 |
|
|
| |
No structure to this strategy:
reports of Treatment Interruption studies
|
| |
| |
Reported by David Margolis, MD, University of North Carolina.
XVI Intl AIDS Conference, Toronto, Aug 2006
"The most dangerous strategy is to jump a chasm in two leaps"
Benjamin Disraeli (1804 - 1881)
There has been deep and sustained interest across the world in discovering ways of controlling HIV infection without resorting to antiviral therapy. The reasons for this are numerous and obvious, and at its heart originate in the natural and universal human desire for healthy and carefree life, unfettered by the specter of chronic disease. However, there is not yet a defined strategy for the administration of intermittent antiretroviral therapy (ART) that is superior to continuous ART when it is clinically indicated.
Initially, there was hope that an immune system reconstituted by ART would be better able to control viral replication and prevent disease, and that high-level viremia, CD4 cell depletion, and immunodeficiency-related disease would not quickly reappear, if therapy is stopped. So far, even if ART is initiated very early after HIV infection, there is no clear evidence yet that a period of antiviral therapy alters the rate progression of disease once therapy is stopped.
Later, as this hope faded, goals were refocused on maintaining the immune system, as measured by the peripheral CD4+ T cell count, at or above a level that would prevent HIV-related disease. The major justification of intermittent therapy, according to this school of thought, was that intermittent therapy would avoid the toxicities, costs, and burdens of chronic ART. The risks of this approach were acknowledged to be that of the loss of efficacy of therapy upon its re-initiation, primarily due to drug resistance that might occur more frequently with intermittent than with continuous therapy.
Several studies examining treatment interruption (TI, no longer STI) were presented at the International AIDS meeting, and add to those published recently and presented at meetings over the last several years. Overall, the studies taken as a whole show that HIV-related opportunistic infections (OIs) and death (as hard an endpoint as one can find) occur at a higher rate when CD4 counts are lower (roughly below 350 cells/μl). As study subjects spend more time with lower CD4 cell counts while on a TI strategy than on a continuous therapy (CT) strategy, more such morbid events occur using a TI strategy unless the criteria for restarting therapy is set somewhere at or above 350 cells/μl.
Surprising to some, TI strategies also appear to incur some risk of death when CD4 cell counts are higher than 350 cells/μl. This unexpected effect, which contributed to the early closure of the SMART study, is presumably caused in some way by ongoing HIV viremia. The frequency of these deaths is much lower than OIs and deaths seen at lower CD4 counts, and so only very large studies like SMART are able to robustly detect and study this phenomenon.
However, these findings would seem to pose significant challenges to further, randomized study of TI in patients who have reached a CD4/viral load threshold and have initiated ART. The contributors to and predictors of death in patients with high CD4 counts and viremia require further study and definition; hopefully study of a retrospective nature. And clearly, studies performed thus far have defined parameters in which ART may be stopped if side effects are intolerable for patients and alternative, tolerable ART cannot be found.
Perhaps in areas of the world where access to ART is limited, a study of TI would be both practical and ethical, as outside of such a study patients might not have access to ART at all. Further, if safe and successful TI strategies could be found, limited resources could treat a much larger group of patients.
SMART: further analysis confirms suspicions
Two podium presentations at IAS expanded on the findings of the international comparison of TI vs. CT, the Strategies for Management of Antiretroviral Therapy trial (SMART). Closed prematurely due to superiority of the CT arm, preliminary findings were presented at CROI 2006 in February. SMART enrolled patients in 33 countries at 318 sites: 57% in the USA, 26% in Europe, and 10% South America. The mean age of volunteers was 46 years, 27% were women, 30% were black, 15% had hepatitis C. 95% had ART experience, and 71% had a viral load under 400 copies/mL when the study began. Median CD4 count at enrollment was 598/μl, and median nadir (lowest ever) CD4 count was 251/μl. 25% of patients in the study had a CD4 nadir below 150 cells and 25% had an AIDS-defining diagnosis. Before the study was prematurely closed, 5472 of a planned 6000 patients, either ART-experienced or naïve, were randomized to: 1) continuous treatment (CT) or 2) an episodic treatment group (TI), in which treatment would be delayed or discontinued until the CD4 cell count fell below 250, and then ART given until the count rose above 350, followed by another TI.
Expecting that ART would increase cardiac and other non-HIV-related disease risks, the investigators defined a unique compound primary endpoint for SMART: HIV clinical disease progression, death, or a defined group of specific severe adverse events (including strokes, heart attacks, kidney or liver damage). When recruitment was stopped by the safety monitoring board in January, SMART had enrolled 5472 of a planned 6000 patients. The mean CD4 count at entry was 598 cells/ul, but the mean nadir CD4 count was 253 cells/ul. Overall, the patients in the TI arm spent 33% of the time on treatment, compared with 93% in the CT arm. The patients had been followed up for a median of 14 months, during which there had been 164 primary endpoints recorded (HIV disease progression, death, or other serious medical complication). Prior to study closure, and the offering of CT to all patients, TI arm subjects spent only 3.1% of the time with CD4 counts below 200 cells/ul, 8.2% time below 250/ul, and 31.7% below 350/ul. CT arm patients spent < 9% of the time below 350 cells/ul.
The study was closed because there were 117 events in the TI arm compared with 47 events in the CT arm, for rates of 3.7 vs 1.5 per 100 person-years (relative risk 2.5). ALL the endpoints---the primary composite endpoint, death, serious events, non-serious HIV disease progression, serious progression and death---favored the CT arm.
Jens Lundgren (abstract WEAB0203) presented a further analysis of SMART at IAS, examining the relationship between CD4 cell count, viral load and the risk of OI or death from any cause during the study. He presented an analysis of the "proximate" CD4 count or viral load before an OI or death event. The proximate value was one obtained during the study-related clinic visit closest in time to an OI or death. This gives important information about how these clinical values relate to risk of an OI or death event, but it was not clear how many of these proximate values were relatively close to the event, and how many were farther away. During SMART, some patients might not be seen for as long as six months, but most of the patients were seen more frequently.
Nevertheless the median proximate CD4 count prior to an OI/death in 120 patients in the TI arm with these events was 343, whereas the median CD4 count in the TI arm overall was 417 cells/μl. In contrast, the median proximate CD4 count prior to an OI/death event was 540 in only 47 patients in the CT arm with these events, and the median CD4 count in the CT arm overall was 617 cells/μl.
The overall hazard ratio (TI/CT) for OI or death was 2.61----that is OI or death was 2.61 times more frequent in the TI arm. This hazard ratio, decreased (that is the risk of OI/death became more similar) when the ratio was adjusted for CD4, viral load, or both. This is not surprising, but is simply to say that patients with CD4 counts of (for example) 350 had more similar rates of OI/death whether in the TI or CT arm, or patients with viral load of 10,000 had similar rates whether in the TI or CT arm.
Part of the higher frequency of OI/death found in the TI group in SMART is similarly unsurprising: patients in the TI group spent more time at a lower CD4 count than those in CT. Although study protocol suggested reinitiation of ART at CD4 <250/μl, implementation of this was, naturally, imperfect. The median CD4 count at ART resumption was 232/μl, with some patients spending time below 200 (interquartile range 192 to 299/μl).
But a critical observation occurred in TI patients with >350 CD4 cells/μl. Although the risk of OI/death was similar in the two arms across CD4 categories up to a CD4 count of 350 cells/μl, the risk of OI/death was higher in the TI arm in groups of patients with more than 350 CD4 cells/μl (p < 0.05). In these patients with more than 350 CD4 cells, median viral load was 10,000 in the TI group that experienced more OI/death but <400 copies/ml (undetectable for the purposes of this study) in the CT group. This very significant finding from this large and well-powered study suggests that chronic viremia, either directly or via some secondary biological effect, increases the risk of death.
These findings were complemented by a second analysis of SMART data presented by Wafa El-Sadr (Abstract WEAB0204). El-Sadr presented an analysis that sought to compare the hazard ratio of the various primary outcomes of the SMART study between the TI and CT arms of the study in different subgroups of patients enrolled.
Overall, the hazard ratio for the combined endpoint of OI/death was 2.6. That is there were 167 such events in the entire study, with a rate of 3.3 per 100 person-years in the TI arm and 1.3 in the CT arm. When broken down to look separately at OI (fatal or non-fatal) and non-OI death the ratios still favored CT. The ratio was 3.6 for OI (95 events, rates 2.6 and 0.6 per 100 person-years, respectively) and 1.8 for non-OI death (72 events, rates 1.3 and 0.7 per 100 person-years, respectively). So the rate of non-OI death, what we would think of in patients on stable HAART as largely death unrelated to HIV, in the CT arm was low and similar to that of OI in patients on stable HAART in this study. The rate of non-OI death was doubled in the TI arm, and the rate of OI was tripled in the TI arm.
The nadir CD4 count (the lowest-ever CD4 count) of patients upon enrollment to SMART did not change the hazard ratio for OI or death significantly, as the CT arm was favored no matter how high or low one's CD4 count ever was. Nor did the CD4 count at entry to SMART change the hazard ratio: no matter how high one's CD4 was at initiation of the TI strategy, the CT strategy was superior.
When the subgroup of only patients who were on ART at entry to SMART (as a significant number in SMART had stopped ART prior to entry) was examined, there were 131 OI/death events. Among these patients who enrolled on ART who actually had HIV RNA suppressed to <400 copies, the hazard ratio for OI/death was 4.0 (rates 3.2 vs. 0.8 per 10 person-years). Within this group the hazard ratio for OI (fatal or non-fatal) was nine-fold, but even the hazard ratio for non-OI death was still 2.1 in favor of CT.
Having had had an AIDS diagnosis, or Hepatitis B or C did not change the outcomes: the rate of OI/death was still lower in the CT arm. Number of ART regimens in the past, and duration of ART exposure did not alter the hazard ratios either.
Although CT was favored over TI in all subgroups examined (CD4, viral load, AIDS, hepatitis coinfection, sex, IDU status, race, ART history or type) there were some differences that were of interest but not yet completely explained. Being on an NNRTI increased the rate of OI/death if assigned to the TI arm (3.7 vs. 2.8 per 100 person-years). Being black increased the rate of OI/death if assigned to the TI arm, compared to Latinos or whites (4.5 vs. 1.8 and 3.1 per 100 person-years, respectively). In summary, Dr. El-Sadr noted that CD4+-guided episodic use of ART based on CD4+ cell count levels, as per the SMART study design, should not be recommended for any subgroup of patients, and that participants who were on ART at baseline and doing well had particularly inferior outcomes on episodic compared to continuous ART strategy.
Other, smaller fry presented findings
Although SMART was by ten-fold the largest TI study presented, other studies reported on the "safety" of the TIs. The Italian BASTA study (basta means "enough!" in Italian) examined the outcomes in 114 HIV-infected patients who underwent TIs, a study nearly 50 times smaller than SMART (abstract WEAB0202). Originally reported at ICAAC in 2003, this study enrolled patients durably suppressed to HIV-RNA <50 copies/ml with stable CD4 > 800 cells/μl. Patients were randomized (2:1) to either interrupt or to continue their ongoing treatment. Therapy was to be restarted if CD4 count <400 cells. Of 114 patients enrolled, 76 stopped therapy while 38 continued it. In 2003 it was reported that at roughly 20 months, about 25% of patients assigned continuous therapy had stopped therapy, presumably due to fatigue or toxicity.
The Italian group felt that CD4-guided TIs were as effective as CT. They observed 2 pneumonias and one aseptic meningitis in the 78 TI patients. The group of CT patients was very unlucky, suffering 2 deaths (not detailed), and one case each of cervical cancer, epilepsy, MI, and a herniated disc! As other groups have found, their analysis found a nadir CD4+ cell count of less than 350 cells/μL to be associated with a shorter time to the CD4 threshold of 40 cells and ART reinitiation.
Of note, in this small study drug resistance was reported more frequently in the TI group. 1 of the 38 CT patients had a resistant genotype (2 RT and 1 PR 90 mutation), whereas 4 such genotypes were found in the TI group (1, 3, 4, and 5 RT mutations).
Some of the conclusions of this study do not differ from the observations of SMART: OIs do not occur frequently after TIs in patients with higher CD4 nadirs and higher baseline CD4 cell counts if ART is reinitiated before CD4 cell counts drop too low. But the small size of this study limits its ability to comment on meaningfully on non-OI deaths that occur at higher CD4 cell counts. For example, while this small study saw more deaths in the CT arm, in ACTG 5170 (Skiest, CROI 2006) saw five deaths after interruption in a similar group of 167 patients.
The French SALTO trial was a prospective, open-label, multicenter trial of TI that enrolled patients on ART for 6 months with CD4 >450/μl and HIV RNA <5000, with no more than one prior treatment failure (abstract WEAB0205). Criteria to resume ART were a decrease in CD4 cell count to less than 300 or the occurrence of a CDC B or C AIDS-defining event. The study enrolled 99 patients (63% men). At 24 months, 88% of patients still had CD4 > 300. As in other studies, CD4 nadir and a history of AIDS were predictive of the need for ART reinitiation. This study also measured total HIV DNA in blood cells at entry, and found that it was also an independent predictor of time off ART. If this test is available, it might be of some use to clinicians if this finding can be further validated in a larger study.
And similarly, the recently published Staccato study:
The Staccato (Swiss-Thai-Australia Treatment Interruption Trial) study (Lancet 2006;368:459-465 also benefited by the use of a higher threshold for the reinitiation of ART. 548 patients with CD4 > 350 cells/ul, and HIV RNA <50 copies/ml were randomized to receive either continuous therapy (CT, n = 154) or a TI (n = 299). Of note, 80% of the cohort was treated with ritonavir-boosted saquinavir, and 80% of the cohort was from Thailand. These results may not therefore be widely generalizable, as the body mass and therefore pharmacokinetics of the study population may differ.
284 patients remained for evaluation in the "staccato" TI arm, stopping therapy when CD4s rose above 350 cells/ul, and restarting when CD4 cells fell below 350. 146 patients were studied while on continuous ART (CT). Median baseline CD4 counts at entry were 470 cells/μl (TI) and 507 cells/μl (CT). The study groups were well-matched: median age 35 years, 40% male, 80% heterosexual/15% MSM, mean pre-ART CD4 cell count of 253-267 cells/ul, median duration prior ARV 13-15 months. TIs were relatively short; a median of 18 weeks.
No AIDS-defining events occurred in either group. Two deaths, from stroke (CT), and colon CA (TI) occurred. Of significance 60.5% of TI subjects maintained CD4 > 350 cells/ul as compared with 96% in CT. At the end of the study TI patients restarted ART, and CD4 counts rose such that 86% of the TI group obtained CD4 > 350 after 12-24 weeks of ART vs 96.9% in CT arm. The proportion of patients with <50 copies HIV RNA/ml after 12-24 weeks of HAART re-treatment was 91% TI, 92% CT. At the end of study, similar proportions of subjects had achieved HIV RNA < 50 copies (ca. 90% in both arms), and although there were 9 virological failures in the TI arm as compared with 6 in the CT arm, this was statistically no different.
Again this is another example of the relative safety of TI in patients with high CD4 cell counts at TI and re-initiation of ART before CD4 counts drop too low. However, the study is again underpowered to examine the question of non-OI deaths in patients with viremia and high CD4 cell counts. Further, much longer term, follow-up will also be needed to fully address the issue of drug resistance and re-treatment failure when a plan of cyclical therapy is used.
|
| |
|
|
|
|
|
