CD4 T Cell Count Reconstitution in HIV Controllers after Highly Active Antiretroviral Therapy

Human immunodeficiency virus (HIV) controllers (HICs) spontaneously control plasma viral load in the absence of highly active antiretroviral therapy (HAART). HICs can be classified as elite or viremic controllers, depending on the degree of virologic control [1, 2]. HICs are generally uncommon, with elite and viremic controllers comprising <1% and 3% of individuals, respectively [3–5].

In the absence of therapy, HICs exhibit characteristics similar to those of noncontrollers receiving HAART, specifically virologic suppression and favorable clinical outcomes. Although the definition is based on virologic criteria, HICs also demonstrate higher initial CD4 cell counts and slower rates of decline than do noncontrollers [5–7]. However, over time a pro- portion of HICs eventually require HAART. The CD4 cell response to HAART in HICs, a population with primarily low viral load set points, to our knowledge has not been previously described. We report here the outcomes after HAART for HICs compared with noncontrollers in the US Military Department of Defense HIV Natural History Study (NHS).

Methods. Participants who received HAART were identified in the database of >5000 patients enrolled in the NHS since 1986, a cohort of consenting military personnel and beneficiaries who were 18 years or older and infected with HIV [8]. Elite controllers were defined as patients with ⩾3 viral loads below the limit of detection for >12 months in the absence of HAART; viremic controllers were defined similarly but with a viral load threshold of ⩽ 2000 copies/mL [5]. HICs (combined group of elite and viremic controllers) initiating HAART were compared with noncontrollers, defined as patients initiating HAART who did not meet the HIC definitions. Patients who had used antiretrovirals before starting HAART were excluded. Because all HICs initiated HAART during 1997–2008, non-controllers were restricted to those starting HAART during the same period.

CD4 cell and viral load values at HIV diagnosis, at HAART initiation, and at 6-month intervals after HAART initiation through 48 months were assessed. Values at HIV diagnosis were the first recorded within 6 months after diagnosis; values at HAART initiation were the last recorded up to 6 months before HAART. Six-month post-HAART interval values were the closest recorded ±90 days. Virologic suppression was defined as a viral load <400 copies/mL; virologic failure was defined as 2 consecutive detectable viral loads after suppression or never achieving suppression. AIDS-defining illnesses were defined using the 1993 Centers for Disease Control and Prevention classification [9]. Seroconverters were patients who had a documented negative HIV test result before their first positive HIV test result.

The t test and the x² test were used to compare groups for factors before and at HAART initiation. Analyses of variance and covariance (for individual visits after HAART) and longitudinal models (incorporating all visits, using the Mixed procedure in SAS software, version 9.2) were used to compare the HIC and noncontroller groups for changes in CD4 cell count and viral load after HAART initiation. Covariates included the year starting HAART (2-year intervals), age and CD4 cell count at HAART initiation, sex, and race/ethnicity (African American versus all other races/ethnicities combined). Viral load at HAART initiation was also included in the model for virologic response to therapy. For the longitudinal model, follow-up visit was also included in the model with a compound symmetry covariance structure. For each visit and overall, the model- based difference between groups ± standard error (SE) is reported. Separate analyses of the change in CD4 cell count were also performed for 3 strata of viral load and CD4 cell counts at HAART initiation. Cox regression models were used to compare groups for time to virologic failure and first AIDS event. Differences with P < .05 were considered statistically significant.

Results. Sixty-two HICs (6 elite and 56 viremic controllers) and 1065 noncontrollers initiated HAART. Compared with non-controllers, HICs had a statistically significantly higher pro- portion of women, were younger at HIV diagnosis, received a diagnosis in an earlier calendar period (Table 1), and had a longer duration between HIV diagnosis and HAART initiation (median [interquartile range], 74 [44–120] vs 5 [2–26] months). Fifty HICs (80.6%) and 894 noncontrollers (83.9%) were seroconverters ( P = .49 ). There were 2 (3.2%) and 74 (6.9%) AIDS events before HAART initiation for HICs and noncontrollers, respectively (P = .26).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1.

Characteristics of Human Immunodeficiency Virus (HIV) Controllers (HICs) and Noncontrollers

As a consequence of the HIC definition, viral load was significantly lower at HAART initiation for HICs than for non-controllers (mean, 3.5 vs 4.6 log₁₀ copies/mL; P < .001 ), with 13 HICs (22%) having a viral load <400 copies/mL at HAART initiation, compared with 29 noncontrollers (3.0%) (Table 2). Nine and 4 HICs had pretherapy viral loads <400 and <50 copies/mL, respectively. Of these HICs, 3 were elite controllers (1 with a pre-HAART viral load <400 and 2 with <50 copies/ mL). On the basis of virologic criteria, 3 elite controllers (50%) and 37 viremic controllers (66%) maintained controller status through the start of HAART. Statistically significant decreases in viral load occurred in both HICs and noncontrollers but were smaller for HICs than for noncontrollers ( P < .001 ); this difference was no longer significant after adjusting for covariates that included viral load at HAART initiation. Rates of virologic suppression were similarly high in both groups (90% vs 82% at 6 months and 84% vs 82% at 24 months). Virologic failure was more frequent in noncontrollers (34.1%) than in HICs (21.8%), but the difference was not statistically significant (unadjusted P = .15). For HICs with an undetectable viral load at HAART initiation, virologic failure was observed in 1 elite controller, who had consecutive viral loads of 566 and 618 copies/mL during a period of treatment interruption.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2.

CD4 Cell Count and Viral Load Outcomes in Various Human Immunodeficiency Virus (HIV) Controller (HIC) and Noncontroller Subgroups

HICs had statistically significantly higher mean CD4 cell counts at HIV diagnosis than did noncontrollers (678 vs 460 cells/μL; P < .001 ); however, values were similar at HAART initiation (376 vs 366 cells/μL; P = .70). Both groups demonstrated statistically significant increases in CD4 cell counts after HAART initiation at each time point ( P < .001 for all); however, CD4 cell count gains tended to be smaller in HICs than in noncontrollers (151 vs 197 and 200 vs 230 cells/μL at 12 and 24 months, respectively) (Table 1). On the basis of longitudinal analyses, the CD4 cell count increase was 46 ± 23 cells/μL smaller (weighted average ± SE) in HICs than in noncontrollers (P = .07); adjusted analyses yielded similar results (47 ± 23 cells/μL; P = .06). The median increase in CD4 cell count over all visits was 39 cells/μL lower in HICs than in noncontrollers ( P < .01 ). With additional adjustment for viral load at HAART initiation, the difference between groups was not statistically significant (weighted average ± SE, 20 ± 23 cells/μL; P = .48 ). Analyses using strata of viral load at HAART initiation (<400, 400–9999, and ⩾10,000 copies/μL) showed non-statistically significant adjusted CD4 cell count differences (weighted average ± SE) of 42 ± 66 (P = .44 ), 21 ± 35 (P = .61 ), and 14 ± 38 cells/μL( P = .74 ), respectively. CD4 cell count increases for HICs with a viral load <400 copies/mL at HAART initiation were a mean ± SE of 109 ± 47 and 138 ± 48 cells/μL at 12 and 24 months, respectively (Table 2).

Outcomes of AIDS (6 vs 86; hazard ratio, 1.3 [95% confidence interval, 0.6–3.0]; P = .53) and death (1 vs 39; hazard ratio, 0.5 [95% confidence interval, 0.1–3.6]; P = .49 ) were similar for HICs compared with noncontrollers. Results were similar after adjusting for covariates.

Discussion. Despite the association with higher CD4 cell counts early in disease, many HICs eventually experience CD4 cell count declines that meet the threshold for HAART initiation. We have previously shown that elite controllers have more stable overall CD4 cell count trends than do viremic controllers [5], with a smaller proportion of elite controllers starting HAART. For the approximately one-third of HICs who received HAART in the NHS, statistically significant increases in CD4 cell counts were observed and tended to be slightly lower than those observed in noncontrollers.

Although HICs often have viral loads that are undetectable by standard clinical assays, a proportion of elite controllers have measurable viral loads by an ultrasensitive single-copy assay [10–12]. If low-level replicating virus is persistent over long periods, then immunologic stability may be compromised. One study showed that CD4 cell count decline was more common among elite controllers with detectable viremia by single-copy assay [12]. Dinoso et al [10] also found the magnitude of viremia for elite controllers to be comparable to that in patients who experience virologic suppression during receipt of HAART, suggesting that viral load-independent mechanisms may be involved as well.

HICs tended to have lower CD4 cell count gains (by ∼50 cells/μL) after receipt of HAART, compared with noncontrollers; much of this difference was related to HICs having lower viral loads at HAART initiation. Analyses stratified by 3 levels of viral load at HAART initiation showed smaller, non-statistically significant CD4 cell count differences of 14–42 cells/μL between HICs and noncontrollers. These results demonstrate that HICs achieve CD4 cell count benefits that are only slightly lower than those noncontrollers achieve. The small number of HICs in the present study limited precision in comparing CD4 cell count changes.

In this study, 9 and 4 HICs had pretherapy viral loads <400 and <50 copies/mL, respectively. This subset of HICs, who had pretherapy viral loads representing the goal of HAART, experienced considerable CD4 cell count increases after starting HAART (although less than those of noncontrollers). The CD4 cell response is in part influenced by viral load as well as by the level of immune activation at HAART initiation, with a higher viral load allowing for greater CD4 cell count reconstitution after suppression. It is likely that noncontrollers with the same viral load as HICs experience greater immune activation and would therefore expect to have greater CD4 cell count gains. Immune activation has been shown to be greater in untreated HICs, compared with that in patients with virologic suppression due to HAART and in HIV-negative patients [7]. A recent study demonstrated both a decrease in immune activation and a reduction in viral load from 7 to <1 copies/ mL in an elite controller receiving HAART; interestingly, the CD4 cell count did not increase, suggesting that additional factors are involved in CD4 cell count reconstitution [13].

A proportion of HICs experience CD4 cell count declines despite having low or undetectable viral loads. The potential for loss of immunologic stability suggests that not all HICs may be ideal candidates for vaccine development. HAART results in CD4 cell count benefits regardless of pretherapy viral load, including in patients with low or undetectable viral loads before the initiation of therapy. Our results support the current treatment guidelines, which emphasize initiating HAART on the basis of CD4 cell count rather than viral load.

• Received October 6, 2009.
• Accepted December 17, 2009.

• © 2010 by the Infectious Diseases Society of America