1. Home>
2. Specialties>
3. HIV/AIDS Clinical Care>
4. Feature

Therapy of Early HIV Infection: Time for the Pendulum to Swing Back?

The use of protease inhibitors (PIs) has led to dramatic decreases in AIDS-related morbidity and mortality. Although very little data exist on the effectiveness of PIs at the early stages of HIV infection, the U.S. Department of Health and Human Services, the IAS-USA, and other organizations have published guidelines recommending their use in patients with early-stage HIV. These guidelines are intended for flexible use, but clinicians often strictly adhere to them. They are even used to gauge the quality of medical care.

Recently in the Lancet, J. A. Levy issued a cautionary note suggesting that in some HIV-positive patients therapy could be deferred. In this article, we take the question of the early treatment of HIV infection one step further asking, if we do treat early, should we always hit as hard as we can? In other words, could we consider initiating therapy with less potent and simpler regimens for those early-stage HIV-positive patients in whom therapy is not deferred? Our basic premise is that overly simplistic models of the pathogenesis of HIV infection and antiretroviral therapy have been used to develop the current guidelines. This article outlines what we believe to be some of the key issues that should influence the formulation of future guidelines.

Must therapy be started before a substantial decline in the CD4-cell count?

In the absence of antiretroviral therapy, it is unusual for opportunistic infections (OIs) to occur in patients with a CD4 count >350 cells/mm³. Furthermore, some immune recovery occurs as a result of antiretroviral therapy, as evidenced both by in vitro data and by significant decreases in OIs in patients whose CD4-cell counts have risen. The CD4-cell counts at which OIs tend to occur in patients on therapy and patients who are not on therapy are similar. Prophylaxis for at least some OIs can be suspended in patients whose CD4-cell counts have increased on therapy. These data suggest that CD4-cell counts are excellent markers for the short-term risk of developing an OI, and that waiting for a CD4 count to decline to about 350 cells/mm³ before initiating therapy may not be harmful in many patients.

Are the viral load levels used to recommend initiation of therapy universally applicable?

Most of the studies that served as the basis for using viral load measurements in making decisions about initiation of therapy were conducted in largely white, homosexual, American male populations infected with subtype B virus. However, variables such as gender, host genetics, and variations in the virus influence HIV progression. For example, in one study, women had a 40% to 60% greater probability of progressing to AIDS or death than men, following adjustment for viral load. A study in Africa indicated that the risk of progression might differ according to HIV-1 subtype. In Brazil, patients infected with the closely related serotypes B and Bbr (an HIV-1 subtype B that differs slightly from the prototype American subtype B; about 40% of Brazilians who are infected with subtype B are in fact infected with Bbr) progressed at different rates to AIDS or death despite having almost identical clinical status, CD4-cell counts, and viral load at study entry. These data indicate that the viral load levels currently used to guide decisions about therapy may not apply to all individuals.

Are we relying too much on too few viral load measurements?

At present, most guidelines recommend initiating therapy based on one or two viral load and CD4-cell count measurements. Given sufficient time, most, if not all, HIV-infected individuals will progress to AIDS in the absence of therapy. Thus, CD4-cell count and viral load measurements are only predictors of the probability of disease progression within a defined time period. In a recent study, progression to AIDS after seroconversion was strongly associated not only with the viral load set point, but also with the rate of viral load increase over the first three years after seroconversion. For those patients who eventually developed AIDS, the slope of viral load was similar in the three years before AIDS, regardless of the starting viral load or CD4-cell count. These data suggest that it may be more appropriate to measure viral load and CD4-cell counts in a serial fashion to better assess the rate of CD4-cell count decline, which is influenced not only by viral load, but also by host and viral factors. Serial testing would allow a more accurate assessment of an individual's immediate risk of developing an OI, intermediate risk of progression, and by extension, the need to initiate therapy.

Can results of clinical trials based on surrogate markers be used to infer survival advantage?

The primary endpoints in recent clinical trials have been changes in viral load and CD4-cell counts, not clinical endpoints. In several other settings, it has been shown that short-term effects on surrogate markers do not always predict clinical efficacy. Despite the undeniable advantages of this "fast-track" process for testing drugs and regimens, this approach fails to provide definitions for durability, therapeutic success, or therapeutic failure. Since there are no data on the effectiveness of treating early HIV infection, these definitions are crucial to our ability to infer that short-term positive effects on surrogate markers will translate into survival advantage.

What constitutes a "durable response"?

In HIV clinical trials, the duration of virologic response is often measured in weeks, and a "durable response" is typically considered to be one that lasts 24 to 48 weeks. However, HIV infection often spans more than a decade, even if untreated. These different time scales may lead to disparities between durable responses based on short-term reductions in viral load seen in clinical trials and survival advantage in actual clinical practice. In HIV care, failure rates are substantially higher in practice than in trials. For example, a recent report from the Johns Hopkins outpatient clinic showed that over 60% of patients on PIs experienced virologic failure after one year. There are many reasons for this disparity, including the exclusion from trials of patients with comorbidities and the inclusion of persons more likely to adhere to therapy than the general population of HIV-positive patients.

What constitutes therapeutic success versus therapeutic failure?

In clinical trials, success is often defined as a reduction in the viral load to below the level of detection. However, viral replication and evolution appear to occur even when viral load is undetectable. The long-term clinical significance of ongoing viral replication, despite a patient's having achieved an undetectable viral load, is unclear. Therapeutic failure is also often defined in virologic terms. When the most sensitive available tests are used, a smaller proportion of patients reach undetectable viral loads and a considerable proportion of these patients experience short-term viral-load rebound. Thus, researchers may judge many patients to have failed on their prescribed therapy, even though the clinical significance of detectable low levels of plasma HIV RNA and short-term rebounds are not known. In fact, in the Swiss Cohort study, prognosis was very similar after up to 30 months of follow-up for complete responders and for transient responders. The definitions of therapeutic success and failure are based on extrapolations from trials of later-stage HIV disease, making their application in early-stage HIV infection potentially problematic, especially when one considers the differing definitions of durable response and the disparities between the results of clinical trials and what is observed in clinical practice.

Why does virologic failure occur?

Virologic failure can occur because of insufficiently potent regimens, the presence of secluded reservoirs, the presence of resistant HIV, or a failure to adhere to the prescribed regimen. Lack of potency was responsible for the failure of monotherapy in the long term, despite its short-term successes. The study of the role of potentially secluded reservoirs of HIV virus is an evolving field. Resistance can play a decisive role in virologic failure. However, one of the most important recent findings is that early virologic rebound may not always be associated with phenotypic or genotypic drug resistance.

Past one year, patient adherence to any prescribed medication averages 50% at best. However, some data indicate that >95% adherence to prescribed medications is essential for the virologic success of PI-containing regimens. In addition, both physicians and patients frequently overestimate adherence. This lack of adherence is partially responsible for the disparity between the high efficacy of PI-containing regimens in clinical trials and the more disappointing results observed in clinical practice. Interestingly, randomly measured plasma drug levels in recent studies suggested that lack of adherence might not always be the cause of virologic rebound with sensitive viruses. Thus, in many instances the reasons for virologic failure are still unknown and need to be elucidated.

Does any detectable viremia necessarily predict a bad outcome?

In clinical practice, a considerable proportion of patients not only do well clinically despite virologic failure, but continue to experience further increases in CD4-cell count. Data from patients with discordant responses indicate that, at least for one to two years, the CD4-cell count response may be better than viral load for predicting prognosis. In addition, even for patients with low CD4-cell counts, the development of OIs during up to five years of follow-up is relatively uncommon with viral loads <10,000 copies/ml. Furthermore, after viral rebound, viral loads do not necessarily return to the set point, and some patients stabilize at relatively low levels. Finally, it has been reported that interruption of antiretroviral therapy is associated with a fall in CD4-cell count regardless of the previous level of viremia. These data suggest that full suppression of viral load is not required to derive at least some immunologic benefit and that even without suppression of viremia, antiretroviral therapy may have a prolonged effect on CD4-cell counts. This, in turn, suggests that PIs may exert benefits independent of their effects on viral load or that escape viruses are less pathogenic.

Does viral rebound always mandate a change in therapy?

The recommendation to change therapy once viral-load rebound occurs is based on the assumption that viral replication is almost exclusively controlled by antiretrovirals. If this were true, once rebound occurs, plasma levels would in time inevitably return to the set point. However, in a still undefined but probably considerable proportion of patients, viral rebound is followed by at least temporary stabilization of viral load at levels below the set point. The size of the pool of potentially infected, replication-competent cells is probably similar in most chronically infected adult patients. Thus, individual differences in viral load must be at least partly due to the capacity of each individual's immune system to modulate viral production and clearance. As a consequence, the magnitude of the rebound must also be proportional to the effectiveness of both the patient's immune system and his or her antiretroviral regimen.

There is evidence that T-cell-mediated cytotoxicity (CTL) plays an important role in controlling HIV viremia. As a general rule, long-term maintenance of CTL requires continued exposure to antigens. Thus, in theory, if antiretroviral therapy were capable of limiting replication to extremely low levels, but not of eradicating infection, CTL response could be jeopardized. Several recent observations appear to corroborate this hypothesis. HIV-1-specific CD4+ T cells are detectable in most HIV-infected patients, but decline with prolonged viral suppression. Levy reported that individuals who elected to stop therapy after several years on PI-containing regimens experienced viral rebound, but the plasma viral load was often higher than when these individuals began therapy. Lori et al. reported that intermittent drug therapy increases the time to viral load rebound, and Ortiz et al. showed that, for recently infected individuals, nonadherence and drug interruptions were generally associated with increased CTL activity. In summary, the long-term immunologic consequences of suppression of viral replication and of a return to viral detectability need to be more thoroughly evaluated.

What is the long-term safety of the newer antiretrovirals?

Long-term side effects of the newer drugs, including abnormalities of fat distribution, hypercholesterolemia, hypertriglyceridemia, insulin intolerance, and diabetes mellitus, are being increasingly documented. Although reported to occur in patients treated with other antiretrovirals, these side effects appear to be more common in patients treated with PIs. The pathogenesis, frequency, and long-term clinical consequences of these metabolic complications, as well as the psychological effects of the characteristic body disfigurations, are still unclear.

In general practice, approximately 40% of the patient population have co-morbidities. Thus, a large proportion of HIV-infected patients would be expected to need treatment for unrelated conditions during the long course of their infection. Significant pharmacokinetic interactions between antiretrovirals (especially PIs and NNRTIs) and other drugs are common. Thus, the treatment of unrelated illnesses may be greatly affected by the use of antiretrovirals. The potential seriousness of the side effects associated with the use of potent antiretrovirals and their potential interference with the treatment of other illnesses must be factored in when devising guidelines for their use.

Are "less potent" regimens still acceptable?

For patients with advanced HIV infection, the most potent available antiretroviral therapy should be started as soon as possible. The more difficult question relates to patients with a low risk of immediate disease progression. Is there any evidence that, for these "early" patients, "less potent" therapies that are easier to take and whose long-term safety profile is better known, may be more appropriate?

There is substantial evidence for a survival benefit of dual-NRTI therapy. In a large observational study conducted in Europe, rates of progression to AIDS or death in patients with CD4 counts <200 cells/mm³ were virtually identical for those on PI-containing regimens and those on dual-NRTI therapy. In addition, recent studies have shown efficacy of NRTI-based, non-PI, non-NNRTI-containing regimens. For instance, in a study in Thailand comparing AZT + 3TC + ddI with AZT + 3TC in treatment-naive patients, the mean increases in CD4 count at 48 weeks were 118 cells/mm³ for those on triple therapy and 73 cells/mm³ for those on dual therapy. Most importantly, 55% of those on triple-NRTI therapy had viral loads <50 copies/ml, a result similar to what has been reported for PI-containing regimens. These and other data suggest that, at least for the first year of treatment, non-PI-, non-NNRTI-containing regimens may be comparable to PI-containing ones as initial therapy.

One of the key factors favoring the use of less potent regimens in some patients is the fact that, in some settings, these regimens may perform better in the long run than more potent ones. In a mathematical model of patients with intermediate risk of HIV-disease progression based on assumptions derived from published data, some non-PI-containing regimens performed equally well or even better than PI-containing regimens at the end of five years of follow-up. These results are explained by the high failure rates caused by non-adherence, limited salvage-therapy options, and poor response to such options in the subgroup whose initial treatment included a PI.

How does one determine optimal initial therapy?

Most of the commonly used regimens have not been directly compared with each other. Therefore, the choice of a particular strategy is based on cross-study comparisons of results. Apart from the obvious limitations of comparing data obtained from different populations under diverse study conditions and selection criteria, the effect of the methods of analysis and of reporting results on the conclusions may not always be fully taken into consideration. For example, a re-analysis of 12 recently reported trials, which used different thresholds to define undetectable viral loads and different methods of reporting, showed that seemingly minor methodological differences may lead to extremely different results. Depending on the method of analysis (e.g., on-treatment vs. intent-to-treat) and on the sensitivity of the viral load assay used, the estimated efficacy of a particular regimen could range from <50% to >90%. Thus, some form of standardization will be essential for making more reliable comparisons of treatment effects in different clinical settings.

Conclusion

The main objective of antiretroviral therapy is to maintain or restore quality of life and prolong survival. There are no published data evaluating the effects of the various possible therapeutic strategies on survival of patients in the early stages of HIV disease. We believe that the present guidelines for the treatment of asymptomatic HIV-positive patients are based on the rather simplistic assumption that the pathogenesis of HIV infection chiefly involves killing off infected cells. In addition, these guidelines fail to acknowledge that, regardless of the regimen chosen, a considerable proportion of patients eventually fail on initial therapy because of the reasons outlined above. Thus, especially for patients at the earliest stages of HIV disease, the ultimate success of antiretroviral therapy will likely depend on the aggregate effectiveness of sequential therapies, not solely on the potency of a particular regimen used as initial therapy.

HIV pathogenesis involves an extraordinarily complex and dynamic interaction between the patient's immune system, the HIV strain in question, and the drugs used. Given the long natural history of HIV infection, the question of when to initiate therapy and with what regimen in patients at the early stages of HIV disease should be viewed as a balancing act between the potential benefits of therapy and its drug toxicities and interactions with other medications, its limiting effects on future therapeutic options, and its interference with the patient's lifestyle. Thus, for patients with a low immediate risk of disease progression, postponing the introduction of therapy or initiating therapy with a less potent regimen (e.g., dual NRTIs) could be more effective in the long run (i.e., lead to longer survival) than immediately starting therapy with the most powerful combination of drugs available (e.g., a PI-based regimen). We hope that our review and discussion of some of the still unanswered or partially answered questions regarding HIV pathogenesis and treatment will stimulate and expand the debate about the design of studies that may help clarify some of these points. Such studies, in turn, should help in the formulation of future treatment guidelines.

— Mauro Schechter, MD, PhD, and Lee H. Harrison, MD

Dr. Mauro Schechter is Professor of Infectious Diseases and Head of the AIDS Research Laboratory at Hospital Universitario Clementino Fraga Filho, Universidade Federal do Rio de Janeiro. Dr. Lee Harrison is Associate Professor of Epidemiology and Medicine at the University of Pittsburgh.

Published in Journal Watch HIV/AIDS Clinical Care October 1, 1999

Citation(s):

Report of the NIH panel to define principles of therapy of HIV infection. MMWR Morb Mortal Wkly Rep 1998 April 24 47 RR 5 1-41.

• Medline abstract (Free)

The AVANTI Steering Committee. Analysis of HIV-1 clinical trials: statistical magic? Lancet 1999 June 12 353 2061-2064.

• Medline abstract (Free)

Cooper DA and Emery S. Therapeutic strategies for HIV infection -- time to think hard. N Engl J Med 1998 October 29 339 1319-1321.

• Original article (Subscription may be required)
• Medline abstract (Free)

Engels et al. Plasma HIV viral load in patients with hemophilia and late-stage HIV disease: A measure of current immune suppression. Ann Intern Med 1999 Aug 17 131 256-264.

• Original article (Subscription may be required)
• Medline abstract (Free)

Fleming TR and Demits DL. Surrogate end points in clinical trials: Are we being misled? Ann Intern Med 1996 125 605-613.

• Original article (Subscription may be required)
• Medline abstract (Free)

Ledergerber B et al. Clinical progression and virological failure on highly active antiretroviral therapy in HIV-1 patients: A prospective cohort study. Swiss HIV Cohort Study. Lancet 1999 March 13 353 863-868.

• Medline abstract (Free)

Levy JA. Caution - should we be treating HIV infection early? Lancet 1998 September 19 352 982-983.

• Medline abstract (Free)

Lucas GM et al. Highly active antiretroviral therapy in a large urban clinic: Risk factors for virologic failure and adverse drug reactions. Ann Intern Med 1999 July 20 131 81-87.

• Original article (Subscription may be required)
• Medline abstract (Free)

Lyles R et al. Natural history of HIV-1 viremia after seroconversion in the multicenter AIDS cohort study. 6th Conference on Retroviruses and Opportunistic Infections, Chicago, Ill, January 31-February 4 1999 abstr. 273 .

Schechter M et al. Protease inhibitors as initial therapy for individuals with an intermediate risk of HIV disease progression: Is more necessarily better? AIDS 1999 13 97-102.

• Medline abstract (Free)

Schneider MME et al. Discontinuation of prophylaxis for Pneumocystis carinii pneumonia in HIV-1-infected patients treated with highly active antiretroviral therapy. Lancet 1999 January 16 353 201-203.

• Medline abstract (Free)