Initial Therapy and Treatment Interruption

Hopkins HIV Report 2006 Mar; 18(2):1-5

Joel E. Gallant, M.D., M.P.H.
Johns Hopkins

At the 13th Conference on Retroviruses and Opportunistic Infections, held in Denver from February 5-8, 2006, there were few presentations on treatment options for antiretroviral-naïve patients. However, there were several important treatment interruption trials as well as a number of studies that have implications for the timing of treatment initiation. Finally, I learned that “Denver” is not just a place where you go to change planes when traveling elsewhere, but is actually a real city, where people live and work, and where there is a superb conference center. Who knew?

Since the approval of atazanavir (ATV) in 2003, there has been a tendency among many physicians to prescribe it using ritonavir (RTV) boosting, despite the lack of specific data on the use of boosted ATV in protease inhibitor (PI)-naïve patients and the absence of any recommendation about boosted ATV in the DHHS guidelines. There are a number of good reasons for this approach, including the demonstrated advantages of RTV boosting for other PIs with respect to pharmacokinetics, efficacy, activity against resistant virus, and emergence of resistance with failure. In the case of ATV, we know that trough levels are higher with RTV boosting, and also that the 2-year efficacy of RTV-boosted ATV (ATV/r) was comparable to that of lopinavir/ritonavir (LPV/r) in PI-experienced patients in the BMS 045 study [Johnson M, et al. AIDS. 2005 Apr 29;19(7):685-94], whereas LPV/r was superior to unboosted ATV in PI-experienced patients in the BMS 043 study [Cohen C, et al. Curr Med Res Opin. 2005 Oct;21(10):1683-92]. By extrapolation, it has been assumed that ATV/r must be at least as effective as unboosted ATV.

The AI424-089 study is the first study to directly compare boosted and unboosted ATV in a treatment-naïve patients [Malan N, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 107LB]. The 200 patients randomized also received lamivudine (3TC) and stavudine (d4T) in the extended release (XR) formulation that was approved but never marketed and is no longer being developed. Median baseline viral load was approximately 5 log₁₀ c/mL, and median CD4 count was approximately 200 cells/mm³. By intention-to-treat (ITT) analysis, 75% of those in the ATV/r arm had a viral load of <50 c/mL at 48 weeks, compared to 70% in the ATV arm. For the <400 c/mL analysis, the results were 86% and 85%, respectively. While the efficacy appeared to be comparable, especially in the 400 c/mL analysis, it is worth noting that there was more virologic failure in the unboosted ATV arm (10% vs. 3%), whereas there was more drop-out due to hyperbilirubinemia in the ATV/r arm. The study was able to conclude that ATV/r was non-inferior to the ATV arm, but it was not powered to answer the more relevant question of whether ATV was non-inferior to the ATV/r am, and in fact the results would suggest otherwise.

There were some other important differences between arms in this study. Hyperbilirubinemia and jaundice were more common in the ATV/r arms, with 3% developing grade 2-4 jaundice compared to <1% in the unboosted arm. Lipid levels were also higher with RTV boosting, although the majority of patients in both arms maintained lipid levels below NCEP targets, and few started lipid-lowering therapy.

Among the patients who failed therapy, resistance profiles were more favorable in the ATV/r arm. None of the 3 patients who failed ATV/r developed PI mutations compared to 3 of 10 failures on ATV, who developed I50L (1) or I50I/L + G73G/S (2). Emergence of M184V also appeared to be more common among those failing unboosted ATV. In the end, these results support both the use of ATV and of ATV/r in treatment-naïve patients. On the one hand, the lower rate of virologic failure and the apparent resistance advantages suggest that ATV/r may be more potent than ATV, and may in fact be preferable if tolerated. On the other hand, the efficacy of unboosted ATV was reassuring, making this a reasonable alternative for patients who can’t tolerate RTV. This study also demonstrates that there is a small price to be paid in lipid levels and the potential for jaundice with the use of even 100 mg of RTV per day.

Monotherapy with Boosted Atazanavir

Most studies looking at boosted PI monotherapy have involved LPV/r, in part because of its record of potency, its high genetic barrier to resistance, and the lack of resistance with failure. However, there is also interest in the use of ATV/r monotherapy, a strategy that is being studied in ACTG 5201 [Swindells S, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 108LB]. This is an open-label pilot trial in which patients with virologic suppression on antiretroviral therapy were first switched to ATV/r plus 2 nucleoside analog reverse transcriptase inhibitors (NRTIs). If they maintained suppression on that regimen, the NRTIs were discontinued after six weeks, leaving them on ATV/r alone. A total of 34 patients completed that second step of the protocol, of whom 1 withdrew consent and 3 experienced virologic failure. Two of the failing patients had undetectable ATV levels and were presumed to be non- adherent. However, the third patient had detectable ATV levels. His viral load resuppressed with continued use of ATV/r alone. None of the three patients developed PI mutations.

This study suggests that this is a viable concept and should be studied further. However, the PI monotherapy studies reported to date have two things in common: (1) there is invariably at least one patient in each study who fails to maintain virologic suppression despite adherence and detectable or therapeutic drug levels, and (2) the sample sizes are never large enough to allow us to definitively conclude that this is a safe and effective strategy. Fortunately, larger, randomized trials are ongoing.

Intensification of Therapy in Patients Taking Trizivir

In the last issue of The Hopkins HIV Report, I discussed Roy Guilck’s presentation of ACTG 5095 comparing zidovudine/ lamivudine/abacavir (AZT/3TC/ABC, Trizivir) + efavirenz (EFV) vs. AZT/3TC (Combivir) + EFV [46th ICAAC, 2005, Abstract H-416a]. That study demonstrated no difference in performance between the two arms. The original study also included a Trizivir arm, which was found to be inferior to the two EFV-containing arms [Gulick R, et al. N Engl J Med. 2004 Apr 29;350(18):1850-61]. Patients who remained virologically suppressed on the Trizivir arm were randomized to intensify therapy with either EFV or tenofovir DF (TDF), and Gulick presented the results of this comparison at CROI [Gulick R, et al., Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 519]. Overall, there was no difference between the arms in time to virologic failure or treatment discontinuation, in the proportion of patients with undetectable viral loads at weeks 24, 48, or 72, or in CD4 cell count responses at those three time points. However, it was noted that treatment failure occurred in 22% of those who added TDF vs. 15% who added EFV. While that difference was not statistically significant, Kaplan-Meier curves demonstrated more failure among EFV recipients in the first 28 weeks, but more failure among TDF recipients between 28 and 88 weeks. A reasonable assumption might be that early failure on EFV was likely to reflect discontinuation due to side effects, whereas later failure on TDF reflects virologic rebound on a less potent regimen. However, this assumption cannot be confirmed by the analysis presented at CROI. Suffice it say that while there is some support for the combination of Trizivir plus TDF, it should still be viewed as an investigational regimen.

A Setback for CD4-Guide Treatment Interruption

Several studies examining treatment interruption strategies were presented at CROI, with the session culminating in the presentation of the large SMART trial, which demonstrated significantly greater morbidity and mortality among patients who interrupted therapy [El-Sadr W, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 106LB]. Prior to SMART, most studies looking at this strategy, in which therapy is stopped when the CD4 count reaches a predetermined threshold and resumed when it falls to another lower threshold, had suggested that this approach was safe, especially in patients with relatively high CD4 nadirs. However, these were much smaller studies than SMART, which involved nearly 5,500 patients. Before discussing SMART, it’s worth mentioning the other treatment interruption studies presented at CROI. I will not discuss trials that exclusively studied fixed-interval interruption strategies [Palmisano L, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 103, Marchou B, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 104], as there has been less enthusiasm for this approach in recent years.

ACTG 5170 assessed treatment interruption in 167 patients who had been on HAART for at least 6 months and who wanted to stop therapy. Enrollment required that they have had a pre-treatment CD4 nadir >350 cells/mm³ and a pre-treatment viral load <55,000 c/mL, which meant that they did not meet criteria for HAART based on guidelines in place at the time. Eligible patients discontinued HAART for up to 96 weeks, with the potential to resume at the discretion of the health care provider. Those on non- nucleoside reverse transcriptase inhibitor (NNRTI)-based regimens stopped their NNRTI two days before stopping their NRTI. After stopping, patients lost approximately 20 cells per week in the first 8 weeks, and 1.7 cells per week thereafter. Viral load rebounded quickly during the first 4-8 weeks and then stabilized. By week 48, 26 had resumed therapy, 12 had CD4 counts ≤250 cells/mm³, and there were 3 deaths. From weeks 49-96, 5 reached a CD4 count ≤250 cells/mm³, 2 each had developed CDC stage B or C conditions, and there were two deaths. None of the 5 deaths were from AIDS-related conditions, and most occurred in patients with relatively preserved CD4 counts. In a multivariate analysis, predictors of time to reaching a study endpoint by week 96 were low CD4 nadir and detectable viral load at entry.

The Staccato Trial has been discussed in several previous issues of The Hopkins HIV Report. This is a trial, conducted mostly in Thailand, in which 548 patients with CD4 counts above 350 cells/mm³ and viral loads <50 c/mL on HAART were randomized to continue therapy (CT, n=154), to interrupt therapy until their CD4 count fell below 350 (TI, n=299), or to take therapy on a week-on, week-off schedule [Ananworanich J, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 102]. As has already been reported, the week-on, week-off arm was terminated early because of high rates of virologic failure. Patients in the TI arm all took continuous HAART during the last 12-24 weeks of the study. At the end of that period, there was no difference in the proportion of patients who had CD4 counts above 350 cells/mm³ (86% vs. 97% in the TI and CT arms, respectively, p=NS). Not surprisingly, those in the TI arm, who spent 37.5% of the study period on HAART, were more likely to experience oral or vaginal candidiasis, thrombocytopenia, or acute retroviral syndrome, while those in the CT arm had significantly more diarrhea and neuropathy, as well as higher total cholesterol levels and more self-reported lipodystrophy. There was no difference in mortality (1 non-HIV-related death in each arm) or in the frequency or patterns of resistance.

ANRS 1269, the Trivican study, enrolled 940 treatment naïve, West African patients with baseline CD4 counts of 150-350 cells/mm³ [Danel C, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 105LB]. After achieving virologic suppression on continuous HAART, they were randomized to remain on continuous HAART (1:6), to interrupt therapy using a fixed-interval approach of 2 months off, 4 months on (3:6), or to interrupt HAART when the CD4 count was above 350 cells/mm³ and to resume when it fell below 250 cells/mm³ (2:6). An interim analysis by the Data and Safety Monitoring Board resulted in discontinuation of the latter arm because of significantly greater morbidity and mortality compared to the other two arms. Mortality was 1.2 per 100 person-years (PY) in the TI arm vs. 0.6 in the CT arm, and the incidence of serious morbidity (any WHO stage 3 or 4 classifying event) was 11.6 vs 6.7 per 100 PY. The incidence of both tuberculosis and invasive bacterial infections was significantly higher in the TI arm, as were hospitalization and clinic visits.

That brings us to the Strategies for Management of Antiretroviral Therapy trial (SMART), a large NIH-funded trial in which 5472 patients from 33 countries and 318 study sites were randomized to what the investigators called the “virologic suppression” arm or the “drug conservation” arm, but which I will refer to as the CT and TI arms, respectively, to avoid confusion. All patients had CD4 counts >350 cells/mm³ at entry. This study was originally designed to look not only at CD4-guided treatment interruption, but also at the question of when to start therapy, so it enrolled both treatment naïve and treatment experienced patients. However, 95% of the participants were treatment-experienced at the time of enrollment. Those randomized to the TI arm discontinued therapy, resuming when the CD4 count fell below 250 cells/mm³, and stopping again when it rose above 350 cells/mm³. Those in the TI arm spent approximately one-third of the study period on therapy, compared to 93% in the CT arm. The study was stopped prematurely because of a 2.5-fold greater risk of clinical progression or death in the TI arm (3.7 vs. 1.5 per 100 PY, p<0.0001). The difference favoring the CT arm was present in both men and women and in blacks and non- blacks. Moreover, outcomes were better in the CT arm regardless of nadir CD4 count, which has always been the best predictor of success with treatment interruption in other studies. Surprisingly, there were trends toward a greater incidence of severe “complications” in the TI arms, defined as cardiovascular, renal, or hepatic deaths or non-fatal events. These endpoints, which can be complications of antiretroviral therapy, were included as it was assumed that they would be less frequent in patients who interrupted therapy.

Interestingly, deaths among patients who interrupted therapy weren’t due to typical AIDS-related conditions; instead, they were the result of events such as myocardial infarction, which we normally don’t think of as being related to treatment interruption. However, the power of this large comparative trial to detect differences in outcome forces us to reconsider the implications of such deaths in other studies, such as ACTG 5170, discussed above. We’ve tended to dismiss those events as being unrelated to treatment interruption, but the possibility of a causal association is not biologically implausible. For example, patients on antiretroviral therapy, some of whom may have developed accelerated atherosclerosis as a result of treatment, will experience inflammation and immune activation when therapy is discontinued, which could lead to an increased risk of myocardial infarction.

Where do these studies leave us? The implications of these findings were controversial enough that the meeting organizers added a panel discussion at the end of the session to help sort out the conflicting data. Interpretations of the data varied widely among the panelists as well as the attendees talking the issue over in the halls. Some took the narrowest approach, concluding that if therapy is interrupted, it should be resumed well before the CD4 count has fallen to 250 cells/mm³, the threshold used in both SMART and Trivican. Others concluded that the entire concept of CD4-guided treatment interruption is dead and should not be studied further. My own opinion falls somewhere between these extremes. Even before SMART, interest in physician-directed, CD4-guided treatment interruption was waning, primarily because options for initial therapy were becoming so much easier and less toxic than they were when the strategy was first proposed. Should a patient taking two pills once a day with no side effects and no toxicity interrupt therapy, even if it’s safe? In parts of the world where access to therapy is not an issue, the standard of care should be continuous therapy without interruption based on the data we have now. Patients who have to interrupt therapy because of toxicity or other reasons should resume before the CD4 falls below 350.

On the other hand, intermittent therapy approaches are attractive in resource-limited settings, where a strategy that allows use of one-third the amount of antiretroviral therapy per patient could allow the treatment of three times as many people. However, the Trivican study points out that in the developing world, tuberculosis and serious bacterial infections can occur at CD4 counts well above the 200 cells/mm³ threshold that we typically think of as marking the onset of risk for opportunistic infections. In such regions, an argument can be made for even earlier resumption (or possibly initiation) of therapy to decrease the risk of these complications.

When to Start Therapy

The SMART trial, while not addressing it directly, also speaks to the question of whether we should be starting therapy earlier. The higher rate of complications and death among patients who interrupted therapy despite having CD4 counts above 250 cells/mm³ is concerning, although we can’t assume that a treatment-naïve patient is the same as a patient who has interrupted therapy. There were several other studies presented at CROI that also addressed this issue, each of which suggested that there may be an advantage to earlier therapy.

Two cohort studies, one from the Hopkins database [Keruly J, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 529] and the other from the Dutch ATHENA cohort [Gras L, et al. Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 530] found that patients who start therapy with higher CD4 counts are more likely to experience normalization of their CD4 counts, despite a greater magnitude of CD4 increase in those with lower baseline CD4 counts. Assuming that it’s good to have a normal CD4 count (which has not been clearly demonstrated), this would argue for earlier initiation of therapy.

An analysis from the ART Cohort Collaborative, which includes data from over 10,000 patients from a variety of clinical cohorts, found that the cumulative probability of progression to AIDS or death was highest among patients who started therapy with CD4 counts below 200 cells/mm³. However, patients who started therapy with CD4 counts of 351-500 cells/mm³ were significantly less likely to progress or die than those with baseline counts of 201-350 cells/mm³, though the difference was much smaller than the difference between either group and those with the lowest CD4 counts at baseline. Weaknesses of this study included the fact that the data were modeled rather than directly observed and the fact that injection drug users were excluded from the analysis.

Kenneth Lichtenstein presented data from the HIV Outpatient Study (HOPS) demonstrating that patients who initiated therapy with higher baseline CD4 counts and who continued therapy without interruption were less likely to experience common drug- associated toxicities, including renal insufficiency, peripheral neuropathy, and lipoatrophy [Lichtenstein K, et al., Conf Retrovir Opportunistic Infect 2006 Feb 5-8;13:abstract no. 769 ]. CD4 responses and rates of virologic suppression were higher and death and clinical progression were lower among those who started therapy earlier. These differences extended to those with pre-HAART CD4 counts of 350-499 cells/mm³ and even above 500 cells/mm³. While initial regimens used today would be less likely to cause complications such as lipoatrophy or peripheral neuropathy, these data do suggest other potential benefits to earlier initiation of therapy.

Of course, observational studies like the ones discussed here are subject to considerable selection bias: for example, the better outcome associated with early initiation of therapy may have more to do with characteristics of patients who are treated aggressively than with baseline CD4 counts. Nevertheless, it should not be surprising that cohort studies originally suggesting no benefit to early therapy are now beginning to detect differences with longer follow-up. These data, together with the continued improvement in initial treatment options, may cause the pendulum to swing once again toward a more aggressive approach to antiretroviral therapy.

2006-03-10
HHR-2006-03-01

Copyright © 2006 - The Johns Hopkins University AIDS Service, Division of Infectious Diseases. Permission to use and reproduce portions of this newsletter is hereby granted provided that author and publication are fully credited and both copyright and permission notice appear with reprinted material. Inquiries may be directed to Sharon McAvinue, Managing Editor. Website: Johns Hopkins AIDS Service.

AEGiS is a 501(c)3, not-for-profit, tax-exempt, educational corporation. AEGiS is made possible through unrestricted funding from Boehringer Ingelheim, Bridgestone/Firestone Charitable Trust, Bristol-Myers Squibb Company, Elton John AIDS Foundation, the National Library of Medicine, Roche and Trimeris, and donations from users like you. Always watch for outdated information. This article first appeared in 2006. This material is designed to support, not replace, the relationship that exists between you and your doctor.

AEGiS presents published material, reprinted with permission and neither endorses nor opposes any material. All information contained on this website, including information relating to health conditions, products, and treatments, is for informational purposes only. It is often presented in summary or aggregate form. It is not meant to be a substitute for the advice provided by your own physician or other medical professionals. Always discuss treatment options with a doctor who specializes in treating HIV.

Copyright ©1980, 2006. AEGiS. All materials appearing on AEGiS are protected by copyright as a collective work or compilation under U.S. copyright and other laws and are the property of AEGiS, or the party credited as the provider of the content.