GMHC Treatment Issues 1995 Sep 1; 9(9): 7
Viral load, the concentration of HIV particles in the blood,
may be a less stable marker in individual patients than some
think, according to one study presented at the ICC Conference
(abstract 0538) by Janet Raboud, Ph.D., and a group from the
University of British Columbia in Vancouver. The researchers
suggest that more must be known about the variability of
assays like PCR (polymerase chain reaction) before they are
used to manage patient care (initiating treatment or
switching therapies), as many doctors are already doing.
The Vancouver group obtained samples from 30 clinically
stable patients who had no apparent infections during the
four week study and for at least three months prior to the
study. Researchers took three samples from each patient over
the course of one week. Four weeks later, they took three
more samples from each patient. Then they noted the highest
and lowest viral load results for each person (as measured by
Roche's Amplicor PCR kit, which measures the concentration of
HIV RNA) and calculated his or her maximum variation over the
The median variation for all the patients was .83 log (a log
is a factor of ten -- one log equals a ten-fold variation and
.83 log equals a 6.76-fold variation). The median range for
patients with more than 500 copies of viral RNA per ml was
.76 log (5.75-fold), and the median range for those with
fewer than 500 copies of HIV RNA per ml was 1.7 log (50.1-
fold). (Roche does not claim that the assay is very accurate
below 400 copies per ml.)
Although many laboratories have reported that the viral load
assay is highly precise (that different laboratories testing
the same sample will get the same result), there is only a
small amount of published data on the stability of viral load
in individual patients. The Vancouver study is the largest
one to address that question.
One other study, by Robert Coombs, M.D., of the University of
Washington, found that four patients had stable viral load
results over 40 to 60 days.1 A study by Mark Holodniy, M.D.,
of the Palo Alto (California) Veterans Affairs Hospital,
reported a less than 0.2 log (1.6-fold) variation in viral
load for seven out of nine patients who had two tests
performed one to three weeks apart.2 Another study by Dr.
Holodniy reported less than a 0.5 log (three-fold) variation
in results from six out of eight patients tested weekly over
four weeks.3 Holodniy claims that the four other people in
these studies are exceptional "outliers."
Roche Molecular Systems presented data from a study of its
assay in a poster at the Second National Conference on Human
Retroviruses (this poster was omitted from the abstract
book). Roche found a maximum two-fold variation in HIV RNA
results in weekly samples drawn over three to six weeks in
eleven out of nineteen clinically stable patients (not on
antiretroviral therapy and with more than 400 CD4 cells). In
four of the nineteen patients there was a maximum three-fold
variation. In three others there was a maximum variation of
three- to six-fold, and one patient had a 6.5-fold jump in
viral load in one week.
Such findings have inspired some clinicians to use a three-
fold (or half log) rule of thumb for monitoring their
patients' progression and the effectiveness of antiviral
therapy. But according to Dr. Raboud, "The variation in viral
load and its causes needs to be better understood before
viral load results can be used for individual patient
management." She does not question the use of the tests in
clinical trials, just the use of a major change detected by a
single viral load assay as the basis for initiating or
switching therapy. A large viral load variation probably
should be confirmed by a subsequent test before any action is
In the New York area, there is a wide range of opinion among
doctors about how and whether these tests can be used.
Treatment Issues' medical consultant, Gabriel Torres, M.D.,
believes that running two tests to establish that there has
been a real change in viral load might be prohibitively
costly for most clinicians.
He continues, "The same variability occurs in CD4 cells, and
people have been using individual CD4 tests to make clinical
decisions for ages, and they are one-sixth the cost. Doctors
are using these [viral load] tests to make clinical
decisions, but whether they are switching on the basis of a
0.5 log, one log or 1.5 log change in viral load depends upon
Meanwhile, James Braun, D.O., president of the Physicians'
Research Network (PRN) in New York, says that "people who are
using the tests to make clinical decisions are jumping the
gun. We really don't know how to use these tests."
Other researchers who have worked with the viral load assays
are somewhat critical of Raboud's findings. "I don't believe
that there is a .83 log variation in clinically stable
patients," says Dr. Holodniy, who argues that viral load is a
very stable marker in the majority of patients. He noted that
there are a number of variables that could affect the test
results, for example how the samples were stored or the
particular technician who ran the assay.
In Roche's and Dr. Holodniy's studies, researchers batched
the samples (stored the samples and sent them in to be tested
at the same time). In the Vancouver study, the samples were
tested as they became available. Dr. Holodniy observes, "If
they didn't batch the samples, that could increase the
variation." In the real world, of course, physicians do not
batch samples from the same patient that are taken at
Dr. Holodniy also mentioned the possibility that subclinical
infections transiently affected viral burden in the Vancouver
study. Dr. Coombs agrees, "You have to consider the biologic
variability in the patient. The precision of the assay is
very good, at about 0.2 log, but when you look at viral load
over time in patients you've introduced a new variable. We
know that things like respiratory infections, herpes
infection, and flu vaccination can dramatically affect viral
Dr. Raboud believes that her patients were clinically stable,
though, so the variations must have arisen from the problems
inherent in the technology of PCR amplification. She claims
that the variations "were basically just noise."
Dr. Holodniy thinks that this variation might not be observed
using other measurement techniques such as bDNA (the branched
DNA test produced by Chiron) or NASBA (Nucleic Acid Sequence
Based Amplification, recently developed by Organon). "I'd
love to able to run these samples through one of the other
[commercial] assays," Dr. Holodniy said.
If the variation is due to the amplification technology, as
Dr. Raboud believes, the other assays may be more reliable
for patient management, but not if the variation is caused by
biologic variability. And if there is this much variation in
clinically stable patients, how is anyone to interpret viral
burden changes in people who are not stable, such as those
with active opportunistic infections?
A number of ongoing and upcoming studies will help resolve
how advisable it is to alter individual patients' therapy on
the basis of viral load results.
The PRN together with the Aaron Diamond Institute is
currently running a pilot study in New York to assess the
strengths of Chiron's bDNA assay. The researchers will track
how helpful it is in making therapy decisions as well as the
correlation between viral load and the course of a patient's
disease. This study will require at least two viral load
tests, performed at least three days apart to establish a
baseline value when changing or initiating therapy.
The Community Programs for Clinical Research on AIDS (CPCRA)
is opening a controlled multicenter study evaluating the
benefits of switching therapies when there is at least a
three-fold change in viral load as compared to switching
therapies as a result of changes in CD4 cell counts.
Currently, the plan is to use Chiron's bDNA assay, but Roche
has asked the CPCRA to use its assay as well. One of the
chief criticisms of this study is that it does not require a
second, confirmatory viral load test before a patient's
treatment regimen is changed. "Whether it's CD4 or some other
parameter, you always need two baselines [at least one
confirmatory test] for making clinical decisions," reiterates
Through the AIDS Clinical Trial Group (ACTG), Dr. Coombs is
running a study with Roche's test that requires two
successive tests showing at least a five-fold change in viral
load before modifying therapy. Even so, he believes that a
smaller change in viral load can be significant based upon
his analysis of the ACTG 116B/117 study (which evaluated
whether it was better for people with AIDS to switch to ddI
or remain on AZT). His analysis, to appear in a forthcoming
paper, found that "even a two-fold [decrease] can have
clinical benefit [improve symptoms or survival] if it is
1. Coombs RW et al. Journal of Clinical Microbiology.
Aug 1993; 31(8):1980-6.
2. Holodniy et al. Journal of Clinical Investigation.
Nov 1991; 88(11):1755-9.
3. Holodniy et al. Journal of Acquired Immune
Deficiency Syndromes. Apr 1994; 7(4):363-368.