Scientific and Regulatory Issues in the Era of Antiretrovirai Polytherapy and a Viral Load-Driven Standard of Care
A Report to the FDA Antiviral Drugs Advisory Committee
14 July 1997
by Spencer Cox, Paul Dietz, Mark Harrington & Theo Smart
Edited by Spencer Cox & Mark Harrington
PREFACE & INTRODUCTION
by Spencer Cox
Two comparisons seem to come inevitably to the minds of writers when they survey the past two years of AIDS research, with the enormous advances that have been made in the theory and practice of AIDS therapy: the first is the miraculous recovery of the biblical Lazarus, and the second is Alice in Wonderland, desperately trying to make sense of a world in which everything changes by the minute. These conflicting metaphors might also describe the position of advocates for people with HIV, trying to determine how patients can get maximal benefit from new drugs and new data, without entirely giving up our ability to continue to get more information about the optimal use of therapies.
This report summarizes development of several selected new anti-HIV drugs, both recently approved (delavirdine and neflinavir) and currently in advanced development (abacavir, adefovir dipivoxil, DMP-266 and GWI4I). In addition, comments are offered regarding proposed changes in the process by which anti-HIV therapies are regulated by the US Food & Drug Administration (FDA). No pretense is made that this report is comprehensive; indeed, the authors do not believe that, at present, enough data are available to make comprehensive judgements about how clinical trials ought to be conducted in this rapidly-changing era of polypharmacy and quantitative viral load measurements. In general, we endorse the trend towards assessment of new therapies based on HIV RNA levels. However, we believe that much work still needs to be done to define regulatory standards which will meet the needs of people living with HIV.
This policy shift represents a significant change for the Treatment Action Group (see Cox 1994), and it has occasioned much discussion amongst our membership. We recognize that there are important arguments both for and against RNA-based assessment of therapeutic efficacy, and we believe that those arguments need to be seriously examined in open public fora. For that reason, we are seriously concerned that the FDA has chosen not to arrange for presentations to the Antiviral Drugs Advisory Committee regarding the historic failures of surrogate markers, including the recent finding that, despite increased clearance of MAC bacteremia, high-dose clarithromycin prophylaxis increased mortality among people at risk for Mycobacterium avium complex (MAC). Safety concerns must be one of the primary issues in considering this transformed regulatory standard.
Nonetheless, we have confidence that these vital discussions can occur, that protections against unsafe and ineffective drugs can be maintained, and that the relevant standards for approval can be altered in ways that improve the utility of data generated by clinical trials. We do not elieve that this will be an easy task, but we look forward to working with FDA, industry, researchers and AIDS advocates to ensure that HIV-inrected individuals can expect continued improvements in their prospects for disease-free survival.
Last month, the Public Health Service published a new draft set of “Guidelines for the Use of Antiretroviral Therapies in HIV-infected Adults and Adolescents.” This document marks the consolidation of a revolution in HIV therapy that has occurred over the past several years (HHS 1997). The improved precision in therapeutic decision-making offered by quantitative viral load tests, and the increased efficacy of treatment have led most researchers to agree that treatment should be initiated considerably earlier in the course of disease, and should aim to reduce plasma HIV RNAto levels as low as possible, ones not detectable by current assays (NIH 1997).
These changes in clinical care necessarily impact the design and conduct of AIDS-related clinical trials. However, the catch is that researchers have broad disagreements about how, specifically, studies need to change: for example, in Jon Cohen’s recent article from Science Magazine, Dr. Douglas Richman, from the University of CalifomiaySan Diego asserted that “Anything that is not designed to completely suppress viral replication is suboptimal.” As a consequence, Dr. Richman and other have opined that studies using two-nudeoside control arms are unethical, exposing patients to undue risk of drug resistance. Their arguments, based on new data about the biology and quantitative dynamics of HIV infection, are often compelling. However, other leading researchers, such as Dr. Michael Saag from the University of Aabama/Birmingham have suggested that only long-term studies, looking at the clinical effects of different “strategies” for treatment of HIV infection, using different combinations of drugs (including, possibly, two-nudeoside regimens) in different sequences, can truly determine how best to treat the disease. Dr. Saag argues that the biology of the disease is only one consideration in determining optimal use of anti-HIV therapy, and that the safety and efficacy profiles of the different therapies, the biological and behavioral variations among patients, and the possible negative sequalae of therapy need to be accounted for in determining how best to make use of potent new drugs (Cohen 1997).
This dilemma was starkly highlighted not long ago when the AIDS Clinical Trials Group (ACTG) set out to design it’s first “strategy trial,” known as the “Strategic Timing of Antiviral Therapy” or START trial. The ACTG leadership agreed that a strategy trial was called for. However, the study proposal, which was designed by Dr. Saag and other top ACTG investigators, was rejected when the leadership could not agree on which strategies for initiation of therapy should be evaluated. Indeed, some investigators felt that the biology of HIV infection had already answered the question: treatment should be initiated as soon as an HIV-infected patient could be persuaded to begin.
Historically, regulatory requirements for new therapies and the clinical care of HIV-infected patients have not diverged greatly. Sequential monotherapy was the clinical standard, and tools for monitoring therapeutic success or failure were extremely limited. In patients with relatively advanced HIV infection, clinical events were likely to occur with depressing frequency, even in the face of antiviral therapy, ensuring that the benefits of therapy could be measured with clinical endpoints.
Now, however, a great divergence has begun to occur. Therapy appears to be becoming capable of delaying illness and death for substantial periods of time, and viral load assays allow an individual patient’s response to therapy to be evaluated rapidly. Drugs are used earlier in the course of infection, and in combinations, which imposes difficulty in assessing the contributions of particular product to therapeutic efficacy. Maintenance of patients on randomized, blinded, properly-controlled trials long enough to evaluate therapeutic effects has become severely problematic — especially when control arms include suboptimal therapy. The requirements of patient care have come into conflict with the goal of substantiating claims about particular treatments.
For some researchers, such as Dr. Joep Lange from the University of Amsterdam, this conflict necessarily implies that the regulatory goals should be abandoned. Dr. Lange believes that the goal of future research should be to “identify maximally suppressive therapeutic strategies that will confer the greatest and longest immunological and clinical benefit at lowest toxicity and cost.” (Lange 1997). Unfortunately, while Dr. Lange has given us an excellent description of current thinking about how to assess optimal care for patients, he has not described an algorithm for assessing the claims made by pharmaceutical companies about their products. The US regulatory system has evolved as it has in order to ensure that companies are not able to make false or misleading claims about the particular health technologies that they manufacture. As we have recently seen with several anti-HIV drugs, such as the cross-resistance profile of Invirase™ brand saquinavir, and the prophylactic effects of high-dose Biaxin™ brand darithromycin against Mycobacterium avium Complex (MAC), such evaluation remains important. Had the companies been allowed indefinitely to make claims based on preliminary data about these drugs, the products would continue to be misused, causing harm to patients.
In response to the changing clinical arsenal, many researchers, regulators, and patient advocates have proposed to move towards a regulatory standard based on quantitative measurements of plasma HIV RNA levels and circulating CD4 cell counts. Dr. Richman observes, “All this talk of HIV RNA and CD4 being surrogate markers, that has always bothered me. They are not surrogate markers. They are the measurement of the disease.”
Such a change seems inevitable given the limitations imposed by today’s standard of care. The conduct of ACTG 320-style studies, in which patients are maintained on treatment regimens while showing virologic – and later clinical – failure seems both practically and ethically untenable. However, alterations in the regulatory standard should account not only for the needs of patient care, but also for the requirements of sound product regulation. As such, these changes need to be carefully thought out, with a close eye on the law of unintended consequences. Changes in the regulatory standard that are ostensibly intended to be responsive to patient care could have the consequence of damaging patient care by permitting unsubstantiated claims about products, leading to misuse of drugs based on preliminary data.
After discussions with industry and AIDS advocates, FDA has suggested a proposal that would permit full marketing approval based on evidence of a durable virologic response to therapy. This proposal seems to change with some rapidity, however a number of suggestions were made during a meeting with AIDS advocates on May 16th, 1997:
- Evidence that a drug confers a reduction of 0.5 log 10 in plasma HIV RNA levels at 16-24 weeks would be sufficient for accelerated approval.
- Evidence that a drug, when used in an appropriate combination regimen, confers a sustained reduction in plasma HIV RNA levels to 48 weeks would be sufficient for full approval.
- Trials demonstrating surrogate response would need to include at least 200 patients treated with the study drug.
- Evidence of safety would be required in at least 300 patients followed for at least six months.
The proposal at the time did not discuss various alternatives for designing RNA-based confirmatory studies. However, various alternatives for such studies address different areas of concern with current trial design, and impose differing obstacles to optimal implementation. For instance, although in May it seemed that FDA had settled on magnitude of virologic response as the optimal measure of therapeutic efficacy, many researchers believe that time to virologic failure may be the appropriate measure of therapeutic efficacy. Depending on how virologic failure is measured, studies may need to become larger and longer, rather than smaller and faster as industry desires. The Antiviral Advisory Committee should carefully consider the different aspects of this proposal, as the implications of particular choices will be important to the design and conduct of future studies.
In particular, the Treatment Action Group believes that the Committee should carefully consider the following questions:
I. How should virologic response be measured? During the period in which FDA relied on CD4+ cell count changes to assess response to therapies for accelerated approval, the measures of CD4 cell count differed from therapy to therapy – in particular, companies looked at both absolute CD4 cell count changes and changes in Area Under the Curve (AUC). Because the agency did not define a standard measure of immunologic response, companies were allowed to perform a number of tests, and to selectively rely on measurements supporting their proposed claims. Before moving to an RNA-based standard for clinical efficacy, the agency needs to define what it means by a virologic response.
a. A number of different measures are available to determine therapeutic effects on plasma HIV RNA levels, including:
i. Magnitude of initial virologic decline
ii. Percentage of patients with at least an O.SIogl 0 decline in plasma HIV RNA levels
iii. Percentage of patients who achieve plasma HIV RNA levels below the limits of detection
iv. Percentage of patients with undetectable plasma HIV RNA levels at a defined time point, or for a defined duration
v. Time to nadir of virologic decline
vi. Time to re-appearance of detectable virus
Each of these measurements have different implications for trial design. For instance, FDA has also suggested that “time to virologic failure” may be the most appropriate measure of clinical response. However, the recent PHS Guidelines list four criteria for virologic failure:
i. Less than a 10-fold (1.0 log) reduction in plasma HIV FUMA by 4 weeks following initiation of therapy,
ii. Failure to suppress plasma HIV FUSJA to undetectable levels within 4-6 months of initiating therapy,
iii. Repeated detection of virus in plasma after initial suppression to undetectable levels, suggesting the development of resistance,
iv. Any reproducible significant increase, defined as 3-fold or greater, from the nadir of plasma HIV FUsIA not attributable to intercurrent infection, vaccination, or test methodology;
In addition, the guidelines also note that persistently declining CD4 T cell numbers as measured on at least two separate occasions, or clinical deterioration may also suggest therapeutic failure in some cases, even in the face of apparent virologic success. These different measures need to be considered, along with their implications for trial design. For instance, measurement of the absolute magnitude of the initial virologc response may have the advantage of being measurable over the short-term in a relatively small number of patients, but may lack long-term predictive capacity, and may be confounded by the assay limits of detection.
RECOMMENDATION I: FDA should define a standard measure of changes in HIV RNA levels to be used in assessing clinical efficacy.
I. What is the minimal evidence of virologic response needed to imply efficacy?
Again, the historical failure of FDA to define minimal criteria for evaluating a CD4+ cell response to treatment meant that therapies were approved based on as little as a ten-cell improvement over AZT monotherapy in a patient population that had been heavily pre-treated with AZT. Assuming that RNA changes are partially predictive of clinical response, the benefit offered by large virologic improvements over acceptable control arms – such as those offered by AZT/3TC/indinavir as compared to AZT/GTC – is unlikely to be overwhelmed by minor or infrequent adverse events. On the other hand, when virologic improvements are small, those differences may easily be swamped by a rare, serious adverse event. FDA would be well-advised to begin by setting minimal criteria for virologic response that is well above the estimate of meaningful clinical changes, allowing a “buffer zone” to account for unidentified adverse events. Therapies that do not meet this minimal standard would still be eligible for approval based on evidence of decreased rates of illness and death in treated patients.
RECOMMENDATION 2: FDA should define a rigorous minimal standard for evidence of virologic response suggesting clinical efficacy.
What is the minimal size of the safety database? Disease and death are responsive not only to therapeutic efficacy, but also to the impact of serious adverse events. Clinical endpoint studies allow us to measure the aggregate effects of diverse biological properties of therapy. If FDA is prepared to sacrifice this measure, then adverse event monitoring needs to be carefully considered. Randomized, controlled trials should be supplemented by an observational database, possibly conducted in the context of an expanded access program, to provide longer-term and potentially more detailed information about serious adverse events associated with therapy. The Treatment Action Group does not believe that 300 people followed for six months is sufficient to allow for the kind of safety assessment required in the absence of clinical endpoint studies, as early experience with ddl pancreatitis and more recent discoveries about protease inhibitor-associated diabetes indicate.
RECOMMENDATION 3: FDA should define minimal standards for the size of the safety database required for approval that is more rigorous that what has been proposed.
What supportive data are needed? In the absence of clinical endpoint data, FDA should use this opportunity to increase the amount of supportive data required – particularly data regarding interactions between the regulated product and products with which it is likely to be coadministered. For instance, both approved non-nudeoside reverse transcriptase inhibitors were approved without substantive safety data on use in combination with HIV protease inhibitors, and Glaxo-Wellcome is currently testing GW-1592 mainly in combination with other products manufactured by Glaxo-Wellcome. If, as is expected, anti-HIV drugs will be used in combination with a diverse array of other anti-HIV products, then minimal pharmacokinetic interaction studies need to be conducted with those products, and more substantial safety and activity data should be generated when an interaction is identified, unless the combination is clearly contraindicated. In addition, interaction data should be available on drugs commonly used to prophylaxand treat opportunistic infections that afflict HIV-infected patients, as well as on anti-anxiety and anti-depressive medications, birth control medications, and methadone.
RECOMMENDATION 4: FDA should require a more comprehensive package of interaction data for drugs approved based on plasma HIV RNA improvements.
What incentives exist to encourage follow-up studies? While FDA’s primary mission is the evaluation of claims regarding the safety and efficacy of particular products, the agency has also recognized the proper safety evaluation continues in the post-marketing setting. If the quantity of data available at the time of approval is to be reduced, then the agency should clearly consider mechanisms, such as limitations on labeling at the time of initial approval, to encourage the conduct of post-marketing studies that will continue to elucidate the safety and efficacy of products as components of different therapeutic “strategies.”
RECOMMENDATION 5: FDA should shape regulations to encourage post-marketing evaluation of therapeutic safety and efficacy.
What decisions should be made in public? Historically, decisions regarding the approval of particular products have influenced overall interpretation of regulations. For instance, when ddC was approved, both the manufacturer and members of the Antiviral Drugs Advisory Committee cited the earlier approval of ddl to support the approval of ddC: both argued that, because allowances had been made for ddl, justice dennanded that it be made for ddC. However, ddC was a different drug, with a different safety and efficacy profile from ddl. The willingness of the Antiviral Drugs Advisory Committee to make commitments regarding acceptable minimal requirements for approval (as was done with 3TQ in private meetings with the company compromises the ability of patient advocates to effectively support patient interests.
RECOMMENDATION 6: Substantive commitments regarding the adequacy of a data set for marketing approval should be made in open public hearings.
Cox S, Davidson D, Gonsalves G, Harrington M, Hogan C, Pringle Smith R. Rescuing Accelerated Approval: Moving Beyond the Status Quo. Treatment Action Group (TAG), New York, 1994.
Cohen J. AIDS Trials Ethics Questioned. Science 276:520-523, 25 April 1997.
HHS, Guidelines forthe Use of Antiretroviral Agents in HIV-infected Adults and Adolescents, June 1997.
Lange JMA Current Problems and the Future of Antiretroviral Drug Trials. Science 276:548-550, 25 April 1997.
Mellors JW, Munoz A, Giorgi JV, etal, Plasma Viral Load and CD4+ lymphocytes as Prognostic Markers of HIV-1 Infection. Ann Intern Med, 15 June 1997.
NIH, Report of the NIH Panel to Define Principles of Therapy of HIV Infection, June 1997.
To read the rest of this report, please download the this Report as a pdf.