 |
|
|
| |
Planned Treatment Breaks: Long Looks at SMART, DART, and BASTA
|
| |
| |
A report from the XVI International AIDS Conference
Toronto, August 13-18, 2006
Mark Mascolini
Researchers who organized SMART, the treatment interruption trial that closed early and decisively in favor of continuous therapy, miscalculated in picking a study name that now leaves them open to lame wordplay. But they've proved savvy indeed in collecting reams of hot data from this huge trial and analyzing it fast. No fewer than four SMART analyses got showcased at the International AIDS Conference in Toronto, and their cumulative mass left two smaller drug-holiday trials, DART and BASTA, in the shadows. But results of all three trials, and related studies as well, hold lessons.
SMART, all will recall, randomized a mostly treatment-experienced contingent of 5472 people to take their antiretrovirals every day or to suspend therapy when their CD4 count topped 350 and to start again when it fell under 250. SMART analysts in Toronto proffered more than a few key conclusions, but for starters:
- As most anticipated, lower CD4 counts and higher viral loads during follow-up in the drug-break group explain "a large proportion of the observed difference in risk" between the two arms in all three clinical outcomes: a new AIDS diagnosis, death from AIDS, or death from some other cause [1].
- But one highly unexpected conclusion emerged from this analysis: The risk of disease progression proved two times higher in interrupters than in noninterrupters among those with a latest CD4 count above 350 [1].
- Blacks in the interruption group had "particularly inferior outcomes" mainly because they suffered more non-AIDS deaths than nonblacks [2].
- When compared with people on steady therapy, people taking drug breaks reported a worse quality of life by several measures [3]. And that expectation-deflating outcome could not be pinned entirely on the higher rate of HIV disease progression in the interrupters.
- Four variables predicted a steeper CD4 drop when people suspended treatment: a high CD4 count at interruption, a low CD4 nadir (a person's lowest-ever CD4 count), a viral load at or below 400 copies, and a prior AIDS diagnosis [4].
Why SMART's drug-break group got sick faster
Jens Lundgren (University of Copenhagen) tackled the key question on SMART--why did people randomized to intermittent therapy get sicker or die so much faster than people who never put treatment on hold? To find an answer, he weighed the impact of the latest CD4 counts and viral loads measured during the course of the trial to see how those values related to death or development of opportunistic disease [1].
Almost all SMART enrollees, 95%, already had antiretroviral experience when the study began, and nearly three quarters had a viral load under 400 copies when they started the study. Median CD4 count at enrollment measured 598 and median nadir 251.
Compared with 47 people in the nonbreak group who had a clinical setback or died, the 120 people with progression in the drug-holiday group had a lower latest CD4 count (343 versus 540), lower latest CD4% (19% versus 25%), and lower latest viral load (4.4 versus 3.1 log, or about 25,000 versus 1250 copies). These numbers should be interpreted in the context of the lower latest CD4 counts in the treatment interruption group: CD4 counts during follow-up were approximately 200 cells lower in interrupters than in noninterrupters.
A closer look at latest CD4 counts during follow-up disclosed a telling difference between break takers and people always on therapy: In CD4 brackets under 350 rates of opportunistic disease or death did not differ significantly between the two study arms, and progression rates were markedly higher at sub-350 counts than at over-350 counts in both interrupters and noninterrupters. The SMART team did not anticipate this finding when designing the trial, but Lundgren noted that it confirms cohort study results [5,6] showing that the AIDS risk for the latest CD4 count drops slowly until the latest count climbs to 350--so 200 CD4 cells should not be seen as an AIDS threshold value.
Because of SMART's design, people who never suspended their antiretrovirals maintained higher CD4s and lower viral loads than people taking drug holidays throughout follow-up (Table 1). Apparently, reaching a higher CD4 tally on treatment and staying there--rather than having a CD4 count that bounces up and down--means a lot in keeping people healthy.
To put it the other way, the treatment-break group spent 32% of total follow-up with fewer than 350 CD4s while the always-on group spent 7% of total follow-up with CD4s under 350. Because progression rates were higher with latest CD4 counts below 350, the SMART researchers suggested this difference "provides a partial explanation for the higher overall risk" of opportunistic disease or death in the drug-holiday group.
Lundgren's analysis yielded one utterly unexpected finding: The risk of disease progression was twice higher in the treatment-break group than in the steady-therapy group for people with latest CD4 counts above 350. Among these people viral loads averaged 10,000 copies in the drug holiday group and below 400 copies in the other group. Lundgren suggested the "excess risk" of progression in the break takers with relatively preserved CD4 counts is "linked to higher HIV-RNA level resulting in impairment of immune function not reflected in peripheral blood CD4 count."
In a second slide talk on SMART, Wafaa El-Sadr (Harlem Hospital, New York) tried to pinpoint HIV progression predictors by sorting trial participants into subgroups when the study began [2]. Risk of an opportunistic disease or death was higher among break takers than people on steady therapy for all three racial/ethnic groups analyzedŃblacks, Latinos, and whites. But relative progression risk for interrupters versus noninterrupters proved higher for blacks (hazard ratio [HR] 3.7) than for Latinos (HR 2.0) or whites (HR 2.1), a difference that approached statistical significance (P = 0.08).
Compared with Latinos and whites, blacks did not miss more study visits and did not spend more time at lower CD4 counts. Instead, El-Sadr attributed the relatively higher overall progression risk in black break takers to their higher rate of non-AIDS deaths.
Among people taking antiretrovirals when they signed up for SMART, having a prestudy viral load at or below 400 copies conferred a 4 times higher risk of progression in interrupters than in noninterrupters. In contrast, among people starting SMART on treatment and with more than 400 copies, the risk of progression was only 1.2 times higher in interrupters than in noninterrupters. The difference between the relative risk rates for prestudy viral loads above and below 400 copies was highly significant (P < 0.001).
That may sound confusing, but the explanation is straightforward: The progression rate among continuously treated people starting SMART with 400 or fewer HIV RNA copies was much lower than the progression rate in break takers starting with a sub-400 load (0.8 versus 3.2 per 100 person-years). In contrast, among people starting SMART with more than 400 copies, progression rates were closer in the steady-treatment group (2.6 per 100 person-years) and the off-and-on group (3.1 per 100 person-years). In other words progression was rare in steadily treated people with a baseline load below 400 copies and somewhat more common in steadily treated people with a starting load above 400 copies, which is what one would expect.
The same sort of disparity held true when SMART statisticians looked at progression risk in people starting the study in different CD4 brackets: Drug interrupters always had a higher progression risk than noninterrupters. But the risk difference between interrupters and noninterrupters was greater in people who started SMART in higher CD4 brackets--because steadily treated people had such a low progression rate if they started with high CD4 counts or percents (boldface numbers in Table 2).
Together these viral load and CD4 findings mean people doing well by these measures when the study began and then assigned to treatment interruptions had a higher risk of progression or death than people doing well at study entry and then assigned to steady therapy. El-Sadr stressed that treatment interruptions--at least the CD4-guided breaks tested in SMART--cannot be recommended for any subgroup. This exhaustive subgroup analysis unearthed no evidence that any subgroup benefited for CD4-steered drug holidays.
Looking at the whole study group, El-Sadr saw a 3.6 time higher risk of fatal or nonfatal opportunistic disease in the treatment-break group and a 1.8 times higher risk of nonopportunistic death in that group. Neither gender nor nadir CD4 count had any impact of relative risk of progression.
When a person's viral load rebounds during a drug holiday, the surge in viral replication stirs up the immune system. Years ago Janis Giorgi from UCLA established this immune activation as an independent correlate of HIV disease progression [7,8]. After El-Sadr's talk an attendee wondered if intermittent bouts of immune activation may partly explain faster progression in SMART's drug-break group. El-Sadr said statisticians are weighing that possibility.
Life proves less than carefree without drugs
SMART's surprises did not end with a dangerously higher risk of progression--including a still unexplained higher risk of "severe complications" like heart disease--among people stopping and starting antiretrovirals at the behest of CD4 beacons. Results from this well-controlled international study also showed that this type of drug break does not do something many assumed it would--make day-to-day living easier for people relieved of their antiretroviral burden [3]. In fact by several quality-of-life measures, drug holidays made things significantly worse.
This SMART subanalysis asked 1225 US study participants to rate themselves on general health perception, physical functioning, pain, mental health, energy, and social functioning periodically over the course of the trial. After an average of more than 2 years of follow-up, current health state and general health perception dropped in the drug-break group and rose in the steady-therapy group, with the difference in general health perception reaching statistical significance (P = 0.02).
Over the first 12 months of the study, then again over the first 36 months, people who did not suspend therapy gave themselves significantly higher scores on current health, general health perception, physical health, physical functioning, and energy. Among people already taking antiretrovirals when SMART began, the ill effects of episodic treatment on quality of life proved more detrimental in people who a sub-400-copy load at study entry. But similar subgroup analyses saw no impact of CD4 nadir, CD4 at study entry, prior AIDS, race, or gender on between-arm differences in quality-of-life scores. Scores for current health dropped when people shelved their antiretrovirals and rose when they started taking their drugs again.
All quality-of-life indices plunged in the visit just before diagnosis of an opportunistic disease or death--not a surprise. But when statisticians removed people who had an opportunistic disease or died from these quality-of-life analyses, they still found consistently worse scores in the treatment-interruption group. The lower quality of life in the treatment interruption group, they wrote in their poster, was "not substantially explained by the higher rate of opportunistic disease [or] death in that arm."
The SMART team noted that this analysis has its limits. SMART participants tended to take long drug breaks (a median of 17 months), so these findings may not apply to people taking shorter drug holidays or interrupting treatment with a different strategy. And because the quality-of-life substudy involved only US trial participants, findings may differ in non-US groups.
Who risked a faster CD4 drop in SMART?
CD4 counts fell fastest in the first 2 months of a SMART drug holiday, and several factors independently predicted a faster fall--higher CD4 count at study screening and study entry but lower CD4% at entry, lower CD4 nadir, prior AIDS, and a sub-400-copy viral load at study entry [4].
Birgit Grund (University of Minnesota) and SMART colleagues from other sites focused on 2025 people who were taking antiretrovirals when the trial began, got randomized to the treatment-interruption group, and stayed off therapy for at least 2 weeks. Median time off treatment stretched to 15 months, and 567 of these 2025 people (28%) kept their antiretrovirals bottled for 12 months or more.
CD4s dropped most steeply in the first 2 months of the treatment breaks:
- First month median CD4 decline: 126 cells (interquartile range [IQR] 13 to 246)
- First 2 months median CD4 decline: 187 cells (IQR 80 to 317)
- Month 2 to 12 median CD4 decline: 12 cells monthly (IQR 3 to 22 cells monthly)
Grund and colleagues built two multiple regression models to sift out independent predictors of faster CD4 tumbles in the first month of a drug break. The first model factored in CD4 count and CD4% at baseline (when a person entered the study), while the second model reckoned CD4 count and CD4% at screening for the trial (a median of 84 days before baseline). Age, gender, race, highest-ever viral load, and duration of antiretroviral therapy did not significantly affect month-1 CD4 drops in either model. But baseline or screening CD4 count or CD4%, screening CD4 count, nadir CD4 count, baseline viral load above 400 copies, and an earlier AIDS diagnosis did propel CD4 changes in the first month of a drug holiday (Table 3).
Another way to look at the 350 CD4 threshold
SMART results show a clear clinical advantage to maintaining a CD4 count above 350 after starting antiretrovirals. Whether starting treatment with more than 350 CD4s--versus 200 to 350--pays off clinically remains a contentious issue. Six-year data from the carefully monitored Johns Hopkins cohort in Baltimore suggest that beginning antiretrovirals above the 350 mark lowers the rate of AIDS-defining illness while offering the best shot at getting CD4 counts back to normal [9].
The study involved 345 people who started their first regimen with fewer than 200 CD4 cells, 152 who started with a count between 200 and 350, and 158 who started with more than 350 CD4s. Everyone in the study began antiretrovirals with a potent regimen (triple-nucleoside regimens were excluded), had at least 1 year of follow-up after starting, and kept their viral load below 400 copies throughout follow-up. People who started with more than 350 CD4s had the lowest viral loads (median 1846 copies versus 6626 copies in the 200-350 CD4 group and 29,441 copies in the under-200 group) and included a lower proportion of injecting drug users. Median age in the three CD4 strata was 39 or 40 years.
After 6 years of follow-up the average CD4 count in the group that started with more than 350 cells leveled off at 800--within the normal range. But average CD4 counts in the two groups that started with fewer T cells plateaued at around 500. That difference appeared to affect rates of AIDS-defining opportunistic illnesses from the time when people started their antiretrovirals. Whereas 14% in the sub-200-CD4 starting group and 11% in the 200-to-350 group had an AIDS diagnosis during follow-up, only 1.5% in the group starting with more than 350 CD4s wound up with an AIDS illness (P < 0.05 compared with the sub-200-CD4 group).
Age over 45 years and injecting drug use independently correlated with a lower CD4 gain during treatment (-61 CD4s, P = 0.04 for age over 45; -101 CD4s, P = 0.009 for injecting drug use). People who started treatment with fewer CD4 cells gained more than those starting with higher CD4 counts, but not enough to catch up after 6 years of treatment.
Richard Moore and Jeanne Keruly believe their findings suggest that beginning antiretrovirals with more than 350 CD4s "may be warranted to achieve full immune recovery."
Different results with drug breaks in Africa and Italy
Two other randomized treatment interruption trials, DART and BASTA, got detailed at the Toronto meeting. With a 12-week off-and-on design in Uganda and Zimbabwe, DART ended early with significantly faster progression in the interruption arm [10]. Using what may be nearly failsafe CD4 stop-and-start signals, BASTA (which means Enough! in Italian) became the first treatment interruption trial to find more clinical setbacks (called "events" in the sanitized parlance of clinical research) with continuous therapy than with treatment breaks [11].
The ongoing DART trial compares monitoring strategies in 3316 people beginning antiretroviral therapy at two sites in Uganda and one in Zimbabwe. Entering DART with fewer than 200 CD4s, study participants started AZT/3TC plus either tenofovir, abacavir, or nevirapine. After 48 to 72 weeks of treatment, 813 people with CD4 tallies above 300 got randomized to 12-week cycles off and on therapy or to nonstop therapy. Median CD4 count at this randomization stood at 358 (range 300 to 1054). While 77% of people in the interruption substudy took tenofovir or abacavir with AZT/3TC, 23% took nevirapine.
Reviewing results through March 2006, DART's safety panel pulled the plug on the interruption substudy because people taking drug breaks were getting sick faster. After a median follow-up of 51 weeks (range 0 to 85 weeks), people in the drug break group had spent 51% of that time off antiretroviral therapy. Rates of a new or recurrent AIDS diagnosis or death measured 8.2 per 100 person-years in interrupters (31 people) and 3.2 per 100 person-years in noninterrupters (12 people). Multivariate analysis showed a significantly higher progression risk by several measures in the treatment-break arm (Table 4).
Earlier this year researchers in Cote d'Ivoire halted a randomized CD4-guided treatment interruption trial for the same reason--faster progression among people taking drug breaks [12]. The CD4-guided BASTA strategy, unlike any other trial of CD4-pegged antiretroviral breaks, found a higher progression rate in the always-on therapy arm than in the off-and-on group. BASTA also differed from other recently reported CD4-guided interruption trials in two other ways--its small size and the CD4 stop-and-go signals.
Franco Maggiolo (Ospedali Riuniti, Bergamo) recruited 114 people on treatment with a viral load below 50 copies and a CD4 count above 800 to continue therapy or to stop until their CD4s fell to 400. The 2-to-1 randomization assigned 76 people to suspend therapy and 38 to continue. So this study is tremendously smaller than SMART and 75% smaller than Staccato, a just-published trial in which 284 people suspended treatment if their CD4s slipped under 350 and 146 continued therapy [13]. Staccato found statistically similar virologic failure rates in the CD4-guided and steady-therapy groups, and no one had a new AIDS-defining diagnosis during the trial.
Some researchers attending Maggiolo's BASTA presentation suggested this study was too small to yield convincing clinical lessons [see note 14]. Maggiolo counted 4 virologic failures in the interrupters (5.2%) and 1 in the noninterrupters (2.6%), but that difference lay within the bounds of a preset 95% confidence interval and thus, according to this analysis, the two strategies were virologically equivalent. All five virologic failures resulted in emergence of resistance mutations.
In contrast the clinical failure rate was significantly worse in the steady-therapy arm, which endured 7 clinical "events" (18.4%) versus 3 (3.9%) in the drug-break group (P = 0.015). These clinical setbacks were two pneumonias and one aseptic meningitis in the interruption group and two (unexplained) deaths, one cervical cancer, one case of epilepsy, one myocardial infarction, and one herniated disk in the steady-therapy group.
Nevirapine levels after treatment interruption
DART's treatment interruption inquest also yielded evidence on how long nevirapine levels last in blood after people stop the nonnucleoside as part of a triple regimen [15]. Bernard Kikaire from DART's Uganda site tracked nevirapine levels in 19 people who stopped the drug, then stopped their two nucleosides a week later. Everyone had taken a nevirapine regimen for 52 weeks and had a CD4 count above 300; no one had a clinical setback in the past 3 months.
Using an assay with a 100-ng/mL lower limit of detection, Kikaire recorded nevirapine levels of 100 to 200 ng/mL in 6 of 19 people 1 week after nevirapine stopped and levels above 200 ng/mL in 5 of 19. Two weeks after nevirapine stopped only 1 person (5%) had measurable levels. Kikaire figured a "conservative" estimated median half-life of 39 hours for nevirapine (range 19 to 81 hours). The estimated time to reach a level of 200 ng/mL stood below 9 days in all but 3 study participants.
Kikaire interpreted these findings as good news meaning that few people risk resistance to nonnucleosides because of lingering low nevirapine levels if they stop the drug and continue two nucleosides for 1 week. But not everyone who heard Kikaire's slide talk was as happy with the results. One attendee noted that continuing 3TC with another nucleoside for 1 week--as everyone in this study did--posed the flip-side risk of resistance to 3TC. And that may be a bigger problem in Africa than elsewhere because of limited nucleoside options there.
Anton Pozniak (Chelsea and Westminster Hospital, London) noted that a 5% rate of lingering nevirapine levels may not sound bad. But multiplied out over a big population of people trying this tactic, that rate would leave lots of people with nonnucleoside-resistant virus. For example, in the larger DART analysis presented in Toronto (see "Different results from drug breaks" above), 187 people interrupted a nevirapine regimen. Applying the same 5% rate to that group means 9 probably wound up with resistant virus.
Nevirapine remains a popular first-line choice in Africa because it comes in low-priced fixed-dose generic combinations. With an estimated 1.3 million people in low- and middle-income countries taking antiretrovirals [16], hundreds of thousands are probably taking nevirapine.
Resistance rates in other nonnucleoside treatment interruption trials have proved much higher than 5%. In the Italian ISS PART trial, for example, 20% of those interrupting nonnucleoside therapy ended up with resistant virus despite continuing nucleosides after the nonnucleoside stopped [17].
Mark Mascolini writes about HIV infection
References
1. Lundgren JD on behalf of the SMART Study Group. Progression of HIV-related disease or death (POD) in the randomised SMART study: why was the risk of POD greater in the CD4-guided ((re)-initiate ART at CD4 < 250 cells/µL) drug conservation vs the virological suppression arm? XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract WEAB0203.
2. El-Sadr W for the SMART Study Group. Inferior clinical outcomes with episodic CD4-guided antiretroviral therapy aimed at drug conservation in SMART study: consistency of finding in all patient subgroups. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract WEAB0204.
3. Burman W for the SMART Study Group. The effect of episodic CD4-guided antiretroviral therapy on quality of life: results of the quality of life substudy of SMART. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract THPE0145.
4. Grund B for the SMART Study Group. Predictors of initial CD4 decline after antiretroviral treatment interruption in the SMART study. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract THPE0144.
5. Phillips A, CASCADE Collaboration. Short-term risk of AIDS according to current CD4 cell count and viral load in antiretroviral drug-naive individuals and those treated in the monotherapy era. AIDS 2004;18:51-58.
6. Podlekareva D, Mocroft A, Dragsted UB, et al. Factors associated with the development of opportunistic infections in HIV-1-infected adults with high CD4+ cell counts: a EuroSIDA study. J Infect Dis 2006;194:633-641.
7. Giorgi JV, Lyles RH, Matud JL, et al. Predictive value of immunologic and virologic markers after long or short duration of HIV-1 infection. JAIDS 2002;29:346-355.
8. Liu Z, Cumberland WG, Hultin LE, et al. CD8+ T-lymphocyte activation in HIV-1 disease reflects an aspect of pathogenesis distinct from viral burden and immunodeficiency. JAIDS 1998;18:332-340.
9. Moore R, Keruly J. Changes in CD4 cell count out to six years in persons with sustained virologic suppression: implications for when to start ARV therapy. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract THPE0109.
10. Hakim J on behalf of the DART Trial Team. A structured treatment interruption strategy of 2 week cycles on and off ART is clinically inferior to continuous treatment in patients with low CD4 counts before ART: a randomisation within the DART trial. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract THLB0207.
11. Maggiolo F, Ripamonti D, Callegaro A, et al. CD4-guided STI: four-years follow-up of a controlled, prospective trial. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract WEAB0202.
12. Danel C, Moh R, Minga A, et al. CD4-guided structured antiretroviral treatment interruption strategy in HIV-infected adults in west Africa (Trivacan ANRS 1269 trial): a randomised trial. Lancet 2006;367:1981-1989.
13. Ananworanich J, Gayet-Ageron A, Le Braz M, et al. CD4-guided scheduled treatment interruptions compared with continuous therapy for patients infected with HIV-1: results of the Staccato randomised trial. Lancet 2006;368:459-465.
14. See the analysis of Staccato, BASTA, and SMART by David Margolis (University of North Carolina) on this Website (http://www.natap.org/2006/IAS/IAS_51.htm).
15. Kikaire B, Walker S, Khoo S, et al. Plasma levels of nevirapine following interruption of ZDV/3TC/NVP in African adults within the DART trial. XVI International AIDS Conference. August 13-18, 2006. Toronto. Abstract WEAB0201.
16. UNAIDS. Report on the Global AIDS Epidemic: Executive Summary. May 2006.
17. Palmisano L, Giuliano M, Bucciardini R, et al. 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. 13th Conference on Retroviruses and Opportunistic Infections. February 5-8, 2006. Denver. Abstract 103.
|
| |
|
|
|
|
|
