AIDS TREATMENT NEWS #220, April 7, 1995

The Problem With the development of protease inhibitors and other new approaches for treating HIV disease, there are more potential treatments, and many more combinations, which are worth trying. At the same time, the resources available for research, both public and private, may be shrinking. Efforts toward small, rapid trials must be encouraged; still, it has always been difficult to get even small trials of leading- edge ideas off the ground. Usually they do not happen at all, or take place only after months or years of delay.

The main task is to address the apathy and lack of commitment that makes it so difficult to get studies done -- not to learn to live with the problem. But also, there is always a place for learning how to accomplish the most with limited resources.

We propose a program for careful, consistent monitoring of experimental treatments which are already being used by patients and physicians. The goal is to greatly reduce the time and cost of getting the first widely credible information about the effects of new treatments, combinations, and treatment strategies in people.

Background The new therapies of interest are already being tried by individual patients and physicians. Reports of results often circulate anecdotally. But for many reasons it is difficult to rely on these reports. For example: (1) Our interview with Dr. Feigal (AIDS TREATMENT NEWS #219, March 24, 1995) noted that even in the most research- intensive settings, sponsored by the largest pharmaceutical companies, there is concern that results of individual cases may be reported selectively.

(2) Most experiences with new treatments are in a clinical, not research, setting. Often there is no advance planning to assure that the data will be credible to third parties.

(3) There is little standardization among clinical physicians regarding which tests and laboratories to use, exactly when to give the tests, how to keep compatible records, etc.

(4) Even if the data is completely in order for research purposes, there is still the credibility problem of convincing others of this. It is much more difficult for each physicians' practice to establish research credibility independently, than for one research organization to build its credibility over time.

New therapies of interest are also tried in pharmaceutical company development programs. Here there are different problems: (1) Leading-edge approaches are seldom considered, since the needs of corporate drug development are different from those of patient care. Usually the best to hope for is monotherapy with one experimental agent, combined with approved drugs; for example, see the current ICC protocol [Inter-Company Collaboration for AIDS Drug Development], which at most combines one experimental drug with two approved ones.

(2) Even unimaginative trials take excruciatingly long to actually begin; again consider the ICC trials, starting only now despite the high-level attention and support long put into them.

(3) Early trials of new agents or new approaches often do not measure viral load or any other indication of activity; or they start by testing doses known to be ineffective. These trials, by design, can only be pieces in a long, expensive development process; no possible outcome could establish a rationale for human use of any treatment.

When considering options in trial design, we should be aware that certain limitations of randomized clinical trials have long been known and analyzed in the medical literature, but ignored in AIDS research.

Arvin R. Feinstein, from the Yale University School of Medicine, examined these limitations in a paper published in 1983 ("An Additional Basic Science for Clinical Medicine: II. The Limitations of Randomized Trials," ANNALS OF INTERNAL MEDICINE, volume 99, pages 544-550); Robert J. Levine, in ETHICS AND REGULATION OF CLINICAL RESEARCH, Yale University Press, 1986, recommends that this article "should be required reading for all persons who conduct or review plans to conduct randomized clinical trials." Feinstein noted that randomized trials have been successful when studying treatments to remedy an existing condition, but have "sometimes created rather than clarified controversy" when used in studies of either primary or secondary prevention of conditions which the patients do not have yet. And there has long been a seemingly irreconcilable conflict between "pragmatic" investigators (usually clinicians), who want the trial to provide useful clinical information, and "fastidious" investigators, usually biostatisticians, who want a clean study.

Feinstein also considered many ethical and logistical limitations of randomized trials. He concluded, "The scientific resolution of most future problems in clinical management will have to come from analyses of events and observations that occur in non-experimental circumstances, during the interaction of nature, people, technologic artifacts, and clinical practitioners. By recognizing the informal "experiments" that are done whenever a clinician treats a patient, and by developing better observational methods for acquiring and analyzing the data, clinical investigators can be liberated from their previous intellectual restrictions and can take the fundamental steps needed to develop an additional basic science for clinical practice." Proposal: Monitoring Project A research organization could implement a monitoring program for systematic data collection, analysis, and reporting about treatments of interest which are already being tried.

One way to both greatly decrease the cost of research, and simultaneously increase its relevance, is to study the treatments which physicians and patients have decided to use anyway. Since the actual treatment is already planned or underway, aside from the research project, the costs and obstacles involved in making a human trial happen are not a problem. Financially, the costs will be largely for lab work and administration. But the real costs of any clinical research includes not only the money, but also human costs (including increased risk and lost-opportunity cost) of asking patients and physicians do to something different from what they would have chosen to do otherwise. By studying treatments which would have been used in any case, this larger cost is also greatly reduced or eliminated.

A protocol for this project would not specify the treatment to be used, as the physicians and patients would choose that in each case. The protocol might include formal selection criteria for the treatments and the patients to be considered. But since different patients' results will not usually be averaged together, inclusion/exclusion criteria need not be severe; the real question is whether there is a clinically interesting question that might be addressed, and this decision can best be made case by case by a medical review board.

For example, such a project might study patients who are trying to lower their viral load (plasma HIV RNA) by using an experimental treatment or combination. Those entered into the study would be given specified baseline and after-treatment viral load tests -- as well as a physical exam, medical history interview, and immunological and other tests widely used in HIV trials. These results could be reported quickly; but there might also be scheduled followup which continues much longer (for example, through interviews at six months and then annually, continuing indefinitely whether or not the treatment is continued).

To avoid any question of selective reporting, everyone accepted for study by the review board would immediately be given a sequence number and be accounted for in the published reports -- even if they never received the tests, or never used the intended treatment. This is only a matter of thoroughness of reporting; it does not affect statistical results, since the different patients are usually not averaged together.

Protocol Outline The particular tests, schedules, and other details will ultimately be chosen by the researchers. This outline is intended as a basis for discussion.

Inclusion/Exclusion: Volunteers who are using antiretroviral therapy must not change that therapy for at least a month before beginning the treatment to be monitored by this program. And they must be starting a treatment which might affect the overall course of HIV disease; it could be an antiviral, or work by some other mechanism, or have an unknown mechanism. But treatments for opportunistic conditions will not be studied in this protocol.

Tests: The primary test to be monitored will be plasma HIV RNA, by the Hoffmann-La Roche quantitative PCR. Other standard panels -- immunological, hematological, blood chemistry -- will also be specified. A physical exam and medical history will be given when the volunteer enters the program, and a physical exam and exit interview (including questions on compliance) at the end of the primary monitoring period.

Visits: Ideally three baseline viral load tests will be given; on the day treatment begins, half a week (3 or 4 days) before, and one week before. Other baseline blood tests will be included as necessary.

Length of Study: The primary monitoring period will be eight weeks. After treatment begins, blood draws will be at half a week, one week, two weeks, four weeks, and eight weeks. Test results will not be batched or blinded, so that the volunteer can use them in making treatment decisions, including whether or not to stay in the monitoring program. Volunteers can change their antiviral regimen and still remain in the program.

Ideally the entire panel of tests would be given at each blood draw; however, the protocol may specify reduced testing, for economic or practical reasons.

Followup: After the primary monitoring period, the treatment being studied may or may not be continued. In either case, follow-ups will be scheduled at six months after treatment began, one year after, and annually after that. Ideally each followup would include the blood tests; but at a minimum, followup could be by phone interview.

At the end of the primary monitoring period, results would be published, or otherwise made available to other investigators and physicians.

An Ethical Concern When an individual case is reported, the person is easier to identify than if they are one of many who are summarized statistically. This problem may be reduced, if the only test results reported are those done specifically for this research; the numbers would not be identical with those in the person's medical record. And some results might be reported in summary form -- for example, percentage changes might be given instead of absolute values in some cases. But ultimately the volunteers will need to understand that this research does not offer the anonymity of a large trial, and accept or reject participation with that in mind.

Specific Questions We believe that patients and physicians will find more interesting treatments to test than we could list by ourselves. But there is no harm in discussing some questions that might be addressed by this kind of screening study.

AIDS TREATMENT NEWS occasionally publishes a list of about 30 treatments which have a valid rationale for testing in HIV disease, are not in a major development project currently, and are already widely used in people for other purposes (making research much easier than with a new chemical); this list last appeared in issue #218, March 3, 1995. In many cases, the possible mechanism of action is not known, and therefore it is not known how relevant a viral load test would be for measuring drug effects. But it may still make sense to include such treatments in an exploratory study based on viral load; there is little cost of doing so, and the results might give some insight into a mechanism of action.

(1) For example, it would be interesting to see if viral load can fall after a patient starts acyclovir (due to an indirect effect of the drug, which is not believed to have anti-HIV activity). Apparently no one has tried this yet. Any effect on viral load might be seen in only some patients -- those who have another virus which may act as a cofactor for HIV and which is susceptible to acyclovir. Therefore, a single negative result obviously does not mean that this line of research should be abandoned; instead, a number of patients will need to be tested.

(2) Another area where we do not know the mechanism concerns micronutrients. Consumption of certain micronutrients (either in food or in supplements) during asymptomatic disease has been found to be associated with substantial delay in disease progression, and increased survival; for other micronutrients, no such effect was found (see "Some Vitamins Associated with Decreased Risk of AIDS and Death." AIDS TREATMENT NEWS # 214, January 6, 1995.) Could a well-designed supplementation regimen (perhaps given by injection to avoid absorption problems) have an effect which could be seen in viral load reduction? (3) Some physicians have suggested doxycycline as possibly beneficial in AIDS treatment, for unknown reasons. Possible mechanisms include controlling possible mycoplasma infections, or treating other infections which may not be diagnosed. These rationales are probably too weak to get this drug into a formal trial, even a small one; but still it would be interesting to see if there are any consistent changes in viral load or other tests when patients start the drug. We should remember that some of the most important leads in the history of medicine have been completely unexpected in advance. The proposed monitoring project brings down the cost of research to the point where such a study could happen.

(4) Hydroxyurea has had a strong rationale as an antiviral at least since the Berlin conference over a year and a half ago, but it has been notoriously difficult to get a formal trial going, in part because no one has exclusive rights to the drug. And the drug is dangerous enough that physicians who are not experienced with it are reluctant to use it without a good reason. Only recently have a number of patients started to try it. With the proposed monitoring study in place, drugs like this could be tested in a research setting as soon as someone wanted to try them; if strong antiviral activity is found, it would bring more attention to a potential treatment. This is a drug that physicians can learn to use safely, if they first have a reason for doing so.

(5) Ribavirin, a broad-spectrum antiviral, was rejected prematurely as an AIDS treatment years ago, after an intense conflict between the Commissioner of the FDA at that time, and the president of ICN, the company developing the drug. Ribavirin might have a role for some patients, especially in combination with ddI or certain other drugs; and today, its antiviral activity could be followed with viral load tests. Despite continuing popular interest, ribavirin has not been revived as a potential treatment because of the lack of commercial motivation for organizing new formal trials. The proposed monitoring project -- based on viral load testing, which was not available when ribavirin was studied previously -- could greatly reduce the cost of getting new credible data about this drug.

(6) The monitoring project could also be used for early information to guide clinical trials of combinations of protease inhibitors or other drugs in mainstream development -- combinations that would not otherwise be tested. One important question here is the effect of these drugs and combinations in people who already had much experience with antiretrovirals, and were unable to lower their viral load further with conventional treatments. The mainstream trials are now focusing on antiviral-naive subjects; but the main use of the new drugs, at least initially, will be for people who have already failed other available treatments. We need to know what combinations are likely to work in these difficult cases, and we may not get that knowledge from the trials currently planned.

The Next Step This monitoring project should probably begin at one site (AIDS Research Alliance [formerly SEARCH Alliance, in Los Angeles] would be ideal); the tests and treatments could be administered either there, or in the offices of physicians associated with the organization. Later, other research institutions or clinical practices might also use the same protocol.

The first critical step is to refine this proposal and develop a detailed protocol, incorporating the views of medical and research experts.

A working monitoring project will mean that as soon as the first patient is ready to try a treatment, that treatment can be tested, with results becoming available almost immediately. While the results from the first few patients will clearly be limited, they will be completely consistent, without selection bias, and collected exactly as specified. If early results are favorable, more people will try the treatment, strengthening the confidence in the results of the monitoring, and also generating early interest in other studies, including larger, controlled trials.