Low CD4⁺ T Cell Nadir Is an Independent Predictor of Lower HIV-Specific Immune Responses in Chronically HIV-1–Infected Subjects Receiving Highly Active Antiretroviral Therapy

Highly active antiretroviral therapy (HAART) has significantly improved the prognosis of HIV-1–infected individuals; however, recovery of CD4⁺ T cells is dependent on the timing of initiation and on the duration of treatment. Although dramatic increases in CD4⁺ T cell counts occur shortly after the initiation of HAART [1–3], there is evidence that low CD4⁺ T cell nadirs may be associated with immune defects that are difficult to correct, and subjects with these defects may not recover CD4⁺ T cell counts to the point where prophylaxis for opportunistic infections can be discontinued [4–6]. However, the relationship between CD4⁺ T cell nadir and the magnitude of HIV-1–specific cellular immune responses is not known. In the present study, we evaluated the relationships among the CD4⁺ T cell nadir at the time of initiation of potent antiretroviral therapy, the frequency of expression of activation markers (HLA-DR and CD38) on CD4⁺ and CD8⁺ T cells, and the frequency of HIV-1–specific CD8⁺ T cells that produce interferon (IFN)–γ and lymphoproliferative responses

MethodsA total of 84 HIV-1–infected individuals were evaluated. These subjects were receiving stable treatment with HAART and had had suppression of viremia to an HIV-1 RNA load <50 copies/mL for at least 3 months. Viral loads were determined using the Roche Amplicor Assay (version 1.5). HLA class I typing was performed at the participating AIDS Clinical Trial Group (ACTG) sites in tissue-typing laboratories using polymerase chain reaction with sequence-specific primers. The CD4⁺ T cell nadir was defined as the lowest recorded CD4⁺ T cell value before the initiation of antiretroviral therapy. All individuals provided informed consent before participating in the study. The cellular assays described below were not performed on every subject, because of the limited availability of samples

Lymphocyte proliferation assays were performed using fresh peripheral-blood mononuclear cells (PBMCs) incubated with p24 protein (1 μg/mL; Protein Sciences), IRC p24 antigen (Immune Response), p24 control protein (Protein Sciences), Candida antigen, phytohemagglutinin (PHA), or medium alone [7]. Results are expressed as Δ-counts per minute (Δcpm)—the difference between counts in antigen-stimulated wells and control wells. There was a strong correlation between Δcpm and the stimulation index (SI; r=0.93; P<.001; data not shown). Only samples with Δcpm>1000 after PHA stimulation were included in the analysis [7–9]

We synthesized 66 peptides (15–19 aa long, 10-aa overlap, consensus sequence clade B 2001; Los Alamos HIV Database, available at: http://hiv-web.lanl.gov) spanning HIV Gag protein (MBS 396; Advanced Chemtech). In addition, we tested peptides corresponding to all HLA optimal HIV-1 cytotoxic T lymphocyte epitopes that were listed in the Los Alamos Database (available at: http://www.hiv.lanl.gov/content/hiv-db/REVIEWS/reviews.html) at the time of the study

HIV-1–specific T cell responses were quantified using enzyme-linked immunospot (ELISpot) assay with frozen PBMCs (0.7–1×10⁵ cells/well) [10, 11]. For the evaluation of Gag-specific responses, pools of 5–6 overlapping Gag peptides were placed in each well, and the total response to Gag was determined as the sum of positive pools after background subtraction. Class I peptides were tested individually at a final concentration of 20 μg/mL of each peptide [11]. Results are expressed as the number of spot-forming cells per 10⁶ PBMCs. The number of specific IFN-γ–secreting T cells was calculated by subtracting the negative control value from the peptide-specific spot-forming cell count. Negative controls were always <30 sfc/10⁶ input cells. The positive controls were PBMCs incubated with PHA. Wells were considered to be positive if they demonstrated at least 50 sfc/10⁶ PBMCs after background subtraction and if the number of spot-forming cells was at least 3-fold higher than background. Only samples with a positive response to PHA and a viability >70% were included in the analysis. In addition, 3-color flow cytometry was performed in accordance with standardized ACTG laboratory protocols with anti–HLA-DR–fluorescein isothiocyanate, anti–CD38–phycoerythrin, and either anti–CD4– or anti–CD8–peridinin-chlorophyll-protein complex (all from Pharmingen)

Bivariate analyses were done using the Mann-Whitney U test, and associations between continuous variables and HIV-1 immune-response variables were assessed using Spearman’s&amp;rank correlation coefficient (rs). For the multivariate analysis, we chose the sum of ELISpot responses to HLA optimal class I peptide response and the magnitude of lymphoproliferative responses to HIV-1 p24 and IRC p24 antigens as dependent outcomes. We did not have well-documented pretherapy viral loads for the majority of subjects, so the impact of this variable could not be tested. For each dependent variable, a multiple linear-regression model was used that included the current CD4⁺ T cell value at the time of the immune assays or the CD4⁺ T cell nadir, the number of days receiving antiretroviral therapy, and CD4⁺ or CD8⁺ T cell activation status (HLA-DR and CD38 positive) for the estimation of the adjusted relationship with HIV-1–specific immune responses

Residual diagnostics were performed to ensure the accuracy of the regression models, and dependent variables were transformed if necessary. For the total magnitude of responses against HLA optimal class I peptide responses, a statistically significant nonlinear relationship with CD4⁺ T cell nadir was found; therefore, a restricted cubic spline was used to fit a nonparametric term in the model [12]. A linear β-coefficient was used for the remainder of the variables. For all analyses, a 2-sided significance level of 5% was considered to indicate statistical significance. We used SAS software (version 8.02; SAS Institute) and R software (version 2.0.0; available at: http://www.r-project.org), and graphs were generated using SPSS software (version 13; SPSS)

ResultsWe found CD4⁺ T cell nadirs to be highly associated with CD4⁺ T cell values at the time of the immune assays (rs=0.54; P<.001) (figure 1a). However, despite this correlation in the group as a whole, there was no correlation between current CD4⁺ T cell counts and nadirs in subjects with nadirs ⩽200 cells/mm³ (rs=-0.04; P=.80), whereas the correlation still held in the group with high CD4⁺ T cell nadirs (>200 cells/mm³) (rs=0.47; P=.001)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

a Relationship between current CD4⁺ T cell values and nadir. CD4⁺ T cell values at the time of immune assays correlate with CD4⁺ T cell nadirs in subjects only with CD4⁺ T cell nadirs >200 cells/mm³. b Lower total magnitude of responses directed against HIV-1 peptides in subjects with lower CD4⁺ T cell nadirs. On average, 20 individual peptides were tested per subject, and samples from each subject recognized an average of 4 peptides (range, 0–16 peptides/subject). In subjects with CD4⁺ T cell nadirs >200 cells/mm³, the median response was 661 sfc/10⁶ peripheral-blood mononuclear cells (PBMCs) (interquartile range [IQR], 338–2433 sfc/10⁶ PBMCs), compared with 363 sfc/10⁶ PBMCs (IQR, 50–532 sfc/10⁶ PBMCs) in subjects with low CD4⁺ T cell nadirs (P=.02). c Total magnitude of responses to pooled HIV-1 Gag peptides, plotted as a function of CD4⁺ T cell nadir. The magnitude was lower in subjects with CD4⁺T cell nadirs ⩽200 cells/mm³ (320 sfc/10⁶ PBMCs (IQR, 144–796 sfc/10⁶ PBMCs) vs. 800 sfc/10⁶ PBMCs (IQR, 458–1060 sfc/10⁶ PBMCs) in subjects with CD4⁺T cell nadirs >200 cells/mm³ (P=.10). White triangles CD4⁺T cell nadir ⩽200 cells/mm³; black triangles CD4⁺T cell nadir >200 cells/mm³. CTL, cytotoxic T lymphocyte; IFN, interferon

There was no association between the CD4⁺ T cell counts at the time of the immune assays and the magnitude of IFN-γ ELISpot or lymphoproliferative responses (data not shown). There was a nonlinear association between CD4⁺ T cell nadirs and the magnitude of HIV-1 antigen–specific immune responses with a threshold of ∼200 cells/mm³; therefore, we used this cutoff value for subsequent analyses. This nadir threshold was convenient because it split the groups relatively evenly and is a clinically relevant CD4⁺ T cell value

We evaluated 46 subjects for HIV-1–specific immune responses to peptides representing the optimal CD8⁺ T cell epitopes determined by each subject’s HLA class I A, B, and C alleles. The sums of responses to optimal peptide epitopes in subjects with high CD4⁺ T cell nadirs (>200 cells/mm³ at the initiation of HAART) were significantly higher than those in subjects with nadirs ⩽200 cells/mm³ (P=.02) (figure 1b). We observed a similar trend with overlapping HIV-1 Gag peptides, but limited sample availability prevented testing of all the subjects with this panel of peptides (P=.10) (figure 1c)

We next evaluated the relationship between CD4⁺ T cell nadirs and proliferative responses to HIV-1 p24, a recall antigen (Candida) and a mitogen (PHA). PHA was used as a positive control in these assays, and only 1 subject was excluded from this analysis for having a Δcpm value for PHA of <1000. There was no relationship between CD4⁺ T cell nadirs and the ability to proliferate in response to Candida antigen (figure 2a). Proliferative responses to PHA were the highest of all the stimulants tested, and these responses were significantly higher in subjects with higher CD4⁺ T cell nadirs (for subjects with >200 cells/mm³, median, 6102.1 Δcpm [interquartile range {IQR}, 4231–8161 Δcpm]; for subjects with ⩽200 cells/mm³, 3547.1 Δcpm [IQR, 2344–6394 Δcpm]; P=.02) (figure 2b). Likewise, HIV-1–specific proliferative responses to a commercially available p24 antigen (Protein Sciences) were higher in subjects with higher CD4⁺ T cell nadirs (for subjects with >200 cells/mm³, 481.7 [IQR, 55–1456]; for subjects with ⩽200 cells/mm³, 67 [IQR, 12–663]; P=.05) (figure 2c). We observed similar results when the IRC p24 antigen (purified from Remune vaccine immunogen) was used (for subjects with >200 cells/mm³, 265.0 [IQR, 48–624]; for subjects with ⩽200 cells/mm³, 68.9 [IQR, 5–292]; P=.04) (figure 2d). Results from the 2 separate preparations of p24 were complementary. These data indicate an effect of CD4⁺ T cell nadirs on HIV-1–specific T cell responses in subjects with optimal viral suppression due to antiretroviral treatment

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2

Association between low CD4⁺ T cell nadirs and lower HIV-1–specific proliferative responses. a, Candida; b phytohemagglutinin (PHA); c HIV-1 p24 antigen (Protein Sciences); d HIV-1 IRC p24 protein. White triangles CD4⁺T cell nadir ⩽200 cells/mm³; black triangles CD4⁺T cell nadir >200 cells/mm³

In a multivariate analysis, the CD4⁺ T cell nadir was the most important independent predictor of the magnitude of HLA optimal class I peptide responses (P=.008) and the level of p24-specific proliferation for both antigen preparations studied (Protein Sciences p24, P=.02; IRC p24, P=.02), after we controlled for the duration of antiretroviral treatment and CD4⁺ T cell or CD8⁺ T cell activation status. Duration of HAART was also independently positively associated with the magnitude of IFN ELISpot responses (P=.05) and proliferative responses to both p24 antigens (for Protein Sciences p24, P=.05; for IRC p24, P=.004). We did not find that the CD4⁺ T cell count at the time of the immune assays predicted the magnitude of HIV-1–specific immune responses, and the duration of HAART was not significantly different between groups with low and high CD4⁺ T cell nadirs (data not shown)

There was an inverse association between higher levels of T cell activation and the magnitude of HIV-1 p24–specific proliferation in the overall cohort (CD4⁺ T cell activation, P=.04; CD8⁺ T cell activation, P=.03). However, immune activation was not independently associated with proliferative or CD8⁺ T cell epitope-specific responses

DiscussionWe found strong independent associations between CD4⁺ T cell nadirs and HIV-1–specific immune responses. Previous studies have evaluated the relationship between CD4⁺ T cell nadirs and proliferative responses to recall and HIV-1 antigens [13], as well as the response to vaccination [14], and those researchers concluded that low CD4⁺ nadirs were associated with a lower degree of immune reconstitution. However, functional studies were limited to cellular proliferation in response to protein antigens. We found a strong correlation between CD4⁺ T cell nadirs and CD4⁺ T cell counts during the study. This is consistent with data published elsewhere [5]. However, the CD4⁺ T cell nadir, and not the current CD4⁺ T cell count, was an independent predictor of immune reactivity. These data suggest that subjects with the lowest CD4⁺ T cell nadirs have variable recovery of their CD4⁺ T cell counts and that, even when there is recovery of CD4⁺ T cells, HIV-1–specific immune responses are not completely reconstituted

We found lower frequencies of IFN-γ–secreting cells in response to HLA class I optimal epitopes in subjects with low CD4⁺ T cell nadirs. One explanation for this finding is an inability to maintain CD8⁺ T cell responses when CD4 help is relatively deficient, which is consistent with the lower levels of proliferative responses to 2 different HIV-1 p24 antigens found in subjects with low CD4⁺ T cell nadirs. However, even among subjects with low CD4⁺ T cell nadirs, there were several with significant responses (Δcpm>1000 and corresponding SI>10). Furthermore, there was no clear threshold nadir that predicted a complete absence of HIV-1–specific immune responses. These results suggest that, although a low CD4⁺ T cell nadir may partially predict less-robust helper responses, prolonged HAART may afford reconstitution even in subjects with CD4⁺ T cell nadirs <200 cells/mm³

We also found a positive association between the duration of HAART and the magnitude of immune responses. Although prior longitudinal studies have found decreasing HIV-1–specific immune responses during HAART [15], our cross-sectional data suggest that a group of subjects with a longer duration of complete viral suppression may be “enriched” for individuals with higher magnitudes of HIV-1–specific immune responses. Further studies will be necessary to evaluate the long-term outcome of subjects with persistent helper and CD8⁺ T cell responses during HAART

The evaluation of immune responses to HIV may be important to determine the risk of viral breakthrough in future studies of subjects receiving successful antiretroviral therapy. If the maintenance of virus-specific CD4⁺ and CD8⁺ T cell responses is important for continued viral suppression during HAART, individuals with low CD4⁺ T cell nadirs may be the most appropriate population for HIV immunotherapy trials

• Received December 8, 2005.
• Accepted March 29, 2006.

• © 2006 by the Infectious Diseases Society of America