Merck Protease Inhibitor: Viral Resistance Suspected, Large Trials on Hold

Protease inhibitors, a new kind of AIDS drug, work
differently than any approved or available treatment; they
block an enzyme, the HIV protease, which is produced by the
virus and required for it to reproduce. About a dozen
pharmaceutical and biotechnology companies are now developing
different protease inhibitors. One of them, code-named L-
735,524 (usually called L-524), being developed by Merck and
Co., has shown very good ability to suppress HIV activity in
people, for about three months. But when the trial was
continued beyond that, a sensitive blood test (for HIV RNA)
showed that viral activity later returned; after about six
months on the drug, the virus was about as active as when
treatment began. Merck researchers suspect that the virus
became resistant to the drug, but they do not know for sure.

After seeing this disappointing result, Merck called off
plans for a large trial, which could have started as early as
March and given the drug to hundreds of people. Instead,
Merck will increase the dose of L-524 in its ongoing small
trial (known within the company as Protocol 006), and also
begin other small trials to test the drug in combination with
other antivirals -- probably AZT, and also in a triple
combination with AZT and ddI. Meanwhile, researchers will
continue laboratory studies of the virus from the volunteers
who have already been treated.

On February 24 Merck held a meeting with treatment activists,
including a number of volunteers who are taking L-524, to
explain the trial results and the company's changing plans;
this writer attended that meeting by telephone. This article
describes some of the information presented, which may be
important for understanding L-524 and other drugs as well.
(Merck also explained its results to reporters from the Wall
Street Journal and the Philadelphia Inquirer; both papers
published articles on February 25.)

[Note: Do not confuse the Merck protease inhibitor, L-524,
with the Hoffmann-La Roche protease inhibitor, Ro 31-8959.
Large trials of the Ro 31-8959, which will involve several
thousand volunteers, are now beginning, as described in our
previous issue, AIDS TREATMENT NEWS #193. Also, do not be
confused by the terminology: the words "protease" (used by
Merck) and "proteinase" (used by Roche) are the same.

[The Roche trials, which will recruit a total of about three
thousand volunteers, are going forward even though "reduced
sensitivity" to Ro 31-8959 has also been found. Why would the
companies respond differently to data suggesting that viral
resistance had developed? There are several possible reasons.
First, Roche is ahead of all other companies in human trials
of its protease drug; early human results were first
presented at the International Conference on AIDS, in June
1993. This means that Roche has had time to study the
problem; while Merck, which has only observed the apparent
resistance to its drug since January 1994, wants to
understand what is happening better before giving L-524 to
hundreds of people. Also, a drug can be useful even if
resistance develops (as it does eventually with almost all
drugs); and Roche has an incentive to preserve its major
advantage in protease inhibitors, its lead in human trials.
But Merck is seeking a major advance, not just a "me too"
drug comparable to what Roche is developing. For the first
three months, L-524 looked like it might be such an advance;
when it failed to hold up later, Merck decided to learn how
to use the drug better, before beginning large trials. (Merck
has other protease inhibitors which work better in the
laboratory than L-524, but they are not yet ready for human
studies.)]

The Blood Tests: P24 Vs. HIV RNA

Protocol 006 was first designed to last for 12 weeks, and to
test for antiviral activity of L-524 in humans primarily by
using a blood test to measure the level of a protein, called
p24, which is produced by HIV. But scientists are becoming
increasingly disenchanted with the p24 blood test (even the
newer 'ICD' version, which Merck is not using). Although this
test can show if a drug has antiviral activity in patients,
long-term studies are finding that it does not predict which
patients will do well in the future. The p24 blood test is
inherently unreliable, because the result depends not only on
the amount of p24, but also on the amount of antibody against
p24 which the patient's immune system is producing.

Because of this concern, Merck researchers decided to also
measure the concentration of HIV RNA in the blood. This new
viral test is not affected by the antibody, so it can be a
more reliable measure of viral activity. (For background, see
"Better Viral Tests: Interview with Mark B. Feinberg, M.D.,
Ph.D.," AIDS TREATMENT NEWS #186, November 5, 1993.) The RNA
test is new and difficult to perform, however, and because of
technical difficulties the results can vary from day to day
even in the same laboratory. To help control this
variability, blood samples were frozen so they could be
tested together in the same batch.

In Protocol 006, no one has yet taken L-524 for more 18
weeks; but Merck has data from a handful of people at 24
weeks, from a small phase I study (Protocol 004) which
started earlier.

RNA data are not yet available from Protocol 006. But
Protocol 004 found that the results of the p24 and HIV RNA
tests were comparable for the first four weeks, but then
became fundamentally different. Within two weeks of beginning
treatment, the p24 had decreased by about 60 to 70 percent of
its starting value; then it continued to decrease until it
became undetectable in most of the volunteers. Then it
remained largely undetectable, although there was a slight
rise eventually.

The HIV RNA was tested only at baseline (before treatment
began) and at every two weeks up to 24 weeks, but samples
were only run in batches at four weeks and then again after
20-24 weeks. During the first four weeks, RNA dropped in
parallel to p24. However, the assays conducted at 20-24 weeks
revealed that RNA had begun to rise back at the 12-week time
point, and by 24 weeks had returned to baseline levels --
apparently because the drug was no longer controlling HIV.
Patients seemed to do well, showing weight gain and T-helper
count increases during the period that the virus appears to
have been controlled; but with only a handful of patients
tested, it is too early to know if the patients benefited as
a result of the drug. If the new test for HIV RNA were not
available, everything else would suggest that L-524 looked
very good.

Why did the two tests give different results? Merck
researchers suspect that the drug initially suppressed the
virus, allowing the levels of antibodies against the p24
protein to become very high. Later, when viral activity
returned, these high antibody levels prevented the p24 from
being detected. But the test for HIV RNA showed that the drug
was no longer stopping the virus.

Trial Design and

Drug Approval Requirements

Protocol 006 is comparing L-524 to an AZT control; after the
trial had begun (in October 1993), it was extended from a 12-
week to a 24-week trial. This change angered some of the
volunteers, who expected to have to be on the blinded study
for only 12 weeks, after which they could switch to "open
label" L-524. Merck acknowledged that there had been
communication problems; clearly the volunteers should have
been told by the investigators before entering the study that
an extension was possible, and should have been told about
the extension as soon as it was officially approved. (Another
issue is that Merck has only given a summary of the study
protocol, not the protocol itself, to the volunteers. Most
pharmaceutical companies do not release their protocols to
the public or the volunteers in the trial; however, the U.S.
National Institutes of Health usually does.)

The evolution of Protocol 006 illustrates some of the
requirements for drug development and approval today. It
began as a 12-week study because the FDA would not approve a
design which called for giving the drug to people for longer
than animals had received it in toxicology tests. As the data
from the first 12 weeks came in, the drug looked like it was
working well; and Merck knew that the FDA would require a 24-
week trial in order to count toward drug approval. Instead of
losing time by starting over with a new 24-week trial, Merck
asked the FDA for permission to extend Protocol 006 for
another 12 weeks. Since there was enough animal experience by
then, permission was granted. (As it turned out, this trial
will not count toward approval in any case, due to the
disappointing 24-week results.)

If L-524 had worked in the six-month trial, what else would
have been needed for approval? Usually the FDA requires two
controlled studies before a new drug can be approved. The
expectation today is that if a new kind of anti-HIV drug
works much better than anything else known at six months (but
as measured only by blood tests, without proof of clinical
benefit to patients), the FDA would require the company to
begin a second study (which would measure clinical benefit,
and therefore could take two years or more to complete)
before the drug could be approved. But while this second
study would have to be started, it would not necessarily have
to be finished before drug approval.

Future Research Plans

Because HIV mutates very rapidly, Merck researchers suspect
that it may be necessary to completely stop the virus in
order to prevent drug-resistant strains from emerging. L-524
did not do this in the doses given (up to 400 mg every six
hours). It seems likely that some viral mutation, probably
already present in the volunteers before treatment began, may
give HIV a low-level resistance to the drug; Merck
researchers believe there is probably no more than a four-
fold reduction in sensitivity (meaning that the drug would
still work if the dose could be increased, by no more than
four times). And even if resistance cannot be overcome by
increasing the dose it might be greatly delayed by combining
two or more drugs with different mechanisms of action -- a
strategy which has been used very effectively against
tuberculosis and other infectious diseases.

While new trials of L-524 are being designed, virological
studies are being done to confirm that viral resistance is
the problem, and to find what mutation or mutations are
responsible. (Drugs can lose effectiveness due to mechanisms
other than viral resistance -- for example, there can be
changes over time in how the body processes the drug.) But
low-level resistance is more difficult than high-level
resistance to test in the laboratory; therefore, these
studies could take months to complete.

Comment: The Past and the Future

Merck has shown consistent commitment to AIDS research, top
quality science, competent operational planning of the drug-
development process, and openness to the interested public --
a rare combination in research today. It has kept its AIDS
program going despite serious setbacks. These began
tragically in December 1988, when Dr. Irving S. Sigal, senior
director of molecular biology at the Merck Research
Laboratories, and the world's leading expert on the structure
of the HIV protease, was killed in the bombing of the
airliner over Lockerbie, Scotland. Other setbacks were the
failure of an early protease inhibitor when it caused biliary
obstruction in dogs in toxicology tests, and the failure of
L-661, a reverse-transcriptase inhibitor, due to rapid
development of high-level viral resistance.

Merck's determination to get it right before going into large
trials is a welcome change from the usual approach of
allowing one's research efforts to become focused on
treatments which clearly do not work very well and never
will. But there is also concern in the community that Merck
may abandon a drug which is better than anything else and
could be useful, if it does not fit with the corporate goal
of a home run. We do need to plan carefully for the
possibility that we may not be able to stop HIV completely at
this time.

And when combination treatment is needed, the obvious drugs
to add are the approved antiretrovirals: AZT, ddI, and ddC.
But the search should not stop there. Other treatment
approaches, aimed at reducing the activation of HIV, should
also be considered. These include LTR inhibitors, tat
inhibitors, certain immune suppressive therapies, certain
anti-inflammatories, and therapies based on providing or
blocking certain cytokines. Also, there are other immune-
based treatment possibilities, which seek to restore or
enhance the immune system's natural ability to control HIV
for years -- which it does far better than any known drug.
This whole cluster of approaches has received far too little
attention and support. But there are some treatments which,
while still experimental, could feasibly be tried now as part
of a combination regimen with L-524.

And if it is true that L-524 shuts down HIV in people more
completely than any other known treatment, even if only
temporarily, then it could be used to answer some important
AIDS research questions, for example:

* In persons with very low T-helper counts, will the immune
system recover on its own (without special immune
reconstitution treatment) if the virus is stopped?

* In those with early HIV infection (for example, with T-
helper counts over 500), would the drug work for a much
longer time, since there is little viral replication and
therefore little opportunity for resistant HIV variants to
evolve?

* Could the drug be used as a temporary antiviral "cover"
during certain therapies (such as IL-2 treatment) which
otherwise can cause a temporary increase in HIV activity?

* Is it feasible to increase the "trough" level of the drug
(the lowest concentration in the body between doses, when
resistant virus has the most chance to emerge) by a time-
release formulation, or by more frequent administration, in
addition to increasing the total dose?

In another area, a major challenge for 1994 and 1995 will be
learning how to use the new viral tests, both for drug
development and for individual patient care. (These are still
research tests, not yet regularly available to physicians.)
Merck's results suggest that the key to controlling HIV
infection may be to reduce viral activity to a low level --
and then keep it low indefinitely, which is more difficult.
The right treatment may be whatever works -- antivirals,
immune-based treatments, drugs to reduce viral activation,
various combination treatments, life-style changes, even
"alternative" treatments where the mechanism of action may be
unknown. Effective strategies may be very different for
different patients -- and for the same patient at different
times.

As better viral tests become more widely available, it will
be feasible to develop "libraries" of treatment strategies,
all of which work at certain times for certain patients, but
do not work at other times. Patients will be periodically
tested for viral activity, and whatever treatments are
necessary will be used to keep the activity down. As the
treatment strategies, guidelines for their use, and viral
tests all improve, it may become possible to control the
disease indefinitely, even without a single treatment that
works for everybody.