Productive Infection Maintains a Dynamic Steady State of Residual Viremia in Human Immunodeficiency Virus Type 1-Infected Persons Treated with Suppressive Antiretroviral Therapy for Five Years

Study subjects.

All study subjects were HIV type 1 (HIV-1)-infected participants of a phase II dose-finding study of efavirenz (600 mg daily) and indinavir. Indinavir was administered at a dose of 800 to 1,200 mg three times daily until month 4, when all patients received 1,200 mg. All patients received indinavir plus efavirenz at the onset of the study, except for two patients who started indinavir first for 6 months and then added efavirenz. Patients 2 and 5 (Table 1) also received lamivudine and stavudine, respectively, 6 months after therapy initiation. Subject 5 discontinued study therapy for drug toxicity (nephrolithiasis) after 5 years. Therapy intensification was offered to all patients after 5 years and consisted of abacavir in all subjects except patient 3, who requested ritonavir. Patients 2, 11, and 12 stopped abacavir at week 10 and 3 and day 2, respectively, for mild side effects. Viral kinetics were not evaluated for subject 15, who had transferred from another institution.

Assay of residual viremia.

Assays for residual viremia between 2.5 and 50 copies/ml were performed according to published methods (25). The Amplicor Ultrasensitive assay was modified to permit the nominal detection of 2.5 copies of HIV RNA (Roche Molecular Diagnostics, Branchburg, N.J.)/ml. Two milliliters (versus 0.5 ml) of frozen plasma was thawed and centrifuged at 23,600 × g for 2 h (versus 1 h) at 4°C to pellet virus. One-half the normal quantitation standard (QS) volume was added to the sample prior to the lysis step, and the RNA pellet was resuspended in 50 μl of diluent volume (versus 100 μl for the Ultrasensitive assay). The entire volume (50 μl) of resuspended RNA was assayed by reverse transcription and PCR amplification. These and subsequent detection steps followed the manufacturer's protocol exactly. Because only half the QS volume is added prior to the lysis and RNA extraction steps but the extracted RNA is resuspended in half the usual volume, normalization for the detected QS value could be performed without modification. The computed value of copies per milliliter was corrected for the protocol modifications by dividing by 8. This included a fourfold correction factor for the larger volume of plasma assayed (2 versus 0.5 ml) and a twofold correction factor for the volume of diluent used to resuspend the extracted RNA (50 versus 100 μl).

To establish the limit of detection and reproducibility of the assay, plasma from two HIV-infected patients with viral RNA concentrations previously determined by the Amplicor or Amplicor Ultrasensitive assays were serially diluted in control plasma from healthy HIV-negative donors to a calculated concentration of 50 copies/ml down to 1.25 copies/ml. Replicate samples at each dilution were assayed by the modified 2.5-copies/ml assay. The correlation of the results based on Amplicor and the modified 2.5-ml assay was excellent with an r² of 0.92 (P = 0.003). Coefficients of variation at individual concentrations ranged from 121% at 1.25 copies/ml to 9% at 50 copies/ml. At 2.5 copies/ml, the coefficient of variation was 37%.

HIV DNA assays.

Peripheral blood mononuclear cell (PBMC) DNA quantitation was performed using a PCR-based system with colorimetric detection according to published methods (6). Based on genomic DNA input into PCRs, the lower limit for reliable quantitation was 5 HIV DNA copies/μg of PBMC DNA. Viral DNA values presented here are normalized to total blood volume by multiplying copy numbers per purified lymphocyte by simultaneous peripheral blood lymphocyte counts. In a natural history study, values computed in this normalization were more stable over the course of HIV disease, including a range of CD4 counts extending below 200 cells/mm³, than when normalized to number of PBMC or CD4⁺ T cells (12). However, normalization of HIV DNA copies to micrograms of PBMC DNA or CD4 DNA gave qualitatively similar results (data not shown).

Viral kinetics.

Initial decline of HIV RNA obeyed biphasic exponential decay to a nonzero constant. Plasma HIV RNA values V(t) from the first year of therapy were fitted to the form

(1)

assuming multiplicative Gaussian noise g(0, s) of zero mean and standard deviation σ.

(2)

by Bayesian methods. The clearance rates δ_I and δ_II are ln(2) times the reciprocal of the first-phase half-life T_I and the second phase half-life T_II, respectively. Simulations were performed using the Markov chain Monte Carlo technique in WinBugs. Estimates for decay half-lives computed by a least-squares fit to equation 1 were not statistically different.

The duration of phase II HIV RNA clearance was defined as the period during which more virus was produced from this pool than from longer-lived (residual) sources, based on the optimal biphasic fit of HIV RNA dynamics. HIV RNA dynamics in patient 8 differed significantly in several regards. In this patient, the measured baseline DNA was higher (22,000 copies/ml), baseline CD4 was lower (81 cells/mm³), and second-phase RNA clearance was slower. In addition, HIV DNA continued to decay significantly (P < 0.05) in this patient throughout the study period. Residual HIV RNA and DNA could therefore not be appropriately defined in this patient, and he was excluded from the analysis of predictors of residual viremia. The inclusion in our cohort of one patient who failed to reach steady states of HIV RNA or HIV DNA raises the possibility that some patients exhibit different long-term viral clearance kinetics; however, much larger cohorts would be necessary to study this intriguing minority.

Decay rates after abacavir intensification were computed by least-squares, treating undetectable values as 2.5 copies/ml. They therefore represent an underestimate of the true decay rate, except in patient 1, who retained detectable viremia. To determine whether transient elevations of HIV RNA at the time of intensification were responsible for the rapid kinetics, clearance rates were also computed, substituting the residual viremia over the previous 4 years as the baseline value.

Semiquantitative microculture assays.

Fifty milliliters of fresh acid citrate dextrose anticoagulated whole blood was processed for CD4 lymphocyte enrichment or CD8 lymphocyte depletion using the Rossette-sep procedure according to the manufacturer's directions (Stem Cell Technologies, Vancouver, British Columbia, Canada). Purity of CD4 fractions was >90% with fewer than 2% CD8⁺ or NK cells or monocytes by fluorescence-activated cell sorting (FACS) analysis. Semiquantitative microculture assays were performed according to previously described methods except for replicate wells at concentrations of 1 × 10⁵, 3 × 10⁵, and 1 × 10⁶ activated patient CD4 cells depending on cell yields (44).

FACS analyses.

Leukocyte differentiation was performed using a Coulter counter (Coulter Electronics, Inc., Miami Lakes, Fla.). Lymphocyte subsets were evaluated by flow cytometric analysis, using 50 μl of EDTA peripheral blood incubated for 30 min at 4°C with fluorochrome-labeled monoclonal antibodies. After incubation, erythrocyte lysis and fixation of marked cells were performed using the Immuno-Prep EPICS Kit (Coulter Electronics).

IFN-γ ELISpot assays.

Ninety-six-well nitrocellulose plates were first precoated with a layer of gamma interferon (IFN-γ) monoclonal antibodies (MABTECH, Nacka, Sweden). PBMC were then added in duplicate wells either with a pool of five previously described synthetic peptides from the gp160 envelope of HIV-1 (20 μM final concentration), with HIV-1 p24 (Protein Sciences Co., Meriden, Conn.) (p24) (0.1 μg/ml) in the presence or the absence of neutralizing anti-CD4 monoclonal antibody (see below), or with no peptide (negative control) (11). The five peptides used in the stimulation are promiscuous as they are recognized by multiple HLA class I molecules (including HLA A1, A2, A3, A9, A25, A26, A29, etc.) and are mostly conserved between different HIV clades. Because these epitopes can also be recognized by HLA class II molecules (10), IFN-γ production by CD4⁺ was blocked by preincubating PBMC with 100 ng of neutralizing recombinant human CD4 monoclonal antibody (R&D Systems, Minneapolis, Minn.)/ml. Plates were incubated overnight at 37°C in 7% CO₂, then the cells were discarded, and the plates were incubated at room temperature for 3 h. A second biotinylated anti-IFN-γ monoclonal antibody (7-B6-1 biotin; MABTECH), followed by streptavidin-conjugated alkaline phosphatase (MABTECH) for 2 h, was subsequently used. Individual IFN-γ-producing cells were detected as dark blue spots using an alkaline phosphatase conjugate substrate kit (Bio-Rad Laboratories, Hercules, Calif.). The spots were counted using a dissecting microscope (×40). HIV-specific responses were reported as number of spot-forming units per 10⁶ mononuclear cells. Baseline and week 24 values were compared using a paired, two-tailed Student t test.

Kinetics of HIV RNA in the plasma.

To characterize kinetics of residual viremia, we examined the decay of HIV RNA in the plasma in our 14 patients sustaining HIV RNA levels of <50 copies/ml for 5 years after initiation of treatment with indinavir plus efavirenz. Prior to treatment, these patients had a median CD4 cell count of 261 cells/mm³ and HIV RNA level of 4.9 log copies/ml (Table 1). HIV RNA values from the first year were fitted assuming biphasic exponential decay to a constant nonzero value. The estimated clearance half-lives of infected cells were 1.2 ± 0.8 days during phase I decay and 24 ± 16 days during phase II decay, consistent with previous reports. We were also able to estimate the duration of phase II decay as between 33 and 260 days (Fig. 1). Prior estimates of the dynamics of the end of phase II have been limited by the sensitivity of the viral RNA assays used. Although the likely complex mixture of cell populations contributing to the pool of plasma virus prevents the association of this time interval with the lifetimes of a particular cellular phenotype, this first direct estimate of the duration of phase II decay provides a quantitative basis for defining phases of HIV treatment.

We next examined whether there was further HIV RNA decline in the plasma after phase II decay during subsequent years of therapy. Prior longitudinal studies of patients with chronic infection have measured HIV RNA using assays with a limit of detection of 50 copies/ml and thus not addressed this issue. Studies using more sensitive HIV RNA assays have detected virus in the plasma but have not examined patients frequently with longitudinal samples (15, 45). In this cohort of patients, using an assay with a limit of detection of 2.5 copies/ml, we were able to measure virus longitudinally and found no decline in levels of residual viremia in 13 of 14 patients over 5 years of observation (Fig. 2). Patients had a median of 13 measures of residual viremia. Of note, intermittent viremia (HIV RNA greater than 50 copies/ml after 9 months of therapy) was infrequent in this cohort. Only four of our patients had any evidence of intermittent viremia after the first 9 months of therapy. The overall frequency of viremia greater than 50 copies/ml after 9 months of antiretroviral therapy was 0.02 episodes per patient-year.

Predictors of residual viremia.

Levels of residual viremia in the plasma varied among patients. Geometric mean RNA values varied from 3.2 to 23 copies/ml, with a median of 6.0 copies/ml (Table 1). When we examined the predictors of the level of residual viremia in these patients, we found that pretreatment levels of total HIV DNA correlated with residual viremia (Fig. 3). Interestingly, baseline HIV RNA and CD4 cell count were not predictors of residual viremia.

Kinetics of HIV DNA.

The kinetics of viral DNA were also examined in this cohort of patients. Pretreatment levels ranged from 313 to 22,000 HIV DNA copies per ml of blood. These levels were not associated with pretreatment HIV RNA measurements, CD4 cell count, prior treatment history, or estimated duration of HIV infection. We measured HIV DNA at 2-month intervals during the first year of treatment and at 6-month intervals over the remainder of the study period. During the first year of antiretroviral therapy, HIV DNA declined by a median of twofold (interquartile range, 1.7- to 3.5-fold). The median half-life during this period (Table 1) was 466 days. This is somewhat longer than previously reported, in part due to the normalization of DNA copy numbers to blood volume, which eliminates the dilutional decrease seen when normalizing to rising total lymphocyte numbers (1, 29, 30). After the first year of treatment, the HIV DNA showed no significant decline. DNA copy numbers remained measurable in all patients throughout the course of the study.

Viral kinetics after treatment intensification.

Given that HIV RNA and DNA remained constant for years after the initial 9 months of therapy, we sought to determine if there was a dynamic steady state of viral production or if plasma virus resulted exclusively from release of particles from latently infected cells. A nucleoside reverse transcriptase inhibitor would block infection of new cells but would have no effect on the release of virus directly from latently infected cells. Therefore, we measured plasma HIV RNA and DNA levels and cell-associated infectivity in patients at frequent intervals for 24 weeks in patients from this cohort who chose to add the nucleoside reverse transcriptase inhibitor abacavir (eight patients) or ritonavir (one patient) to their regimen after 5 years of therapy. The five patients from this cohort who did not add abacavir were also monitored prospectively.

Plasma HIV RNA levels rapidly declined by 4 weeks in four of five patients with detectable HIV RNA who received abacavir (Fig. 4) (Table 1). We were unable to quantify the magnitude of decline because of the limit of detection of our assay. The estimated half-life of infected cells was 6.7 days (range, 4.4 to 11 days). In two of the five patients (subjects 7 and 8) who did not add abacavir, HIV RNA levels did not decrease. Steady-state levels below the limit of detection or missing data precluded evaluation in the other three patients Total HIV DNA levels did not decline over the 24 weeks of observation (data not shown). There was no change in semiquantitative cultures of HIV infectivity (data not shown).

Cellular activation and HIV immune responses after therapy intensification.

Reductions in HIV RNA levels in patients initiating therapy are associated with rises in CD4 cell counts, reduction in HIV-associated immune responses, and decline in cellular activation (2). Initial CD4 cell increases are attributed to redistribution of memory cells (36). In this group of patients already treated for 5 years, we did not observe increases in CD4 cell counts after therapy intensification (data not shown). However, we did observe a reduction in some markers of cellular activation in both CD4⁺ and CD8⁺ cells (Table 2). CD8-mediated HIV Gag and p24 antigen responses were also lower in patients receiving abacavir, although the differences between the groups were not statistically different when the lower background activation rates in the abacavir group were included in the analyses. These observations provide support for a reduction in HIV antigen load with therapy intensification, suggesting perturbation of a dynamic equilibrium by the addition of an additional antiretroviral agent.