Protease Inhibitors and Beyond: Interview with David Feigal, M.D.


David Feigal, M.D., is Director of the Division of Antiviral
Drug Products of the U.S. Food and Drug Administration. After
the February 23-24 meeting of the National Task Force on AIDS
Drug Development, a meeting focused on protease inhibitors,
we asked Dr. Feigal for his perspective on issues related to
the current development of these experimental drugs.

Access and Supply Problems

ATN: There is wide agreement that the protease inhibitors,
despite their known limitations, are the most promising
single approach to HIV treatment today. Yet it will be some
time before many patients have access to these drugs. What
should we as a country do in this situation?

Feigal: At the National Task Force meeting, the companies
developing protease inhibitors were fairly frank about their
manufacturing problems; one of the companies even had an
activist group send in a consulting chemist to verify the
problems. That shows how generally open they have been. These
compounds are difficult to produce -- and to make matters
worse they are not very bioavailable [meaning that a large
oral dose must be used, so a larger supply must be produced].
A company which is manufacturing only a few hundred
kilograms, which is typical in early drug development, may
only be able to supply the drug for three or four hundred
patients, if a year's supply per patient is needed. Many of
the candidate protease inhibitors require multi-step
synthesis processes with low yields, and for some of them the
synthesis takes a long time, and there are other problems
with scale-up to large quantities. These problems will be
solved; but it has meant that companies had to make decisions
to go ahead with a much larger scale of production than is
usual early in drug development.

This supply problem will pass with time; if these products
meet their potential, they will be manufactured on a large
scale, and then this period of poor availability will go
away. The question is what to do in the meantime -- and how
get the clinical answers at the same time as we provide the
product to people who cannot participate, or choose not to
participate, in clinical trials.

Several suggestions have been proposed, some from the
companies. One is a tiered approach, to make drug available
to patients with the greatest need. Very few companies have
much supply, and even that would be used up very quickly.

This is an area where we could use input from the community.
What do those affected by HIV think would be the fairest way
to deal with these problems? In a different area, when
Betaseron was approved for multiple sclerosis, there was a
serious supply problem, and a lottery was used to distribute
it. This was very unpopular; there was much unhappiness about
using a lottery.

The companies are much aware of the difficulties this limited
supply creates; and there could be a partnership between the
community and the companies, to figure out how to distribute
a drug in very short supply. What about a product that could
meet less than half the need? The largest supply possibility
we heard about at the National Task Force was enough product
for several thousand people; that is not as large as the d4T
expanded access, or the currently ongoing 3TC expanded
access.

There must be some way of cooperatively dividing up these
scarce drugs; there needs to be a partnership between the
community that needs the access and the companies. To the
extent that the companies feel they are being asked [by the
community] to do things that the FDA would not approve, we
are happy to get involved. But many of the issues revolve
around such matters as logistics, and protecting trials.
These issues are between the company and the community
affected by the disease more than they involve the FDA. If
the companies are saying the FDA will not allow wider use of
the drug at a certain stage of development, then we should
get involved; it usually will not be true. We have said with
the proteases for a long time that we would welcome an
expanded access proposal. Manufacturing remains the real
problem.

Other Treatment Options

Aside from waiting for the protease inhibitors: I agree with
your statement that these are the single most promising
agents to come along. But also, if you look closely at some
of the combination data now developing [using other drugs
which are more available], some of these are obtaining
preliminary results in the same general magnitude as the
protease inhibitors. Most of them, like the proteases, are
most effective for patients who have not been pre-treated
with many other drugs. But some of them were capable of
getting fairly dramatic drops in viral load and rises in CD4
count, even in patients who have been on prolonged zidovudine
[AZT]. So I believe it would be a mistake to focus only on
the protease inhibitors, particularly since we know that
resistance to these drugs does develop over time; that is an
early signal that the protease inhibitors, too, are probably
going to be most useful in a combination.

ATN: So people should not rule out other options that might
be available now?

Feigal: I would agree that there are other options now. Most
of these results I referred to have been presented publicly,
at the Human Retroviruses and Related Infections conference
in Washington [January 29 - February 2, 1995]; the 3TC data
and the nevirapine data were very interesting. Each of these
has different pluses and minuses, and a different story in
terms of availability; but these and some others are showing
fairly similar types of preliminary evidence.

Individual Variation

It is difficult to evaluate these preliminary studies,
however, because some of them are based on small samples.
Researchers often find that the most compelling way to
present their data is to show individual patients' results.
But there can be much selectivity in picking the best
responders. There have always been single patients in almost
any trial who have had dramatic responses at the start of the
trial.

What is more important is that the average responses are
better. We do not know yet how good they are in broader
populations and how long the effects will last, and we still
do not know how to translate changes in viral load to long-
term clinical outcome. We have had these tests for a short
period of time.

ATN: I remember that when the AZT trials started, there were
miracle stories then.

Feigal: There was a patient in the original BW 002 [AZT]
study who had CD4 count under 50, who went up to a thousand,
in the first 16 weeks of the study. We attributed this to
problems with the T-cell test at that time. But even though
the average difference between the placebo and AZT was only
40 T-cells in that original trial, there were individuals who
had very dramatic responses. The same was true for patients
who started ddC after having been on AZT for periods of time.

There always have been individuals who may have responded
unusually well to a drug. And just as we have learned about
the pathogenesis of the disease by studying long-term
survivors, we would probably learn something about drug
therapies by taking a look at some of these dramatic
responders, and seeing what is different about their virus,
or their drug absorption, their intracellular triphosphate
levels, etc. Some of the variation might be due to chance,
but still I think there is potential to learn much about
therapy from these individuals.

T-Cell Responses

What is different about the protease inhibitors (compared to
the gloom we felt a couple years ago, as we looked at
developing a family of me-too nucleoside analogs) is that we
have seen people who we thought could not mount much of an
immune recovery, take their T-cells from under 50 into the
several hundreds, and sometimes almost back to near-normal
levels. The caution, of course, is that we do not have enough
followup to know how much of your immunity is really restored
by having your counts rise. There have been cases of patients
who have developed pneumocystis with high CD4 counts
(although there have always been such cases, even in the
original pneumocystis trial). Despite such concerns, I still
think we are on the verge of making some real progress with
the protease inhibitors, and with drug combinations [with or
without protease inhibitors included].

Business Incentive Issues

ATN: There is also concern in the community that the rules
for drug development have made it too difficult for companies
to justify their financial investment.

Feigal: The businessmen would be better at addressing that
issue than I. But one clear message that has been sent -- and
we certainly said it again at the National Task Force meeting
-- is that we would welcome a good accelerated approval
package for drugs in new classes. [The FDA's "accelerated
approval" regulations not only provide drugs sooner to
patients with serious or life-threatening diseases; they also
help companies by reducing the investment required before a
drug begins to generate revenue.]

When accelerated approval was first proposed, there were
discussions about whether it would be used primarily just to
approve new drugs in the same class. With a new class of
drugs, would you have to prove clinical benefit [without
using accelerated approval] for the first drug, and use
accelerated approval only for later drugs in the same class?
But today, so long as the clinical confirmation studies are
in place, we do think that accelerated approval is a viable
option [even for the first drug in a new class]. It knocks a
couple years off of the development time, compared to
traditional drug development. That allows the companies to
begin financially recouping their development costs for the
product at an earlier stage; if they still have patent life
on the product, they have exclusivity for more years. They
can continue to do the clinical studies funded in part on
revenue from the product, rather than having to put more
money into a product that is not yet generating income. And
there is also the issue of the cost of money; when companies
talk about the $250 million that it costs to develop a
product, about half of that is the cost of money [which is
reduced by having a faster development track].

Accelerated Approval and

New Indications

ATN: One of the criticisms I heard at the National Task Force
hearings is that accelerated approval would work for initial
approval of a drug, but was not available for additional
indications [because the need is not as urgent].

Feigal: That is not true; in fact, there are examples where
we have already used accelerated approval for an additional
indication. For example, clarithromycin was on the market for
non-AIDS indications, and we gave it accelerated approval for
MAC.

Part of the issue here is that the cycle of developing new
indications is relatively long, compared to the patent life
of the product. If you look at the time frame, of usually
taking about a year to design and launch the trial (not just
FDA time, but the whole time of the community and the
researchers in setting up a trial), and then enrolling the
trial for about a year and a half, and then having another
year of followup, and another half year to write up the
results if it is a big study, and then another four to six
months for us to review the results and get the indication
changed, you are looking at a cycle that can easily run two
to five years for an additional indication. So one of the
issues that has been raised in AIDS and in cancer is whether
that is a standard which results in the drug label not
keeping up with medical practice.

Viral Load and Early Approval

The other time that discussion comes up [of whether we will
always need to have clinical-endpoint studies eventually] is
the issue of what does it mean to have a validated surrogate
marker? We, at the FDA, already accept the fact that the
virus is relevant; and that changes in viral load mean that
you have an active drug; and that that, along with favorable
immunologic changes, form a reasonable basis for accelerated
approval. What we do not know is whether or not you can
individualize someone's therapy and get a better result by
changing their therapy as their viral load changes. We do not
know if early changes in viral load will predict a sustained
response some years later -- consider the Concorde
disappointment with the early changes in CD4 count [which
failed to predict better outcome later]. So we still need
clinical endpoints to be confirmed.

But in the future, we might reach the stage where the viral
load, or some other measure, or some combination measure, is
so highly predictable that it is a reasonable basis to change
clinical practice. Then we would be in a situation where we
are with cholesterol-lowering agents, and anti-hypertensive
agents; they are not labeled to prevent heart disease or
stroke, they are labeled to lower cholesterol or to lower
blood pressure. Right now we would not quite know what a
label would mean if it said that this drug was indicated to
lower HIV RNA in the serum. We are still not quite there yet,
although we are all excited by how promising these measures
look.

Another consideration is that many of the nasty surprises
from basing treatment on surrogate markers [such as a
notorious study of heart drugs, in which two drugs which
successfully suppressed abnormal heart rhythms also increased
the chance of death] did not come from the fact that the
markers were necessarily poor, but from the fact that the
drug had some other slow, damaging effect that was only
detectable by doing a clinical study and seeing what the
actual results were. So I think, for drugs in new classes, we
eventually need to know how good they are clinically. The
incentive for the companies is that accelerated approval gets
them marketing use of their product, at a time when they are
still pursuing the clinical question.

Here again we need a partnership with the community; the
community needs to remain convinced that it is important to
find out clinically how good these things are, and has to be
willing to participate in trials. We have passed the era of
placebo-controlled trials, but we can still learn a lot from
active-controlled trials, or dose-response trials, and we
need to have the commitment to do that.

Some companies are discouraged by how difficult it is to
clinical trials in HIV in the U.S., with the rapidly changing
therapy, with people discontinuing early in trials because of
the availability of a new trial or a new expanded access.
There needs to be that commitment together to find answers. I
think that commitment is still there; it just gets frayed
around the edges at times.

FDA Clinical-Trials Workshop

ATN: Kessler mentioned plans to set up a workshop on AIDS
clinical trials. What issues will this meeting address?

Feigal: The quickest way to organize this meeting may be to
use the existing advisory-committee mechanism; we can turn
the meeting into more of a workshop than a hearing. We will
probably try to do this in late spring or early summer.

A number of questions need to be aired, and relatively
quickly. For example, what are the most relevant comparison
arms? That depends on whether you do a traditional standard
therapy vs. new therapy, or whether you extend that design
into combinations (which we at the FDA already believe should
be done more vigorously), or whether you modify it even
further by taking a treatment-strategy approach, and leaving
some parts of protocols very wide open, in terms of use of
concomitant therapies.

Another important area is the role of larger trials, vs.
smaller, intensive, high-tech trials. This is not an either-
or, but how you best integrate those approaches.

Then there is the question of non-survival clinical
endpoints. Is it still relevant to use the old CDC
infectious-disease endpoints? Or are there more efficient
measures, such as counting them in a weighted fashion, or
evaluating more than just the first endpoint and also
counting subsequent ones?

On viral load, there is still much to be learned about which
viral load markers predict which clinical events, and how far
into the future do they predict. We know that CD4 counts have
predicted only for about a year. But we need to take a look
and see what we can learn, and design further end-point
research into the trials that are ongoing right now.

There are a number of interesting confirmatory trials
underway from the past accelerated approvals. The d4T trials,
the ddC trials, and many of the ACTG studies, will
potentially give us much data fairly quickly about CD4
counts, viral load, and clinical events.

We also need continued focus on special populations: the
design of pediatric trials, the issue of vertical
transmission, and design for the use of drugs in pregnancy,
including treating women with advanced disease who are
pregnant. Some of the newer compounds do not have the
genotoxic effects that many of the nucleosides have; they may
be safer for use in pregnancy, even aside from the issue of
vertical transmission.

I would see this as a two-day workshop to have people present
designs, and have panels; and probably use the advisory
committee mechanism at the end to figure out what the
immediate take-away messages are, and where the Antiviral
Drug Products Advisory Committee would like to comment --
messages that commercial sponsors might pay attention to as
they are designing trials.

There has been some misunderstandings that the reason that
large trials are not used, for example, is because the
sample-size calculations are not done correctly. That may
happen in some cases, but more often there is disagreement
among the designers of those trials over whether they are
ready yet to look for small or moderate effects, or whether
they are still screening drugs for larger effects. This
workshop will be a chance to put some of these issues on the
table and talk about some strategies.

We will not come up with a cookbook. There are many different
ways to develop drugs; we need to continue to use all the
ways we can, as we continue to find surprises when we modify
our designs, or try a different population. For example, we
are getting a new appreciation that we too rapidly concluded
that it was hard to benefit patients with very low CD4
counts.

I think this is an exciting time, as there is a renewed
interest not only in the proteases, but other combinations of
non-nucleosides, and the AZT plus 3TC combination, that give
us some reason to expect that we could make important
treatment advances if we try some of these new approaches in
trials.

The Empty Pipeline?

ATN: Many people are saying that after the protease
inhibitors, the "pipeline" for future development of AIDS
drugs is empty. How do you see this?

Feigal: It is hard to judge the pipeline. We thought the
pipeline was fairly empty after many of the non-nucleoside RT
inhibitors were abandoned three or four years ago; but
interestingly, the non-nucleoside RT inhibitors are still
around, and may, at the end of the day, in the right
combinations, have an important role. The proteases bloomed
rather rapidly at a time when people though the pipeline was
largely empty. If it seems empty at times, history shows that
it is cyclical.

Many advances that have been made in other fields, in chronic
infections and malignancies, have been through careful
construction of rational combination therapies. Even without
new drugs there is still plenty of work to be done on
combinations for HIV. For example, 3TC did not look
attractive before the combination result with AZT; it is
fortunate that the combination trial got done.

I do not see the pipeline as being terribly empty.

And there is always the left-field factor. Although we now
focus a lot on drugs that have a direct antiviral effect, it
is possible that some of the immunomodulatory therapies, or
therapies that try to target programmed cell death, or other
mechanisms of disease pathogenesis, might make a major
difference, particularly in some combinations. If you look
more broadly than just the direct antivirals, there is lots
of work to be done.

ATN: People are concerned that small companies are excluded
from drug development, because the rules make the process so
expensive.

Feigal: The possibility of a drug coming out of left field
does raise the issue of the small-company problem. The
biggest need of little companies has always been to have big-
company partners, just to keep the business going. A lot of
how well the little companies do will depend on the big
companies' interest in investing [which has clearly
diminished recently, in all of biotech, not just AIDS].

In terms of the FDA's response to small companies, there is
only one set of laws and regulations, and they apply equally
to all companies. However, we are given the flexibility to
interpret and to modify the requirements. We look for
opportunities to do problem solving with small companies.
Typically that has involved allowing them to try and get some
initial evidence of efficacy as efficiently as they can,
because that is usually the make-or-break for a new product.
If they say that their compound is an "antiviral," but when
they give it to someone, it does not affect any parameter
that they can measure, then probably the substance is not
worth pursuing as an antiviral. On the other hand, if it is
dramatically antiviral, the company will probably be able to
find partners to develop it.

One way we can help speed the development process is to allow
companies to keep the animal toxicology studies just ahead of
the human trials, so they do not have to do the whole
toxicology profile evaluation before they even start in
humans. The big companies sometimes use that strategy, too;
although often when they are fairly committed to a product,
they will do the whole animal workup before they even bring
it in [to the FDA]. It is case-by-case problem solving; it
depends on how much is known about the product, and the class
of products, and how worrisome the potential toxicity. We are
committed to try to find ways to solve problems with these
companies, and give them answers quickly. Problem solving and
rapid review is how we see our role.

ATN: What else was significant at the recent protease-
inhibitor meeting of the National Task Force on AIDS Drug
Development?

Feigal: Just the ability to get the companies to come and
talk about drug development, and to talk about their NDA
(drug approval) filing times in public, was an unusual event;
some people may not have appreciated that companies do not
usually like to do that, they sometimes even keep NDA filings
completely confidential. Kessler [FDA Commissioner David
Kessler, M.D.] was asked by the community to take a hard look
at the protease inhibitors and see what could be done.
Whether the meeting was an overall success or not, there were
things taken away, including a commitment by Merck to try and
find expanded access in their drug development, and to do
pediatric trials.

The companies are asking us what is the minimum safety
database [required for approval]. An unfortunate result of
the supply problem is that we do not have much safety
experience with some of the protease inhibitors; often there
are not even three dozen patients who have received the
product more than three or four months. This issue was
addressed at the meeting; if we cannot have expanded access
with ten thousand patients to look at safety [before drug
approval], what will we settle for? What came out was a
recommendation for several hundred patients with at least
half a year or so of exposure and followup. I think that can
happen relatively quickly, with the studies and the use of
these compounds already in the pipeline. A good outcome from
the meeting was preliminary discussion of what the
requirements should be, what are the parameters. This helps
the companies in their planning.

Kessler took the commitment to that meeting very seriously,
including personally lobbying the companies to get them to
come forth and talk about these issues fairly frankly. It was
interesting to me, knowing what they have told the FDA in
confidence, how willing the companies were willing to be
frank and open. If they seem to be excited about these
products, I think it is because of a sincere belief by the
companies that these could be potentially breakthrough
products for this disease.