Ruiz, Lidia^a; Martinez-Picado, Javier^a; Romeu, Joan^b; Paredes, Roger^a; Zayat, Mohd Khalil^a; Marfil, Silvia^a; Negredo, Eugenia^b; Sirera, Guillem^b; Tural, Cristina^b; Clotet, Bonaventura^a,b

Introduction

Highly active antiretroviral therapy (HAART) has proved to be notably effective at reducing HIV-1 replication, improving the clinical course and inducing a continuous increase in CD4 T cells in people infected by HIV-1 ^[1,2]. The sustained recovery of CD4 T-cell counts is thought to be bimodal, with an initial phase attributable to CD4 memory cell redistribution and a second phase in which naive CD4 cell expansion predominates ^[3]. Functional studies have also revealed an increase in T-cell proliferative responses to recall antigens and mitogens, which has been associated with a decreased risk of opportunistic infections ^[3,4]. However, to our knowledge, no recovery of HIV-1-specific T-cells responses has been consistently demonstrated in chronically infected individuals after persistent suppression of viral replication by HAART ^[5].

The possibility of CD4 naive cell reconstitution with HAART and the lack of recovery of HIV-1-specific immunity have provided a rational for pursuing postinfection therapeutic immunization ^[6]. Eventually, HIV-1-specific immunity boost might be achieved through periods of controlled replication of the infecting virus. Antiretroviral treatment interruption in early and advanced infection has been systematically associated with a rapid rebound in plasma viraemia ^[7–10], as well as with significant decreases in CD4 T lymphocytes ^[7,9]. A few anecdotal cases of patients treated early following the acute infection who transiently discontinued their antiretroviral drugs, were able to control the expectable increase in plasma viraemia after a second treatment interruption ^[11,12]. This phenomenon was found to be associated with persistent and vigorous T-helper cell and cytotoxic T lymphocyte precursor responses.

Although preliminary studies have faced treatment interruption in patients who had been on successful first-line regimens for periods of up to 1 year ^[7,9,10], little is known about plasma HIV RNA and CD4 T-cell responses in multiply pre-treated subjects stopping their medication after long term plasma viral suppression below < 20 copies/ml and sustained CD4 : CD8 ratios > 1. A precise understanding of HIV dynamics and immunological behaviour during controlled treatment interruptions is crucial to design studies focused on enhancing immune responses by using intermittent therapy.

This study investigates the safety and the evolution of virological and immunological parameters in a group of 12 chronically HIV-1 infected patients who discontinued their antiretroviral treatment after 2 years of successful virus suppression.

Methods

Study design and subjects

Twelve asymptomatic adults with laboratory evidence of HIV-1 infection were included in the study. All patients must have been on antiretroviral treatment for > 2 years. In addition, all patients must have maintained a CD4 : CD8 ratio > 1 for ≥ 6 months and HIV RNA plasma levels < 20 copies/ml for > 24 months prior to study entry. Patients discontinued their current antiretroviral treatment during a maximum period of 30 days until viral rebound exceeded 3000 copies/ml. All subjects gave written informed consent to take part in the study and the protocol had approval from the hospital's ethical committee.

Sample collection

Plasma viral load was evaluated at day 0 and on 3 days per week thereafter during the interruption period. When patients reinitiated their prior antiretroviral regimen, plasma samples were obtained at day 0, 3 and then once per week for the following 2 months. Flow cytometry and cell proliferation tests were performed on peripheral blood mononuclear cells (PBMC) both at baseline and at the end of interruption period. Additional proliferation assays were determined after treatment resumption when plasma viral load regained undetectable levels. Virus genotype of HIV-1 reverse transcriptase (the first 259 codons) and protease was determined at the end of the structured treatment interruption (STI) period.

Quantification of plasma viraemia

HIV-1 RNA levels were measured with the ultrasensitive reverse transcription (RT)–PCR assay (Amplicor Monitor kit; Roche, Barcelona, Spain) with a threshold sensitivity of 20 copies/ml ^[13]. Results were reported within 48 h of the sample collection.

Direct sequencing of reverse transcriptase and protease coding regions

A 1310-bp fragment encompassing the protease and the first 259 amino acids of the reverse transcriptase gene was generated by RT–PCR. Nested PCR products were sequenced using the dRhodamine terminator cycle DNA sequencing kit (PE Applied Biosystems, Madrid, Spain) and electrophoresis in an automated sequencer (ABI 3100, PE Applied Biosystems).

CCR5 genetic analysis

Total genomic DNA was extracted from PBMC aliquots as described previously ^[14]. A portion of the CCR5 gene was amplified by PCR as describe elsewhere ^[15].

Flow cytometry analyses

The lymphocyte subpopulations were determined on freshly isolated blood at baseline and at the end of the interruption period as described previously ^[16]. The direct immunofluorescence analysis included the following monoclonal antibody panel: X40 [negative control γ1–fluoresceine isothiocyanate (FITC)] X39 [negative control γ2a–phycoerythrin (PE)], negative control polyclonal antibody [IgG2a–peridinin chloropyll protein (PerCP)], SK1 (CD8–PE and CD8–PerCP), SK3 (CD4–FITC and PerCP), SK7 (CD3–PerCP), 2D1 (CD45–FITC), MΦ P9 (CD14–PE), L48 (CD45RA–FITC), SK11 (CD62L–PE), UCHL-1 (CD45R0–PE), HB7 (CD38–PE), L243 (HLA-DR–PerCP) and L293 (CD28–PE) from Becton Dickinson (Madrid, Spain).

Lymphocyte proliferative assay

Heparin-treated venous blood samples were collected in CPT Vacutainer tubes and PBMC were isolated by Ficoll-Hypaque density gradient within 4 h. Lymphocyte proliferation assay was performed as described previously ^[3]. The antigens used in the assay were cytomegalovirus (BioWhitaker, Barcelona, Spain), tetanus toxoid and Candida albicans (Sanofi Diagnostic Pasteur, Madrid, Spain), purified tuberculin (3 μg/ml; Serum-Staat-Institut, Madrid, Spain) and HIV-1 p24 recombinant protein (AutogenBioclear, Madrid, Spain). Positive antigen-specific responses were defined as SI > 5. The SI was calculated as described elsewhere ^[5]. Viability of unstimulated and stimulated cultured cells was quantified by flow-cytometry at day 6 before the addition of [³H]-thymidine.

Data analysis

Linear regression analysis of log-transformed HIV-1 RNA plasma concentrations > 20 copies/ml within the treatment interruption period was used to calculate the slope of viral rebound. RNA values reported as undetectable (< 20 copies/ml) were considered equivalent to 20 copies/ml. The doubling time of plasma viral load was calculated as described elsewhere ^[17]. Quantitative data of lymphocyte subtypes and proliferative SI were compared between groups with a Student t test for paired samples.

Results

Twelve HIV-infected subjects fulfilling the inclusion criteria were included in the study (Table 1). Patients did not experience any adverse events throughout the interruption period and only two subjects showed transient drug-related adverse effects (patient 7, dizziness and patient 12, diarrhoea) when treatment was reinitiated.

Table 1

Kinetics of plasma viraemia

Following therapy interruption, HIV-1 viral burden increased exponentially in 10 out of the 12 patients during the established interruption period whereas two subjects (patients 9 and 12) maintained HIV-1 RNA levels below the limit of detection (< 20 copies/ml) (Fig. 1). In five out of the 10 patients experiencing viral rebound, HIV-1 RNA became detectable between days 10 and 15. In the five remaining individuals, viral load increase did not occur until 18–21 days after the interruption of therapy. Viral increase gradients varied from 0.3 to 0.8 per day (mean 0.45 ± 0.05), yielding a mean half-life (t_1/2) of 1.6 ± 0.15 days (Fig. 1). A two-phase decay in plasma viraemia below detectable levels occurred once treatment was restarted. The t_1/2 of viral decay during the first phase (0–7 days) was 1.37 ± 0.29 days whereas in the second phase the t_1/2 was 6.16 ± 2.63 days (mean, 27 days; range, 12–47 days).

Fig. 1

Genotypic resistance

No drug-resistance mutations were found in plasma from our patients except for subject 8. Zidovudine-resistance mutations (D67N, T215Y/C and K219Q) were detected in this patient, who had received zidovudine monotherapy for 26 months before the initiation of HAART (Table 1). The same pattern of mutations was characterized in PBMC at baseline with the exception that residue 215 was detected only as tyrosine.

Characterization of CCR5 deletion (Δ32)

Four subjects (patients 5, 6, 8 and 9) were heterozygous for the Δ32 deletion (Table 1). However, no correlation was found between longer viral doubling times or longer times to viral detection (> 20 copies/ml) and the presence of CCR5 Δ32 mutation.

CD4 and CD8 lymphocyte subsets

The mean percentage of CD4 and CD8 T lymphocytes (56 ± 2 and 41 ± 27) did not vary at the end of the interruption period with respect to the baseline values (54 ± 2 and 42 ± 2) (Table 1). No changes in the percentage of CD45RA/CD62L and CD45RO on CD4 and CD8 T cells were observed in our patients when antiretroviral treatment was transiently discontinued. Of note, these immunological markers were normally expressed in contrast with HIV-1 seronegative people (data not shown). The level of expression of T-cell activation antigen CD38 on CD8 T cells increased significantly (64 ± 4 to 70 ± 3;P = 0.006. However, the percentage of CD8 T cells expressing HLA-DR did not differ significantly from the mean baseline value (15 ± 2 to 15 ± 1). The percentage of CD8/CD28 T cells decreased significantly compared with baseline values (60 ± 4 to 53 ± 4;P = 0.003).

Lymphocyte proliferative assay

None of our patients showed a specific CD4 T-cell response to Candida or cytomegalovirus when treatment was discontinued. However, four patients improved significantly their proliferative responses to tuberculin (patients 3, 5, 6 and 8) and two their response to tetanus toxoid (patients 6 and 8). Interestingly, patient 8 also showed a consistent increase in HIV-1 p24 T-helper response during the interruption period (the SI was 6 and 7 in two consecutive measurements). Moreover, three additional individuals (patients 3, 11 and 12) improved their proliferative responses to tetanus toxoid and tuberculin when they reached undetectable viral loads once therapy was reinstituted. Among them, patient 3 showed a significant recovery of HIV-1 p24 T-helper cell response (SI, 15).

Discussion

In this pilot study, the controlled interruption of HAART in chronically HIV-infected patients with long-lasting viral suppression was not associated with clinical or immunological deleterious effects. Viral load remained undetectable in two patients after 30 days of treatment interruption. These two individuals did not have differential characteristics with respect to the remaining patients when several parameters were analysed. Moreover, they could not be considered long-term non-progressors because both showed high viral load levels at the time of therapy initiation (Table 1). In addition, none of them showed HIV-specific proliferative responses at study entry. The remaining 10 individuals reached detectable HIV-1 RNA levels (> 20 copies/ml) within a median of 14 days after stopping treatment. This period was slightly longer than those presented in previous studies ^[7,8]. However, once plasma viraemia was detectable, rebound slopes were similar to those reported previously ^[7–9]. Furthermore, in contrast with previous data ^[7], we did not observe flatter viral rebound slopes in patients heterozygous for the Δ32 mutation in CCR5.

According to the design of our controlled treatment interruption, therapy was successfully restarted when viral load levels exceeded 3000 copies/ml. This threshold was set arbitrarily with the aim of safely balancing the deleterious exposure to active viral replication and the eventual degree of immune-activation – potentially helpful to avoid uncontrolled viral rebound. Regardless of the plasma viral burden achieved, all patients returned to undetectable viraemia level within the 30 days following recommencement of the same pre-interruption regimen. The decay rates were comparable to others published previously ^[9,10,17,18] and followed a biphasic dynamic ^[19]. No clinical disturbances or adverse events were seen during treatment interruption. Transient mild adverse effects related to the reinstitution of medication were seen in only two patients.

Genotypic resistance was an important issue to assess in the present study because five patients were drug-experienced before they first initiated HAART. Consistent with prior studies, none of our drug-naive patients selected resistance-conferring mutations ^[7,9]. However, zidovudine-resistant mutations were detected in one of our five drug-experienced individuals. The presence of an identical pattern of mutations in proviral HIV DNA before treatment interruption suggests that such mutations had been selected during earlier non-suppressive monotherapy with zidovudine. Of note, the partial transition from tyrosine to cysteine at residue 215 had been documented in other patients with viral replication in the absence of zidovudine pressure ^[20]. In subject 8, T215C most probably evolved from replication-competent virus existing in long-lived cells as a result of viral replication during treatment interruption. Nevertheless, we cannot totally exclude its emergence due to ongoing viral replication below detectable HIV-1 RNA levels during HAART.

We did not observe significant variations in CD4/CD8 T-cell percentages nor in the respective proportion of memory and naive subsets when therapy was stopped. This suggests that treatment interruption in our cohort was not associated with any immunological impairment. Conversely, other studies have demonstrated increases in plasma viraemia appended by CD4 T lymphocyte declines during transient interruptions of antiretroviral therapy ^[7,9], as well as significant variations in memory and naive T-cell subpopulations ^[7]. Such differences with other studies may be explained by our selection of a group of patients with significantly prolonged HIV-1 suppression, higher CD4 T-cells count and a CD4 : CD8 ratio > 1. In terms of immune activation markers, a significant increase in the proportion of CD8/CD38 cells was detected as a response to plasma viral rebound during the interruption period. This results are in agreement with other previously published data ^[7].

A sustained CD4 HIV-1-specific response has been observed only in patients who receive HAART early following acute infection and in HIV-infected individuals with non-progressive HIV disease ^[21]. Whether a recovery of this HIV-1-specific CD4 T-cell responses is attainable in chronically HIV-infected patients initiating HAART is currently uncertain. On the other hand, it has also been proposed that prolonged HAART reduces HIV-1-specific immune responses because there is insufficient HIV antigen to drive these responses ^[22,23]. In our group of patients, there was an absence of HIV-specific T-helper cell responses to p24 at study entry, which was recovered following therapy discontinuation in two subjects. In one patient this HIV-specific response was detected during the peak of viraemia and in the other when viraemia was controlled after reintroducing antiviral therapy. This HIV-specific response after initial drug-interruption episodes had never been previously described in chronically HIV-infected individuals, although similar levels of response were described after 18 days of a third episode of drug-interruption in a patient who controlled viraemia ^[12]. Eventually, a specific CD4-cell expansion or restoration may have occurred during the interruption period following the brief antigenic stimulation. Furthermore, a maintenance or even an improvement of this response could be detected in our study when viral load was suppressed again below 20 copies/ml after the reinstitution of the prior HAART. Nevertheless, a potent enough and durable response may require more cycles of treatment interruption. Of note, subject 8 was the unique patient who reached plasma viraemia > 5 log₁₀. Thus, the lymphoproliferative responses seen in this individual may suggest that significant increases in viral load of limited duration could be useful for priming the immune system.

In conclusion, the interruption of antiretroviral treatment in this group of patients is not associated with a decline in CD4 T lymphocytes and viral rebounds are effectively controlled with treatment reinstitution in HIV-infected patients with long-lasting viral suppression. Further cycles of intermittent controlled treatment interruption might enhance immune response. Additional immune-based interventions could be required to obtain stronger specific T-helper and cytotoxic responses to control virus replication in patients with established HIV infection. However, the brief duration of the interruption and the reduced number of patients included in our pilot study, argue for the development of further studies to extend and confirm our results before STI may be considered as an alternative therapeutic strategy in the clinical setting.

Acknowledgements

The authors thank the patients who participated in this study. We thank A. Garcia, J. Miranda and J.C. Martínez for technical assistance and M. Pujol, Blood Department and M.A. Fernández, Cytometry Department, Hospital Germans Trias i Pujol for providing human serum from healthy donors.

References