AIDS TREATMENT NEWS Issue #204, August 5, 1994
During the last two years, new blood tests for HIV viral load have
increasingly been used in clinical trials and other scientific
research. There is great interest among researchers, as the early,
preliminary results of major studies are showing that changes in viral
load due to change in therapy can predict clinical benefit in
patients. Meanwhile, the tests are about to come into use (outside of
formal studies) in medical practice for individualizing patient care;
the challenge will be learning when to use them and how to interpret
the results. But it is widely agreed that a reliable blood test for
viral activity in the body -- if it can help predict which drugs will
be successful for a particular patient, as these tests do seem to do
-- could revolutionize HIV medicine by improving treatment with
existing drugs, as well as shortening the time required for new
treatments to be proven.
We believe that this development is more important than most
physicians, treatment activists, or even the scientists working with
the tests may realize, for the following reasons:
* Existing treatments do seem to work well at certain times for many
people. A scientifically validated way to guide each patient to the
best treatment for him or her could substantially improve patient
care now, without the need to wait for better drugs to become
* A reliable viral test would enable antiviral drugs -- and even some
immune-based and other kinds of therapies to be tested much more
rapidly and less expensively than under the current system. This
should greatly shorten the time required to make better treatments
Now it usually takes years, and hundreds (if not thousands) of
patients to show that an AIDS treatment improves survival, or reduces
the number of opportunistic infections. In contrast, blood tests often
show clear results in weeks, and relatively few patients may be enough
to produce statistical proof. But first, it must be shown that a
decrease in viral load caused by a change in drug therapy indicates a
real benefit for the patient.
* "Alternative" treatments -- those without a rich, well- connected
institution behind them -- could also be tested scientifically,
perhaps in small, community-based studies. Those which are worthless
can be discarded, and those which are found to be effective can be
targeted to the particular patients most likely to benefit.
* Similarly, traditional medicines being used around the world in HIV
treatment could be screened with these tests, potentially making
effective treatment available for the first time for millions of
people who cannot obtain high- priced pharmaceuticals. At the same
time, validating traditional remedies may provide new options to
improve treatment for those who already have access to conventional
* We believe that the biggest obstacle to AIDS treatment development,
the major reason it has not been productive, is that the combination
of high regulatory hurdles and lack of national will has made it
impossible for low-priority treatment ideas, those with only a few
champions and little or no industry or government support, to begin
to be tested so that they can build credibility if they work. Since
most major medical advances start as low-priority ideas, and prove
themselves through a series of surprises, the current system --
which limits the field to projects which already have major support
-- chokes off the wellspring of innovation, virtually guaranteeing
stagnation. This problem does not correct itself, due to peoples'
natural tendency to focus on what they already consider important,
instead of cultivating development paths for ideas which are
currently outside of their understanding and therefore may not
appear attractive. Giving the power to determine what works to
individual patients and physicians can allow new ideas to proceed,
breaking the existing monopoly on drug development and the resulting
choke hold on our future.
All these benefits depend, of course, on whether what is being
measured by the new tests is a useful indicator of how well a
treatment is working. It would seem logical that lowering the amount
of HIV in the blood would be an improvement. But experts are cautious,
because prior viral tests (especially the p24 antigen test) have been
too inaccurate to be very useful in drug development. Much remains to
be learned; but the information now available is encouraging.
Background: What Is HIV RNA?
The genetic information for almost all living things is stored in the
nucleus of cells, in a chemical called DNA. In the body, the same
information in DNA can be transcribed into RNA; then the information
in the RNA is translated into proteins, which determine the structure
and function of the cell.
Retroviruses (such as HIV), however, have their genetic information in
RNA. When the virus gets inside a cell, the information is transcribed
in reverse into DNA, which then becomes part of the cell's genetic
inheritance. The cell can then produce new viruses, or be damaged in
various other ways, sometimes producing abnormal cytokines which can
Each individual HIV virus has two copies of the RNA which specifies
its genetic information. The new tests for viral load detect this RNA,
and the test results are usually given as number of copies per
milliliter of blood plasma. For example, if someone get a test result
of 100,000 (a fairly high number), it means they have 100,000 copies
of the RNA (or 50,000 virus particles) per milliliter of plasma.
What a Blood Test Needs to Do to Be Useful
First, of course, any test used in research or in clinical care must
be accurate and reliable. Quality assurance is necessary; it can be
done by sending known samples to the labs that run the test. The
samples are coded so that the labs do not know what is in each; their
answers are then compared against the correct values.
But also, to be useful for drug development or for individualizing
patient care, a test must be validated -- that is, we need to know
that what the test measures has clinical usefulness. Different tests
are useful for different purposes.
For example, the T-helper count (CD4 count) clearly is useful for
prognosis -- predicting how an individual patient is likely to do. For
example, pneumocystis almost always occurs when the T-helper count is
under 200 or 250, and CMV retinitis when the count is under 50. (These
numbers are for adults; the numbers for young children are entirely
different.) But while the T-helper count clearly provides predictive
information, this does not automatically mean that a change in
T-helper count caused by a drug will indicate a corresponding change
in the prognosis of patients. In fact, the well-known Concorde study
of early use of AZT found that those taking the drug had higher
average T-helper counts than those who did not, and this difference
persisted for the three years of the study; however, for reasons not
well understood, it did not make much difference in death rates or
other clinical outcomes -- casting doubt on the use of the T- helper
count as an early signal of whether a treatment is working.
In analyzing a clinical trial where some patients are randomly
assigned to one treatment regimen and some to another, it is easy to
tell if one treatment raises the T- helper count (or other blood-test
result) more than the other. Also, it is statistically straightforward
to determine if one treatment group does better than the other. But it
takes more complex statistics to tell that the blood test is actually
showing that the drug is working -- to rule out the possibility that
the drug helps the patient, and also just happens to change the blood
test independently. In the latter case, the test could be worthless
for indicating in advance whether another drug is also going to work.
This means that a trial with clinical endpoints (which may take
several years to generate enough deaths or serious illnesses to
produce statistical proof that one treatment is better than the other)
may be necessary to validate a new blood test. Fortunately, the tests
for HIV RNA can be run with frozen blood samples, which have already
been saved from past trials, where the outcome for each patient has
already been recorded. This retrospective validation is now well
underway, and preliminary results from three major trials, presented
by different groups at a recent scientific meeting, have suggested
that changes in HIV RNA do indeed predict changes in clinical outcome
cause by drug treatment. These results are preliminary, and questions
[Note: An alternative strategy would be to not wait for the validation
step, by making the reasonable presumption that, other things being
equal, a low viral load in patients is better than a high viral load.
Then a drug which reduced viral load could be considered effective for
purposes of approval, unless there was other information which
rebutted the presumption; safety, of course, would have to be proven
separately, but proof of efficacy, not safety, is the bottleneck in
drug development. The benefit of this approach is that small companies
would have a chance to develop and market AIDS treatments; and large
companies would have the ability and incentive to develop their drugs
rapidly. We believe that this strategy would serve the public better
than current approaches for discovering and developing better AIDS
treatments; but for various reasons it would be difficult to sell
What we do unquestionably need is better information on how to use and
interpret these tests. There is a danger that future studies will
focus too much on definitive academic proof that the tests can be
useful -- which we are learning anyway. Instead, they should focus on
getting practical information for physicians who are caring for
Testing for HIV RNA -- Two Different Technologies
Two completely different methods, quantitative PCR (currently being
developed as a standardized test kit by Roche Molecular Systems,
Somerville, New Jersey) and branched DNA (developed by Chiron
Corporation, Emeryville, California), are now being widely used in
research to test for the amount of HIV RNA in blood. A short
explanation of how they work appeared in "Better Tests for HIV
Activity; Interview with Mark B. Feinberg, M.D., Ph.D., AIDS Treatment
News # 186, November 5, 1993. These two different kinds of tests
measure the same thing, and usually they give comparable results.
But each kind of test has different strengths and weaknesses. At
present, the branched DNA test has a cut-off at 10,000 copies of HIV
RNA per milliliter; it cannot measure values lower than that. (Soon
the cutoff will be reduced to 5,000 copies per milliliter.) The
quantitative PCR testing service which is now commercially available
through reference laboratories can go down to 200 copies per
milliliter. No one yet knows if keeping the counts under 5,000 is good
enough for maintaining health; if not, some patients will need to use
PCR to track lower counts.
On the other hand, the branched DNA test appears to be more consistent
than quantitative PCR in measuring different subtypes of HIV (although
Roche points out that what matters is change in time within each
patient, so this might not be an issue). Which test is more accurate
is in dispute. Branched DNA is much easier than PCR to run in the
laboratory, which should help to make it more accepted in widespread
Eventually one of these testing technologies may prove better overall
than the other. But at this time, researchers using the tests in
clinical trials are about equally interested in both.
Anyone using HIV RNA tests now, before they are part of standard
medical practice, should understand that, because of how medical tests
are regulated, both quantitative PCR and branched DNA tests are
becoming available to physicians and patients in a test format that
does not require FDA approval. Therefore, the customer must be extra
careful about getting a good-quality test.
In the U.S., medical tests are traditionally regulated in two
different ways. Usually, the test is provided in the form of a
standardized kit, which contains all the materials and instructions
required so that any qualified laboratory can test samples. The FDA
must approve such a kit before it can be widely used in the U.S. --
and therefore it takes longer for the standardized test kit to become
available for medical practice.
But a company can also set up its own sophisticated laboratory,
sometimes called a reference laboratory, which can perform the testing
service. To offer this service without a kit is much more difficult
than running a kit produced by somebody else, since the lab has to buy
or make all of the materials itself, check their purity, etc. In this
case, the company can offer the test commercially as a service, not a
product, and this testing service is not regulated by the FDA. Because
FDA approval can take some time, the availability of the standardized
kit can be a year or more behind the availability of the same test
through a testing service by a reference laboratory.
During this time before FDA approval, how can one be confident that
the test is accurate? We suggest purchasing the test through the
companies which are experts in this area -- Roche Biomedical
Laboratories for quantitative PCR, and Chiron Corporation for branched
DNA. Other companies can run these tests through various agreements
with the patent holder, setting up their own reference laboratories;
but then the patent holder does not run these laboratories and cannot
guarantee their quality. Both Roche and Chiron have immense
investments in their technologies (which not only can test for HIV,
but will also have many other uses in medicine); and both know the
technologies at least as well as anyone else. They have the incentive
and also the means to make sure to have good quality control before
offering their tests to physicians. Also, government and other
researchers have done considerable work with their tests and would
quickly notice any serious problem. These protections are not
available if one selects a "brand X" HIV RNA test.
And later, even after the FDA-approved kits become available, there
will still be concern that the laboratory personnel who run the tests
be properly trained, to avoid significant variation due to lack of
The Challenge: Interpreting Test Results
When one has one's blood tested for HIV RNA, the result comes back as
a number of copies of viral RNA per milliliter; this number can range
from a low of 200 to a high of over a million. Sometimes the result
will be that the number of copies was below the cut-off for the test
-- 200 copies for PCR, 10,000 copies (soon to be 5,000) for branched
DNA. A single number is hard to interpret; it is more important to
watch how the number changes over time.
HIV RNA is a very sensitive measure, and can change considerably in
response to many things. Some persons with HIV can have as much as a
three-fold variation in the number from day to day, for no known
cause; others have less variability. Immune stimulation -- for
example, from a flu shot -- can cause a large temporary increase.
Because of this variability, a small change -- even a two-fold or
three-fold change -- in a single number might not be meaningful.
AZT and other approved AIDS treatments often cause a drop of about one
log (ten fold) in HIV RNA level; however, this drop is usually not
sustained, probably because the virus develop resistance to the drugs.
Saquinavir, the experimental Hoffmann-La Roche proteinase (protease)
inhibitor, did about the same -- close to a one-log drop, but not
Many researchers believe that, to be a major advance in AIDS
treatment, a new drug, combination, or other treatment regimen should
produce at least a two-log (100 fold) drop in HIV RNA levels, and that
this drop must be sustained -- for many months, hopefully for years.
The experimental protease inhibitor now being tested by Merck & Co.
has produced a two- log or greater drop in a few patients, but again
the decrease was not sustained; within a few months, the HIV RNA
levels went back up. (For a look at indications of clinical benefit
from even the temporary drop, see "Antivirals and Immune Recovery:
Interview with Michael S. Saag, M.D., AIDS Treatment News #200, June
Since an ideal treatment will be hard to find, we will have to settle
for less at first. Perhaps several treatments which each alone have a
smaller effect could be combined to give better results. As HIV RNA
tests come into wider use, physicians and patients will be trying all
kinds of treatment regimens while watching the viral load. This could
lead to important advances in the search for better treatments or
Still there are questions. The HIV RNA test only tells how much virus
is in the blood; it does not tell how pathogenic that virus is. Also,
the blood level of the virus might or might not be a good indicator of
what is going on elsewhere in the body. Much remains to be learned;
but meanwhile, it does seem reasonable to use the level of HIV RNA in
the blood to help guide treatment decisions, with the goal of lowering
the viral load and keeping it low.
The quantitative PCR test for HIV RNA is now available to physicians;
in fact, it has been available since late 1993, although this has not
been widely known. Chiron has announced that its branched DNA test
will be available to physicians starting August 15.
A major scientific meeting on these issues -- Surrogate Markers of
HIV: Strategies and Issues for Selection and Use -- will be held
October 12-14 near Washington D.C.; it is being organized by Cambridge
Healthtech Institute, Waltham, Massachusetts. It's conclusions may
soon become obsolete, however, as important new data from ongoing
trials will become available in the months after the meeting. These
new trials, prospectively designed to include testing for HIV RNA (as
compared to previously-run trials which fortunately happened to have
some frozen blood in storage) are important for various reasons. For
example, the old trials did not process the blood properly for the new
HIV RNA tests; the tests can still be run, but some of their
sensitivity is lost. HIV RNA testing may work even better in ongoing
trials, where this is not a problem.
A major future regulatory issue is whether HIV RNA will have to be
re-validated for each new class of drugs -- requiring a large,
long-lasting trial with "clinical endpoints" (death or major
opportunistic infections) before the viral load tests would be
accepted as proving efficacy of the new drug class. [For example, if
the viral load as measured by HIV RNA is shown to predict patient
benefit from nucleoside analog treatments (AZT, ddI, etc.), can it
then be used similarly for protease inhibitors, or will another major
trial be required to validate it for protease inhibitors first?] If
the FDA decides that the answer is yes -- and it may be leaning that
way at this time -- then the result could be to add years to the
development time of every new class of drugs, and create a
multimillion dollar disincentive to the development of new kinds of
How to Order HIV RNA Tests
* Quantitative PCR for HIV RNA: Physicians can order this test through
Roche Biomedical Laboratories, Research Triangle Park, North
Carolina; the customer service number is 800/533- 0567, 8 a.m. - 6
p.m. Monday through Friday, 8 a.m. - noon Saturday, Eastern time.
* Branched DNA for HIV RNA: Chiron Corporation will offer this test to
the HIV clinician starting August 15, 1994. Sample collection,
shipping, and other client services will be handled by Nichols
Institute, a commercial laboratory which can serve clients
throughout the U.S. Information about this test can be obtained by
Note: Recent preliminary information on the validation of HIV RNA (as
an indicator of benefit of drug therapy) is not yet public, so it is
not described in the references below.
Bagnarelli P, Valenza A, Menzo S, and others. Dynamics of molecular
parameters of human immunodeficiency virus type 1 activity in vivo.
Journal of Virology. April 1994; pages 2495-2502.
Dailey P, Lindquist C, Collins M, and others. Detection and
quantitation of HIV RNA from lymph node tissue using a branched DNA
(bDNA) signal amplification assay (Chiron Quantiplex* HIV-RNA assay).
The First National Conference on Human Retroviruses and Related
Infections, December 12-16, 1993, Washington, DC [abstract #313].
Detmer J, Collins M, Zayati C, Irvine B, Kolberg J, and Urdea M.
Comparative quantification of gold standard HIV-RNA representing
subtypes A-E using the Quantiplex* HIV-RNA assay. The First National
Conference on Human Retroviruses and Related Infections, December
12-16, 1993, Washington, D.C. [poster #204].
Gupta P, Kokka R, Neuwald P, Kern D, Rinaldo C, and Mellors J.
Expression of HIV-1 RNA in plasma correlates with the development of
AIDS: A Multicenter AIDS Cohort Study (MACS). Ninth International
Conference on AIDS, June 7-11, 1993, Berlin [abstract #2481]
Holodniy M, Eastman PS, Mole L, and others. Reduction of plasma HIV
viral load during ddI/ZDV combination therapy in the presence of ZDV
resistance. The First National Conference on Human Retroviruses and
Related Infections, December 12-16, 1993, Washington, D.C. [abstract
Holodniy M, Mole L, Winters M, and Merigan TC. Diurnal and short-term
stability of HIV virus load as measured by gene amplification. Journal
of Acquired Immune Deficiency Syndromes. April 1994; volume 7, pages
Mellors J, Kingsley L, Gupta P, and others. Detection of plasma HIV
RNA by branched DNA (bDNA) signal amplification predicts early onset
of AIDS after seroconversion. The First National Conference on Human
Retroviruses and Related Infections, December 12-16, 1993, Washington,
D.C. [abstract #274].
Mulder J, McKinney N, Christopherson C, Sninsky J, Greenfield L, and
Kwok S. Rapid and simple PCR assay for quantitation of human
immunodeficiency virus type 1 RNA in plasma: Application to acute
retroviral infection. Journal of Clinical Microbiology. February 1994;
Pachl C, Saxer M, Elbeik T, and others. Quantitation of HIV-1 RNA in
plasma using a branched DNA (bDNA) signal amplification assay:
Evaluation of specimen collection and stability. The First National
Conference on Human Retroviruses and Related Infections, December
12-16, 1993, Washington, D.C. [abstract #312].
Pachl C, Elbeik T, Saxer M, and others. Quantitation of HIV-1 RNA in
plasma using a signal amplification branched (bDNA) assay. 93rd
General ASM (American Society for Microbiology) Meeting, May 16-20,
1993, Atlanta, GA.
Saksela K, Stevens C, Rubinstein P, and Baltimore D. Human
immunodeficiency virus type 1 mRNA expression in peripheral blood
cells predicts disease progression independently of the numbers of
CD4+ lymphocytes. Proceedings of the National Academy of Sciences,
USA. February 1994; volume 91, pages 1104-1108.
Sninsky JJ and Kwok S. The application of quantitative polymerase
chain reaction to therapeutic monitoring. AIDS. 1993; volume 7
(supplement 2), pages 529-534.