Testing for tuberculosis
A gourmet menu
Advancing molecular testing
William Check, PhD
Nucleic acid-based assays are “very powerful but very expensive tools,”
said Mary Jane Ferraro, PhD, MPH, director of microbiology laboratories,
Massachusetts General Hospital, convening a symposium on molecular
methods in infectious diseases this past fall at the 39th Interscience
Conference on Antimicrobial Agents and Chemotherapy.
Symposium speakers discussed the application and interpretation
of these tests for human papilloma virus, hepatitis C virus, human
immunode-ficiency virus, and cytomegalovirus, but they noted that
the tests are useful for other pathogens as well.
Human papilloma virus. Discussing molecular methods for
detecting human papilloma virus, Nancy Kiviat, MD, stated that many
women are infected with HPV, but only a few get cervical cancer.
Dr. Kiviat is chief of pathology, Harborview Hospital, and director
of cytopathology, University of Washington Associated Hospitals,
"Some women’s immune systems can’t handle the infection, and we
think those individuals are at highest risk of HPV-related neoplasia,"
Dr. Kiviat said. Viral risk factors include the proteins E6 and
E7, expressed during infection with high-risk HPV types. These proteins
reduce the activity of the tumor-suppressor proteins p53 and RB105.
"These viral genes set the stage for accumulation of chromosomal
abnormalities that, in some cases, lead to progression to malignancy,"
Dr. Kiviat told CAP TODAY.
HPV testing, Dr. Kiviat added, plays a definite role in assisting
triage for women with ASCUS, or atypical squamous cells of undetermined
significance. Widespread Pap screening has reduced the rate of cervical
cancer from 40 to seven cases per 100,000 women. But in addition
to revealing squamous intraepithelial lesions, Pap smears also identify
ASCUS in five to 10 percent of women screened, and all of these
women are asked to return for a repeat Pap smear. Many eventually
have colposcopy and biopsy.
Most of these examinations are unnecessary because the risk of
cervical cancer with ASCUS alone is extremely low. Yet more than
one-third of high-grade SILs identified during screening come from
ASCUS smears. Testing for HPV can help identify the minority of
women with ASCUS who have underlying disease, without requiring
all women with ASCUS to undergo invasive examinations.
Evidence for this contention comes from studies, such as the one
performed at Northern California Kaiser Permanente Medical Group
(JAMA, 1999;281:1605-1610). The Kaiser study compared use of HPV
testing versus repeat cytology for triage of women with ASCUS. Of
46,000 women screened with standard cytology, 995 who had ASCUS
returned for colposcopy and repeat cytology. HPV testing was performed
for these 995 women using a hybrid capture assay (Digene Corp.)
on specimens collected in liquid medium during the initial screening.
Two major findings emerged. First, approximately the same fraction
of women39 percentwould have been sent for colposcopy
and biopsy based on detecting HPV DNA in the initial cervical specimen
as established by findings at repeat cytology. Second, based on
colposcopy results, "The sensitivity and specificity of HPV testing
for identification of women with underlying high-grade SIL was essentially
the same as that of repeat Pap smear," reported Dr. Kiviat.
The investigators proposed that women with ASCUS who are HPV-positive
undergo immediate colposcopy and all other women who have ASCUS
undergo repeat Pap testing, which would provide 97 percent sensitivity
for high-grade SIL. "Most important," the investigators concluded,
"our proposed algorithm allows HPV testing and a subsequent triage
decision based on the initial [ASCUS] Pap specimen." With this scheme,
many women could avoid the cost and inconvenience of repeat physician
This approach requires using the same sample for Pap smear and
HPV testing. Part of the sample is used for cytology, and part is
held in liquid medium and probed for HPV if ASCUS is found.
To prepare for this tack, Dr. Kiviat told CAP TODAY, the University
of Washington Hospitals switched to liquid-based cytology a few
years ago. Now any woman with ASCUS automatically will have an HPV
test: Those who are positive will be referred for colposcopy, and
those who are HPV-negative will follow a routine Pap smear schedule.
To reduce the cost of the HPV assay, Dr. Kiviat uses only high-risk
probes. "Virtually all samples with CIN II or III have a high-risk
HPV type," she said.
Based on preliminary results of studies she and her colleagues
have conducted in Seattle and Africa, Dr. Kiviat believes HPV testing
also will play a role in primary screening for cervical cancer.
Detection of HPV DNA appears to have sensitivity and specificity
as good as cytology, if not better, for high-grade SIL (CIN II or
III). So, in the future, it may be more cost-effective and better
for the purpose of detection to use HPV as a primary screen and
then do a Pap smear on HPV-positive women.
Some laboratories are in an earlier stage of implementing HPV
testing, however. Says Stephen Dumler, MD, director of the Division
of Medical Microbiology, Johns Hopkins University Hospitals, "We
are evaluating the Digene assay, predominantly for high-risk strains,
in collaboration with cytologists and gynecology groups around the
university." Physicians have adopted a liquid-based collection method
and, for samples with ASCUS, the laboratory will reflexively test
for high-risk HPV types. "We are assessing the feasibility of that
approach," Dr. Dumler told CAP TODAY.
At Massachusetts General, "Clinicians are not knocking down the
doors yet" for HPV testing, said James Versalovic, MD, PhD, the
hospital’s assistant director of microbiology. "We are definitely
talking about it," he commented to CAP TODAY. "Cytologists are still
content with routine Pap smears. But the question of whether or
when to do HPV testing for ASCUS smears is on the table."
Hepatitis C virus. Turning to molecular tests for detecting
and quantitating hepatitis C virus, symposium speaker David Gretch,
MD, PhD, associate professor of laboratory medicine, University
of Washington, stated that during the past few years, the burgeoning
epidemic of HCV in the United States increasingly has been recognized.
About 4 million persons are infected. Approximately 85 percent of
infections enter a chronic phase, and perhaps 20 percent of those
progress to cirrhosis over about 20 years. Infection with HCV is
the leading cause of the need for liver transplant, Dr. Gretch noted.
Qualitative detection of HCV viremia predicts liver disease in
persons who are seropositive for HCV, said Dr. Gretch, referring
to an earlier study. All 16 seropositive patients in the study who
had qualitative evidence of HCV viremia had liver disease on biopsy,
regardless of whether liver enzymes were elevated, and not one of
seven seropositive patients with normal enzymes and no detectable
plasma HCV RNA had biopsy evidence of liver disease. "In the presence
of normal enzymes, viral RNA testing is very important to guide
decisions about biopsy," Dr. Gretch concluded.
Reviewing available tests, Dr. Gretch noted that the qualitative
RT-PCR assay for HCV confirms viremia and active infection and is
the most sensitive. An isothermal qualitative assay based on transcription-mediated
amplification technology is used in blood banking. Quantitative
RT-PCR and bDNA assays help monitor and guide therapy. Treatment
of HCV infection with interferon, now typically combined with ribavirin,
is producing sustained five-year responses in 40 percent of patients,
Dr. Gretch said. Virological responsereduction of viral load
to below 100 copies/mLis the most accurate indicator of response.
If patients clear viremia after one week of therapy, 76 percent
have a sustained response; if viremia clears by two weeks, sustained
response is seen in 36 percent; by three weeks, sustained response
is seen in 13 percent.
HCV RNA testing is also used to determine length of treatment:
If viremia does not clear by three months, therapy is stopped. (Six
months is the cutoff for combination therapy.) Terminating therapy
in the absence of a treatment response is especially valuable for
patients who are not tolerating the drugs.
An additional application of molecular testing in HCV infection
is for sequencing isolates to determine their genotype. About two-thirds
of U.S. patients are infected with type 1 virus, which has only
about a 10 percent response rate, while virus types 2 and 3 have
a 35 to 40 percent response rate. "But typing should not be used
to deny therapy," Dr. Gretch- cautioned. "Response rate for type
1 virus can be doubled by extending duration of interferon therapy
to 48 weeks." Therapy can be stopped for patients with types 2 or
3 virus whose viremia clears by six months.
Massachusetts General’s Dr. Versalovic sends out samples for HCV,
"But we are moving towards implementation of molecular HCV testing,"
he added. He initially plans to use quantitative viral load measurement
to follow up patients in therapy. For budgetary reasons, quantitative
testing will come firstright now that assay is in highest
demand from gastroenterologists.
"I’m not entirely in agreement with our volume distributions in
HCV testing," Dr. Versalovic told CAP TODAY. But once he sets it
up and gets it under control, he plans to introduce qualitative
testing and genotyping for diagnosis and followup. "I agree that
qualitative testing and genotyping are probably more important than
quantitative testing," he stated. The qualitative test is more sensitive
than the quantitative assay by one order of magnitude, 100 copies
versus 1,000 copies.
"Qualitative testing may become the preferred confirmatory diagnostic
test for HCV," Dr. Versalovic said, although he noted it is not
yet approved for that purpose.
Human immunodeficiency virus. Assays for HCV are the highest-volume
molecular tests in many laboratories. In others, HIV testing takes
the top spot.
Quantitative HIV assays are used to monitor viral load during
therapy and soon may be adopted in some centers for diagnosing acute
HIV infection, explained Angela Caliendo, MD, PhD, medical director
of microbiology/molecular diagnostics, Department of Pathology and
Laboratory Medicine, Emory University, Atlanta.
The only FDA-approved quantitative assay for HIV is the Roche
RT-PCR, which is used for an estimated 70 percent of tests. The
other primary tests for HIV viral load are Quantiplex bDNA from
Bayer and NucliSense (formerly NASBA) from Organon Teknika.
In their standard versions, all three methods are sensitive down
to 400 to 500 copies/mL. In their ultra sensitive versions, the
lower limit of detection is 50 to 100 copies/mL. Even though there
is good correlation between bDNA 3.0 and Roche’s ultrasensitive
method, Dr. Caliendo said, "I still do not recommend that you switch
technologies for monitoring an individual patient."
A qualitative assay (that detects proviral DNA) is available that
is used primarily for neonatal diagnosis. Even for this application,
Dr. Caliendo prefers a quantitative RNA assay.
One difference among assays is dynamic range: the ultrasensitive
RT-PCR has an upper limit of 75,000 copies/mL, while bDNA version
3.0 goes up to 500,000 copies/mL, and NucliSense goes up to 10 million
copies/mL. Therefore it would be necessary to use both the standard
and ultrasensitive versions of RT-PCR, while one could use the ultrasensitive
versions of the other two assays alone.
"The need for a standard and ultrasensitive assay may be a problem
for laboratory workflow and turnaround time," said Dr. Caliendo,
particularly for laboratories that have a small volume of HIV testing.
Even though her laboratory does not do a high volume of HIV testing,
Dr. Caliendo uses RT-PCR, arranging workflow to accommodate it.
Dr. Caliendo chose RT-PCR for HIV testing because she had considerable
research experience with the assay and was satisfied with its performance.
She uses the NucliSense assay for various research studies. She
likes having an assay with an internal control. "It is very important
to have a handle on efficiency of extraction," she said.
Inherent in the RT-PCR and NucliSense assays is the ability to
evaluate efficiency of specimen extraction. "When you get a negative
result," Dr. Caliendo commented, "you know this is not due to inhibitors
or poor specimen extraction."
The methods also differ in specificity. The specificity is nearly
100 percent for RT-PC-R and NucliSense and is 95 to 98 percent for
RT-PCR and bDNA are less labor-intensive to perform than NucliSense.
But an automated extractor, priced at about $60,000, has simplified
A precaution that should be taken with all assays is not to overinterpret
small changesfor example, from undetectable to 200 copies/mL.
For all assays, intra-assay variation and biologic variation of
viral load together are about 0.5 log. So a change in viral load
doesn’t represent clinical change until it exceeds 0.5 log, about
a threefold change. "This is especially important at lower values,
below 500 copies/mL," Dr. Caliendo stressed. "You might want to
ask for a four- to fivefold change at that level." Patients have
"blips" where they go below 50 copies/mL, then go back above that
level, and then come down again. "Make sure clinicians repeat a
test before changing therapy," Dr. Caliendo advised.
In acute infection, less than six weeks after exposure, there
is a mononucleosis-like syndrome with negative or indeterminate
enzyme-linked immunosorbent assay and Western blot and very high
viral load. In one study, viral load during acute infection ranged
from 250,000 to 100 million copies/mL of HIV RNA. Bruce Walker,
MD, Eric Rosenberg, MD, and colleagues at Massachusetts General
Hospital focused on whether there is a pool of acutely infected
persons who could be identified with appropriate testing. They tested
560 individuals who presented to physicians with an acute mononucleosis-like
syndrome that turned out to be Monospot negative. Seven (approximately
one percent) patients had definite or possible acute HIV infection.
"There is more acute HIV infection out there than people appreciate,"
Dr. Caliendo concluded.
To find such cases, the clinician should take a history for HIV
risk factors when someone presents with a viral syndrome and then
determine whether to test for HIV. If the viral load is greater
than 200,000 on two consecutive tests, the clinician should follow
the patient to verify seroconversion and make a treatment decision.
"You need to appreciate that you are using this test in a way
for which it was not licensed," Dr. Caliendo noted, "so the clinician
needs to get patient consent and send a sample for concurrent ELISA."
If the patient does not have HIV infection, the chart should reflect
a negative ELISA along with the negative viral load result.
From the laboratory’s viewpoint, Dr. Caliendo said, "Clinicians
are using viral load testing for the diagnosis of acute infection,
and the laboratory doesn’t always know. Clinicians and laboratories
need to work together on this."
At Massachusetts General, HIV is the third most ordered DNA amplification
test performed directly on a primary clinical specimen, with more
than 4,000 tests conducted annually, according to Dr. Versalovic.
(First is chlamydia; second is gonococcus.) For diagnosis, Dr. Versalovic
uses ELISA confirmed by immunoblot or Western blot. "Quantitative
HIV viral load is not a primary diagnostic tool," he emphasized.
"It is not FDA-approved for that."
Quantitative viral load is done to follow patients on therapy
or to establish a pretherapy baseline. Some physicians also order
quantitative RT-PCR for diagnosing acute infection. "It is useful
in the acute HIV setting because you see sky-high viral loads,"
Dr. Versalovic said, often above the 750,000 copies/mL upper limit
of detection for the RT-PCR assay. He agreed that one must also
obtain informed consent and a diagnostic ELISA in this context.
Cytomegalovirus. Nucleic acid-based assays for detecting
cyto-megalo-virus encompass qualitative and quantitative PCR, hybrid
capture, NucliSense, and bDNA, said Vincent Emery, PhD, reader in
virology, Royal Free and University College Medical School, London.
They are used in transplant recipients and HIV-infected persons.
PCR and NucliSense are used most often, in Dr. Emery’s experience.
Hybrid capture and bDNA will be used more in the future because
the companies that market them have assays for other viruses, and
it is easier to add one more virus by the same method.
"But working out specificity and sensitivity for disease for those
assays is in [its infancy]," said Dr. Emery, who runs Bayer bDNA,
Roche PCR, and an in-house assay for CMV. "That is historicwe
were interested in CMV for several years, and commercial assays
have only come online in the last few years," he told CAP TODAY.
With culture, CMV background is usually negativesensitivity
is only 25 percent. Even with a seropositive recipient of a seropositive
organ, one would not expect to culture virus unless an infectious
virus was on board. "So there is a high threshold of detectionif
you do detect virus in culture, it has a high negative predictive
value," Dr. Emery said.
For molecular assays, most sampleswhole blood, plasma, leukocyteswill
work, although viral load in plasma is about 20-fold lower than
in whole blood.
Volume is another issue. With PCR, if you make it very sensitive
and take a large amount of sample, then you will find positivity
in most people who are antibody-positive. "That is clearly of no
value," Dr. Emery commented. "What you want is high levels of DNA
post-transplant indicative of active infection. You can do that
two ways: take a small amount of blood and have a very sensitive
assay or a rather large amount of sample and have a relatively insensitive
assay." Dr. Emery has chosen the former route.
With this strategy, a viral load below 200 genomes/mL provides
reasonable confidence that there is no replication or very low-grade
replication. "With herpes viruses and their latent infection," Dr.
Emery said, "there may be a low level of reactivation all the time
that a healthy immune system controls." Only with immunocompromise
does loss of control occur.
Another sampling issue surfaces when monitoring response to therapy.
Viral load goes below the limit of detection with plasma more quickly
than with whole blood. To look at small changes in a short period,
Dr. Emery’s laboratory uses whole blood and plots viral load decline
Earlier detection of CMV infection has become an important clinical
issue with the advent of effective chemotherapy. Identifying persons
at risk of disease and targeting them for pre-emptive therapy increases
survival time and allows low-risk individuals to avoid treatment.
Median viral load is significantly higher in those who have symptoms
or will develop symptoms, although, for unknown reasons, some persons
sustain viral load up to 1 million copies/mL and don’t get disease.
Most clinicians will consider pre-emptive therapy with two consecutive
positive qualitative PCR results. But because of earlier and more
frequent treatment, the rate of CMV disease has declined. "So it
is important to validate your approach to make sure you don’t call
a positive too early or your clinicians will intervene too often,"
Dr. Emery cautioned.
In HIV-infected patients, quantitative viral load of CMV predicts
progression to CMV disease and is related to survival even more
strongly than HIV viral load. But therapy can’t be denied on the
basis of a low CMV viral load. As a result, Dr. Emery said, "We
don’t routinely use VL to trigger treatment. We use two consecutive
[qualitative] results, even though we know that leads to overtreatment."
In the future, Dr. Emery said, "Hopefully where we are heading is
individualized patient management."
Although newer, more active therapeutic regimens for HIV, such
as highly active antiretroviral therapy, or HAART, have reduced
CMV disease, Dr. Emery’s laboratory still monitors patients with
a history of CMV disease or a history of CD4 cells less than 100/mL.
But CMV disease is very low in those who respond to HAART, so it
is difficult to justify screening unless CD4 cell count deteriorates
or viral load rises. "We have had patients with a CD4 level of 25
who now have 250 cells," he noted.
"Molecular assays for CMV will play an increasingly important
role in management of CMV infections," Dr. Emery concluded. He recommended
testing for CMV in all post-transplant patients and selected persons
infected with HIV.
Dr. Emery also routinely monitors post-transplant patients for
human herpes virus 6 and 7. "We have had a few cases where marrow
engraftment was problematic, almost certainly due to HHV 6 infection,"
he told CAP TODAY. "But in order of importance, CMV is still the
dominant infection post-transplant."
Johns Hopkins’ Dr. Dumler tests for CMV on cerebrospinal fluid
or vitreous humor. His view of clinical context and optimal assays
differs from that of Dr. Emery. "I don’t like the PCR qualitative
assay for CMV," he stated. He believes CMV is ubiqitous, so it can
be amplified under many circumstances. "And the qualitative test
doesn’t correlate with disease so well," he added. "Antigenemia
measures viral load quantitatively and correlates well with disease.
To get a molecular assay for viral load that quantifies virus and
much more closely correlates with clinical disease is a very important
objective for the future."
Standardization of assays. Molecular testing may have become
too successful too fast. Such methodology was discussed at the 1999
meeting of the Association for Molecular Pathology. "What was very
striking was the great variation in how people prepare tests," said
David Hillyard, MD, director of molecular pathology and associate
medical director for infectious disease testing, ARUP Laboratories,
Salt Lake City. "We need to begin to consider the issues that contribute
to a good test. People need to exchange samples and see how tests
compare. And for quantitative tests, we need to see whether we are
all getting the same values."
"It is essential for us to get standardization of these assays,"
agreed Dr. Caliendo. "It is especially critical right now for hepatitis
C." The two quantitative assays, RT-PCR and bDNA, provide very different
results. "We anticipate that both companies will standardize their
assays to the World Health Organization standard that is now available,"
Dr. Caliendo is involved with a working group of the AIDS Clinical
Trials Group, which is evaluating CMV assays, and an NCCLS group
writing a general guideline for quantitative molecular diagnostics.
"With HCV and CMV, we are back to where we were with HIV assays
three to five years ago," she said.
William Check is a freelance medical writer in Wilmette, Ill.