Two-way street for HCV test algorithm
April 2002 William Check, PhD
For several years now Robert Dufour, MD, has coordinated a workshop on
hepatitis viruses at the annual meeting of the American Association for Clinical
Chemistry. "Based on attendee comments, we expanded the workshop from a half-day
to a full-day program after the first year," says Dr. Dufour, chief of pathology
at the Veterans Affairs Medical Center, Washington, DC, and professor of pathology
at George Washington University Medical Center. "Probably the main thing that
people come to learn," Dr. Dufour says, "is how to approach patients who are suspected
of having hepatitis—which laboratory tests are helpful for diagnosis, followup,
and monitoring of treatment."
With regard to laboratory testing for hepatitis C virus specifically, two
questions predominate. One is whether and when the strip immunoassay, also known
as the recombinant immunoblot assay (RIBA, Chiron), should be used to confirm
a positive result for anti-HCV antibodies on enzyme immunoassay.
The other, larger issue is whether it is preferable to use the qualitative
or quantitative polymerase chain reaction assay for HCV RNA to confirm chronic
HCV infection following a positive EIA result. Some laboratory professionals
strongly advocate the qualitative assay, while others believe the less sensitive
quantitative assay is acceptable, provided its limitations are understood.
Georg Lauer, MD, research fellow in medicine at Massachusetts General Hospital,
was co-author of a review article that recommended the qualitative test (N
Engl J Med. 2001; 345: 41-52). "I am quite comfortable with the other approach,"
he says. "It is perfectly fine to do the quantitative assay first if you have
no easy access to the qualitative test and if you are willing to send the small
number of negative quantitative tests for qualitative testing."
After consulting with clinicians, Leonard Grosso, MD, PhD, associate professor
of pathology and director of the DNA diagnostics laboratory in the division
of molecular pathology, St. Louis University School of Medicine, set up the
quantitative-qualitative sequence. "From my perspective, we could just as easily
have gone the other way," he says. "But I didn’t think we could manage the logistics
of offering both sequences." Dr. Grosso initiated discussions with clinicians
after a review of testing patterns showed that gastroenterologists and hepatologists
were by and large ordering quantitative and qualitative tests for HCV on every
specimen.
Confusion is common, says Nizar Zein, MD, associate professor of medicine
and pediatrics and consultant in gastroenterology, hepatology, and liver transplantation
at the Mayo Clinic. "I lecture on HCV and I am amazed how much clinicians are
confused about the utility of qualitative and quantitative tests, especially
their dynamic range," he says. "I think clinicians, particularly in private
practice, probably don’t have time to learn the details about these tests."
Understanding optimal testing algorithms for HCV becomes increasingly important
as a wave of occult HCV infections becomes clinically manifest. "From the literature,
the magnitude of HCV infection is even larger than I would have expected," says
Richard A. McPherson, MD, chair of the division of clinical pathology, Medical
College of Virginia, Richmond. Approximately 2.7 million people in the United
States have active chronic HCV infection. Each year 30,000 new infections occur,
of which perhaps 80 percent will progress to chronic viremia. Without effective
intervention, mortality from HCV is expected to triple from the current rate
of 8,000 deaths annually. Dr. McPherson attributes the increase in HCV cases
to "a bolus of persons infected in the ’60s, ’70s, and even the ’80s through
youthful indiscretions—intravenous drug use and tattoos—and through
blood transfusions prior to institution of blood screening." As a result, he
says, "Lots of solid citizens these days have HCV."
Dr. McPherson points out that, while the disease itself has many clinical
complications—serum cryoglobulin, immune complex disease, vasculitis—current
treatments are also "complicated and expensive with many side effects," putting
a premium on testing algorithms that select appropriate patients for treatment.
Test results may help a clinician determine whether to perform liver biopsy.
"The role of liver biopsy is unclear," says Dr. Zein. "It is not obvious that
every patient needs one. But I would definitely not biopsy unless the patient
is documented to have HCV RNA by PCR."
Laboratory testing, along with clinical examination and biopsy results, influences
also whether treatment is initiated. "Criteria for interferon treatment are
changeable," Dr. Zein says. "They are not set by any formal council or body."
Therapy generally would be given to patients who have chronic HCV infection
and viremia and elevated liver enzymes and who show evidence of liver damage
associated with HCV infection. However, "views on therapy change from physician
to physician, as well as with advances in therapy," says Dr. Zein. "When therapy
was ineffective and toxic, it was hard to justify treating many patients. As
it became more effective and we became more sophisticated at managing complications,
therapy is now considered in more patients."
Sustained response rates (disappearance of viral RNA for six months after
the end of treatment) with pegylated interferon (interferon attached to polyethylene
glycol to extend its half-life) and ribavirin are seen in about 30 percent of
patients infected with type 1 HCV and in more than 70 percent of persons infected
with non-type 1 viral strains. (In the United States, type 1 makes up 80 percent
of HCV infections.) Trials are underway that test therapy in pediatric populations
and patients with extrahepatic manifestations of HCV—vasculitis, kidney
diseases, or lymphoproliferative disorders.
Laboratory testing for HCV infection typically is initiated in persons being
worked up for abnormal liver function tests. "We become even more suspicious
if a patient had high-risk exposure, such as blood transfusion before 1990 or
a history of IV drug use," Dr. Zein says. Individuals with normal liver enzymes
and a known high-risk exposure may also be investigated because one-third of
persons chronically infected with HCV may have normal liver enzymes.
The VA system, says Dr. Dufour, has a mandatory screening questionnaire for
all patients seeking care. The questionnaire asks about IV drug use, multiple
sexual partners, and transfusions or organ transplants performed before 1992.
"We know that, in our network, the overall rate of asking these questions is
about 90 percent," he says, "and two hospitals are over 95 percent. When this
screening instrument went into place in August 2000, we saw our testing numbers
increase about fourfold."
The Centers for Disease Control and Prevention has recommended that all physicians
question all their patients about risk factors for HCV infection. Few primary
care doctors comply with this recommendation, says Miriam Alter, PhD, acting
associate director for science in the CDC’s division of viral hepatitis.
Once a clinician decides that a patient needs laboratory workup, the
ensuing steps are somewhat standard. Frederick Nolte, PhD, associate professor
of pathology and laboratory medicine and director of the clinical microbiology
and molecular diagnostics laboratories, Emory University Hospital, describes
his testing protocol. Specimens with a positive EIA for anti-HCV antibody are
analyzed with qualitative PCR for HCV RNA to define whether the patient is actively
infected or has resolved the infection. "If RNA is negative but the clinical
index of suspicion is high," he says, "it might be prudent to repeat it later,
since there are reported cases of chronically infected patients with occasionally
negative RNA tests." While Dr. Nolte personally has not found intermittent viremia
to be a major problem, he notes that up to 15 percent of chronically HCV-infected
individuals followed over time have occasional RNA dropouts.
If the RNA test is positive, the clinician may order liver biopsy to assess
the amount of liver damage. If the clinician decides to treat, that is the time
to do quantitative PCR for viral load and genotyping. After 24 weeks of treatment,
a qualitative RNA test to assess whether the patient has cleared the virus is
recommended. If viremia has not clear-ed, treatment might be stopped for a patient
with a poor prognosis. Patients whose treatment is continued undergo qualitative
RNA testing again at 48 weeks to look for end-of-treatment response. Those who
test negative return six months later for qualitative RNA testing to determine
whether they have achieved a sustained virological response.
The first step in this algorithm is the screening EIA. Two generations of
enzyme immunoassays are used in the United States and both are FDA approved.
The third-generation version gained modest improvement in sensitivity at the
cost of specificity, Dr. Nolte says, "so its performance characteristics are
more appropriate for blood banks. The second-generation test has a better performance
profile for diagnostics."
While RIBA used to be done routinely to confirm a positive EIA, many laboratories
no longer offer it. "Our clinicians are very plugged in to molecular diagnostics,"
says Dr. McPherson. "Their preferred way to confirm a positive EIA is with RT-PCR.
If RIBA is requested, we send it out. It is still extremely expensive."
Says Dr. Nolte: "RIBA has always been termed a supplemental serological test.
In many settings you screen with EIA, then reflex all positive samples to RIBA,
which provides additional information about the validity of that initial antibody
result." In blood banking, this can be helpful. Current serological tests, however,
are rather specific, reducing the need for RIBA confirmation, says Dr. Nolte.
In addition, he says, "In most diagnostic settings we are not so concerned with
establishing whether this is a true antibody response. We are more interested
in whether the patient is still infected." That is what a PCR assay shows.
Over time, RIBA probably will lose ground to PCR, Dr. Zein predicts. "PCR
is not standardized very well at this point, and some laboratories may have
problems with contamination." But in time, Dr. Zein sees most laboratories adopting
standardized PCR kits as they become available, which, he says, "will hasten
RIBA falling out of favor."
But Dr. Dufour suggests RIBA can still be helpful in selected patients. "We
have found that positive results on EIA fall into low-positive and high-positive
categories," he says. Many low-positive results turn out to be false-positives—over
90 percent of them are negative on RIBA. (In Dr. Dufour’s population, 15 to
20 percent of patients fall into the low-positive group.) Conversely, in those
with high-titer EIA values, RIBA is almost never negative.
"We believe that it is important to identify false-positive cases so those
individuals aren’t labeled as having had HCV infection," Dr. Dufour says. As
a practical matter, he defines low-positive results as those with a signal-to-cutoff
ratio of less than 3.5. "The average ratio in our laboratory is about five,"
he says, "so there is not much overlap or gray zone."
The CDC, too, has found that positive EIA results fall into two categories,
Dr. Alter says. "All screening assays have false-positives in certain populations,"
she says. "While the EIA for anti-HCV antibody performs very well as a diagnostic
tool in persons who have liver disease, it may perform less well in persons
who are asymptomatic."
Because screening assays have false-positive results, they typically have
confirmatory assays, such as the Western blot test for HIV. Confirmatory testing,
however, is not routinely used with the anti-HCV EIA, says Dr. Alter. "In a
clinical setting where most patients being tested have liver disease, that is
probably not an issue," she says. "But since laboratories have no idea whom
they are testing, CDC would like to implement a more standardized fashion of
reporting results."
Because cost is a barrier to universal confirmatory testing, Dr. Alter and
her colleagues sought a way to improve reliability of test interpretation while
keeping cost down. Selective confirmation is one feasible approach: Dr. Alter’s
group found that 95 percent or more of persons with signal-to-cutoff ratios
of 3.8 or greater on EIA were RIBA-positive, whereas signal-to-cutoff ratios
of less than 3.8 were confirmed only about 15 percent of the time.
"We are suggesting that laboratories could report out positive EIAs having
a signal-to-cutoff ratio of 3.8 or greater with the caveat that they have not
been confirmed but are 95 percent likely to be positive," she says. Samples
with a lower signal-to-cutoff ratio would undergo confirmatory testing. "We
are planning to publish such a recommendation," Dr. Alter says. The group is
coordinating with the appropriate professional organizations.
"This recommendation is not intended to change test ordering by clinicians,"
Dr. Alter says. Rather, it establishes an algorithm for laboratories to use
independent of who is being tested. Only laboratories testing a high proportion
of low-risk persons would do much confirmatory testing. According to the CDC’s
1998 recommendations, cited by Dr. Alter, laboratories that go straight from
a positive EIA to PCR should already be doing RIBA on samples that are PCR-negative.
Using RIBA in this way makes it more likely the patient was never infected,
says Dr. Lauer. "You do a favor for the patient and clear the record," he says.
But, he points out, a negative RIBA is not a perfect proof. In one study, 18
of 43 persons who cleared viremia by 20 years after infection also became completely
negative for anti-HCV antibodies (Nat Med. 2000;6: 578-582).
Even stronger divergence of opinion attends the second question—whether
qualitative or quantitative PCR should be the primary assay for HCV RNA. In
one camp are those who believe, as Dr. Nolte does, that qualitative RNA detection
is the mainstay of molecular diagnostics for hepatitis C. "There is really no
value to quantitation except in one situation—as part of the pretreatment
evaluation to establish a baseline viral load," he says.
Dr. Nolte succinctly states his argument against using quantitation as the
primary PCR assay: "HCV is not HIV." Unlike with HIV, HCV viral load does not
predict disease progression. High viral load is an indicator of poor response
to therapy with interferon and ribavirin and may influence length of therapy
(24 versus 48 weeks). But it is only one of five important factors, along with
genotype 1, liver biopsy results (evidence of fibrosis), age greater than 40
years, and male gender.
Moreover, Dr. Nolte says, "There is no demonstrated value to monitoring HCV
viral load over the treatment course. We are not trying to suppress virus below
some cutoff or magic number—we are trying to eliminate it. The test of
virologic cure is absence of viral RNA from serum or plasma using the most sensitive
test available." A patient’s viral load could be under the lower limit of the
quantitative assay, 600 IU/mL, but not the lower limit of the qualitative assay,
100 IU/mL. (International units do not translate simply into copies per milli-liter.)
"Since clinicians are asking for quantitation, some laboratorians have adopted
the quantitative assay as their basic PCR method," Dr. Nolte says. "From a medical
and patient management perspective, that is probably not the best course."
Dr. Nolte’s laboratory performs many more qualitative tests than quantitative
tests. "We may be somewhat unusual that way," he acknowledges. "I think a lot
of the preference for quantitative testing has to do with simple misunderstanding."
Clinicians may check the quantitative box because they don’t fully grasp the
lack of value with quantitation and because it looks like a quantitative test
provides more information, he says. More recently, he adds, "It looks like clinicians
may be coming around to understand that HCV is not HIV."
One qualitative PCR assay for HCV RNA is FDA approved, the Roche Amplicor.
"Roche has a good program to make sure the laboratory knows how to do the test,"
says Dr. Nolte, whose laboratory was one of the sites for the assay’s Phase
III clinical trials. CAP Survey results indicate that 80 percent or more of
laboratories that test for HCV use the Roche Amplicor assay. GenProbe has developed
a qualitative transcription-mediated amplification assay for HCV, but it has
not yet been approved by the FDA.
Two quantitative assays for HCV RNA exist, Bayer’s branched-chain Versant
kit and Roche’s Amplicor Monitor, but neither is FDA cleared. "That demonstrates
the problem with quantitative tests," Dr. Nolte says. "To get FDA clearance
requires demonstrating some clinical utility, and there is not a lot of clinical
utility for quantitative tests aside from pretreatment evaluation."
Dr. Zein expresses similar views. "I agree that a quantitative RNA assay should
not be used to determine whether or not a patient has viremia because it is
not as sensitive as the qualitative assay," he says. "After a patient is confirmed
to have viremia by qualitative assay, it is fine to use quantitative testing
to get a pretreatment viral load value." Dr. Zein believes a quantitative assay
also could be useful to measure the decline in viral RNA with therapy. But,
he says, a qualitative assay should be used "to determine end-of-treatment response
and whether to stop therapy at 24 weeks or continue for 48 weeks." Using a quantitative
assay as the primary method would be acceptable if negative values are confirmed
by qualitative testing.
For several years, Dr. Dufour also used the qualitative assay as standard.
Now, however, his VA network (VISN 5) is adopting quantitative RNA as its initial
test, mostly based on "a significant improvement in sensitivity." The quantitative
assay’s lower limit of detection, 500 to 600 IU/mL, is low enough, he estimates,
to detect virtually everybody who is viremic. "We see very few individuals with
RNA values lower than 100,000 IU/mL," Dr. Dufour says. In fact, for a period,
one hospital in the network was doing confirmation with qualitative RNA and
another with quantitative RNA. In patients thought to be highly probably infected
(those with multiple elevated liver enzymes), 98 percent were positive by quantitative
assay and 94 percent by qualitative assay.
Dr. Dufour prefers quantitative RNA values for two reasons. First, he says,
patients whose viral load is higher than the median value of those infected
almost never respond to six months of treatment. Second, with regimens that
include pegylated interferon, if viral load has not fallen by at least two logs
by 12 weeks, you can predict that the patient will not respond to treatment
and thus discontinue therapy. "We still believe that for monitoring efficacy
of treatment at 24 and 48 weeks of treatment, you need to use the most sensitive
assay, which is the qualitative assay," he says.
After an evaluation period, Dr. Grosso also adopted quantitative testing as
the primary RNA assay, with reflexive qualitative RNA determination for samples
that were negative in the quantitative assay (Arch Pathol Lab Med. 2002;126:
100-102).
"Our clinicians felt they needed quantitation of HCV RNA when it was present
but needed the lower limits of detection of the qualitative assay to make sure
they would get a clinically useful result in patients with low viremia," Dr.
Grosso says. To meet these goals, clinicians were ordering both tests simultaneously
on all patients because they wanted to be sure results would be available at
the next clinic visit. This tactic produced redundant information and increased
costs.
Dr. Grosso discussed this situation with the gastroenterology division and
offered to set up a reflex protocol starting with either assay. The clinicians’
preference was quantitative testing followed by qualitative testing if the quantitative
assay was negative. During the evaluation period, it was demonstrated that this
sequence was cost-effective and provided the information the clinicians wanted
by the next patient visit.
In this situation, turnaround time was an important component. "Our initial
concern," says Dr. Grosso, "was if we do one test, then we have to do a second
test; will the elapsed time from getting the specimen to reporting the second
result provide the information to the clinician in time for the patient visit?
I didn’t expect that to be a problem, and it -hasn’t been." Turnaround time
was just over eight working days. (Both tests are done in-house.) Clinical visits
are scheduled one to three months ahead, so eight working days is not a problem.
In the review Dr. Lauer co-authored, qual-itative tests are called "tests
of choice for the confirmation of viremia and the assessment of treatment response."
Dr. Lauer considers this a minimal and cost-effective approach since, if quantitative
testing is done first, samples that are negative would have to be confirmed
with a qualitative assay "just in case they are in that small window." But that
doesn’t happen often, he says. "We will see maybe five patients per year who
are negative on quantitative and have to be tested on qualitative."
On the other hand, proposed advantages to quantitation are mostly illusory,
in his view. "Current guidelines do not recommend altering standard treatment
if a patient has higher or lower viral load," Dr. Lauer says. When new therapies
first came out, some specialists discussed the possibility of shorter treatment
for patients with lower viral load, even with genotype 1. "But that has not
come into practice," he says.
Using kinetics of viral load decline during therapy to decide whether treatment
should be continued is also premature. "Maybe in the future, it will be important
to consider these quantitative data," Dr. Lauer says, "but where the evidence
is right now, you don’t need it." He sees no hard data to support continuing
treatment if viral load has dropped by two logs at 24 weeks. These decisions
depend very much on the treating physician, Dr. Lauer says. "Some of my clinical
colleagues in the GI unit use that approach sometimes, but it is not supported
by trials data. Why two logs and not one or three?"
The remaining molecular assay, genotyping, has one clear value: as a predictor
of response to therapy. Among patients infected with type 1 virus, about 30
percent have a sustained response, while 60 to 70 percent of those infected
with non-type 1 strains have a sustained response. "There is no evidence that
genotype is important in disease progression or severity," Dr. Nolte says.
There are no FDA-cleared tests for genoptyping. Most U.S. laboratories doing
genotyping use the line probe assay (Lipa) developed by Innogenetics and now
distributed by Bayer Diagnostics. "Lipa functions as a Western blot strip,"
says Dr. Nolte. "It has 19 type and subtype probes. Many laboratories—including
us—like it because it can use a PCR product." At present, identifying
subtypes has no clinical utility.
Dr. Nolte recently evaluated Visible Genetics’ sequence-based system for HCV
genotyping. "This system may be an attractive alternative for laboratories that
are already using Visible Genetics’ system to do HIV-resistance genotyping,"
he says. Like Lipa, the sequencer takes a PCR product. Sequencing is a bit more
expensive and more complex than Lipa, in Dr. Nolte’s estimation, but it is potentially
more informative, and its database is easier to update. "I think we might switch
to sequencing," he says. "We already have the equipment. And it would be more
comforting to me to see raw sequence data and make my own decisions."
Dr. Dufour is taking a different approach. "We have been sending out samples
to Quest for the Roche direct sequence method," he says. "But we recently decided
to switch to Lipa." The theoretical advantage of direct sequencing is that it
can detect multiple strain infections, he says. "But in almost 300 individuals
we tested, we didn’t see a single patient who had multiple strain infection."
And he finds Lipa to be significantly less expensive.
Looking at the overall picture of HCV management, Dr. Zein says, "I
am unsatisfied, but I am extremely pleased and happy with the progress that
has been made. Progress in HCV is tremendous. Both academia and the pharmaceutical
industry are to be congratulated for really turning things around in a very
short time, from not being able to do anything five years ago to being able
to cure 40 to 50 percent of patients today.
"But," he continues, "that still leaves more than half of patients with no
options. It is hard to be satisfied when every other patient we see gets no
benefit from therapy and when therapy is expensive and associated with severe
side effects. The battle won’t be over until we improve response rates even
more."
Laboratory issues remain as well. Dr. Dufour points out that the role of liver
enzymes, such as alanine aminotransferase, is not yet clear. "Many people who
normalize their ALT do not have a virologic response," he says. "Conversely,
we see occasional patients who clear the virus but do not normalize ALT. So
we get ALT levels, but I am not sure how useful or reliable they are."
Whether clearing viremia is an absolute measure of response is another major
question. Some data suggest that interferon therapy helps not just by clearing
virus, but by decreasing liver damage as well. "In retrospective analyses, patients
treated with interferon had lower fibrosis scores even if they didn’t clear
virus," Dr. Dufour says. "And in followup they had less probability of developing
hepatic carcinoma and lower likelihood of decompensated cirrhosis."
In the new-test arena, Dr. Nolte is evaluating an assay that measures hepatitis
C core antigen using a mixture of monoclonal antibodies. Its lower limit of
detection is about 10,000 copies/mL, not sensitive enough to define treatment
response but possibly sensitive enough to supplement screening EIA. While PCR
for HCV RNA measures down to about 100 copies/mL, Dr. Nolte says "it is rare
to find anyone who has a viral load of less than 100,000 when presenting for
viral diagnosis." If the antigen-detection test proved diagnostically useful,
it potentially would be less costly and easier to perform than PCR.
Dr. Lauer points to another evolving area of laboratory medicine—whether
the dynamics of viral load decline during early therapy will give an idea of
the probability of response. "Obviously," he says, "it would be nice to know
early in therapy whether a patient is responding. If not, we could spare them
from prolonged therapy." Large trials will be needed to answer that question.
William Check is a medical writer in Wilmette, Ill.
|