AIDS TREATMENT NEWS Issue #209, October 21, 1994
John S. James

We attended the meeting to review the existing factual knowledge in this area, and to see what professional consensus has developed at this time. There is widespread agreement on the major technical facts about markers of HIV disease, what is known and what is not known. But while we share that factual consensus, we left the meeting seriously concerned that the research community is about to make a major strategic mistake, to follow a path leading to several more years of missed opportunities and lack of progress in finding better treatments for HIV. This article will review the facts, and also summarize our concerns.

What Are "Markers," and Why Are They Important?

HIV disease develops slowly, usually over many years, including a long period without symptoms or with only minimal symptoms, followed by unpredictable episodes of worsening and improvement. This pattern makes it hard to tell if a treatment is doing any real good -- either for an individual patient, or on the average for a group. Because of lack of reliable ways to measure treatment effect quickly, researchers have traditionally considered survival improvement as the "gold standard" for proving that a treatment really helps. And, with some continuing reluctance, most have also accepted "clinical endpoints" -- meaning not only death but also major opportunistic infections -- as a valid measure that a treatment is working.

Relying on clinical endpoints causes several problems, however. First, since HIV develops slowly, it usually takes many years to get statistical proof that one treatment leads to more deaths or serious illness than another. Such trials not only present obvious ethical problems, they also are difficult to complete, since patients and physicians will usually sense what is going on long before full statistical proof is obtained, and drop out of the trial if they have been randomly assigned to the treatment which appears to be worse. And it is virtually impossible to test a potential treatment for EARLY HIV disease using clinical endpoints, because of the long time it will take (up to ten years or more) for many people to get sick.

Also, "clinical endpoints" do not help much to guide the treatment decisions for an individual patient. By the time the endpoint has occurred, serious damage has already been done.

Because of these problems, researchers have searched for a "surrogate marker" for HIV disease progression -- usually a blood test which, hopefully, will tell quickly if a treatment is helping or not. (The word "surrogate" refers to the use of the marker as a surrogate, or substitute, for death or major illness, as an outcome in clinical trials.) For several years, the T-helper count (and, to a lesser extent, the p24 antigen test) have reluctantly been used as surrogate markers. But as more has been learned from clinical trials, it has become clear that small changes in the T-helper count -- and also the p24 antigen test, in most cases -- are not reliable as measures of whether or not a drug is working. (The T-helper count, however, does remain a very good indicator of disease prognosis, telling who is at risk for certain opportunistic infections and needs to use prophylaxis to prevent them.)

A new test, for the number of copies of HIV RNA in blood plasma, has been used by researchers for several years, and recently has become commercially available to practicing physicians. [For background, see "HIV RNA: New Blood Test for Individualized Therapy and Faster Trials," AIDS TREATMENT NEWS #204, August 5, 1994.] This test, essentially a measure of how much the AIDS virus has been reproducing in the body during the last several days, appears to be much better than either the T-helper count or p24 antigen as a measure of whether an antiviral treatment is working in a patient. Much remains to be learned, however; and while any physician can order this test, few are using it today outside of research, because of lack of guidance and collective medical experience in interpreting what the numbers mean.

Tests for plasma HIV RNA will almost certainly become more widely used in the future. The issue now is what additional evidence may be needed before they should come into widespread use. Clearly this depends on what use is being considered:

* For early clinical trials, such as for SCREENING drugs and combinations for antiviral effects in people, and for gaining insights into drug action, there is widespread agreement that the tests are ready now.

* For definitive proof that a blood test can be a complete substitute for clinical benefit -- such that the FDA could use the test alone to approve a drug for marketing, without ever needing clinical evidence that patients did better -- it is widely agreed that no marker of HIV disease progression (including plasma HIV RNA) is ready now.

* For individual patient management, there are different views -- often from the same researcher. Almost everyone personally believes that lowering the level of HIV RNA in the plasma (which means the number of viral particles in the blood), and keeping it down, would be a good thing. But to get definitive, unequivocal proof for this will take years, since new trials will have to be started to relate the RNA change to clinical endpoints (death and major illness), which take a long time to occur. (There have already been RETROSPECTIVE studies, using frozen blood saved years ago in clinical trials, so that researchers can know how well the patients will do in the future without waiting for years to find out. Retrospective results released in the last few months have shown a strong relationship between lowering the level of HIV RNA by drug therapy, and later benefit to patients. But to be completely sure, researchers also want to do PROSPECTIVE trials specifically designed to answer the question about HIV RNA or other proposed markers of HIV disease progression. Those trials -- still in early discussion stages today, nowhere near beginning -- will take a long time. The central controversy is to what extent the general use of HIV RNA testing should be delayed until the prospective trials are done.)

Statisticians have calculated what it would take to get a completely satisfactory proof that a particular blood test (such as HIV RNA) does, indeed, work reliably as a surrogate marker. The bottom line is that the treatment effect must be about four times the "standard error"; and with the poor treatments (such as AZT) that we are familiar with today, which have small treatment effects, this will require huge trials, with thousands of patients (since a large number of patients reduces the standard error of the average). And, years later, after these trials are done, the surrogate marker being tested will be definitively proven only for one class of drugs; if tested with AZT, for example, it would also be accepted for showing the effect of ddI, or of ddC. But then a new trial, with thousands of other people and also taking years, would be needed to prove that HIV RNA also works with protease inhibitors, or with any other class of drugs.

Note on HIV RNA, and "Viral Load" Terminology

A number of different markers of HIV disease progression have been proposed, but the one generating most interest today is the measurement of the number of copies of HIV RNA per milliliter of blood plasma. These copies are contained in virions, free virus particles in the blood; each virion has two copies of the RNA. If a patient gets a test result, for example, of 100,000 copies per milliliter (a fairly high number), this means that there are 50,000 virus particles per milliliter of blood plasma. (One minor source of confusion is that some testing companies report the results in terms of one milliliter, but others report in terms of a different amount of blood plasma, meaning that the numbers have to be corrected arithmetically in order to be comparable.)

This measurement (of the number of copies of HIV RNA per milliliter of plasma) is often called the "viral load," we will use that terminology in this article. But there has been some confusion, since the term "viral load" can mean two different things.

It is now known that when HIV infection occurs, a great many cells in the lymph nodes are silently infected. The genetic information of HIV has been integrated into the genetic inheritance, the DNA, of the cell; but in a large majority of the cells, this information is not being expressed. That means that the HIV in those cells is causing little or no damage at the time; but it also means that the immune system cannot detect those latently infected cells, and therefore cannot eliminate the infection. Those cells, therefore, form a reservoir of silent infection; and years later, for reasons not entirely understood, the cells can become activated and start producing virus. When this virus gets into the bloodstream, it can be measured by tests for HIV RNA, or by other kinds of viral tests.

The term "viral load" is usually used to mean the amount of free virus in the blood (or in the lymph node, if a lymph node biopsy is done as part of a research study). We will follow this usage. The confusion is that sometimes "viral load" has also been used to mean the silent reservoir of infection. We think that "viral reservoir" or "proviral load" would be best for that meaning, but we will wait to see what usage develops.

The viral load in the blood, not the viral reservoir, is the important measurement today. The viral reservoir cannot be affected much by any known treatment; also, it cannot be measured except by sophisticated research tests, which are far from general availability. The treatment strategy today is to use antiviral and other treatments to reduce the viral load -- by blocking the various steps of viral replication, by killing infected cells which have become activated, or by preventing them from becoming activated in the first place. Future treatments may be able to attack the viral reservoir directly, ultimately eradicating the infection; but that is not possible now.

Another terminology question involves the phrase "surrogate marker." Some believe that a viral measurement should not be called "surrogate" marker, since it is not a substitute for anything else, but a direct measurement of the virus. We agree, but will use the term "surrogate marker" in this article, since it is part of the name of the meeting we are reporting.

The Consensus Document

On the afternoon of the second day of the three-day Surrogate Marker meeting, an appointed Scientific Advisory Committee met privately to develop a consensus statement on surrogate markers. This committee, of 17 members altogether, included well-known AIDS researchers, government and industry experts, and two AIDS treatment activists. A "Consensus Statement, Scientific Advisory Committee on Surrogate Markers of HIV" was drafted that afternoon and immediately distributed to the other attendees of the conference. The three-page statement is too long to reproduce here; instead, we will quote from it and summarize. (The complete consensus statement can be obtained from Cambridge Healthtech Institute, which organized the conference. To request a copy, write to: CHI-Consensus Statement, 1000 Winter Street, Suite 3700, Waltham, MA 02154, or send a fax to 617/487-7937.)

In the text below, we have quoted key sections of the consensus document, and included explanations for non- technical readers. [Also, we have expressed our own concerns about what we believe is the unfortunate direction in which the professional consensus seems currently to be moving. To separate our reporting of the meeting from our evaluation of it, we have enclosed the latter in brackets -- or placed it in the Comment section, below.]

From the introduction to the consensus document: "The Committee believes that the impact of viral infection on the immune system drives the AIDS disease process and the Committee determined that its principal goal should be to recommend ways to test this hypothesis. The appropriate use of surrogate markers in clinical trials has the potential both to accelerate the evaluation of therapeutic interventions and to provide critical insights into the pathogenesis of HIV infection."

[Note that the main goal is to test a hypothesis, one that the committee already agrees with -- not to use the knowledge we already have to develop new treatments quickly.]

Next, the document listed a number of potential surrogate markers:

"All of the following are correlates of progression, but none can work in isolation of the others:

"Immunologic markers: CD4 count; CD4 percent; Beta-2 microglobulin; Neopterin; Lymphocyte function (proliferative responses, specific and nonspecific; cytokine elaboration).

"Virologic markers: p24; ICD p24; viral culture titer; DNA PCR; RNA PCR; Genotypic or phenotypic drug resistance; Biologic phenotype.

Explanation: "Genotypic" drug resistance refers to tests for particular mutations which are known to convey some degree of resistance to particular drugs -- for example, tests for the mutation at position 215 which conveys resistance to AZT. "Phenotypic" drug resistance is tested by growing the virus in the laboratory and seeing if it is resistant, regardless of what known or unknown mutations might be involved. "Biologic phenotype" refers, for example, to the difference between "SI" (syncytia inducing) and "NSI" (non syncytia inducing) virus; SI virus develops in some but not all people with AIDS, and is associated with faster disease progression.

The Committee agreed that "a change in any of the above markers indicates an effect of a therapeutic agent IN VIVO. However, the absence of an effect in HIV DNA or genotypic or phenotypic resistance is insufficient evidence of the absence of antiviral activity.

The Committee then listed nine factors which "should be considered in assessing the relevance of these surrogate markers as they relate to clinical efficacy." The first two -- which the discussion at the meeting clearly showed were considered most important by the Committee -- were "Consistency across trials in correlation with clinical outcomes," and "Relationship to the putative mechanism of action of a therapeutic agent and proposed disease pathogenesis" (meaning that the marker "must make sense"). The other seven factors to consider in judging a proposed surrogate marker are: "Degree to which a surrogate marker explains a treatment effect; Magnitude of effect: Duration of effect; Precision/reproducibility; Dynamic range; Breadth of patient populations to which it is appropriate; Practicality and cost."

Then the Committee made four recommendations:

(1) It asked, "Which surrogate markers currently deserve the greatest commitment of investigative resources to validate them in a clinical setting?" It listed "HIV RNA in plasma" as the most important one; in the next most important category, it listed: "IUPM (infectious units per million PBMCs [peripheral blood mononuclear cells]); CD4; Immunologic function; HIV DNA; PBMC mRNA; Lymph node loads; Point mutations for resistance genotype; and Viral phenotype.

The discussion (although not the consensus text) made it clear that, within this second category, the greatest interest is in immunologic function -- the use of certain experimental tests which ask, not how many T-cells someone has, but how well those T-cells work. [In reaching this determination, the committee had oral reports of a potentially important finding which was not available in written form because of publication delays. Clearly much more research is needed in developing immunological tests of HIV disease progression. For years this work has been held back by lack of funding, largely because it has not had a clear relationship to particular pharmaceutical products.]

(2) "What is the most efficient and effective way to assess surrogate markers?

"Designing and executing trials of therapeutic agents that will establish a clear link between the effects of such agents on surrogate markers and clinical progression should be a high priority activity. Options include: A large "clinical difference" trial with a nested subset; Two medium- sized trials: one in early disease and one in late disease.

"In these trials, adequate samples should be banked for later use in case-controlled studies."

Explanation: Basically this is a recommendation to start new, large, clinical trials to link changes in a potential marker (such as HIV RNA) with changes in progression to "clinical endpoints" (death or major opportunistic infections). Since thousands of people are likely to be required for these trials, researchers know that they will not have the budget to run the blood tests on everyone; HIV RNA, for example, now costs about $200 per test. Therefore, they are proposing to draw blood samples from all the volunteers and store them in freezers; later, when they find out who reaches a "clinical endpoint," they will test their blood, and compare those who did reach an endpoint with a random sample of those who did not. This is the "case-controlled" study design -- here used to save money, compared to the cost of testing everyone.

(3) "Can RNA load be used as a variable in a clinical trial in today's climate?

"Such a trial could be done if it were well designed."

[This simple-sounding statement hides a lot. It refers to the ethical and practical problems of having a control group for a study of plasma HIV RNA -- problems the researchers are very much aware of.

[In order to get absolute statistical proof that keeping the HIV RNA low does help to prevent death or deterioration, there probably has to be some form of comparison between patients whose HIV RNA is kept low (by drug therapies, and changing drugs when necessary), and another group in whom it is not kept low. Would it be ethical, with the current state of knowledge, to leave clinical-volunteers with a high viral load, without available treatments, until something bad happens to them -- or to try to avoid the issue by not running the test, so that no one would know if their viral load was high or not? Could such a trial be completed -- or would participants get their own tests, and treat themselves if necessary, preventing the trial from getting the "clinical endpoint" differences it was designed to get? We suspect that the glib answer really means that the Committee was not ready to deal with the question.]

(4) "What should people with HIV infection and clinicians caring for them be told about viral quantitation now?

"Viral quantitation is prognostic of progression, but it is not known yet whether measurements of viral load can be used to inform therapeutic decisions and so enhance treatment."

[What this leaves unsaid is that researchers are very much worried that widespread use of plasma HIV RNA testing will make it impossible to run their large, clinical-endpoint trials to "validate" the HIV RNA against clinical endpoints. First, if the test becomes part of the standard of care, with patients routinely using it to guide their therapeutic decisions, who will be available to volunteer for trials to see whether patients whose physicians use the test do better than patients whose physicians do not? Second, who will fund those trials? The government has less and less money for AIDS trials (and for research in general); and the companies that run the tests will have little incentive to finance a large, expensive trial, if they can sell their tests without it.

[When one of the participants at the meeting reminded the others of the urgency of the current situation, he was not referring to the fact that people are dying. The urgency was because people can get the HIV RNA test "in California" (actually nationwide, and in other countries as well -- the California location of one facility is coincidental); therefore the trial must be started soon if it is to be started at all.]

The Committee also agreed on five more technical points. One concerned using changes in the virus to drive change in therapy. Another concerned the need to standardize study methodology, even internationally, so that data can be combined to answer certain questions which may require thousands of "clinical events." Another suggested that viral load, markers of immune function, and T-helper counts are likely to be the best markers to use in trial of immune-based therapies.

Comment

Our main concern is not that unethical trials may be run, letting some peoples' viral load stay high without changing their treatment, to show that bad things happen. We are more fearful that no trials will be run -- that effort and attention are being misdirected into attempts to run huge, years'-long trials to "validate" HIV RNA as a surrogate marker. In fact these trials will probably not be financed, could probably not be run successfully even if they were financed, and would only be marginally useful even if they were conducted successfully.

They would not be very useful because they would take several years to validate each proposed surrogate marker for each different class of drugs. In other words, no marker will be "validated" until several years after the first drug of the class is ready to be given to thousands of people. In a rapidly moving field, with progress in biotechnology driving continuous acceleration of change, this means that the marker would only be available for approving non-critical advances, when the leading-edge research has moved into areas in which no marker could yet have been validated.

What should be done instead is to start now, using plasma HIV RNA and other markers, to screen dozens of potential HIV treatments which are already in human use. We need many small trials -- often 20 volunteers or fewer per treatment arm could be enough. These trials should publish their initial results quickly, within months, but also continue long-term followup lasting for years. The goal is to find new elements of combination treatments -- and later new combinations -- which lower the viral load as much as possible, in as many patients as possible, and keep it down for many months or years.

But very little of this work is being done. The problem is that the kinds of studies which are easiest to justify in academic meetings are not the ones we need to save lives.

Note:

Aside from the consensus document and the issues around it, a number of scientific papers were presented at the conference. We are still reviewing them, and preparing a report for a future issue of AIDS TREATMENT NEWS.

941021
ATN20901

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, 1994. 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. .