Therapeutic vaccinations and treatment interruptions

Gareth Hardy, HIV i-Base

In the 14 February issue of AIDS, Yves Levy et al and the ANRS 093 Study Group reported on the long-term impact of a therapeutic vaccination strategy on viral load following treatment interruption in chronic infection. [1] The short-term success of this strategy was reported previously [2] demonstrating the induction of HIV-specific T cell responses in the vaccinated group for whom a higher proportion maintained viral loads below 10,000 copies/ml than the unvaccinated control patients at three months. The present paper describes the findings of two years follow up in these patients.

70 asymptomatic patients with undetectable plasma viral loads (<50 copies/mL) and CD4 counts above 350 cells/mm3 who had received >1 year of HAART were enrolled in the original study. Patients were randomised to continuing HAART alone (n=37), or continued HAART plus the vaccine ALVAC vCP 1433 and HIV-LIPO-6T (both Aventis Pasteur) at weeks 0, 4, 8 and 12 followed by three cycles of IL-2 (Chiron) at weeks 16, 24 and 32 (4.5×106 IU, bid for 5 days) (n=33).

If viral load was below detection at week 40, HAART was stopped. Re-initiation of HAART commenced if viral load went above 50,000 copies/ml at 4 weeks after interruption or above 10,000 copies/ml at 8 weeks or anytime afterwards. During this initial 52-week study standard tritium incorporation lymphocyte proliferation assays were carried out at weeks 0, 16 and 36 using P24 antigen and 11 long HIV peptides from nef, gag, pol and env and IFN-gamma ELIspot assays at the same time points using 18 pools of overlapping peptides covering gag, RT and nef proteins.

The long-term follow up lasted from week 52 to week 100 and included all patients, except one who withdrew consent. Further treatment interruptions were permitted if viral load was again below detection on any 2-monthly visit, following HAART resumption. The criteria for re-initiation of HAART remained the same throughout.

The primary endpoint of this study was the cumulative number of days without HAART during treatment interruptions. Secondary endpoints were percentage of patients with viral load rebound 4 weeks following treatment interruption, CD4 count, AIDS-related events and clinical events. Time off HAART was estimated using the Kaplan-Meier method. The association between time off HAART and age, gender, prior IL-2 use, nadir and baseline CD4 counts, HIV-specific lymphocyte proliferative and IFN-gamma ELISpot responses and treatment group were analysed by regression analysis.

In the control group 34 patients discontinued treatment compared to 32 patients in the vaccinated group at week 40. At the second interruption (week 69), 24 patients interrupted in the control group and 22 in the vaccinated group. At the third interruption (week 92), 11 patients in each group interrupted.

In an intent-to-treat analysis, the cumulative median number of days off HAART (with viral load below 10,000 copies/ml) during subsequent treatment interruptions was significantly greater in the vaccinated group at 177 days, than the control group at 89 days (P=0.010). Likewise, the longest treatment interruption and time duration off treatment following the last treatment interruption were also greater in the vaccinated group (69 and 64 days) compared to the control group (53 and 47 days) (P=0.010 and P=0.006 respectively). The median CD4 nadir during the week 40 to week 100 period was 510 in the vaccinated group (change from baseline -57 cells/mm3 blood) and 400 in the control group (change from baseline -179 cells). The mean proportion of time off HAART was 0.27 in the control group vs 0.43 in the vaccinated group. By the study end 38% of the vaccinated patients and 19% of the control patients did not meet the criteria for resuming HAART (P=0.085).

Viral loads also significantly differed between the two groups. Twenty eight days following each treatment interruption median viral loads in the vaccinated group were 4.48, 4.00 and 3.66 log copies/mL after the first, second and third interruption respectively. In the control group these median viral load values were 4.81, 4.44 and 4.43 log copies/mL respectively. Differences in viral loads between the two groups were highly significant after the second and third treatment interruptions (P=0.015 and P=0.024) but not after the first interruption (P=0.42). In the vaccinated group viral loads decreased by a median of 0.82 log copies/mL between the first and third treatment interruption, but only by 0.28 log copies/mL in the control group.

In a sub-analysis of 22 patients (11 from each of the two groups) who successfully completed three treatment interruptions, the median decrease in viral load between the first and third treatment interruption was 1.32 log copies/mL for the vaccinated group and 0.39 log copies/mL for the control group. Therefore, for the 11 patients in the control group who successfully completed three treatment interruptions, their viral load dynamics did not differ to the other patients in the control group who could not complete three treatment interruptions.

While this study was not powered to determine differences between the two groups in terms of clinical outcomes, it was noted that five patients in the control group experienced class B or C AIDS defining events during treatment interruptions: two cases of thrombocytopenia, one case of Kaposi’s sarcoma and two cases of herpes zoster infection. No events were recorded in the vaccinated group (P=0.05).

Analysis of baseline characteristics that may have been predictive of time off HAART did not find any such relationship for baseline CD4 count or baseline HIV-specific CD4 or CD8 T cell responses. A significant correlation existed between randomisation to the vaccination treatment group and increased time off HAART (P=0.016). Following all three vaccinations at week 16, fifty-eight percent of vaccinated patients demonstrated CD4 lymphocyte proliferative responses to all multiple HIV peptides, compared with 25% in the control group (P=0.006). These responses positively correlated with time off HAART (P=0.0027). Furthermore, the breadth of HIV-specific CD8 T cell responses tended to increase at week 36 for vaccinated patients with 23% responding to at least 3 separate pools, compared to 5% of control patients (P=0.066). The breadth of HIV-specific CD8 T cell responses correlated with time off HAART (P=0.037).


The authors argue that these results strongly suggest that the improved outcome of patients receiving vaccination resulted from the induced HIV-specific T cell responses. However, their data do not prove a direct relationship between induced T cell responses and better controlled viraemia.

On the one hand, no control groups existed for IL-2 showing the effect of IL-2 without vaccination in these patients, or vaccination without IL-2. Secondly, it can not be determined whether improved T cell responses resulted from improved control of virus, or whether improved control of virus resulted from improved T cell responses.

If IL-2 were to augment the effect of treatment interruption, which this study design did not address, then improved T cell responses may simply have been an outcome of a modified virological response to treatment interruption by IL-2. While these results are encouraging, the relationship between induced T cell responses and improved viral control is still only an association and may be circumstantial.

A clear link between induction of T cell responses and subsequent control of virus needs to be determined in order to prove that HIV-specific T cell responses induced by therapeutic vaccines actually have clinical benefits.


  1. Levy Y, Durier C, Lascaux A-S et al. Sustained control of viremia following therapeutic immunization in chronically HIV-1-infected individuals. AIDS. 20(3):405-413, February 14, 2006.
  2. Levy Y, Gahery Segard H, Durier C et al. Immunological and virological efficacy of a therapeutic vaccine strategy combining HIV Lipo-6T and recombinant ALVAC-HIV 1452 vaccines and subcutaneous recombinant interleukin-2 in HIV-1-infected patients: results of a multicentric randomised study. AIDS. 2005; 19:279-286.

Links to other websites are current at date of posting but not maintained.