With the prolonged demise of HIV eradication hypotheses and the growing appreciation of the various unpleasant realities of lifelong HAART, the issue of the optimal time to initiate treatment with antiretroviral therapy in chronically HIV-infected individuals is enjoying something of a renaissance. While the "Concorde" of triple cocktail combinations may still be a ways off (yet, unsurprisingly, less far fetched in the U.K. than in the states), AIDS experts in the U.S. have at least recently allowed themselves to consider the possibility that "hitting it early" just might not be the irreproachable paradigm it was once believed to be. At last month's AIDS Update in San Francisco, TAG's Mark Harrington was invited to share his thoughts on this increasingly timely topic. A transcription of Mark's "When to start?" remarks appears below.
Paul Joseph Corser
The AIDS Update was held in San Francisco's Bill Graham Civic Auditorium, attended by several hundred front-line AIDS service providers. Some of my relatives attended my talk, which was given after Neal Nathanson described the priorities of the NIH Office of AIDS Research (OAR).
Advertisement"Good afternoon. I would like to salute my mother, Judith, and my aunt, Peggy, who were kind enough to come today, and to thank my family for their strong loving support of my AIDS work and of TAG. These remarks are dedicated to two activist colleagues of mine who died in January 1999: Paul Corser and Rod Sorge. In each of their cases, while antiretroviral therapy may have helped, its complications no doubt intensified the complexities and difficulties they faced living with AIDS.
Paul Corser worked at amfAR for over ten years. He was a founding program director of amfAR's Community Based Clinical Trials network. He also helped pioneer amfAR's support of needle exchange programs, and served on the board of the Lower East Side Harm Reduction Center. In addition, he worked closely with TAG and helped to shape amfAR's recent innovative pilot research grants for immune reconstitution and vaccine research. Paul lived with AIDS for this entire decade, with many ups and downs. Ironically, he had finally achieved an undetectable viral load when, on January 4, he died in his sleep of a heart attack.
Rod Sorge joined ACT UP/New York (the AIDS Coalition to Unleash Power) in the late 1980s, when he was in college. He was a founding member of ACT UP's needle exchange program, which won several critical legal victories in the early 1990s, both in New York City and in New Jersey. Later he helped found the Harm Reduction Coalition and several needle exchange programs around New York. He was a national and indeed an international leader in the harm reduction movement. He was also an active heroin user, and as such experienced the worst our nation's health care system has to offer. When he was diagnosed with HIV-associated tuberculosis, the rifampin prescribed for him reduced the methadone levels in his blood to sub-therapeutic levels and he went through withdrawal. This happened many times, and only demonstrates the hostility of the health care system to drug users in this country. Rod died on January 26 of this year.
In the case of Paul Corser, we will never know whether his HAART (highly active antiretroviral therapy) contributed to his heart attack. In Rod's case, it's clear that the complex interactions between HAART, rifampin and methadone made all three less effective, and that his providers could have done far more to reduce his suffering. Their lives inspire us to continue their struggle; their deaths show us how far we still have to go.
The question "When to start?" is one of the most important unanswered questions facing AIDS research. Therefore, it would seem likely that some of the best, smartest AIDS researchers would be trying to answer this question. Unfortunately, for a variety of reasons, this is not the case. Unfortunately, therefore, we do not have much more hard evidence than we did three years ago [at the time HAART was introduced] about the best time to start antiretroviral therapy.
My talk will have five sections:
1. What do we know now?
2. Why don't we know more?
3. The reasons for starting early are theoretical.
4. The reasons for starting later are practical.
5. What can we do to answer the question?
We know that viral load and CD4 levels predict the rate of progression and time to AIDS. Higher viral load equals faster progression. Lower CD4 count means a shorter time to AIDS. We know this from MACS [Multicenter AIDS Cohort Study] data published by John Mellors in 1996 and 1997. However, the MACS data included only men. In addition, we know from newer data that women progress with a somewhat lower viral load then men. However, the difference was not judged great enough to warrant a change in Public Health Service (PHS) guidelines at this time.
We know from several randomized, controlled trials with clinical endpoints (ACTG 175, CPCRA 015, Delta) that if you start with a CD4 count below 350/mm3, it's better to start with two nucleoside analogues (AZT+ddI or AZT+ddC) or ddI alone than to start with AZT alone. We know that this approach actually slows time to AIDS and reduces mortality when compared with AZT monotherapy over several years of follow-up.
We know from ACTG 320 that if you take an AZT-experienced person with a CD4 count below 200/mm3, adding 3TC and indinavir (Crixivan), your chance of developing AIDS or dying is half that of those who added only 3TC -- for the period of time studied.
We also know from Abbott's pivotal licensing study that if you give ritonavir [Norvir] to people with a CD4 count below 100/mm3 who are already on one or two nucleoside analogues, the rate of AIDS illness and death is also halved, at least in the immediate term. We know from retrospective, observational studies carried out in the U.S., Canada, France and elsewhere in the developed world, that AIDS and death rates have plummeted since the introduction and widespread use of HAART.
It's important to note, however, that most of the deaths which have been averted have been among people whose CD4 counts were already quite low when HAART was introduced. For example, the twice-published studies of Frank Palella included only people whose CD4 counts were below 100/mm3 at baseline. Among this group, mortality has dropped by more than half since 1996. But we have no reliable clinical evidence that starting any kind of antiretroviral therapy with a CD4 count over 350 cells/mm3 prolongs either health or life. We have no reliable clinical evidence that starting HAART with a CD4 count over 200 cells/mm3 prolongs health or life. Thus, if we restricted our dataset to randomized, controlled clinical trials, we would have no compelling reason to put two thirds of the HIV infected population -- those with a CD4 count over 350 -- on treatment at this time.
The Health and Human Services (HHS) Guidelines Panel (of which TAG's Spencer Cox and Mark Harrington are members) evaded the issue of "When to start" by saying that, in the asymptomatic population, the decision to start treatment should be made by the patient and physician taking into account the risk of progression as measured by viral load and CD4, the patient's willingness to undertake therapy, and other factors. "Treat or observe" for the group with a CD4 count below 500/mm3 or a viral load higher than 10,000-20,000 copies/mL. "Observe or treat" for the group with higher CD4 counts or lower viral load.
Of course, experts and blue ribbon panels have been wrong before. In 1990, after the results of ACTG 019 were released, The Public Health Service (PHS) recommended the use of early AZT for all people with CD4 counts below 500/mm3. This was refuted by the Concorde study in 1993. In the early 1990s, experts commonly recommended the addition of ddC (or less commonly ddI) to a failing AZT regimen. This approach was disproved by ACTG 155 in 1993 and by later studies.
As late as 1995, many "experts" were adding new protease inhibitors -- most commonly Roche's hard capsule saquinavir (Invirase), as it was the first to be approved -- to failing nucleoside regimens. By 1996, it was clear that this was the wrong strategy for starting a protease inhibitor (it should, instead, be given with a new background of nucleoside analogues) and that Invirase, the weakest drug of its class, predisposed people to developing resistance to other, stronger protease inhibitors.
Of course, since no one is carrying out definitive large, long-term randomized controlled trials with clinical endpoints of different starting times, we are unlikely to witness the direct refutation of the current paradigm -- although it is likely to be displaced by new developments.
What are some of the reasons, or excuses, why we don't know more about when to start, and why more research isn't being done to answer this question? First, the field is rapidly changing. New drugs and new concepts are constantly emerging.
Secondly, answering this question conclusively would take several years and several thousand patients. However, hundreds of thousands of "early" patients (for the purposes of this talk, defined as those with a CD4 count over 350-500/mm3) are already on HAART, and we don't know if it's helping them in the long run. Moreover, no one flinches from a several thousand patient study to validate new approaches with new mechanisms of action, e.g., interleukin-2 (Proleukin) or the Salk Remune HIV immunogen. Why shouldn't we undertake one which has such profound public health and cost implications?
Third, doctors like giving pills. Fourth, drug companies like selling pills. Fifth, the short-term benefits of HAART in advanced disease are dramatic.
Sixth, the large NIH-funded clinical trials networks are busy struggling to be refunded. The AIDS Clinical Trails Group (ACTG) and the Community Programs for Clinical Research on AIDS (CPCRA) are both up for renewal in the year 2000, and they're too busy worrying about whether they'll be refunded to launch an ambitious trial such as "When to start."
The ACTG claims it lacks the patients. The CPCRA claims it lacks the resources. The NIH, which funds them both, claims it lacks the authority to make the networks do the study. The drug companies lack the incentive, and so won't supply drug for such a study. The insurance companies and HMOs, which do have an incentive to have this question answered, seem indifferent to the potential gold-mine of data (and possible savings) which such a study could provide. Finally, some of the best and brightest AIDS researchers lack the patients and the resources and are more interested in biological mechanisms than in large public health studies. Obviously, this is an untenable and unacceptable collective evasion of responsibility.
- Hitting early preserves immune function. In general, however, immune function is fine until the CD4 count goes below 350 cells/mm3, and, in any case, guidelines recommend starting therapy whenever someone becomes symptomatic.
- Hitting early delays resistance. Actually, since antiretroviral therapy imposes selective pressure on the virus to develop resistance, early therapy -- especially if associated with non-adherence or incomplete viral suppression -- may actually select for resistance. According to retrovirologist John Coffin, the only time hitting "early" may actually create a bottleneck in viral diversity is extremely early -- within the first six months of infection.
- Hitting early may speed time to eradication. So far, unfortunately, there is no evidence that this assertion is true. With HAART alone, estimates of the time to eradication range from twelve to thirty years. [NB: Since this speech was delivered, Finzi et al. published in Nature Medicine (5:5, 512-517, May 1999) their most recent estimates of the time to eradication with HAART, which range from 60 years -- if there are only 100,000 "reservoir" cells -- to over 70 years if there are one million or more of these cells.]
- Hitting early preserves anti-HIV immune function, or anti-HIV CD4 cells. Again, the evidence to date from Bruce Walker's group, the pioneer in this field, suggests this is true only when one treats very early -- during primary infection and before seroconversion, in fact. Moreover, his dataset is a small one. Other researchers, however, report greater variability in this cell population, suggesting it may be preserved into chronic HIV infection. [NB: In another paper in the same issue of Nature Medicine (5:5, 518-25, May 1999), Picker and colleagues from the University of Texas Southwestern Medical Center in Dallas showed that "HIV-1-specific CD4 T cells are detectable in most individuals with active HIV-1 infection, but decline with prolonged viral suppression." This implies that HAART actually suppresses this important CD4 population.]
Eradication of HIV from an infected individual's body is still a worthwhile goal, but it's one which appears increasingly far off and, with HAART alone, unlikely. In the meantime, could we settle -- at least temporarily -- for drug-free remission? I don't see why not. It has already been achieved in a handful of cases, starting with the notorious Berlin patient (see Lisziewicz et al., "Immune control of HIV after suspension of therapy," abstract 351, Sixth Retrovirus Conference, Chicago, 1999), and continuing with a number of individuals treated during primary HIV infection and in a few treated (e.g., at the Aaron Diamond AIDS Research Center in New York; see Ortiz et al, "Containment of breakthrough HIV plasma viremia in the absence of antiretroviral drug therapy is associated with a broad and vigorous HIV specific cytotoxic T lymphocyte (CTL) response," abstract 256, Sixth Retrovirus Conference) during chronic infection.
Ironically, in all these cases, unplanned "drug holidays" appear to have permitted a limited viral rebound which stimulated the expansion of anti-HIV CD4 and CD8 cells, which were then able to control -- but not eliminate -- the virus. This argues for an examination of structured drug holidays to investigate the potential of pulsed dosing to allow the immune system a chance to catch up with if not overtake the virus (see Waldholz & Tanouye, "Studies will see if drug 'holidays' for HIV patients could lead to a vaccine," Wall Street Journal, 25 January 1999; and "Pulsed therapy and structured interruptions of treatment," Project Inform Perspective 27, April 1999).
Some might object that such drug holidays might encourage the development of drug-resistant HIV. So far, evidence from both Franco Lori's group and from the COMET study does not support this objection (see Lori et al., "Intermittent drug therapy increases the time to HIV rebound in humans and induces the control of SIV after treatment interruption in monkeys," late breaker abstract LB5, Sixth Retrovirus Conference, and Neumann et al., "HIV-1 rebound during interruption of highly active antiretroviral therapy has no deleterious effect on reinitiated therapy," AIDS 13(6):677-83, April 1999):
In the COMET Study, 10 antiretroviral-naïve patients initiated therapy with zidovudine, lamivudine, and indinavir for 28 days, followed by interruption of all drugs for 28 days and then reintroduction of the same regimen. No new resistance mutations developed during the study. The authors conclude that a 1-month interruption of all drugs in a HAART regimen does not adversely affect the virologic efficacy of the same regimen once reinitiated (Neumann 1999).
- HAART does not completely suppress HIV replication. Therefore, once it is started, the potential for the evolution of drug-resistant virus is omnipresent. Moreover, virologic failure is common, occurring in between 20%-60% of patients within two years. Although there appears to be a temporal delay, virologic failure will lead eventually to clinical failure.
- Due to resistance and cross-resistance, the number of plausible, realistic, sequential, potent, tolerable and non-cross-resistant HAART regimens is either two or three. Drug intolerance -- e.g., anemia, diarrhea, kidney stones, pancreatitis, peripheral neuropathy, psychosis, rash, Stevens-Johnson syndrome, etc. -- can reduce this number still further in some individuals.
- Over 90% of the new drugs in the pipeline are "me-too" drugs which inhibit reverse transcriptase or protease. They will not work against many current or future drug resistant or cross resistant viral isolates.
- The drugs are hard to take. Fewer than 95% adherence virtually assures the evolution of drug resistance, as Paterson and colleagues from the University of Nebraska demonstrated earlier this year:
Adherence to protease inhibitors (PIs) was assessed electronically in 84 subjects at two sites using MEMScaps. Baseline CD4 counts ranged from <50 cells/mm3 >500 and baseline viral load from <400 copies/ml to >100,000. At three months, a highly significant association was found between adherence and virologic suppression (p=0.00001). 81% of subjects with >95% adherence had complete viral suppression compared to 64% with 90-95% adherence, 50% with 80-90% adherence, 25% with 70-80% adherence and 6% with <70% adherence. (Paterson et al., "How much adherence is enough? A prospective study of adherence to protease inhibitor therapy using MEMSCaps," abstract 92, Sixth Retrovirus Conference 1999).
- Long-term side effects -- elevations in cholesterol and triglycerides, lipodystrophy, perhaps even cardiac disease -- are emerging. Some appear serious. They may require additional treatments, adding to the complexity of therapy and the risk of treatment failure.
- HAART is expensive. If more people waited to start treatment until "later," the public health system could afford to treat more HIV-infected people who are not currently in care. The cost of HIV care is rising for outpatients with the addition of expensive new drugs and laboratory tests.
- We will have better drugs and strategies in the future.
- We will have better information in the future.
- Regimens appear generally most potent if you are antiretroviral naïve. As with virginity, you're only drug naïve once.
- HAART-induced immune restoration appears impressive in many who started therapy quite late. (See Tubiana et al., "Immunologic reconstitution after antiretroviral treatment," Presse Med. 28(8):424-8, Feb. 27 1999). Obviously, however, it's better to avoid irreversible immune system damage and end-organ disease. The cutoff of 350 CD4 cells/mm3 I mentioned before, however, is earlier than either irreversible immunological damage or most serious opportunistic infections.
What strategies to activists have to encourage researchers to find out what is the best time to start antiretroviral therapy? We are encouraging NIH funded researchers and others to:
- Develop simpler HAART regimens -- twice or, optimally, once daily.
- Develop more potent HAART regimens.
- Develop rational sequencing strategies.
- Develop strategies to encourage adherence.
- Develop new antiviral compounds targeting drug resistant HIV.
- Develop new antiviral classes which inhibit new viral or cellular targets.
- Study immune activation to flush out latent HIV reservoirs.
- Study immune reconstitution to restore or augment anti-HIV immunity.
- Study immune de-activation to slow down or silence HIV reservoirs.
- Develop cheaper, quicker drug resistance assays.
- Develop cheaper, quicker assays for anti-HIV CD4 and CD8 cells.
- Study structured drug holidays.
- Study immediate versus deferred therapy in primary HIV infection (PHI).
- Study immediate versus deferred therapy in chronic HIV infection.
- Be willing to admit the uncertainty of current approaches and be willing to encourage people with HIV to make decisions to start or defer according to their own values, needs and preferences.
Ultimately, we need to develop alliances between people with HIV, doctors and health care workers to pressure NIH, academic researchers, drug companies and HMOs to work together to design research studies which answer the question -- when is the best time to start antiretroviral therapy, and what is the best regimen or strategy to start with. Thank you.
Arguments for Immediate Antiretroviral Therapy
Starting early preserves immune function.
Immune function does not appear to become significantly compromised until CD4+ T-helper cells fall below the 300-350 cells/mm3 mark.
Starting early forestalls the development of resistance.
Under conditions of less than complete suppression of viral replication, selective drug pressure on HIV by early drug therapy actually facilitates the development of drug resistance.
Starting early prevents widespread dissemination of the infection and thus shortens the time required eventual eradication.
There is no evidence that this is true and some evidence that the exact for opposite is closer to being the case. (In a 1998 study at Johns Hopkins University, individuals with the most advanced HIV disease prior to initiation of HAART showed the fewest number of persistently infected memory T-cells after treatment.)
Starting early spares or preserves the HIV-specific T-cell subsets.
Because potent antiretroviral therapy dramatically reduces the quantity of circulating virus, HIV-specific immune responses virtually disappear on HAART due to the relative absence of antigen.
Arguments for Delayed Antiretroviral Therapy
- HAART does not completely shut off viral replication, and thus the development of drug resistance is virtually guaranteed.
- Cross-resistance greatly limits the number of viable treatment options after the first combination regimen fails -- or can no longer be tolerated.
- Immune restoration has been quite good even in patients who were quite late to initiate antiretroviral therapy.
- Close to 80-90% adherence to complicated and onerous treatment protocols is required in order to enjoy only a 50% chance of successful viral suppression.
- There are precious few non-"me too" drugs in development.
- Long-term side-effects which are beginning to emerge can be serious and are poorly understood.
- HAART is expensive, and if more people waited to start antiretroviral therapy the public health system could afford to treat more HIV-infected individuals who are not currently in care.
- HAART has its most dramatic effects in individuals who have not received any prior antiretroviral treatment.
- Better drugs, superior knowledge, longer range strategies and easier dosing schedules are likely to be available in the future.