AIDS TREATMENT NEWS No. 110 - September 7, 1990
John S. James

Dozens of potential AIDS antivirals are now coming out of laboratories, and there is no way that all could get the full-scale clinical trials required to convince the FDA that the drugs are good enough for general use. Even if more money were available, there are not enough experienced scientists, research nurses, or patients meeting entry criteria, to run so many large trials. The proposed new system could quickly screen all the promising drugs, so that the truly important ones could quickly be moved into larger, definitive trials.

The importance of the Statistical Working Group is that the people needed to make the proposal work are part of the process, so most issues could be resolved immediately. At a key meeting on July 10, 1990, statistical experts from NIAID (the U.S. National Institute of Allergy and Infectious Diseases), made sure that ideas were scientifically acceptable. Sometimes they did not know what the FDA would think, but then they could ask Ellen Cooper, M.D., head of the antiviral division of the FDA, who was at the same meeting. And the AIDS activists there knew what was or was not likely to be acceptable to patients. Also at the meeting were statisticians from the U.S. National Cancer Institute, NIAID's statistical contractors from Harvard and the University of Minnesota, and at least two pharmaceutical companies involved in AIDS trials, Bristol-Myers Squibb and Hoffmann-La Roche.

Pharmaceutical companies seldom run clinical trials to compare drugs, because each company only develops its own, and at any given time seldom has more than one drug at the same stage of development for the same disease. The screening trials must be sponsored by a neutral party, for example a government agency or a community-based or academic research organization. Pharmaceutical companies must be consulted early, however, because they will be involved when their drugs are tested. There is concern that some companies may be reluctant to put their drugs to an early test.

The Statistical Working Group would not have succeeded without good ideas going in. The basic elements of the screening trials were already familiar to the scientific community, as a similar system has been used successfully in cancer research.

How the Trials Will Work

(1) Outcome Measurements

For over a year, scientists have debated the use of viral or immunological "markers" to tell if a drug is working against HIV. At this time the scientific mainstream has not been willing to accept improvement in available markers, like T-cells or p24 antigen or antibody, as good enough proof of efficacy to lead to drug approval. As a result, trials take years instead of weeks, because in order to prove a treatment works, enough people must die or suffer disease progression to generate statistically significant numbers in favor of the treatment. (Note that even a perfect drug which cured instantly would also take a long time in this system, since the delay depends on what happens to the volunteers who are not treated.) Also recruitment is difficult when patients need to enter a long (e.g., two year) trial in which their treatment must be controlled by a protocol written in advance. Such trials can require long-term sacrifice of medical options which otherwise would have been chosen.

However, the anti-HIV drugs which have been tested and are believed to benefit patients -- AZT, ddI, and ddC -- all cause average T-helper cells to rise within a short time, usually eight weeks of starting the treatment. There is no reason to believe that these drugs themselves raise T-cells directly; instead they appear to be inhibiting HIV, allowing the immune system to recover. Therefore other drugs which inhibit HIV at least as well would also probably cause the T-cells to rise, within the same time frame. By using T-helper count not as definitive proof of efficacy but only as a preliminary indication, the Statistical Working Group sidestepped the currently-deadlocked debate on whether such markers are good enough for final, definitive trials.

(2) Placebos

Brainstorming among clinical-trials experts at the July 10 meeting led to an idea for avoiding placebos in the screening trials without sacrificing their scientific power. The first suggestion was to test many drugs at the same time. Most proposed drugs will not work, so they will in effect be the placebo. As one scientist put it, if five drugs are being tested, then if there are one or two winners, these should stand out from the others (in T-helper improvement). And if by luck one of the drugs was a superwinner, that would be obvious, too.

The trials would be kept short, probably two or three months, to avoid exposing participants to ineffective drugs for long.

Alternatively, in some cases a placebo or no-treatment arm might be acceptable in these trials. Since many drugs would be tested, patients would have only a small chance, probably less than one in five, of receiving no treatment. And the trial would not last long. Persons who were not critically ill might be willing to risk entering a no-treatment arm for two or three months.

(3) Blinding

An administrative problem with a trial comparing many drugs is the difficulty of making all the pills look the same, to maintain "blinding" -- not let patients, physicians, or staff know who is getting what drug. But recent thinking among clinical-trials experts is de-emphasizing the importance of blinding for studies with objective outcome measures like T-helper counts.

What is important is randomization -- assigning patients randomly among different arms in the trial. If volunteers choose which drug they will get instead of being randomly assigned, it would be impossible to be sure there were no hidden biases caused by certain groups of patients, with better or worse prognosis, choosing certain drugs. Without randomization it would be impossible to fully trust the result.

If trial designers can avoid blinding the rapid-screening trials -- no decision has yet been made -- it would greatly simplify trial administration, especially since these trials will be testing many drugs.

(4) How Many Patients?

Statisticians are still working out the details, but preliminary discussions suggested that trial size might be about 30 to 50 patients for each drug being tested.

(5) Efficiency:

Rapid Screening Trial and Master Protocol This system would not need to be limited to comparing any particular group of drugs. New drugs could be added at any time, provided there were enough volunteers. Drugs would be finishing at different times, as their respective study arms were filled and the volunteers completed the two- or three-month test.

Administration could be greatly streamlined by use of a master protocol -- a concept which has been used in cancer research. The protocol for the trial is developed in advance, even before all the drugs to be tested are known. With each new study arm, only the drug and method of administration must change. Everything else -- entry criteria, blood work, data collection forms, data analysis, etc. -- stays the same.

Of course each new drug would have to be approved by the FDA, and by all IRBs (institutional review boards) involved. But it's much easier and faster to approve just one drug, under the master protocol, than an entire study design.

Another advantage of using a master protocol is that all the different drugs can be compared, because entry criteria, measurements made, etc. are the same. Because the epidemic changes over time, exact comparisons can only be made between two or more treatments when patients have been randomized among them at the same time. But even when some drugs become available later and are tested months or even a year or more after others, comparisons between them will be much better than comparisons between drugs tested in different, unrelated studies. To a large extent, each new drug tested could be compared with all previous ones in the same master protocol.

Overall Picture

This drug-screening system would be testing several different drugs at any one time. Any volunteer who met the entry criteria would be randomized to one of the currently available drugs, and take it for a short time, probably about 12 weeks. About 30 to 50 patients would be given each drug.

Improved T-helper count would be the primary measure of efficacy, although other data would of course be collected and analyzed. The drugs would be compared to pick those few which stood out from the others. These would quickly go into larger, more conventional trials designed for fast-track drug approval. They might also be made available for early access, through a system like parallel track.

Interviews with the Developers

We attended the July 10 meeting, but this article is based mainly on phone interviews with some of the people who have developed the screening-trial concept. There are different versions of the idea, however, and to simplify this presentation we combined them into the unified picture above.

So far only one article has been published on this rapid-trial idea -- "New Screening Proposal Could Speed AIDS Drug Trials," by Sari Staver, American Medical News (published by the American Medical Association), August 10, 1990. Ms. Staver interviewed Daniel Hoth, M.D., director of the AIDS program at the National Institute of Allergy and Infectious Diseases, and Ellen Cooper, M.D., from the FDA, who was at the meeting. Both endorsed the idea. Dr. Hoth said it might be implemented in 12 to 18 months.

We spoke with Susan Ellenberg, Ph.D., Chief of the Biostatistics Research Branch of the Division of AIDS at NIAID. She is one of the organizers of the Statistical Working Group, and clarified several points about the rapid screening trial idea:

* It is still in the discussion phase, primarily among statisticians. Physicians and pharmaceutical companies still need to be convinced of its value.

* Testing four or five drugs at at time, in a short trial looking at virological or immunological markers for a preliminary screen, seems "a very sensible idea; clearly many people were interested." While some concerns were raised at the meeting, the general consensus was positive; no one said the idea could not work.

* The main concern is that some drugs may be valuable but not show benefits early. Pharmaceutical companies may be afraid that their products could be rejected too early. But any drug showing potential would be followed up; the screening trials are to help set priorities, not to kill drugs which have other evidence in their favor.

Another key person in the development of this rapid screening proposal is David Schoenfeld, Ph.D., a statistician at Harvard Medical School and Harvard School of Public Health, who has worked with a similar system in cancer research for several years. The cancer trials tested many drugs for melanoma, brain tumors, and other cancers. They used tumor size as the outcome measure, but Dr. Schoenfeld thinks that the T-helper count may be a more useful measure, potentially making the rapid screening trials even more successful in AIDS than they have been in cancer. We asked Dr. Schoenfeld about the contributions of AIDS activists in the development of the screening-trials concept. He said that the activists had been "very important and helpful in the AIDS Clinical Trials Group. They have been in a sense the initiative" for ideas like the screening trials. And they bring to the statisticians knowledge of what patients are thinking, what their concerns are. Usually the statisticians who design trials do not have contact with patients; and yet if trials are not acceptable to potential volunteers, they will not be able to recruit, and cannot be run. The activists "keep things moving in the ACTG, reminding us this is a crisis. They are very knowledgeable about clinical trials, and have good ideas."

Dr. Schoenfeld explained that a key advantage of the rapid screening trials is to speed the development of really new therapies -- that now there is no good vehicle for quickly testing them.

Another member of the Statistical Working Group saw in the rapid screening trials a rapprochement between the short-term needs of patients and the long-term needs of research. "They are short enough to ask questions without putting people in a bad situation.

For example, in a 12-week trial patients could risk a sub-active dose. Trials using means such as response-surface methodology could show optimum doses for combination therapies." She noted that the United States is not doing well at this time in trial design, and that the "pre-trial trial" of the screening system will help us minimize the consequences. (Note: response-surface methodology, a technique developed in chemical engineering, is a way of designing studies to find the best doses of two drugs used in combination.)

Memos written by Bob Huff of the Treatment and Data Committee of ACT UP/New York, and other activists in the Statistical Working Group list some advantages and problems of the concept. "A screening trial could identify early activity in a high-quality study before phase II trials begin." It would "provide information to support prioritization decisions for phase II trials," "give the earlier definition of a drug's efficacy profile," and "provide much useful information to phase II trial designers." The system would also allow combination therapies to be tested as easily as single drugs. The single study would cost less than multiple small trials. The trials could easily find volunteers, both because of the short time commitment required, and because it would be easier to publicize one trial than many separate ones.

Limitations and problems include lack of long-term data, the possibility that entry and safety criteria unique to one drug might have to be applied to others to allow randomization, and the fact that "the trial only identifies 'winners' according to the criteria of the 'AZT model,' i.e., early T-cell count increases," meaning that a better drug that takes longer to be effective might be missed. There is also the question of whether drug companies will be willing "to go head to head with competitive drugs" and support the program.

Another issue (not mentioned in the above document) is whether there are enough promising new drugs that a screening system is even needed. Today some researchers are even saying that they do not know what drugs to test. There are many scientifically promising compounds; the problem is the great bottleneck in getting them through (or even into) the pre-clinical development required for the all-important IND (Investigational New Drug approval, the FDA's permission to test the drug in humans) and then through the early phase I human test. The new proposal for screening trials will not solve this problem, which must be handled separately (for example, by improving incentives for pharmaceutical companies by facilitating earlier marketing approval for critically important drugs). Screening trials cannot overcome all the bottlenecks in medical research, but they do address one of the major points of delay.

What Happens Next?

Dr. Schoenfeld is now developing technical details of the proposal, and is preparing a paper for presentation to the next ACTG meeting in November. Notes from the July 10 meeting will also be available. Activists are circulating concept sheets among members of the Statistical Working Group for their comments.

Community-based research organizations and others should work with the rapid screening trial proposal; with more support, it may not take the predicted year or two to implement. Also, research groups which are too small to run the full system and test many drugs at once might still be able to borrow parts of it -- for example, by testing one particularly important treatment against a no-treatment arm, using the same study design which the full screening proposal will use. Few people have known about the concept until now; we hope this article helps get the word out.

To contact the Statistical Working Group, which is developing the proposal, call Bob Huff, ACT UP/New York, 212/674-8381.

900907
ATN11002

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, 1990. 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. .