Sustained control of viremia following therapeutic immunization in chronically HIV-1-infected individuals

Objective: Viral rebounds inevitably follow interruption of antiretroviral treatment in HIV-1-infected individuals. The randomized ANRS 093 aimed at investigating whether a therapeutic immunization was effective in containing the long-term viral replication following discontinuation of antiretroviral drugs in patients.

Methods: Seventy HIV-1-infected patients effectively treated with antiretroviral drugs were randomized to continue treatment alone or in combination with four boosts of ALVAC 1433 and HIV-LIPO-6T vaccines followed by three cycles of subcutaneous interleukin-2. The impact of vaccination on viral replication was assessed by interrupting antiretroviral drugs first at week 40 and thereafter during follow-up until week 100. Antiretroviral drugs were re-initiated according to predefined criteria.

Results: The median cumulative time (days) off treatment was greater in the vaccine group (177) than in the control group (89) (P = 0.01). The proportion of time (mean, SE) without antivirals per-patient was 42.8% (5.1) and 26.5% (4.2) in the vaccine and control groups, respectively (P = 0.005). Viremia (median log₁₀ copies/ml), 4 weeks following the first, second and third treatment interruption was higher in control patients (4.81, 4.44, 4.53) in comparison with vaccinated patients (4.48, 4.00, 3.66) (P = 0.42, 0.015 and 0.024, respectively). HIV-specific CD4 and CD8 T-cell responses elicited by the therapeutic immunization strongly correlated with the reduction of the time of antiviral therapy (P = 0.0027 and 0.016, respectively).

Conclusion: Our findings provide evidence that therapeutic immunization significantly impacts on HIV-1 replication. This translated into a decrease of up to 40% in the duration of exposure to antiretroviral drugs over 15 months of patients' follow-up.

Introduction

Highly active antiretroviral therapy (HAART) allows for a significant and sustained reduction of plasma viremia in HIV-1-infected individuals. HAART has consequently profoundly altered the course of HIV-1 infection by reducing morbidity and mortality ^[1,2]. However, the lack of short-term realistic hopes of viral eradication ^[3-6] with antiretroviral drugs, requires uninterrupted therapy and changes the disease into a chronic infection that requires long-term medical management. A viral rebound is indeed observed in the majority of patients following discontinuation of antiviral drugs ^[7,8]. In addition, HAART is not devoid of adverse effects, including drug toxicities, selection of drug-resistant viruses and high costs, and thereby raising the concerns about its long-term use over decades. The development of additional therapeutic strategies that contribute to the control of viral replication and minimize HAART exposure has thus become critical.

The rationale for therapeutic immunization in HIV-infected patients is based on several lines of evidence indicating that the immune system contributes to the control of HIV-1 replication ^[9-12]. Specifically, cellular immunity has been shown to inhibit HIV-1 replication ^[9,10] and an inverse relationship has been demonstrated between specific antiviral CD4 and CD8 T-cell responses and plasma HIV viral load ^[11,13,14]. In a subgroup of patients, referred to as long- term non-progressors, preserved immune responses are associated with control of HIV replication to exceptionally low levels in the absence of treatment ^{[11,12,15,16]}. In most patients however, although HIV-1 specific CD4 and CD8 T cells may be detected at different time points of the disease, the cells appeared to be functionally impaired and unable to control viral replication following treatment discontinuation ^[17-21]. Therapeutic immunization aims at improving the strength and breadth of immune responses against HIV and to keep the balance in favor of the immune system ^[22]. By inducing and/or maintaining long-term effective immunity, it is hoped that therapeutic vaccines would help to contain viral replication in chronically infected individuals.

In the present study, we examined the long-term impact of a therapeutic immunization strategy on viral replication following treatment interruption in chronically-infected individuals who have successfully been treated with HAART. The first phase of the randomized ANRS 093 study has recently been reported ^[23]. Therapeutic vaccination was shown to markedly increase the percentage of patients who responded to HIV antigens as compared with patients in the non-vaccinated control group ^[23]. A higher proportion of patients in the vaccinated group maintained a viremia below 10 000 HIV RNA copies/ml, in comparison with the control group, 3 months after having interrupted HAART. To assess long-term outcomes, the protocol included an extended follow-up. We now report on the long-term 2-year clinical, virological and immunological outcomes of the study.

Methods

Patients and study design

Seventy patients over 18 years, with asymptomatic HIV-1 infection and CD4 T-cell counts > 350 cells/μl and plasma HIV RNA < 50 copies/ml and who have been previously treated with HAART for at least 1 year were enrolled in fifteen centres in France. Patients who received interleukin (IL)-2 therapy in the past were also eligible to participate to this study. Patients were randomized to continue either HAART alone (n = 37; control group) or combined with ALVAC vCP 1433 and HIV-LIPO-6T (both given by Aventis Pasteur) administered at weeks 0, 4, 8, 12 followed by three cycles of subcutaneous IL-2 (given by Chiron Europe) at weeks 16, 24, 32 (4.5 MIU, twice daily for 5 days) (n = 33; vaccine group) (Fig. 1a). The characteristics of the vaccines have been previously reported ^[23]. Patients with HIV RNA < 50 copies/ml stopped HAART at week 40. Following treatment interruption, HAART was reinitiated if HIV RNA was > 50 000 cp/ml at 4 weeks after interruption or > 10 000 copies/ml at 8 weeks after interruption or thereafter at any protocol follow-up visit. All patients, but one in the vaccinated group, agreed to participate in the long-term follow-up which extended from week 52 to week 100. At each visit, every 2 months, patients who exhibited plasma viremia < 50 RNA copies/mL after HAART resumption were allowed to subsequently discontinue HAART. Following each treatment interruption criteria for resuming HAART were the same throughout the study (see above). For data analysis, we also considered as failures patients who resumed HAART for clinical symptoms or any other reason than virological failure. Safety, laboratory, routine immunological measures and plasma viral loads were performed locally. HIV-1 DNA levels in peripheral blood mononuclear cells (PBMCs) were determined in a central laboratory using a modified version of the Amplicor Monitor test (Roche, Molecular Systems, Branchburg, New Jersey, USA) as previously described ^[24]. Clinical, immunological and virological data were obtained in all on-study patients, at each time point, according to the protocol design.

Fig. 1

The study was approved by the institutional ethics review boards and all patients gave written informed consent.

Specific immunological assessment

HIV-1 specific responses were measured at weeks 0, 16 and 36 as previously reported ^[23]. Lymphoproliferative responses (LPR) were determined from isolated PBMC in a standard [³H]-thymidine uptake assay in quadruplicates with 1 μg/ml of P24 HIV antigen or one of 11 HIV-1 long peptides derived from Nef: Nef 66-97 (N1), Nef 117-145 (N2), Nef 182-205 (N3); from Gag: Gag 183-214 (G1), Gag 253-284 (G2), Gag 17-35 (G3); from Pol: Pol 325-353 (P1), Pol 325-342 (P2), Pol 335-355 (P3) and from Env: Env 303-335 (Env), Env MN 303-335 (EMN) (Neosystem, France). A stimulation index (SI) ≥ 3 and CPM ≥ 3000 was considered as positive response. Interferon-γ Elispot assays were performed from PBMC with 18 pools of 15mer-peptides covering HIV-1 Gag, RT and Nef proteins at 2 μg/ml (NeoSystem). Responses against one pool was considered positive if the mean of spot-forming cells (SFC/10⁶ PBMC) in triplicate wells was higher than the mean of six unstimulated wells (background) plus 100 and was higher than 2 × background.

Statistical analysis

Analysis was performed in an intention-to-treat approach which included all patients randomized who agreed to participate in the follow-up until week 100 (one patient refused).

Primary end point was the number of days without HAART cumulated during subsequent treatment interruptions according to the protocol criteria (see above). The proportion of time without treatment indication was calculated for each patient by dividing the number of days off-treatment indication by the number of days of follow-up between weeks 40 and 100. Secondary endpoints were: plasma HIV RNA rebounds 4 weeks following treatment interruptions, CD4 lymphocyte counts, AIDS-related events and clinical events.

Time without HAART indication was estimated by the Kaplan-Meier method and analysed by the log-rank test with censored times for patients off treatment indication at the end of the follow-up and time set up to 0 for non-interrupters. The proportions of time without treatment and CD4 cell counts were analysed by the Wilcoxon rank sum test. Comparative analysis of viremia (plasma HIV RNA log₁₀ copies/ml) following each interruption was performed for all patients from the two treatment groups who underwent at least one treatment interruption (Wilcoxon rank sum test) and for 11 patients from the control and the vaccine groups who underwent three interruptions (Wilcoxon signed rank test). Frequencies of AIDS-related events were analysed using Fisher's exact test.

The association between baseline or on-study characteristics and the time duration off HAART was analysed by regression analysis. Baseline and on-study characteristics included in the model were: age, gender, prior IL-2 treatment, nadir and baseline CD4-T-cell counts, HIV-specific LPR and IFN-γ Elispot responses and treatment group assigned to patients (therapeutic vaccination or not). Categories of continuous variables were based on groups of equal size. Univariate general linear models were used for statistical comparisons. Two-sided P values less than 0.05 were considered statistically significant. Analyses were completed using SAS version 8.2 (SAS Institute Inc., Cary, North Carolina, USA).

Results

At randomization, between June 2000 and April 2001, patients from both groups were indistinguishable in terms of plasma HIV-1 viral load, CD4 cell count, cellular HIV-1 DNA levels, pre-therapeutic nadir values of CD4 T cells, and duration of antiviral therapy (Table 1). The median follow-up was 102 weeks (range, 99-110).

Table 1

Thirty four patients in the control group and 32 patients in the vaccinated group discontinued treatment at week 40. The second and third treatment interruptions were concordant in the two groups, occurring at weeks 69 and 92 of follow-up. At week 69, 24 patients in the control group and 22 in the vaccinated group interrupted therapy. Eleven patients from each group discontinued treatment at week 92.

We first determined whether therapeutic immunization had an impact on the patients' ability to contain viral replication following treatment interruption, throughout the follow-up period of the study. We therefore calculated the number of days off HAART (i.e.: the number of days with plasma viremia < 10 000 RNA copies/ml) accumulated during subsequent treatment interruptions, in an intent-to-treat analysis (Table 2 and Fig. 1b). The calculated median length of time off treatment was significantly greater in the vaccinated group 177 days; 95% confidence interval (CI), 52-118) than in the control group (89 days; 95% CI, 113-142) (P = 0.010 log-rank test) (Fig. 2a). The longest treatment interruption and the time duration off treatment following the last treatment interruption were significantly longer in the vaccinated patients (69 and 64 days) than in the control group (53 and 47 days) (P = 0.010 and P = 0.006, log-rank test, respectively) (Table 2). At the end of the study, 38% (12 of 32) of vaccinated patients and 19% (seven of 37) of patients in the control group did not meet the criteria for HAART resumption (P = 0.085, chi-squared test) (Fig. 1b).

Table 2

Fig. 2

To further determine the gain in HAART sparing, the cumulated number of days/patient follow-up off HAART was divided by the total number of days/patient follow-up from week 40 (first treatment interruption) to week 100. This ratio was 0.27 (mean; 4247 days off HAART/15959 days of follow-up) for the 37 patients in the control group, whereas it was 0.43 (5871/13780) for the 32 patients in the vaccinated group (P = 0.005, Wilcoxon test) (Fig. 1c and 1d).

To assess the impact of therapeutic immunization on viral rebounds, we took advantage of the design of the study that included a determination of plasma viremia 28 days following each consecutive treatment interruption. In the control group, the median plasma viremia was 4.81 (range: < 1.3-6.04), 4.44 (< 1.7-5.30), and 4.53 (3.65-5.27) log₁₀ cp/ml following the first, second and third treatment interruption, respectively. In vaccinated patients, these values were 4.48 (< 1.3-6.17), 4.00 (1.80-4.96) and 3.66 (3.18-4.58) log₁₀ cp/ml. Viremia following the first treatment interruption did not differ significantly between the control and the vaccinated groups (P = 0.42; Wilcoxon test), whereas differences between viral rebounds following the second and third treatment interruption were highly significant (P = 0.015 and 0.024, respectively) (Fig. 2a). Globally, in the control group, plasma HIV RNA decreased by 0.28 log₁₀ copies/ml between the first and the third treatment interruption, whereas this decrease was of 0.82 log₁₀ cp/ml in vaccinated patients (Fig. 2a).

We next analyzed the dynamics of rebounds in plasma viral load in eleven patients from each group who underwent three treatment interruptions. The eleven patients in the control group did not behave differently from the entire control group in that their viremia increased to 4.92 (< 1.3-6.04), 4.69 (4.08-5.3) and 4.53 (3.65-5.27) log₁₀ copies/ml following the three subsequent treatment interruptions. The median viremia decreased by 0.39 log₁₀ between the first and the third treatment interruption (-1.24 to 3.57) (P = 0.70; Wilcoxon signed rank test). In contrast, in patients from the vaccinated group, viremia increased to 4.98 (4.07-6.18), 3.97 (1.80-4.96) and 3.66 (3.18-4.58) log₁₀ copies/ml after the three subsequent treatment interruptions. The median viremia between the first and the third treatment interruption decreased by 1.32 log₁₀ copies/ml (2.39 to 0.26) (P = 0.001; Wilcoxon signed rank test) (Fig. 2b).

Analysis of changes in CD4 T-cell counts following successive treatment interruptions between weeks 40 and 100 showed that the nadir of CD4 cell count reached 400 cells/μl (range, 228-1016) and 510 cells/μl (range, 315-819) for patients in the control and in the vaccinated groups, respectively (Table 2) (P = 0.14). The latter values represent a median loss of -179 and -53 CD4 T cells in comparison with baseline in the control and the vaccinated groups, respectively (P = 0.11) (Table 2). Interestingly, seven patients in the control group and none in the vaccinated group reached a CD4 T-cell count below 300 cells/μl during the treatment interruption period (P = 0.013; Fisher test).

Although the study was not powered to detect differences between groups in terms of clinical outcome, all clinical events related to the vaccination or occurring during treatment interruptions were rigorously collected. Vaccination was well tolerated throughout the follow-up of the study. However, during treatment interruptions, five patients in the control group developed class B or C AIDS-defining clinical events: two cases of thrombocytopenia, one case of Kaposi sarcoma and two cases of Zoster infection. There was no event recorded in the vaccinated group (P = 0.05; Fisher's exact test).

We next assessed the relationship between baseline and on-study virological and immunological characteristics of patients and the proportion of time off treatment by means of analysis of variance and correlation analysis (Table 3). Neither CD4 T-cell counts at baseline nor ex vivo HIV-specific CD4 or CD8 T-cell responses were predictive of the length of time off treatment (Table 3). We observed however, a significant correlation between the treatment group assigned to patients and the mean length of time off treatment (P = 0.016) (Table 3). At week 16, after the four boosts of vaccines, 19 of 33 (58%) vaccinated patients exhibited a multiepitopic HIV-1-specific CD4 LPR as compared with nine of 36 (25%) in controls (P = 0.006). These responses were positively correlated with time off treatment (P = 0.0027) (Table 3). A post-hoc analysis showed that the breadth of HIV-specific CD8 T-cell responses tended to increase at week 36 in the vaccinated group in which seven of 31 (23%) of vaccinees recognized at least three pools of HIV peptides as compared with two of 37 (5%) of control patients (P = 0.066). We further found an association between the breadth of HIV-specific CD8 immune responses and time off treatment (P = 0.037) (Table 3). Finally, in the multivariate model, the comparison between treatment groups was no longer significant when HIV-specific LPR (to at least 1 HIV-1 peptide) responses were included in the model (P = 0.14), demonstrating that the effect of the vaccine regimen on viremia was correlated with elicited immunological responses. Taken together, these results strongly suggest that the improved outcome in patients who received the vaccination strategy resulted from the elicited HIV-specific immune responses.

Table 3

Discussion

In the present randomized study we demonstrate that therapeutic immunization results in a sustained control of viremia following HAART discontinuation in chronically infected patients. The successful containment of viremia translated into a prolonged time off antiviral therapy over 15 months of patients' follow-up. Assuming that the theoretical maximal time off HAART extended from week 40 to week 100, namely 420 days, the results indicate that therapeutic immunization allowed the patients to minimize by 42% the time of HAART exposure in comparison with 21% for patients in the control group. This randomized study further made it possible to show that this latter effect correlated with the breadth of elicited specific antiviral CD4 and CD8 T-cell immune responses. Our observations strongly suggest that enhanced viral-specific immune responses associated with a sustained control of viremia may be achieved through therapeutic immunization in chronic HIV-1 infection.

Viral rebounds inevitably follow antiviral treatment interruptions in HIV-infected patients. Using rigorous prospectively defined criteria for the definition of HAART resumption throughout the long-term follow-up of this study, we found that vaccinated patients experienced a significant decrease of plasma viral rebounds following subsequent treatment interruptions. The fact that about one-third of these patients experienced a decrease by 1.32 log₁₀ copies/ml in median viremia between the first and the third treatment interruption is noteworthy. In contrast, in non-vaccinated patients no significant decrease in the viral rebounds following subsequent treatment interruptions was noted. The latter observation is consistent with other studies showing the lack of effect of one, or subsequent structured treatment interruptions, on viral load rebounds in chronically-infected individuals ^[8,25-31].

The first phase of the ANRS 093 study has shown that therapeutic immunization led to a higher percentage of patients with HIV-specific CD4 and CD8 T-cell responses ^[23]. The extended follow-up of patients gave us the opportunity to show that these elicited immune responses were strongly associated with a significant decrease in the duration of exposure to antiretroviral drugs. These data favorably contrast with several previous reports showing a lack of correlation between immune responses and control of viral replication following treatment interruptions in patients treated during chronic or acute infection ^[21,25-36]. Previous studies suggested that the preservation of HIV-specific immune responses in patients treated with antiviral drugs during the primary infection could be associated with a better control of HIV replication following treatment interruption ^[32]. However, in the long term, durable virologic control occurred infrequently in these patients despite the presence of detectable HIV-specific immune responses ^[33]. Other studies, of antiviral treatment interruptions in patients treated during the acute infection either with antiviral drugs alone or combined with therapeutic immunization have shown limited benefits ^[34-36]. However, in the light of new data showing that therapeutic immunization may enhance HIV-specific immune responses in the chronic phase of the infection, there is a need to re-evaluate new and more potent immunogens in this context.

The observation that five out of 37 non-vaccinated patients, but none from the vaccine group, developed a class B or C AIDS-defining events during treatment interruptions indicates that supervised treatment interruptions, not preceded by therapeutic immunization, were not without risks. These clinical events were related to the high levels of viral rebounds in patients as well as to a loss of CD4 T cells during treatment interruptions. Seven out of 37 non-vaccinated patients, and none from the vaccine group, reached CD4 T-cell counts below 300 cells/μl (P = 0.01), the threshold level below which therapy is usually recommended. Although not significant, the nadir of CD4 T-cell counts during treatment interruptions was lower in non-vaccinated patients in comparison with that of vaccinated individuals. In these latter patients, it is plausible that both the containment of HIV replication by therapeutic immunization and IL-2 administration contributed to minimize the loss of CD4 T cells during treatment interruptions.

There are as yet no consensus criteria for assessing the clinical efficacy of therapeutic immunization. The presence of HIV-specific immune responses in patients cannot be considered as a valuable surrogate marker for clinical benefit. The present demonstration that an immune intervention significantly impacts on viral replication may however be considered as a potential surrogate for efficacy ^[37]. Control of viral load would predict a survival advantage and/or a reduction in the time of exposure to antiviral drugs. Given the long-term toxicities of antiretroviral drugs, an overall reduction of the time of HAART exposure could be of subjective and objective clinical benefit. However, the evaluation of the impact of therapeutic immunization on clinical outcome requires further long-term clinical studies. These studies may represent a good opportunity for more in-depth analyses of immune functions such as the role of HLA haplotype or functional analyses based on the cytokine profile and cytotoxic activity, in order to better understand immune correlates of T-cell protection.

Our findings support further efforts to develop therapeutic vaccination strategies in HIV infection as an approach to contain viral replication and minimize exposure to antiviral drugs.

Acknowledgement

Sponsorship: This work was supported by a grant of ANRS. Aventis Pasteur provided ALVAC-HIV (vCP1433) recombinant and HIV LIPO-6T vaccines and a grant; Chiron Europe provided Interleukin-2, and a grant. Neither ANRS nor any pharmaceutical company was involved in design, data collection, analysis of the study, or the writing of the report.

Contributors

Y. Lévy, J. P. Aboulker and J. F. Delfraissy were responsible for the concept and the design of the study. Y. Lévy coordinated the study. Y. Lévy, M. Kazatchkine and C. Durier wrote the report. H. Gahéry-Ségard and J. G. Guillet were responsible for immunological data. C. Rouzioux was responsible for virological data. A. S. Lascaux, C. Goujard and J. P. Cassuto enrolled patients. J. P. Aboulker, V. Meiffrédy and M. Resch were responsible for data collection and validation (coordinating centre). J. P. Aboulker and C. Durier performed the statistical analyses. All authors were responsible for critical revision of the report and approved the final version submitted.