Treatment interruption strategy reports from the XIV International AIDS Conference

Mark Dybul MD, for

The meeting saw some shifts in how investigators currently view treatment interruption strategies, particularly those strategies for patients treated during chronic HIV infection.

A ‘debate’ on treatment interruption strategies including those for patients treated during acute and chronic infection as well as those in salvage therapy revealed remarkable consensus that, at this point, most strategies should not be applied broadly in clinical practice pending the results from ongoing studies.

There was also a remarkable lack of data presented on auto-immunization in chronic infection throughout the meeting and in the debate there was general agreement that auto-immunization as a strategy in chronic infection is not a viable clinical strategy. Yet there was much interest in many other treatment interruption approaches and much interesting data. Although the topic is becoming very diverse, with increasing overlap between topics, it remains easiest to divide things along the standard categories of acute, chronic and salvage.

Acute HIV infection

There is increasing evidence for a division of responses to treatment interruptions among patients treated during acute HIV infection, prior to seroconversion, and those treated during recently acquired HIV infection (generally defined as those within 90 days to six months of acute infection).

During the ‘debate,’ Bruce Walker [Abstracts WeOrA197 and ThOrB259] updated his group’s well-know study of 14 patients who were treated with HAART prior to seroconversion (or soon thereafter in two of the 14) for a mean of about a year and have since undergone 1-4 cycles of treatment interruptions of variable duration with resumption of HAART when the viral load exceeded 5,000 copies/mL three times or 50,000 copies/mL once.

Ten of 14 patients continue to maintain viral loads < 5,000 copies/mL for up to three years or more after one to four interruptions and eight of 10 maintain viral loads in the lowest quartile compared to data of patients in the MACS. This continued positive effect is not due to HLA B27 or B57 (known to be factors in long term non-progressors) and in fact several patients are HLA B35 positive (a known marker of rapid progression). Thus, there continues to be evidence for auto-immunization phenomena as a result of treatment interruptions in the majority of these patients treated with HAART during acute HIV infection, prior to seroconversion.

In contrast, Marty Markowitz [Abstract ThOrB260] presented his data from 16 patients treated during recently acquired infection (mean time from symptoms two months), five of whom received HAART alone and 11 of whom received HAART plus the vaccine candidate ALVAC.

In contrast to Walker’s patients, only two had significant CD4 T cell proliferation responses to p24 Antigen and all patients had a rebound of plasma viremia to relatively high levels within three weeks of treatment interruption that was indistinguishable from patients treated during chronic infection.

Interestingly, all patients had a peak viremia followed by nadir set-point viremia that decreased about 1.5 log10 copies/mL in both groups; this indicates the ability of the immune system to do a pretty good, if not complete, job of controlling HIV. Markowitz also showed an ebbing of virologic control over time with 10 of 12 patients having a viral load > 5,000 copies/mL after 21 months of treatment interruption.

In a similar way, Miro and colleagues [Abstract ThOrB1437] reported on 12 patients who initiated HAART within 90 days of infection and continued for greater than one year. All had viral loads < 20 copies/mL before beginning four cycles of two months off HAART followed by two-four months on HAART. Again, similar to the Markowitz patients, none of the 12 had CD4 T cell proliferation responses to p24 Antigen when HAART was initially started. During sequential interruptions, nought out of 12, four out of 12, four out of 12 and two out of six had viral loads < 3,000 copies/mL at the end of each interruption despite both CD4 and CD8 T cell responses increasing during interruptions.

Thus, there is increasing evidence for a different degree of auto-immunization effect in patients treated during acute versus recently acquired infection. This may have to do with the relatively rapid loss of HIV-specific CD4 responses since Walker’s patients have high responses while Markowitz and Miro saw low or absent responses.

In addition, a study by Vogel and colleagues of SIV [Abstract MoPeA3083] showed that 19 of 21 macaques infected with Mac239 had at least one CTL escape mutation during acute SIV infection. Therefore, it is possible that treatment early, but not during acute infection, may lead to immunologic losses that limit the auto-immunization effect. In the debate session, Walker and Markowitz gave good summaries of the potential pros and cons of early therapy including a potential for auto-immunization for those treated very early and also possible limitation of the evolution of HIV in many quasispecies that may be reflected in the escape mutations. There was general agreement that more studies were needed.

Finally, and very importantly, Walker presented follow-up on a patient in his group’s studies who was doing very well with control of plasma virus to < 5,000 copies/mL for around 1,000 days who had then had a rebound to >50,000 copies/mL. The initial suspicion was viral escape, but after sequencing the virus itself it was discovered that the patient had been super-infected with another virus (that was 12% different than the original virus the patient had been infected with) and had a sexual history that was consistent with super-infection. Thus, a patient who could control one virus very well had difficulty controlling a new virus. These data, combined with those from Vogel, indicate the significant challenges for vaccine, and therefore, therapeutic immunization, strategies.

Chronic infection: auto-immunization

As noted, there were almost no reports of auto-immunization studies in chronic HIV infection. In the debate session Gunthard [Abstract ThOrB262] gave further evidence from the SITT study that auto-immunization is likely to have limited clinical significance in patients treated during chronic infections with the response rate around 15%. In addition, he provided interesting evidence that the proviral DNA levels may help predict responders and non-responders, and that the CD8 CTL response levels do not correlate with the ability to control virus: good evidence for a lack of significant auto-immunization effect.

In addition, Garcia [Abstract ThOrB1438] presented data from 44 patients treated during early chronic infection who underwent treatment interruptions. Comparing the responders (viral load <5,000 copies/mL and >0.5 log10 copies/mL decrease compared to baseline) to non-responders, they found that a combination of low CD4CD38 and naïve cells and a high memory cell count with good proliferation responses to tetanus toxoid and p24 Antigen predicted responders 97% of the time.

There were several presentations that may explain the lack of auto-immunization effect in many patients treated during chronic HIV infection.

Poccia [Abstract ThOrB1442] presented a study of 26 patients who underwent a single treatment interruption of one month followed by re-initiation of therapy and evaluated patients who had delayed versus rapid viral rebounds. Those who had delayed rebounds did not have evidence for significant increases in HIV-specific responses, indicating that the slow rebound was not related to high levels of immune response. In contrast, patients with rapid rebounds of viral load did have significant increases in CD8 T cells producing IFN-gamma. However, when these cells were analyzed, they were mostly CCR7-CD45RA-, an indication of pre-terminally differentiated, or less effective, CTL responses.

In addition, Oliva and colleagues [Abstract ThPeA7135] showed that in these same patients an increase in viral load was associated with a significant decrease in NK cell secretion of IFN-gamma, indicating significant dysfunction, that was restored when HAART was resumed.

Finally, Van Lunzen [Abstract ThOrA1180] showed that, while there is a broadening of CTL responses to HIV in the periphery following treatment interruptions in chronic infection, there was no significant broadening of responses in CTL responses isolated in lymph nodes in most patients evaluated. Thus, in chronic HIV infection, auto-immunization may fail due to poor qualitative CD8 and NK cell responses during treatment interruptions.

Less total time on therapy

At this conference, there seemed to be a real shift from using treatment interruptions in chronic HIV infection for auto-immunization to using these interruptions for the sole purpose of reducing time on therapy. This strategy is employed to try to reduce time on therapy, and therefore diminish the unwanted toxicities and decreases in quality of life in patients. The most clinically relevant of these approaches in resource rich settings may be stopping therapy in patients who started HAART under previous guidelines who would not meet current recommendations for starting therapy.

Gallant [Abstract ThOrB1439] presented his group’s study of 101 patients. Originally, the study included only patients who never had a CD4 count < 200 or an OI. However, in the current cohort, 21 patients had AIDS –defining criteria. 68 of 101 (67%) patients, with a mean CD4+ T cell count of 483 cells/mm3 and a mean viral load of 12,439 copies/mL, have been off ARV therapy for a mean of 62 weeks.

The greatest predictor of remaining off therapy was a CD4 T cell count > 350 cells/mm3 PRIOR to initiating HAART. The viral load prior to therapy was not a significant predictor of resuming therapy. Similarly, Krolewiecki [Abstract ThOrB1440] presented data on 28 patients who started HAART with CD4 > 350 cells/mm3 and had been on ARV therapy for > 6 months. Half of the patients remained on continuous therapy and half discontinued HAART.

At 24 weeks, in the discontinuation group the CD4+ count had dropped from 610 to 544 cells/mm3 but the LDL cholesterol also dropped from 132 to 90 mg/dL; there was no significant change in either marker in the continuous group. Although two patients resumed HAART, one patient in the continuous arm stopped due to toxicity. Thus, relatively long-term interruptions may be safe in patients who began therapy when the CD4 T cell count was > 350 cells/mm3, but patients need to be monitored carefully for declines in CD4 counts.

In contrast, there were numerous reports confirming that treatment interruptions in patients who have controlled viremia but CD4 counts <200 cells/mm3 can lead to significant drops in T cell counts and even clinical events. Finally, Fumaz and colleagues [Abstract ThPp2135] showed a significant increase in quality of life in patients who underwent a single interruption.

In contrast to single interruptions for patients with high CD4+ counts, multiple interruptions may not have a significant benefit and may carry risks of resistance.

Dybul [Abstract ThOrB261] showed data from a study of 24 patients randomized to maintain continuous HAART and 21 who underwent cycles of four weeks off HAART and eight weeks on HAART. While at week 40, after the fourth off HAART period, there was a significant reduction in total and LDL cholesterol and triglycerides in the intermittent compared to the continuous arm, by week 48 (or 8 weeks back on HAART) there were no significant differences between the groups.

In this regard, although it was in 12 patients treated during primary HIV infection, Milinkovic and associates [Abstract TuPeB4532] evaluated lipid levels in patients who underwent four cycles of two months off followed by two-four months on HAART. During the period of continuous therapy for a year after diagnosis, triglyceride and cholesterol levels increased 70% and 46%, respectively. During interruptions, triglyceride levels decreased 23% and cholesterol 12% (which was not significant) but rose to baseline levels during resumption of treatment.

However, short cycle therapy that does not allow for viral rebound still looks good. Dybul updated the NIH pilot study: eight patients who remain on a study of seven days off/ seven days on HAART with Zerit (stavudine) + Epivir (lamivudine) + Crixivan (indinavir)/Norvir (ritonavir) are maintaining suppression of plasma viremia to < 50 copies/mL with infrequent blips of < 500 copies/mL in some patients, for 72-104 weeks.

In addition, the significant reductions in LDL and total cholesterol and triglycerides that had been reported up to 24 weeks have been extended to 53 weeks and up to 72 weeks in those that have been evaluated. Dybul noted that the clinical application might be limited due to difficulties with adherence. In this regard, he reported an eight person pilot study of seven day on/seven day off therapy with a once a day regimen of Videx (didanosine) + Epivir + Sustiva (efavirenz). Out to 24-40 weeks, there has not been a single value > 50 copies/mL.

However, Gunthard showed that in approximately 50 patients evaluated at day seven of the initial interruption in the SITT study, a substantial number had > 50 copies/mL. There was no clear explanation for the discrepancy: Dybul noted in the discussion that they had evaluated approximately 70 patients and had not found any patients with a significant rebound at day seven.

Salvage therapy

Two large, controlled studies have been completed. Unfortunately, they have conflicting results.

Katlama and colleagues [Abstract WePeB5887 and ThOrB263] presented data from the GIGHAART study, which evaluated immediate six-seven drug salvage therapy versus a treatment interruption of eight weeks followed by six-seven drug salvage therapy in 70 patients; the endpoint was viral load changes at weeks 12 and 20 (i.e., after all patients had been on 12 weeks of GIGHAART). The delayed group did substantially better with 26% vs. 62% having > 1 log10 copies/mL reduction in viral load (mean –0.37 vs. –1.91 log10 copies/mL), and 15 % vs. 38% < 400 copies/mL.

However, Clotet [Abstract ThOrB264] presented the Retrogene study of salvage with triple PIs – Norvir + Fortovase (saquinavir) + Kaletra (lopinavir/ritonavir) – on an NRTI backbone in patients randomized to receive a treatment interruption or not. At 48 weeks there was no difference in viral loads between the interruption vs. non-interruption groups.

The differences in the studies may relate to the number of mutations at baseline (more in the Katlama study so interruption maybe more helpful) or the duration of follow-up, but the data are unclear at this point regarding the benefits of interrupting therapy in patients with a highly-resistant virus starting antiretroviral salvage therapy.


All references are to the Program and Abstracts of the XIV International AIDS Conference. July 7-12, 2002. Barcelona, Spain.

Copyright 2002 by HIV and All Rights Reserved


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