliferation of these cells, and (ii) ongoing viral replication that included the selection of new drug-resistant mutants. These observations provide a plausible explanation for the divergent clinical outcomes previously associated with LLV.

FIG. 1.

HIV-1 RNA and CD4 cell levels over time. Plasma HIV-1 RNA and CD4 cell levels of subjects are shown over time. The lower limits of detection of the plasma RNA assays used are shaded. Times at which HAART was initiated, intensified, or stopped are indicated with the symbols defined in the inset. The following very low LLV values of two subjects are not discernible: B2, 51 c/ml at 12.6 years of age; V1, 53 c/ml at 8.5 years of age. LPV/r, lopinavir-ritonavir combination.

FIG. 2.

Genetic divergence from the most recent common ancestor of infection during HAART. Genetic divergence from the MRCA of pol encoding RT (left side) and env encoding C2-V5 (right side) is shown for representative subjects. The upper three panels (subjects F1, G2, and T2) show viral sequences diverging from the MRCA during HAART in both plasma (gray diamonds) and PBMC (black dots), suggesting ongoing viral replication despite a median plasma HIV-1 RNA level of <50 c/ml. In the lower panels (subjects H2, H4, M2, and P1), plasma and PBMC virus did not diverge from the MRCA, suggesting that viral replication was inhibited by HAART.

FIG. 3.

Divergence of PBMC-associated HIV-1 env sequences after initiation of HAART. Viral sequence divergence from the pre-HAART specimen was plotted (vertical axis) for each specimen analyzed during HAART (horizontal axis). The point of reference for calculating divergence was the mean distance of each subject's viral sequences from his or her specimen collected immediately prior to the initiation of HAART. While the divergence slopes for F1 and G2 are similarly positive, suggesting ongoing viral replication, only the slope of G2's data was statistically significant (P = 0.00002) in the linear random-effect model, likely due to the greater amount of G2 data and a longer period of observation.

FIG. 4.

Neighbor-joining phylogram of pol encoding PR with drug-resistant genotypes selected during HAART shown for subject G2. Plasma (gray diamonds) and PBMC (black dots) viral sequence evolution of G2 is shown. Selection of drug-resistant mutants (with codon and encoded amino acid listed) contributed significantly to the evolution from the MRCA (circled). Sequences are rooted to subtype B sequences from GenBank and the percentage of substitutions per site indicated by the horizontal line.

FIG. 5.

Amino acid alignment of HIV-1 pol sequences encoding RT of subject G2. An alignment of a subset of G2 viral sequences is shown to demonstrate the increasing prevalence of mutants over time in PBMC. Sequences are shown in an abbreviated format to feature amino acids that confer resistance to antiretroviral drugs (nucleoside and nonnucleoside RT inhibitors). Dots indicate amino acid identity to the subject's MRCA (G2 Ancestor Seq). Sequences from PBMC are labeled with specimen dates and cell and sequence indicators, and plasma sequences are similarly labeled with dates and plasma and sequence indicators.