Yet the highly publicized cancellation of the SMART trial has hardly put an end to the STI debate; individuals on both sides can find evidence in that study and others to bolster their positions. Some declare that STI is too dangerous to consider, while others claim it will inevitably occur at some point in a lifelong course of medication. In spite of the SMART trial findings, data from several other STI studies also reported at the Retrovirus conference showed that the effectiveness of treatment interruption, as well as its safety, cannot be so easily dismissed.
|How the Body Reacts to a Treatment Interruption|
Viral load rebound.
Acute retroviral syndrome.
CD4 cell decrease.
However, the first studies of STI did not focus on practical benefits or quality of life; rather, they investigated whether treatment breaks actually improve the body's ability to fight the virus (see "Structured Treatment Interruptions: Protocol or Wishful Thinking," in the Spring 2000 issue of BETA). One theory -- called "autoimmunization" -- held that intermittent treatment could in fact boost the immune system's response to HIV by exposing it to a series of viral rebounds. Since effective HAART successfully suppressed the virus to extremely low levels, the hypothesis went, the immune system never had the chance to mount a response. These early studies took participants off of their medications for fixed periods to allow viral load to increase and then be resuppressed, measuring immune response over time. Despite some small studies indicating that this strategy might prove effective, this avenue of research was all but abandoned after the large Swiss-Spanish Intermittent Treatment Trial yielded negative results.
A second area of interest in STI concerned its possible uses for treatment-experienced individuals who had developed multidrug-resistant HIV. Various studies showed that when treatment was interrupted in people who had developed resistance, their virus often reverted to wild type (i.e., without resistance mutations) in the absence of medication. In a number of these individuals, the wild-type virus could then be suppressed once again with reinitiation of HAART. In the GigHAART study, Christine Katlama, MD, of the Pitié-Salpêtrière Hospital in Paris and colleagues successfully employed this strategy prior to initiating an aggressive salvage therapy regimen. However, during the treatment interruption period, study participants were highly susceptible to opportunistic infections (OIs). Further investigation has shown that resistance mutations are quick to reestablish themselves once HAART has been restarted.
Because the early STI studies were small, they offered provocative but hardly conclusive evidence. By the time the larger studies such as SMART got underway, the theories of autoimmunization and STI as a pre-salvage strategy were already fading. Nevertheless, investigators hoped to learn more about the safety of treatment interruptions and how they might be used to spare HIV positive people the costs -- of all types -- of being on continuous HAART. For that reason, the first reports from these larger trials have been awaited with great anticipation.
Participants were randomly assigned either to maintain a continuous HAART regimen or to stop and start treatment based on their CD4 cell counts. In the treatment interruption arm, participants stopped taking HAART at enrollment until their CD4 counts fell to 250 cells/mm3, at which time they resumed treatment. When their CD4 counts rose to 350 cells/mm3, they again stopped treatment. These participants had a median of three treatment interruptions, with a median length of 18 months per interruption. The hope was that people who received treatment intermittently would maintain good health with fewer side effects and less drug toxicity.
The DSMB halted enrollment in the study in January after a median of 14 months of follow-up, and recommended that all participants in the STI arm shift to continuous treatment. The incidence of serious medical events, HIV disease progression, and death were found to be more than twice as high in those undergoing intermittent HIV treatment, with 117 cases (3.7%) compared with 47 cases (1.5%) in the continuous treatment arm. Because the study was large, this difference was highly statistically significant. The DSMB concluded that the risk to participants was serious enough to recommend stopping the study.
Overall, participants in the STI arm spent 33% of the time on treatment, compared with 93% in the continuous treatment arm; however, they spent over three times longer at CD4 levels below 350 cells/mm3. It is not entirely surprising that this group experienced a higher incidence of AIDS-related conditions, such as candidiasis. But one truly unexpected result was the increased number of non-AIDS-related medical complications -- such as cardiovascular, liver, and kidney problems -- in the interruption arm: 59 episodes (2.1%) in the STI arm compared with 37 (1.4%) in the continuous treatment arm. These kinds of complications are typically associated with HAART, and it was expected that participants in the continuous treatment arm would experience them more frequently.
Principal Investigator Wafaa El-Sadr, MD, MPH, Chief of the Infectious Diseases Division at Harlem Hospital, suggested that increased levels of inflammation, caused by higher viral loads in the treatment interruption group, might be responsible for the increased incidence of severe complications, such as cardiovascular disease. However, this is very preliminary speculation; the cause is unknown and still mysterious.
The overall risk of disease progression did not differ based on gender, race, CD4 cell count at enrollment, or prior AIDS diagnosis. Another surprise was that it was also unaffected by nadir CD4 cell count. This was unexpected, since previous STI studies have shown that patients with low nadir CD4 cell counts are at greater risk for disease progression.
One factor that was linked to an increased risk of disease progression was HIV viral load at enrollment. At the start of the study, 70% of participants had viral loads less than 400 copies/mL. Among these individuals, the CD4-guided treatment interruption arm had a risk of disease progression nearly four times greater than that of the continuous therapy group. In contrast, among participants entering the trial with viral loads above 400 copies/mL, there was no difference in disease progression between the two study arms.
Although recruitment into the SMART study has stopped and those in the treatment interruption group are being urged to switch to continuous therapy, the investigators will continue to follow study participants and monitor the effects of treatment interruption over the long term. A substudy of SMART will also analyze quality of life as perceived by participants in the trial; these results may help elucidate the larger life context and benefits for those attempting treatment interruption.
The study, sponsored by the French National Agency for AIDS Research (ANRS), recruited 840 participants in the Ivory Coast capital of Abidjan. In this trial, 70% were women and 35% had advanced HIV disease. All participants were treatment-naive at enrollment, with CD4 cell counts between 150 and 350 cells/mm3. Participants were initially started on HAART for at least six months, until CD4 count rose above 350 cells/mm3 and viral load fell to less than 300 copies/mL. Individuals were then randomly assigned to one of the three arms: 1) continuous treatment with antiretrovirals; 2) CD4-guided intermittent treatment, stopped at 350 cells/mm3 and resumed at 250 cells/mm3; and 3) fixed cycles of treatment interruption, with two months on followed by four months off.
Participants were followed for 96 weeks. After 19 months of follow-up, those in the CD4-guided interruption arm were found to be more than twice as likely to experience serious illness -- including invasive bacterial infections, candidiasis, and tuberculosis -- as those on continuous HAART. This STI arm also had twice the death rate. The Trivacan study's DSMB recommended halting the CD4-guided treatment interruption arm; the fixed-length treatment interruption and the continuous treatment arms are ongoing.
Median CD4 counts at enrollment were 470 cells/mm3 for the STI arm and 507 cells/mm3 for those on continuous HAART -- much higher than in the Trivacan study. Another important distinction between Staccato and the SMART and Trivacan trials concerns the CD4 cell level at which those undergoing treatment interruption resumed therapy. In both SMART and Trivacan, participants restarted HAART when their CD4 count fell to 250 cells/mm3; however, in the Staccato trial, participants restarted at 350 cells/mm3. The results were strikingly different. In this study, patients in the episodic therapy arm managed to take drugs for 62% less total time than those in the continuous therapy arm, while avoiding the serious illnesses encountered in the other trials. Though AIDS-related illnesses, including vaginal and oral candidiasis and thrombocytopenia, were more frequent in the treatment interruption arm, no AIDS-defining illnesses were observed in either group. Nor was there evidence of emerging drug resistance mutations in the STI arm.
Aside from the criteria for starting and stopping anti-HIV treatment in the STI groups, there are a number of other differences between SMART and Staccato that might explain their discordant results, including the makeup of the study populations, prior treatment experience, extent of disease progression before the study, and drug regimens used by participants.
After stopping treatment, participants' CD4 cell counts declined following a pattern seen consistently in other studies: an initial rapid decline (of 198 cells/mm3 on average) during the first eight weeks, followed by a much more gradual decline (a mean of 1.74 cells/mm3 per week) over the ensuing months. Treatment interruption was found to be relatively safe in this group, with a few AIDS-related and non-AIDS-related events occurring after treatment was stopped (including thrombocytopenia, herpes zoster, bacterial infection, and hypertension). Five of the study participants died, but none of the deaths were directly attributed to HIV/AIDS. However, at least three of these five deaths involved cardiovascular disease; in light of the SMART trial findings, it is not entirely certain that these deaths were unrelated to stopping treatment. In contrast to the SMART study, in ACTG 5170 having a CD4 nadir below 400 cells/mm3 was highly predictive of faster CD4 decline or clinical events following the interruption of treatment.
The study recruited 403 participants, 80% of whom were men, who had CD4 counts greater than 450 cells/mm3 and HIV viral loads below 200 copies/mL for six months prior to enrollment. The median CD4 count was 745 cells/mm3 and the median CD4 nadir was 280 cells/mm3. Study participants were randomly assigned to two groups: one receiving continuous therapy for 96 weeks, the second receiving intermittent therapy for eight weeks, followed by eight weeks of treatment interruption.
Participants in the STI arm experienced a median CD4 decline of 155 cells/mm3; 3.5% of them fell to CD4 counts below 300 cells/mm3, compared with 1.5% in the continuous treatment arm. Yet there were no AIDS-defining illnesses in either group, and both groups experienced roughly equivalent -- and curiously high -- rates of emergent drug resistance (17% of STI participants, 14% of those on continuous HAART). Here again, more instances of AIDS related conditions -- thrombocytopenia, acute retroviral syndrome, and candidiasis -- were seen in the STI group, as was the case in other studies. Overall, however, a treatment interruption of 96 weeks was shown to be clinically safe in this trial, while reducing drug exposure by nearly 50%.
Unsurprisingly, a higher CD4 cell count at the beginning of the study was predictive of slower disease progression during treatment interruptions. However, unlike most of the other trials presented at the conference, the ISS-PART study found a very high rate of drug-resistance mutations in the STI arm. Of the 136 individuals undergoing STI, 33 (30%) developed such mutations, and they were more than twice as likely as participants who did not develop resistance to experience virologic failure when they resumed therapy. Use of an unboosted protease inhibitor (PI) was strongly associated with the emergence of resistance.
In March, shortly after the Retrovirus conference, the treatment interruption arm of DART was terminated (although the main aspect of the study on monitoring HIV disease progression will continue), and all participants were advised to switch to continuous treatment. This occurred for much the same reason as the early termination of SMART; the study's DSMB found in its review that participants taking treatment breaks were much more likely to experience HIV related illnesses than those on continuous treatment.
A total of 799 HIV positive individuals were recruited into the treatment interruption substudy. At enrollment, participants were treatment-naive and had CD4 counts less than 200 cells/mm3. All began antiretroviral therapy, and after 12 to 18 weeks achieved CD4 cell counts above 300 cells/mm3. At that time, they were randomly assigned either to stay on continuous antiretroviral therapy or to begin cycled STI.
Participants in the STI arm experienced four times as many instances of HIV-related disease -- candidiasis being the most common -- compared with those in the continuous therapy arm (25 vs. 6 subjects). There was one death in each of the two arms. Further analysis of this study will be presented at the International AIDS Conference this August in Toronto.
Risks of Treatment Interruptions
Benefits of Treatment Interruptions
The primary difference between the cancelled STI trials and the CD4-guided Staccato trial, as mentioned above, is that participants in the cancelled trials did not resume HAART until their CD4 levels fell to 250 cells/mm3, while the Staccato trial participants restarted antiretroviral treatment at 350 cells/mm3. It is possible that individuals living for longer periods with lower CD4 counts are more susceptible to infections and illnesses. However, in her remarks about the SMART study, El-Sadr emphasized that it is still premature to conclude that STI is safer at the higher threshold.
Another notable factor is that participants in CD4-guided STI studies are typically off treatment for much longer periods than those in fixed length studies. For example, in the ACTG 5170 trial, participants went off HAART for a median of 96 weeks, while in the fixed-length trials the treatment breaks ranged from one to three months. In fact, the trial of this type that had the longest treatment hiatus -- the DART trial, which cycled in periods of three months on and three months off -- also showed the greatest number of adverse events.
Therefore, it was very surprising to find in the SMART trial that CD4 nadir did not predict HIV disease progression. Because the study was so large, these results certainly open the door for new hypotheses about the long-term effects of low CD4 count, even if they do not completely overturn current beliefs.
The shocker came in the SMART trial findings. To see elevated rates of clinical events such as cardiovascular, kidney, and liver problems in the STI group -- when it has generally been agreed that these are side effects associated with HAART -- turns common wisdom on its head. No one really knows yet what this means, and a number of investigators may now take a new look at their data. For example, the deaths attributed to non-AIDS related cardiovascular disease in the ACTG 5170 trial could turn out to be something more disturbing. Unfortunately, because this finding is so recent, it will be some time before more is known.
At enrollment in the PART study, participants had been on antiretroviral therapy for an average of two years; 70% were using a non-nucleoside reverse transcriptase inhibitor (NNRTI)-based regimen, and the rest were on unboosted PI-based HAART. Due to the long half-life of NNRTIs, efavirenz (Sustiva) was stopped six days earlier than the other drugs, and nevirapine (Viramune) was stopped three days earlier. It is unclear why 30% of participants developed resistance mutations, though the emergence of mutations was associated with the use of an unboosted PI; 50% of participants on unboosted PIs developed mutations, compared with 20% of those on NNRTIs.
In contrast, few resistance mutations were recorded in the Staccato trial, in which 80% of the participants were on a boosted-PI regimen: ritonavir (Norvir)-boosted saquinavir (Invirase) along with two nucleoside reverse transcriptase inhibitors (NRTIs). Boosted PI-based HIV treatment shows a high barrier for resistance.
An important consideration for anyone interested in a treatment break is the half-life of HIV medications. It takes a variable amount of time to metabolize and clear different antiretrovirals from the body. If all medications in a combination regimen are stopped abruptly, the levels of some may diminish more quickly than others, opening the door for the development of resistance to those drugs that linger longer at low amounts. The NNRTIs efavirenz and nevirapine and the NRTI tenofovir (Viread) can take up to a week after treatment is stopped for levels to decline below the minimum inhibitory concentration. Therefore, when contemplating a treatment interruption, people should work with their doctors to create an appropriate schedule for discontinuing the specific drugs in their HAART regimen.
The most worrisome issue raised by SMART -- the increase in cardiovascular, liver, and kidney problems in those taking a treatment break -- has opened up a new area of investigation. The cause could turn out to be, as El-Sadr suggested, an increased inflammatory response due to higher HIV viral load, but it is still too early to know. Whatever the reason for these clinical events, it underlines once again that beliefs about HIV disease -- in this case, that these medical problems are only associated with HAART use -- are always open to question.
Few of the researchers investigating STI argue that treatment interruption is preferable to continuous HAART for controlling progression of HIV disease. Rather, over the last decade, thinking on STI has shifted more toward a harm reduction strategy. Even with the risks involved, people will continue to take treatment breaks, and the need to understand the best ways to conduct them will continue to drive further studies. As the DART investigation team stated in the press release announcing the closure of its STI substudy, "the need to interrupt therapy for multiple reasons, e.g., drug toxicity, is likely to remain an integral part of life-long treatment of HIV disease. Finding out how and when therapy can be interrupted safely and which patients may benefit from such strategies are still important questions and should be pursued in future trials."
Michael Sledge is the guest editor of the Winter 2006 issue of BETA.
Ananworanich, J. and others. Failures of 1 week on, 1 week off antiretroviral therapies in a randomized trial. AIDS 17(15):F33-F37. October 17, 2003.
Danel, C. and others. The CD4-guided strategy arm stopped in a randomized structured treatment interruption trial in West-African adults: ANRS 1269 Trivacan trial. 13th CROI. Abstract 105LB.
El-Sadr, W. and others. Episodic CD4-guided use of antiretroviral therapy is inferior to continuous therapy: results of the SMART study. 13th CROI. Abstract 106LB.
Fagard, C. and others. A prospective trial of structured treatment interruptions in human immunodeficiency virus infection. Archives of Internal Medicine 163(10):1220-26. May 26, 2003.
Ghosn, J. and others. No benefit of a structured treatment interruption based on genotypic resistance in heavily pretreated HIV-infected patients. AIDS 19(15):1643-47. October 14, 2005.
Katlama, C. and others. Long-term benefit of treatment interruption in salvage therapy (GIGHAART ANRS 097).10th CROI. Boston. February 10-14, 2003. Abstract 68.
Mackie, N. and others. Clinical implications of stopping nevirapine-based antiretroviral therapy: relative pharmacokinetics and avoidance of drug resistance. HIV Medicine 5(3):180-4. May 2004.
Marchou, B. and others. Structured treatment interruptions in HIV-infected patients with high CD4 cell counts and virologic suppression: results of a prospective, randomized, open-label trial (Window-ANRS 106). 13th CROI. Abstract 104.
Oxenius, A. and others. Stimulation of HIV-specific cellular immunity by structured treatment interruption fails to enhance viral control in chronic HIV infection. Proceedings of the National Academy of Sciences. 99(21):13747-52. October 15, 2002.
Oxenius, A. and others. Human immunodeficiency virus specific CD8+ T-cell responses do not predict viral growth and clearance rates during structured intermittent antiretroviral therapy. Journal of Virology 76(20):10169-76. October 2002.
Palmisano, L. and others. Final results of a randomized, controlled trial of structured treatment interruptions vs. continuous HAART in chronic HIV-infected subjects with persistent suppression of viral replication (ISS-PART). 13th CROI. Abstract 103.
Skiest, D. and others. Predictors of HIV disease progression in patients who stop ART with CD4 cell counts >350 cells/mm3 (ACTG 5170). 13th CROI. Abstract 101.