Partial treatment interruption of protease inhibitors augments HIV-specific immune responses in vertically infected pediatric patients

Background: Although highly active antiretroviral therapy has significantly reduced morbidity and mortality in HIV-infected children, it often fails to completely suppress viral replication, thereby allowing the emergence of drug-resistant variants. Protease inhibitor (PI) based therapy has been hypothesized to depress cell-mediated immune responses by reducing antigen presentation.

Objectives: To determine the effects of partial treatment interruption (PTI) of PI on HIV-specific cellular immune responses in children.

Methods: We conducted a retrospective longitudinal study of HIV-specific cellular immune responses in 13 children who were vertically infected with HIV. All had detectable plasma viremia and had undergone PTI for a median of 1.0 year (range, 0.41-3.35 years) while continuing nucleoside reverse transcriptase inhibitor and non-nucleoside reverse transcriptase inhibitor therapy.

Results: No significant changes in viral load were observed in the immediate time-point before and during PTI (P = 0.84) as well as in the overall period before and during PTI (P = 0.17). CD4 T-cell levels declined slowly immediately before and during PTI (P = 0.07) as well as during the overall PTI period (P = 0.0002), but the rate of CD4 T-cell decline was not significantly increased during PTI. Immediate to PTI, HIV-specific CD4 and CD8 T-cell responses increased by 70% (P < 0.0001) and 92% (P < 0.0001), respectively, and CD4 and CD8 T-cell activation levels (P = 0.6834 and P = 0.6081, respectively) remained unchanged.

Conclusion: HIV-specific cellular immune responses are boosted in children who have interrupted PI-based therapy.

Introduction

Highly active antiretroviral therapy (HAART), consisting of a regimen of a nucleoside reverse transcriptase inhibitor (NRTI) and either a non-nucleoside reverse transcriptase inhibitor (NNRTI) or a protease inhibitor (PI), has contributed to the significant diminution of morbidity and mortality in adults and children with HIV/AIDS ^[1-5]. Although HAART suppresses viral replication and increases CD4 T-cell levels, complete eradication of HIV is not achieved because virus remaining dormant in latent reservoirs and tissue sanctuaries cannot be reached by antiretroviral therapy ^[6-8]. Treatment is therefore sustained indefinitely. While many patients tolerate antiretroviral therapy, others experience drug-related toxicities such as lipodystrophy, dyslipidemia, insulin resistance, hepatoxicity, and hypertension ^[9-13]. These factors, compounded with the complex treatment schedules and dosing regimens associated with antiretroviral therapy, lead to poor adherence ^[14] and predispose patients to the emergence of drug-resistant virus. The adherence problems observed in HIV-infected adults are magnified in children, resulting in a large population of children with extensive drug resistance and limited treatment options ^[15].

Structured treatment interruption (STI) has been advocated for patients with multi-drug-resistant (MDR) virus who are intolerant of combination antiretroviral therapy. STI in these patients often results in the emergence of a drug-susceptible viral population that may theoretically improve responses to subsequent combination antiretroviral therapy. Although the data are inconsistent, most studies have shown that STI in those patients with MDR virus does not result in an improved virologic response to salvage therapy ^[16-18]. Moreover, the emergence of wild-type HIV has been associated with significant immunologic harm, as defined by reduced CD4 T-cell levels and increased immune activation ^[19].

Most recently, focus has been placed on simplified therapy or partial treatment interruption (PTI) for patients with drug-resistant HIV. In a pilot study of adults with MDR virus and detectable viral loads, interruption of the PI component of the regimen was often associated with stable viremia and CD4 T-cell counts, at least through 48 weeks of observation ^[20]. Although the processes acting to maintain stable plasma HIV RNA levels were not fully defined, the persistent partial activity of the nucleoside analogues and the durable maintenance of PI-associated mutations that decrease viral 'fitness' were suggested as two possible mechanisms.

Cell-mediated immune responses, especially cytotoxic T lymphocytes (CTL), play an important role in the defense against HIV. In addition to their antiviral properties, PI may cause immunosuppression by impeding antigen presentation, an integral process in cell-mediated immune responses. Specifically, the PI ritonavir has been shown to modulate proteasomal activity and MHC class I-restricted peptide presentation of several lymphocytic choriomeningitis epitopes in vivo and in vitro ^[21]. Additional studies have also implicated PI in hindering antigen presentation and adversely affecting β-chemokine levels ^[22,23].

We therefore hypothesized that PTI of PI may affect the magnitude and breadth of HIV-specific cell-mediated immune responses. We performed a longitudinal retrospective study of cryopreserved peripheral blood mononuclear cells (PBMC) from 13 HIV-1 vertically infected children who had undergone PTI. The detailed clinical profiles of these patients are described elsewhere (Abadi et al., unpublished data). We assessed virologic and immunologic parameters before and during therapy interruption, as well as post-therapy interruption for those reinitiating PI-based HAART. In addition, we measured HIV-specific T-cell responses as well as T-cell activation levels at these times.

Methods

Patient samples and viral load measurements

Perinatally HIV-1-infected pediatric subjects were followed and treated at the Jacobi Medical Center (Bronx, New York). These patients had been placed on a simplified treatment regimen by their supervising clinicians as a result of virologic failure, worsening adherence to the complex treatment regimen, difficulty with the palatability of PI, and/or therapy associated complications such as hyperlipidemia and lipodystrophy. Heparinized whole-blood samples (2-5 ml) were collected during scheduled monthly visits after obtaining informed consent, based on local Institutional Review Board-approved protocols. PBMC were isolated by Ficoll-Paque PLUS density gradient centrifugation (Amersham Pharmacia Biotech, Uppsala, Sweden) and cryopreserved. Plasma HIV RNA was measured with Amplicor HIV-1 Monitor (version 1.5) with a lower limit of quantification at 400 copies of RNA/ml (Roche Diagnostic Systems, Branchburg, New Jersey, USA). Absolute levels of CD4 T-cells were determined by flow cytometry with the BD MultiTest CD3/CD4/CD8/CD45 Reagent Kit and analyzed on a FACSCalibur (BD Biosciences, San Jose, California, USA).

Immunophenotyping

Cryopreserved PBMC were thawed, washed, and incubated overnight at 37°C in 5% CO₂. PBMC (1 × 10⁵) were resuspended in 170 μl phosphate-buffered saline (PBS; Media Tech, Herndon, Virginia, USA) and 1% bovine serum albumin (Sigma, St Louis, Missouri, USA) and stained with 20 μl 7AAD (VIAPROBE, BD Biosciences) for 10 min at 4°C in the dark to discriminate living cells from cells that had lost their membrane integrity (late-apoptotic or dead cells). Next, PBMC were stained with each of the following monoclonal antibodies (BD Biosciences): 1 μl CD3-phycoerythrin-Cy7 (PE-Cy7), 2 μl CD4-allophycocyanin (APC), 1 μl CD8-allophycocyanin-Cy5.5 (APC-Cy5.5), 1 μl human leukocyte antigen (HLA)-DR-fluorescein isothiocyanate (FITC), and 20 μl QuantiBRITE CD38-PE. Cells were incubated for 30 min at 4°C, washed, and analyzed on a FACSDiva flow cytometer (BD Biosciences), according to the manufacturer's specifications. Calibrated QuantiBRITE fluorescent beads were used to construct a standard curve for quantification of CD38. FlowJo software (TreeStar, Ashland, Oregon, USA) was used to convert the measured sample mean fluorescence intensity to antibodies bound per cell (ABC).

Cytokine flow cytometry

Cryopreserved PBMC were thawed, washed in RPMI-1640 medium (Media Tech) supplemented with 15% fetal bovine serum (FBS; Gemini, Woodland, California, USA) and incubated overnight at 37°C in 5% CO₂. HIV specific responses were determined by using pools of overlapping HIV-1 clade B 15-mer peptides spanning the Gag (123 peptides), Nef (49 peptides) and Pol (249 peptides) regions. Peptides were obtained from the AIDS Research and Reference Reagent Program (NIAID, NIH). Staphylococcus enterotoxin B (5 μg/ml; Sigma) served as a positive control antigen. Briefly, 2 × 10⁵ PBMC were resuspended in 200 μl RPMI-15% FBS and incubated with each peptide pool (5 μg/ml for each peptide) and 2.5 ng/ml intereukin (IL)-7 and IL-15 (R&D Systems, Minneapolis, Minesota, USA) at 37°C in 5% CO₂ for 2 h. Brefeldin A (10 μg/ml; Sigma) was added, and the cells were incubated for an additional 5.5 h at 37°C in 5% CO₂. Cells were then washed in PBS with 0.02% EDTA and 1% BSA and transferred to a 96-well V-bottom plate, treated with FACS permeabilizing solution (BD Biosciences), and surface stained with the following monoclonal antibodies (BD Biosciences): CD3-perdinin-chlorophyll-A protein (PerCP, 2 μl), CD4-APC (1 μl), tumor necrosis factor (TNF)-α-PE (5 μl) and IFN-γ-FITC (1 μl) for 30 min at 4°C. Finally, cells were washed, fixed with 1% paraformaldehyde, and analyzed on a FACSCalibur flow cytometer. The data were analyzed with CellQuest (BD Biosciences) and FlowJo software. Samples were gated on CD3+CD4+ or CD3+CD4- (CD8+) lymphocytes and analyzed for TNF-α and interferon (IFN)-γ expression. Results were expressed as the percentage of CD3+CD4+ or CD3+CD4- (CD8+) expressing TNF-α and IFN-γ.

Statistical analysis

Statistical analyses were performed with GraphPad Prism (release 4.0, GraphPad Software, San Diego, California, USA). Comparisons of viral load, CD4 T-cell levels, and immune parameters before and during PTI were performed using paired t tests. The rate of decay of CD4 T cells was estimated using linear regression.

Results

We analyzed 13 vertically infected pediatric patients with a median age of 9.9 years (range, 4.0-15.1 years), who were placed on a simplified PI-interrupted regimen for a median of 1.0 year (range, 0.41-3.35 years; Table 1). Nine subjects were male and four were female. Seven patients were of Hispanic and six of non-Hispanic black origin. All patients had detectable plasma viremia (median, 4.66 log₁₀ RNA copies/ml; range, 2.92-5.22 log₁₀ RNA copies/ml). Genotypic analysis before PTI indicated extensive NRTI and PI resistance and revealed that these mutations persist during PTI (Table 2). During PTI, no child had Centers of Disease Control and Prevention (CDC) defined clinical disease progression. Patients 1 and 4 resumed PI-based therapy after a period of 1.39 and 0.67 years, respectively, due to significant declines in their CD4 T-cell counts, increasing viremia, and therapy-associated complications. Upon reinitiation of therapy, patient 1 received Amprenavir as a substitute for Nevirapine, whereas patient 4 received the same regimen as before PTI (Table 1).

Table 1

Table 2

Maintenance of viral load and a slow decline in CD4 T-cell levels during PI-interruption

Viral load and CD4 T-cell levels were measured longitudinally for all patients (Fig. 1a). To better evaluate the effect of treatment interruption on viral load and CD4 T-cell levels, we compared changes in these parameters in two ways. First, we compared viral loads at the time-point immediately before and during treatment interruption (immediate samples). Second, using a broader analysis, we compared mean CD4 T-cell counts and mean viral loads over the entire sampling period before PI-interruption (median, 83 weeks; range, 63-221 weeks) and during PI interruption (median, 43 weeks; range, 4-157 weeks). Immediately after the initiation of PTI, we observed no significant change in viral load (mean, 3.90 log₁₀ RNA copies/ml before PTI versus 3.93 during PTI, P = 0.84; Fig. 1b, d). In addition, during the overall PTI period, we also detected no significant change in viral load (mean, 3.87 versus 4.00 log₁₀ RNA copies/ml, respectively, P = 0.17; Fig. 2c, e). Of the 13 patients studied, only two had a sustained increase in plasma HIV RNA levels of at least 0.5 log₁₀ copies RNA/ml. Patients 1 and 4, who eventually resumed PI-based therapy, had an increase in plasma HIV RNA levels immediately after initiation of PTI (0.374 and 0.280 log₁₀ RNA copies/ml, respectively). However, both patients responded well to their subsequent PI-containing regimen with markedly lower levels of plasma viremia (4.31 versus 2.62 log₁₀ RNA copies/ml for patient 1, and 4.73 versus 3.15 log₁₀ RNA copies/ml for patient 4).

Fig. 1

Fig. 2

In general, CD4 T-cell levels declined at the time points immediately before and during PTI (mean 658 versus 585 cells/μl, P = 0.07) and during the overall PTI period (mean 681 versus 501 cells/μl, P = 0.0002). However, this should be viewed in the context of the overall slow decline in CD4 T-cell levels in these patients seen both prior and subsequent to PTI (Fig. 1). We assessed CD4 decline by measuring the slope of CD4 T-cell counts before and during PTI in all patients by linear regression (excluding patient 9, for whom there were only 2 points during PTI). The mean slope before PTI was -0.8813 cells per week and -2.226 cells per week during PTI. These slopes were not significantly different (P = 0.347).

T-cell activation levels remain unchanged during PTI

T-cell activation in HIV-infected individuals is measured with the markers human leukocyte antigen (HLA)-DR and CD38. Immediate to PTI, we did not detect a significant change in CD4 and CD8 T-cell activation levels as measured by the absolute number of CD38 molecules (P = 0.6834 and P = 0.6081, respectively) or the percentage of cells expressing both CD38 and HLA-DR (P = 0.4738 and P = 0.2021, respectively; Table 3). Only a significant rise in HLA-DR expression on CD8 T-cells was detected immediate to PTI (P = 0.0431). No general trend could be observed between levels of T-cell activation and interruption of PI-based treatment.

Table 3

Increased CD4 and CD8 T-cell responses during PTI

The strength of CD4 and CD8 T-cell responses to HIV gene products was measured by intracellular cytokine secretion. We utilized pools of 15-mer overlapping peptides to HIV Gag, Nef, and Pol to measure production of IFN-γ and TNF-α in the peripheral blood. Cell-mediated immune responses were measured just before the initiation of PTI and after a mean duration of 29 weeks off PI therapy. Total IFN-γ secretion by CD4 T-cells in response to the three peptide pools increased significantly during PTI as compared to immediately prior to interruption (0.96% versus 1.63%, P < 0.0001). Before PTI, the responses to Gag, Nef and Pol contributed 41.4%, 31.7%, and 26.9%, respectively, to the total response. During PTI however, the Gag response decreased to 13.8% (P < 0.0001) and the Nef response increased to 54.9% (P < 0.0001). The proportion of the overall response directed to Pol was not significantly different (P = 0.14). In the two patients in whom therapy was reinitiated, the overall CD4 T-cell response declined to below the levels seen before PTI (from 1.13% to 0.6% in patient 1, and from 0.79% to 0.66% in patient 4). Although the responses to Gag and Pol declined to around half the initial levels in most cases, the response to Nef increased in patient 4 (0.19% to 0.30%), and declined only slightly in patient 1 (0.28% to 0.26%), suggesting that the increased contribution of anti-Nef responses during PTI continued after reinitiation of therapy.

Total IFN-γ secretion by CD8 T cells in response to the three peptide pools also increased during PTI as compared to immediately prior to interruption (mean, 0.82% versus 1.58%, P < 0.0001). The overall response to individual antigens was marginally altered during PTI, with a slight increase in Gag (27.7% to 33.5%, P = 0.014), a decrease in Pol (38.1% to 29.5%, P = 0.007), and no change in the Nef response (P = 0.25). To investigate a change in the phenotype of virus-specific cells, we calculated the percent of CD8 T cells that produced both IFN-γ and TNF-α, a correlate of cytolytic cells ^[25]. No significant differences in the frequencies of these cells were seen before and during PTI (66.2% versus 72.1%, P = 0.12). In the two patients who reinitiated therapy, total IFN-γ secretion by CD8 T cells declined (1.06% to 0.7% in patient 1, and 0.82% to 0.47% in patient 4). This was accompanied by decreased responses to all antigens.

The changes in the HIV-specific immune responses immediate to PTI did not appear to be driven by increases in viral load. No significant correlation was detected between the fold change in the number of IFN-γ producing CD4 T cells and changes in mean viral loads either before and during PTI (r = -0.35, P = 0.24, Spearman correlation) or on the specific day of immune response measurement (r = 0.088, P = 0.78). The CD8 T-cell IFN-γ response did not correlate with either the mean viral load before and during PTI (r = -0.01, P = 0.97) or the viral load determined on the same day as the immune response measurement (r = 0.06, P = 0.84).

It has been shown that PI exhibit large differences in their ability to inhibit the proteosome. Ritonavir is a potent inhibitor, whereas nelfinavir and indinavir cause minimal inhibition ^[26]. We performed a detailed analysis of each patient's treatment regimen. Subjects were divided into two groups, those treated with ritonavir (n = 5) and those treated with either nelfinavir or indinavir (n = 8). Total CD4 and CD8 IFN-γ, TNF-α, and double positive IFN-γ/TNF-α responses were compared for each group prior to and during PTI. CD8 IFN-γ, TNF-α, and double positive IFN-γ/TNF-α immune responses were augmented in both ritonavir and nelfinavir/indinavir treated groups. However during PTI, CD4 and CD8 TNF-α immune responses were substantially greater (Mann-Whitney test, P = 0.045) in ritonavir as compared to nelfinavir/indinavir-treated patients.

Discussion

PTI of a single therapeutic class for patients with MDR HIV has been hypothesized to maintain viral suppression, prevent the expansion of wild-type HIV, halt the accumulation of PI drug resistance, and reduce total drug exposure and its resulting toxicity and cost ^[24]. Here, we examined the effects of PTI of PI on HIV-specific immune responses in a pediatric population. We performed a longitudinal retrospective study of immune responses in 13 vertically infected children from stored samples. Although our study was limited in sample size, we observed significant immunological and virological benefits in this cohort: a slow decline in CD4 T-cell levels, limited changes in viral load, and augmented HIV-specific CD4 and CD8 T-cell immune responses.

Overall, the patients who interrupted PI remained immunologically and virologically stable. Residual activity of nucleoside analogues and maintenance of mutations within the viral protease that most likely decreased the replicative capacity of the drug-resistant variant may have contributed to this clinical outcome. We also hypothesized that HIV-specific immune responses could be enhanced by the selective removal of PI, a class of drugs thought to variably impair cellular immune responses. We found that HIV-specific CD4 and CD8 T-cell responses increased immediately after interruption of PI and thereafter. These enhanced responses were not driven by increased viral replication and antigen load. Moreover, generalized immune activation (as defined by HLA-DR and CD38 expression) remained stable during the interruption suggesting that the immunosuppressive effect of PI may have been limited to antigen presentation.

Others have shown that a simplified therapeutic protocol, in which didanosine and hydroxyurea were substituted for HAART, enhanced HIV-specific immune responses ^[27]. Additionally, in a study of chronically infected adults who had NNRTI substituted for PI, increased proliferative responses to recall antigens were found ^[28]. In contrast, in a control group of patients continuing on PI-based HAART, no immunological or virological improvement was observed, even after 65 weeks of therapy ^[28,29]. The substitution of NNRTI for PI was suggested to induce an autovaccination effect as well as improve antigen processing and presentation. Support for this hypothesis comes from in vitro studies in which antigen processing and presentation in relation to class I and class II major histocompatibility antigens was shown to be defective in the presence of PI ^[21]. Moreover, PI have been also demonstrated to interfere with lymphocyte cell-cycle progression and proliferative responses ^[22,30].

Until now, the majority of work on the effect of PI on proteosomal processing has been performed in vitro; little data exist on the potential in vivo immunomodulatory role of PI. However, evidence is now accumulating implicating the role of PI, especially at high concentrations, in the in vivo inhibition of proteosomal function. It has been proposed that the hyperlipidemia seen in patients on PI treatment may be caused by PI-induced proteosomal inhibition that in turn impairs apoB degradation ^[31]. In addition, it was recently reported that the PI ritonavir selectively inhibits RANKL induced nuclear factor (NF)κB activation and downstream signaling events in vivo ^[32]. It is notable that the mechanism by which ritonavir impacted NFκB was by preventing the degradation of its major cytosolic-binding protein, IκBα. Ritonavir has also recently been shown to activate the chymotryptic activity of the 26S proteasome which may lead to enhanced antigen presentation in some situations ^[33].

Other non-immunologic factors may also have contributed to the stable viral loads observed in these patients ^[34]. Viruses bearing drug-resistant mutations have been shown to persist despite the absence of specific antiretroviral drug pressure ^[35]. We contend that the continued use of NRTI in these children prevented the emergence of wild-type HIV and allowed for the durable maintenance of linked PI mutations. Theoretically, the persistence of a viral population which can stimulate T-cell responses but cannot deplete HIV-specific CD4 T cells might be critical. The significant improvement in HIV-specific immune responses upon PI interruption in this group may reflect an autovaccination event by a less fit virus with diminished replicative capacity and cytopathicity ^[36,37]. Upon the removal of PI-based therapy, antigen presentation may have been enhanced allowing for amplified immune responses. These enhanced responses then persist because the viral population being targeted is intrinsically less pathogenic. In addition, the lack of changes in immune activation markers also speaks to the reduced cytopathicity of these MDR viral populations, and may directly or indirectly positively impact HIV immune responses by decreasing bystander cellular activation and apoptosis. Moreover, transient viremia at the time of the switch may have also served as an autovaccination event that in turn augmented immune responses. Finally, certain PI have been shown to exert other immunomodulatory effects on host immune function. For example, as indinavir is known to inhibit host proteases necessary for dendritic cell (DC) migration and trafficking ^[38], it may have altered many of the basic immune functions of DC. We noted a significant association between ritonavir interruption and enhanced specific CD8 T-cell production of TNF-α when compared to the interruption of nelfinavir/indinavir. In healthy volunteers, TNF-α cooperates with CD40L to maximize the ability of DC to expand virus-specific CTL responses ^[39]. In contrast, cooperative effects between TNF-α in combination with CD40L were variable in HIV-1 infected patients. We speculate that the interruption of ritonavir may have played a role in expanding CTL responses by enhancing antigen presentation by DC via the upregulation of TNF-α secretion by CD4 and CD8 T cells.

This retrospective single-arm study has some limitations. First, as the study was not randomized, it is difficult to define precisely what effect PTI had on outcomes. A prospective randomized study with clearly defined outcome measures is needed. Second, the immunologic assays used here measured only cytokine production in response to consensus sequences. In future studies, other assays will be needed to fully explore the potential effects of PI on antigen-specific responses.

In conclusion, simplification of therapy by PTI may provide a durable option (although not permanent) in patients who cannot tolerate PI-based HAART. In addition to limiting toxicity and the evolution of additional PI-associated mutations, this strategy may boost HIV-specific CD4 and CD8 T-cell responses and may increase the durability and clinical benefit of simplified combination antiretroviral therapy in vertically infected children.

Acknowledgments

This work was supported by the Elizabeth Glaser Pediatric AIDS Foundation and the NIH (AI060379). D.F.N. is an Elizabeth Glaser Scientist of the Elizabeth Glaser Pediatric AIDS Foundation. M.P.D. is supported by the James S. McDonnell Foundation 21^st Century Science Awards/Studying Complex Systems. F.A.L. is supported by the University of California Office of the President Postdoctoral Fellowship. The AIDS Research and Reference Reagent Program, NIAID, NIH generously donated peptides. We thank Drs. Mette Hazenberg and Becky Schweighardt for helpful discussions.