New Trial Proposed for When Drugs Fail



On May 11 this writer and other community representatives met
with members of the Clinical Trials subcommittee of the
Inter-Company Collaboration for AIDS Drug Development (ICC).
The ICC is a group of 17 major pharmaceutical companies which
have agreed to share information to facilitate more rapid
development of AIDS treatments, especially combination
antiretrovirals. The meeting was chaired by David Barry,
M.D., who is Group Director of Research and Development for
Burroughs Wellcome Co., and chairman of the Clinical Trials
subcommittee of the ICC.

Dr. Barry said that the kind of innovative trial design which
would get the most support in the ICC today would be studies
of combinations of drugs (often entirely new ones) in
patients who already have resistant virus. In practice, this
will usually mean studies in patients who have already used
AZT and/or other antiretrovirals, and for whom the drugs are
no longer working well. There are different kinds of
laboratory tests to tell if a patients' virus is resistant to
certain drugs; patients might be selected based on those
tests, or in some cases they might be selected based on
clinical failure of the drugs they have been using.

This emphasis on "experienced" patients -- those who have
already used AZT or other drugs for some time -- is important
for several reasons. First, many of the trials conducted
until now have sought "naive" patients as the most desirable
-- for example, the only ICC trials proposed so far insist on
naive patients. This means that many people who most need a
new drug are disqualified from studies (unless they lie), and
also that there are large numbers of people for whom new
treatments are not being developed.

In addition, it is difficult to recruit naive patients for
trials. This problem may seem unexpected, because there are a
great many people with HIV who have never used AZT or any
other treatment for their condition -- probably many more of
them than of those who have received treatment. But usually
these people are untreated for a reason. Many do not know
they are infected. Others have chosen not to use any of the
available drugs, due to concern about side effects, fear of
being disqualified for clinical trials for treatment-naive
patients in the future, or for other reasons. It is difficult
to persuade people who have not wanted conventional treatment
in the past to now be willing to enter trials of experimental
drugs. By contrast, there are many who are now in the
healthcare system who are willing to take drugs and know that
they need new ones.

There is also a scientific reason why trials with patients
who already have resistant virus are important. Dr. Barry
said that by studying them it may be possible to learn much
that will apply to naive patients -- but much faster than by
studies in the naive patients directly. This could provide
what might be called a "surrogate marker" for DURATION of
benefit -- a very important indication of drug usefulness
which is not otherwise available. [This is possible because
patients who have developed resistant virus will have
developed a very large number of different "quasispecies" of
HIV. If a new drug combination can suppress all of them, well
enough to bring the viral load down to about 100 copies per
ml or less for at least six months, then that same
combination will probably suppress virus for very much longer
in naive patients who have not yet build up many resistant or
cross-resistant quasispecies.]

Dr. Barry suggested a trial design for this purpose. First,
before planning the trial in detail, it will be necessary to
decide what tests to use to define who has resistant virus;
today it is not certain whether "genotypic" tests (which look
in the DNA of the virus for a mutation which is known to
confer resistance) or "phenotypic" tests (which grow the live
virus in the laboratory and see how much drug it can
tolerate) would be best for this purpose. Dr. Barry left the
question of definition for later.

The kind of trial he proposed could use a very small number
of patients (probably well under 30) with resistant virus.
These patients could get one of several different drug
combination regimens (ideally consisting of new drugs which
the patient had never used before). The choice among the
several available regimens could be made in several ways: it
might depend on which particular mutation the patient had, or
it might be random, or it might be the choice of the patient
and treating physician, or of the experimenters.

While on the new-drug regimen, both the CD4 (T-helper) count
and viral load would be followed. As long as a patient does
well -- as long as the viral load remains low and the CD4
count does not decline -- he or she would remain on the
treatment. But whoever stops doing well -- whether in six
days, six weeks, or six months -- then moves to another of
the available regimens.

Dr. Barry believes that if the data is recorded carefully,
and if the original resistance is known, then this trial
could be of great benefit not only to the patients
themselves, but also to others in the future, because it
would tell which resistance patterns respond to which drugs.

This design implies that the patients in the trial will have
to be treated and followed individually. This is because
every patient has a number of different quasispecies of HIV;
every patient is different from every other in the collection
of quasispecies they have. This means that even if two
patients have the same resistance mutation (in some of their
quasispecies), and are the same by all other available tests,
they may still react very differently to the same drug
combination. This individualizing of care makes the trial
more in the patient's interest than conventional trials which
use standardized treatment protocols to see which does better
or worse on the average.

Note also that in this proposed trial, patients and their
physicians will receive their viral load, CD4, and other
blood work results immediately -- not after the study is
over, as in many other trials. They will need to know those
results as soon as possible in order to know when it is
necessary to change treatments.

Other Proposals

Three other proposals were also made at the meeting.

(1) Small, uncontrolled trials. Bill Bahlman, of the
Treatment and Data Committee of ACT UP/New York, proposed
small, uncontrolled screening trials for initial tests of new
combinations. (An earlier draft of this proposal was
published in AIDS TREATMENT NEWS, issue #221, April 21,
1995.) Each trial will have about 30 patients, and will last
about six to ten weeks, but with an additional followup phase
lasting for at least a year, for those patients who choose to
stay on the treatment. Bahlman would study both
antiretroviral naive patients with CD4 count above 300, and
antiretroviral experienced patients with CD4 under 300. A
change from the earlier version of the study is that there
will be no requirement that patients have a minimum viral
load -- since modern tests can detect very low levels of the
virus, even those who start with a low level will still be
able to show if they are getting benefit from starting a new
treatment regimen.

This proposal is controversial because it does not randomly
assign patients to two or more different treatment groups;
therefore, when the results are analyzed, there will not be
different groups in the same study to compare. But these are
preliminary, pilot studies, to quickly screen combinations to
see which are worth testing in larger, more formal trials.
And these screening studies are looking for large effects,
not for small differences. It should be possible to spot
combinations which have outstanding short-term antiviral
activity in people, by looking for large changes in viral
load and CD4 count.

(2) Combining different protease inhibitors. Jules Levin of
the New York University Community Advisory Board brought a
proposal by two NYU investigators, Roy Gulick, M.D., and Fred
Valentine, M.D., to compare MK-639 (the Merck protease
inhibitor) alone, vs. saquinavir (the Hoffmann-La Roche
protease inhibitor) alone, vs. the two together. This would
be a randomized, pilot study with about 90 patients, lasting
24 weeks, in persons with CD4 count 50 to 500 who had not
previously used any protease inhibitor. It would look for
safety and toxicity, changes in viral load and in CD4, and
resistance patterns.

The main point of this proposal was to combine two protease
inhibitors; the drugs selected are those which are furthest
along in clinical development. Combining protease inhibitors
has been suggested, but has not yet been done in trials; the
only way for a person to obtain two of the drugs would be to
register for different trials under different names, and it
is believed that a few people have done so.

Both Merck and Roche representatives said that they have been
discussing a combined protease inhibitor trial. But Roche is
waiting for its new formulation, which will have about three
times greater bioavailability than the current version of
saquinavir. When that is available, then there will be a
dose-finding study, and then a PK (pharmacokinetic) study to
look for interaction between the drugs (there is concern that
protease inhibitors might interact with each other in unknown
ways, because most of them are metabolized by the liver).
Only when these studies are done will the combined trial be
run.

An obvious way to speed this process would be to start now
and use the existing formulation; when the new one becomes
available, switch to it and adjust the dose in the ratio of
the average bioavailabilities. Once it reaches the
bloodstream, the new formulation is the same drug; the
difference is how much of it is absorbed when taken orally.
But companies are reluctant to begin a new trial with a
substance they expect to abandon, and this strategy for
saving time was not discussed at the meeting.

(3) Computerizing medical records. Existing projects are
already computerizing medical-record information of HIV
patients. But almost all of these are extracting data from
existing paper records, then entering it into a database.
This approach leads to transcription errors; also, much of
the data is never recorded and is lost to research.

Dr. Barry outlined major advantages for research if most HIV
physicians would keep their primary patient records by
computer, and then share data for research purposes, after
names and other identifying information had been removed. For
example, it would in theory be possible to find patients who
had (secretly) taken more than one protease inhibitor at the
same time, and check for any possible side effects which
occurred; this information could make future studies safer
and faster. Many other drug combinations could be checked in
the same way, to look for treatment regimens which might be
working unexpectedly well, or unexpectedly poorly.

Confidentiality will be a major concern. It is important to
bring all of a patient's records together in the research
database; this could be done anonymously, by using a code
number, so that the institution holding the combined data
would not have the names of patients. But some problems would
remain. For example, when a research team studying protease
inhibitors looked for the records of all patients taking more
than one, they might be able to identify one of their
volunteers because some of the blood work would match their
records exactly, and then they might no longer want that
volunteer in their trial. Unless the system can guarantee
that patients will not be harmed by participating, it will be
difficult to get their cooperation. And with medical care as
fragmented as it is in the U.S., it would be impossible to
enforce cooperation effectively, even if this were desired.

Note: Over two years ago AIDS TREATMENT NEWS reported on
Medsys, an early system which computerized patients' medical
records very effectively, primarily for use within the
primary-care office or clinic; physicians could also share
the information anonymously for research purposes, if they
chose to do so. What impressed us most about this system,
developed by a team led by Larry Bruni, M.D., at the National
Community Research Initiative in Washington, D.C., was its
ease of use; in one afternoon, medical office employees could
learn how to use it effectively. But medical-record
computerization has not yet caught on in AIDS practices,
either public or private.

For background on the NCRI system, see "New Software
Available for physicians' Offices and Community-Based
Research," AIDS TREATMENT NEWS #165, December 18, 1992. For
current information, call Medical Management Systems,
202/546-0887.