Convergent Combination Therapy

With the recent news about a three-drug combination which
stopped HIV infection in laboratory tests, an approach called
"convergent combination therapy" has generated more interest
among patients, physicians, and the public than any other recent
development in AIDS treatment. The media handling of the story
has angered some scientists and activists, who point out that the
idea is not new, and that there are a number of reasons why it
may work less well in people than in the laboratory. There seems
to be a rough consensus among informed observers that the
development may indeed be important, that it has been overplayed
by the media, and that there is nothing magic about the number
three or about the particular three-drug combinations being
discussed so far -- certain other combinations might work equally
well.

What Is Convergent Combination Therapy?

Drug combinations have long been important for many diseases
which are difficult to treat -- for example, cancer and
tuberculosis -- and AIDS experts have long suspected that HIV
treatment, too, will require more than one drug. But there are
different theories guiding the selection of which drugs are
likely to work well together.

The usual approach has been to combine drugs which work at
different stages in the life cycle of the target organism. (One
successful example of such a combination is called co-
trimoxazole, best known by its brand names Septra or Bactrim,
which is used for treating or preventing certain infections. It
consists of two drugs, trimethoprim and sulfamethoxazole,
combined in a fixed ratio, which block two different enzymes
required by the target organisms.)

In AIDS, this conventional approach has called for combining
drugs which work at different stages in the life cycle of HIV, in
order to make more powerful treatments, and in particular to slow
the development of drug resistance (since a mutant virus which
happens to be resistant to one drug will, by chance alone, most
likely be susceptible to the other; in that case, since the new
virus still cannot grow, the dangerous mutation will be lost).
Examples of this approach might be to combine AZT with a protease
inhibitor, or with a tat inhibitor. Unfortunately this theory
has had little practical test in AIDS, because all of the
approved anti-HIV drugs, as well as many of the experimental ones
now being tested in people, target only one point in the viral
life cycle, the enzyme reverse transcriptase (which is essential
for HIV reproduction, but not found at all in uninfected human
cells).

The idea behind convergent combination therapy is that
carefully chosen drugs against the same target, the reverse
transcriptase enzyme, may work even better together than drugs
against different targets. The theory is that reverse
transcriptase must perform critical functions for HIV, and if it
is changed too much (in order to evade various drugs), it may not
work as well, and perhaps may not work at all.

For example, any HIV which is resistant to AZT has been
found to have one (or more) of only five different mutations in
the gene which codes for the reverse transcriptase enzyme -- at
codons (positions in the gene) 41, 67, 70, 215, and 219; the one
at 215 is the most serious cause of AZT resistance. Of course
there might also be others which are currently unknown. But the
point is that there are only a few different mutations which can
block the effect of AZT and still allow the virus to reproduce at
all. Other mutations can convey resistance to ddI or to other
drugs.

The idea of convergent combination therapy is to find
particular combinations of drugs, all targeted against the
reverse transcriptase enzyme, such that there is no combination
of mutations which could convey resistance to all the drugs
without also making the reverse transcriptase unable to function.
Then any variant of HIV would either be susceptible to one or
more of the drugs, and stopped by it, or would be unable to
reproduce because of the incompatible mutations required to
provide the various drug resistances. If the theory works,
therefore, there would be no evolutionary pathway for the virus
to develop resistance to all of the drugs.

The reason this idea is fairly new is that it may be most
applicable to viruses -- not to other disease-causing organisms
such as bacteria, which have much more complex genetic machinery
than viruses, including various ways to disable drugs without
necessarily affecting the function which the drugs are targeting.

History: The Harvard Experiments

According to a detailed news release dated February 16 from
Massachusetts General Hospital, convergent combination therapy
was first proposed by Yung-Kang Chow, an M. D. -Ph.D. student at
Harvard Medical School, working under the supervision of Martin
S. Hirsch, M. D., a noted AIDS researcher and Professor of
Medicine at Harvard. Results of an early test of the idea,
conducted at Massachusetts General Hospital, were presented at
the VIII International Conference on AIDS in Amsterdam, in July
1992.(1) This poster reported that a genetically engineered
virus, artificially given several mutations to confer
simultaneous resistance to AZT, ddI, foscarnet, and pyridinone
(an experimental antiviral being developed by Merck & Co., also
known as "L661" or "the L-drug") was no longer able to reproduce.

A recent paper in Nature(2) led to the current public
interest. It describes two different series of experiments. In
one, viruses were artificially given various mutations, for
example at positions 215 and 219 (AZT resistance), 74 (ddI
resistance), and 103 (resistance to pyridinone and certain other
experimental drugs with a similar mechanism of action, such as
nevirapine). When all these mutations were combined in a single
virus, the reverse transcriptase (RT) activity of that virus was
undetectable, and the virus became non-infectious. But when only
some of these mutations were combined (for example, the first
three, giving resistance to AZT and ddI but not to pyridinone,
etc.) , then the virus was still infectious (although somewhat
less so than the original virus with none of the mutations).

The second series of experiments detailed by Chow and others
in the recent Nature paper tested combinations of two or more
anti-HIV drugs, at concentrations which can be achieved in the
body, to see whether they could prevent or stop an HIV infection
in a laboratory culture of human cells. The combinations (1) AZT
plus ddI; (2) AZT plus alpha interferon plus the Hoffmann-La
Roche protease inhibitor Ro31-8959; (3) AZT plus alpha interferon
plus soluble CD4; and (4) AZT plus pyridinone, all failed to do
so (although AZT plus pyridinone almost worked). However, the
three-drug combination AZT plus ddI plus pyridinone completely
prevented the infection in the laboratory test. Even when the
drugs were added at the peak of HIV infection, seven days after
the virus was added, the three-drug combination of AZT plus ddI
plus pyridinone stopped spread of infection in the laboratory
cultures.

After 49 days of the three-drug treatment, the cells were
placed in a drug-free medium for an additional 45 days, giving
ideal conditions for HIV to grow. But after the 45 days, there
was no evidence of the virus; nothing was found even with a PCR
test so sensitive that it could detect a single copy of the HIV
genetic information within DNA from 100,000 cells. This does not
mean that the virus would similarly be reduced to such low levels
in people -- for a number of reasons, some of which are discussed
below. But it does suggest that certain combinations (for
example, AZT plus ddI plus pyridinone) might work much better as
antivirals than combinations now in use (for example, AZT plus
ddI).

The virus used in this experiment was a clinical isolate
(i.e., obtained from a patient), not a laboratory strain or a
product of genetic engineering. The particular patient's virus
was selected because previous research had suggested that it
included a number of variants of HIV.

The Nature article mentioned two other three-drug
combinations which stopped HIV infection in laboratory cultures:
AZT plus ddI plus nevirapine (BI-RG-587); and AZT plus ddI plus
foscarnet (an intravenous drug approved for anti-CMV use, which
is also known to have some anti-HIV effect). No details were
given, however; a manuscript on these results is in preparation.
The success of these other combinations (in the laboratory)
illustrates that "breakthrough suppression is not unique to a
particular convergent combination. Regardless of the mechanism
of breakthrough suppression, as reverse transcription is
necessary to create HIV-1 variants, complete inhibition of this
process using convergent combination regimens should prevent
emergence of multiply resistant viruses as well as any immune
escape mutant viruses." (page 652)

The work at Harvard/Massachusetts General Hospital was
funded by three different units of the U. S. National Institutes
of Health: the National Cancer Institute, the National Institute
of Allergy and Infectious Diseases, and the Medical Scientist
Training Program.

Related Research

Some of the drug-resistance mutations in the reverse
transcriptase enzyme are known to interact with each other, so
that viruses containing certain combinations of mutations may
behave differently than expected in laboratory experiments. For
example, a mutation at position 74 which causes ddI resistance
can also suppress the AZT resistance caused by certain other
mutations.(3) This data suggested that AZT resistance and ddI
resistance might not be able to co-exist in a single virus. But
unfortunately, by the time of the VIII International Conference
on AIDS in Amsterdam last July, some groups reported early data
showing that virus resistant to both AZT and ddI could be found
in some patients who had first received AZT treatment and then
switched to ddI. (4,5)

Other experiments at the Wellcome Research Laboratories in
Kent, U. K., found another mechanism by which HIV may be limited
in its ability to develop multi-drug resistance to reverse-
transcriptase inhibitors.(6) In these tests, a virus was
constructed with mutations associated with resistance to AZT,
ddI, and non-nucleoside reverse transcriptase inhibitors
including pyridinone and nevirapine -- as was done by Chow and
others in the Harvard group -- except that this virus was given a
different mutation (at position 181) to confer resistance to the
non-nucleoside drugs, and it did not contain one of the AZT
resistance mutations (at position 219) that was present in the
virus made by the Harvard group. Unlike the multiply-resistant
virus in the Harvard study, the virus with these mutations could
reproduce. However, the mutation at position 181 suppressed the
effect of an AZT- resistance mutations, so that virus could be
inhibited by AZT. Although at first this result would seem to
weaken that of the Harvard group, in that virus with (different)
mutations known to cause resistance to the same three drugs was
able to reproduce, the practical result is the same, since that
virus had become susceptible to AZT again, so the three-drug
combination should be effective against it.

Confusion in the News

One widespread confusion among readers of the news-media
reports of convergent combination therapy concerns the identity
of the drugs being discussed (other than AZT and ddI). The media
picked up the name pyridinone as the third drug used in the
three-drug combination for most of the experiments in the recent
Nature article. That name had not been widely known before the
current publicity, and people have often not realized that it is
another name for Merck's "L-drug," better known as L661 (or
L697,661), as mentioned above.

Another confusion is that while the Nature article largely
reported on AZT plus ddI plus pyridinone, the major clinical
trial of convergent combination therapy, planned to start in a
few months, will use a different convergent combination, AZT plus
ddI plus nevirapine. Nevirapine (also known as BI- RG-587), an
anti-HIV drug being developed by Boehringer Ingelheim
Pharmaceuticals Inc., is similar to pyridinone in that both are
non-nucleoside reverse-transcriptase inhibitors; and the drugs
show cross resistance, meaning that virus which has become
resistant to one will also be resistant to the other even before
being exposed to it (suggesting that the mechanism of action of
the two drugs is similar). The combination with nevirapine was
tested by the Harvard group and found to work; this was reported
in the Nature article, but few details were given.

Another confusion stems from a short article in the current
Business Week (March 8, page 37) which includes the statement
that "The three-drug approach worked with only certain strains of
HIV -- variations of the most common form, known as HIV-1 -- but
it didn't work with other strains." We called the reporter, who
told us that this statement referred only to the tests of whether
a virus artificially given certain combinations of mutations
could reproduce (not to the more direct and more important tests
of whether a three-drug combination given to wild-type virus
could stop its growth). However, the Harvard group only did this
test on one kind of virus, and it did not have any failures
(cases in which the virus could reproduce despite the mutations
which conferred three-drug resistance).

While we could not definitively track down this confusion by
press time, we suspect that the Business Week report may have
referred to the Kent, U. K. results mentioned above. Here a
virus remained able to reproduce despite having artificially-
inserted mutations associated with resistance to the same three
drugs that the Harvard group used. (Some of the mutations were
different from those used at Harvard, however, so the different
results concerning the viability of the virus are not
contradictory.) But in practical terms, the mutant virus was
still susceptible to the three-drug combination, because one of
the mutations canceled out the resistance provided by another,
making the virus become susceptible to AZT again. Therefore the
three-drug convergent combination therapy should still work
against the virus tested in Kent, since no known combination of
mutations leaves it able to reproduce and also resistant to all
three of the drugs.

Why Convergent Combination Therapy Might Not Work in People

To keep perspective on the potentially important but still
unproven idea of convergent therapy, it is important to remember
that most drugs and therapies which work well in the laboratory
do not turn out to be useful in people. There are many reasons
for this. Here are only some of the reasons why convergent
combination therapy might not work:

* In the body, HIV infects a number of different kinds of
cells -- and drugs can work differently in different cells. For
example, AZT works much better in T-helper cells than in
macrophages, which also are infected with HIV. Before AZT can be
effective, it must be chemically transformed to its triphosphate
form after it enters the cell, and some cells are better than
others at producing this chemical reaction.

If any of the three or more drugs used together for
convergent combination therapy fail to reach effective levels in
certain kinds of cells, then the virus could still grow in those
cells, and the therapy would not be completely effective.

* New, unknown mutations might enable HIV to become
resistant to any particular combination therapy. The large viral
burden now known to exist in the lymph nodes during HIV infection
gives plenty of chances for such a mutation to develop, if one is
biologically possible.

* Different combinations of known mutations, other than
those which have been tested so far, might also be able to confer
simultaneous resistance to drug combination without preventing
the virus from reproducing.

* Reverse transcriptase inhibitors will not stop viral
activity in cells which have already been infected. Even if
convergent combination therapy prevents further spread of
infection, the virus in chronically infected cells could continue
to cause problems. These cells could stay in the body for a long
time. (In the laboratory cell culture experiments, in contrast,
infected cells die and stop producing virus -- meaning that
eliminating the virus in cultures does not imply that the same
drugs will eliminate it in the body.)

* Until drug combinations are tested in people, no one can
be sure they will be safe -- a question which is especially
critical with drugs which will have to be taken indefinitely.
There could be drug interactions which are harmful, or which make
one or more of the drugs ineffective.

In the press release from Massachusetts General Hospital on
convergent combination therapy, repeated cautions about not
misinterpreting the results are underlined or in bold type.

Clinical Trial Plans

The major clinical trial of convergent combination therapy,
named ACTG 241, in expected to begin in a few months at sites
throughout the U. S. -- hopefully starting enrollment by May.
Volunteers will be randomly assigned to receive either AZT plus
ddI plus nevirapine, or AZT plus ddI plus a placebo. The trial
will last one year. Originally it was scheduled to recruit 200
volunteers, but that number has been increased to 400. Sixteen
sites are expected to participate.

A small study now ongoing at the University of Alabama in
Birmingham is testing the combination in eight volunteers. This
study is giving nevirapine to three groups of volunteers: one of
which had been taking AZT alone, another taking ddI alone, and
the third taking the combination of those two; only this last
group will serve as a test of the convergent combination when the
nevirapine is added. The main goal of this trial is to check
that the combinations are safe, as ddI plus nevirapine has not
previously been given to patients. The study is also measuring
p24 antigen levels and is doing viral cultures, in order to get a
quick estimate of whether or not the convergent combination
therapy is working.

Another study, which is recruiting now in Philadelphia,
Pittsburgh, and Providence, may test a different convergent
combination (with pyridinone, the Merck "L-drug," combined with
AZT and ddI), but the decision to try all three drugs has not
been made yet. For now, the study will only test two drugs,
comparing AZT alone with AZT plus pyridinone. Merck recently
decided to begin a pharmacokinetic study to show that ddI and
pyridinone can safely be used together; if this study is
successful, then ddI might be added later to both treatment
groups, giving a comparison of AZT plus ddI with the three-drug
convergent combination.

To be eligible for this trial, volunteers must have a T-
helper count under 250. Volunteers do not need to have been on
any antiretroviral therapy in the past; they can qualify whether
or not they have used AZT, ddI, or ddC, but if they are currently
using ddI or ddC, they will need to stop those drugs two weeks
before entering the trial. (There might also be a two-week
washout for AZT, and a four-week washout for d4T.) This trial
will last 24 weeks, and weekly visits to the medical center will
be required for the first eight weeks, then less frequent visits
will be required. After the 24 weeks, L-661 should be available
on an open label basis to participants, if the drug still looks
promising at that time.

Forty volunteers are now being recruited at each site. For
more information, call the University of Pennsylvania in
Philadelphia, 215/662-2473; or the Pitt Treatment Evaluation Unit
in Pittsburgh, 412/647-8125; or the Miriam Hospital/Brown
University in Providence, 401/331-8500 x2928.

Comment: The Future

We are concerned that, no matter how well convergent
combination therapy might turn out to work in people, there seem
to be no plans or institutional means to get it to many patients
for two years or more. The major trial may take several months
to begin and then it will run for a year, with little chance of
data being released before that year is up. Then it will likely
take months for all the data to be collected from the trial
sites, checked, analyzed, delivered to the FDA, and published.
Then, if that trial shows that the treatment works well,
nevirapine could probably be approved for marketing, a process
likely to take several more months at least.

What about parallel track, or accelerated approval, as
faster means to get the treatment to patients in the meantime?
The problem is that, as far as we know, there are no plans to
obtain the data needed to support either of these programs, until
after the major trial is finished and analyzed. It is hard to
believe that parallel track, let alone accelerated approval,
could be based on only the eight patients now being studied in
Birmingham. (The 120-patient trial discussed above may or may
not decide to test the three-drug combination.)

What about the treatment underground? We have heard that a
nevirapine underground will be difficult, because the published
information on how to synthesize the drug would yield too
expensive a product. Therefore, expert chemists would first need
to develop a better synthesis procedure, and it seems unlikely
that this substantial effort will come together underground when
there is so little clinical evidence to prove that the treatment
is useful. (We do not know how difficult it would be to
synthesize pyridinone, if the combination including this drug
should turn out to work.)

What should be done? Two different plans should be pursued.
First, we need a rapid trial of convergent combination therapy
with more than the eight patients in Birmingham but fewer than
the several hundred in the major study now being planned. This
proposed trial would look at safety, at markers of viral
activity, and at T-helper count and other markers of immune or
general health. It should run for several weeks and then release
the data available at that time, while offering the patients
ongoing treatment and collecting longer-term followup data. One
central question this trial would ask would be whether patients
who continue to deteriorate on other therapies are likely to
improve when switched to this one. If so, the unapproved drug in
the combination (probably nevirapine or pyridinone) could be
given accelerated approval, or made available through parallel
track if the company is willing to pay the expense. (If not,
there is no legal way to allow patients to pay the cost of
parallel-track distribution, and governments are very unlikely to
pay, so patients will be deprived of the drug. It is rumored
that there have been a number of cases in the past where the FDA
has pleaded with a company to make its drug available under an
early-access program but the company has refused; the information
remains confidential, however, so the public does not know, and
cannot bring pressure to obtain the drugs or to reform the
system. The only way we know around this problem is to push for
accelerated approval, which companies are very eager to cooperate
with.)

Second, we should do more laboratory testing to find out
whether combinations of reverse transcriptase inhibitors already
approved or otherwise available (AZT, ddI, ddC, d4T, foscarnet)
show promise as convergent combination therapy. If combinations
of existing drugs work in the laboratory and then work in small
human trials, physicians could start using them for patients with
no alternatives, without waiting for official action to be
completed.

Making the system work so that it does the right thing with
convergent combination therapy could be an early task for the new
"AIDS czar" (AIDS coordinator, or special assistant to the
president) when one is appointed. (For news about possible
candidates for this job, see story in The New York Times,
February 28, 1993.)

References

1. Chow Y-K, Hirsch MS, Merrill DP, and others. Replication
incompatible and replication compromising combinations of HIV-1
RT drug resistance mutations. VIII International Conference on
AIDS, Amsterdam July 19-24 1992, abstract number PoA 2450.

2. Chow Y-K, Hirsch MS, Merrill DP, and others. Use of
evolutionary limitations of HIV-1 multidrug resistance to
optimize therapy. Nature. February 18, 1993; volume 361, pages
650 to 653.

3. St. Claire MH, Martin JL, Tudor-Williams G and others.
Resistance to ddI and sensitivity to AZT induced by a mutation in
HIV-1. Science. September 27, 1991; volume 253, pages 1557-
1559.

4. Eron JJ, Chow Y-K, Bechtel LJ and others. Interactive
effects of AZT- and ddI-selected HIV-1 reverse transcriptase (RT)
mutations. VIII International Conference on AIDS, Amsterdam July
19-24 1992, abstract number PoB 3580.

5. McLeod G, Mayers D, McCutchan F, Sanders-Buell E, Hammer, S.
Dideoxynucleoside resistance patterns in clinical isolates of
HIV-1. VIII International Conference on AIDS, Amsterdam July
19-24 1992, abstract number ThA 1569.

6. Larder, BA. 3'-Azido-3'-deoxythymidine resistance suppressed
by a mutation conferring human immunodeficiency virus type 1
resistance to nonnucleoside reverse transcriptase inhibitors.
Antimicrobial Agents and Chemotherapy. December 1992; volume 36,
number 12, pages 2664-2669.