Contamination and Causality
William Check, PhD
In a well-known test of creative intelligence, people are asked to list as many uses for a brick as they can think of. Research scientists and clinical laboratorians have been engaged in a similar activity for more than 20 years: trying to figure out how many medical uses they can identify for molecular diagnostics—methods based on hybridization, amplification and, most recently, sequencing of nucleic acids. Impressive success has rewarded this endeavor, particularly in the area of infectious disease. Molecular technology, it seems, can do virtually anything a clinical microbiology laboratory already does. In addition, nucleic acid-based tests can detect and identify many organisms that elude conventional methods or that current methods can find only slowly and imperfectly.
However, there is a major difference between listing uses for a brick and devising uses for molecular technology. In the intelligence test, every use counts, no matter how farfetched. In the clinical arena, a test has to pass stringent criteria before it’s adopted. Accuracy is important, of course, but so are ease of performance, cost-effectiveness and—above all—clinical impact.
"Many people confuse analytic sensitivity and diagnostic sensitivity," says Michael Wilson, MD, director of the Department of Pathology and Laboratory Services at the Denver Health Medical Center. "They think that because you have a [technically] better test, it results in better patient care. But this is not always true." Amplification assays can detect minute quantities of nucleic acid, but that does not always translate to detecting more infected patients, Dr. Wilson says. He cites the example of a patient with a pulmonary problem: An amplification assay might detect mummified mycobacterial DNA, while the patient’s symptoms might actually be due to a fungal infection. (See "Contamination and causality," side story.)
Right now only a handful of microbiological applications of molecular technology can be considered to have proven clinical value. These tests include assays for Chlamydia trachomatis and Neisseria gonorrhoeae detection, HIV viral load, hepatitis C virus detection and viral load, and detection of human papillomavirus, cytomegalovirus, and Mycobacterium tuberculosis.
In terms of volume, the top tests in clinical infectious disease done by commercially available molecular amplification methods for direct identification of microorganisms in clinical specimens are for Chlamydia trachomatis and Neisseria gonorrhoeae, often performed as a combined assay. Says James Versalovic, MD, PhD, director of molecular diagnostics at Massachusetts General Hospital, "Nucleic acid amplification-based detection of pathogen DNA is the gold standard" for these sexually transmitted diseases.
Amplification of HIV-1 RNA is a common molecular approach to the quantitative assessment of a pathogen. "Development of this assay allowed quantitation of the amount of HIV-1 in plasma, so that clinicians could follow patients on antiretroviral therapy," Dr. Versalovic says. HIV-1 drug resistance testing, including genotypic and phenotypic methods, is being used more often to manage patients who have failed an antiretroviral drug regimen, as recommended by the International AIDSSociety and published in 2000.
On the menu of molecular tests for HCV Dr. Versalovic lists qualitative and quantitative detection of HCV RNA and HCV genotyping. Food and Drug Administration approval of a qualitative nucleic acid-based test for HCV is expected soon, which Dr. Versalovic believes will increase the demand for testing further.
Turning to HPV, Dr. Versalovic says, "That situation has changed in a big way." In February, the National Cancer Institute issued a formal recommendation that HPV DNA testing be performed in all patients with ASCUS or AGUS determinations on a Pap test. Now, instead of a woman with an ASCUS or AGUS Pap test result being called back months later for another Pap test or even colposcopy, HPV DNA testing would be performed immediately on the same specimen, assuming a liquid collection method was used for the original cervical sample. "The key benefit from HPV testing, both for individuals and for the health care system, is that women may avoid unnecessary Pap smears and colposcopy if they are HPV DNA-negative," Dr. Versalovic says.
Nucleic acid-based testing is also used for detecting cytomegalovirus infections and infection with Mycobacterium tuberculosis, two applications that will be discussed in the June issue of CAP TODAY.
Dr. Versalovic calls the Chlamydia-Neisseria story "an example of rapid stepwise evolution" in diagnostic testing. Culture of both organisms is "challenging," he notes, so antigen testing was the preferred method of screening for most of the past two decades. More recently, nucleic acid hybridization made it possible to detect both pathogens directly with greatly increased specificity and improved sensitivity. "Finally," Dr. Versalovic says, "nucleic acid amplification techniques developed in the last decade have at least partially supplanted hybridization, due to increased sensitivity." But technical improvement isn’t always the last word. Amplification’s benefits are more apparent in high-prevalence areas; in low-prevalence areas, higher cost becomes a factor.
Better therapy for chronic HCV infection has stimulated molecular testing. In one trial, sustained (24-week) virologic responses were seen in 38 percent of patients with chronic HCV treated for 48 weeks with interferon alfa-2b plus ribavirin (N Engl J Med. 1998;339:1485-1492). Treatment with long-acting pegylated interferon (administered once weekly) appears to be a promising alternative (N Engl J Med. 2000;343:1666-1672, 1673-1680, 1723-1734). "We are basically talking about a curable disease," says David Gretch, MD, PhD, associate professor of laboratory medicine at the University of Washington. "If you get rid of virus, the liver will regenerate and histology will improve." Successful treatment also reduces the risk of hepatocellular carcinoma. And data are accumulating that show, even in patients who don’t get a sustained virologic response, treatment slows disease progression.
But treatment with interferon is expensive and can have serious side effects, so selecting appropriate patients is important. Criteria include elevated alanine aminotransferase, abnormal histology on liver biopsy, anti-HCV antibody, and viremia. "Some doctors treat patients without doing liver biopsy," Dr. Gretch says, "but that is not the practice at our center." In general, the presence of fibrosis on liver biopsy suggests the need for treatment. However, it is those patients with advanced fibrosis (stage 3 is bridging fibrosis while stage 4 is cirrhosis) who are likely to show the most immediate benefit from therapy, he says.
Molecular testing has a growing place in verifying HCV infection, Dr. Versalovic says. Confirming a positive ELISA for anti-HCV antibody is a bigger problem with HCV than with HIV-1, because of the relatively large number of indeterminate results with the HCV immunoblot test (RIBA). "Indeterminate results with HCV immunoblot are now confirmed by qualitative testing for viral RNA," Dr. Versalovic says. He predicts that qualitative molecular testing (RT-PCR, for example), which is more accurate than HCV immunoblotting, will be increasingly used to confirm a positive ELISA and may eventually replace RIBA.
Dr. Gretch already favors PCR for HCV RNA to confirm a positive ELISA. A false-positive ELISA is unusual in the tertiary-care setting because of the high-prevalence population. "RIBA confirms 96 to 97 percent of positive ELISA results," Dr. Gretch says, even when PCR is negative. Almost every antibody-positive, PCR-negative case represents resolved infection, which has a favorable prognosis. "Rarely will that individual show up infected later," he says.
(Dr. Gretch characterizes confirmatory testing among blood donors as "a whole different story." In this population, 40 percent of positive ELISA tests will be false positive, that is, RIBA negative.)
Verifying a positive ELISA test with PCR can be especially helpful for managing persons with normal ALT levels, who are seen largely as the consequence of a national campaign to promote awareness of HCV infection. (Many of the estimated 2.7 million HCV-infected persons in the United States are undiagnosed.) Persons who injected illicit drugs when young might come for screening, even though they are not sick. If they are antibody-positive, a negative PCR assay can reduce their worry; a positive PCR result might indicate the need for a biopsy. In one study of persons who were ELISA-positive but had normal ALT, all who were virus-free by PCR had normal liver biopsy, while all who were virus-positive by PCR had hepatic disease activity.
Controversy surrounds the management of HCV-infected persons with normal enzymes. The benefits of therapy have been shown so far only in patients with elevated ALT. Some studies even suggest that HCV-infected persons with normal enzymes don’t respond as well to therapy. Some physicians assert that such patients have mild disease, so they don’t need to be biopsied. But at the University of Washington, Dr. Gretch says, biopsy shows a range of disease activity in these patients. Their approach is to put PCR-positive patients with normal enzymes into a study, in accordance with a National Institutes of Health consensus conference recommendation.
For detecting HCV viremia, Dr. Gretch favors qualitative PCR, which is an order of magnitude more sensitive than quantitative PCR. Quantitative PCR still detects more than 93 percent of patients with viremia, because most patients have >700 copies/mL of HCV RNA. But Dr. Gretch prefers the qualitative assay for following patients on therapy because some fall in the zone where the qualitative test is positive but the quantitative assay is negative. Patients with viremia at 24 weeks of therapy may clear viral RNA if treated for an additional 24 weeks.
Dr. Gretch’s laboratory was one of the trial sites for the Roche Amplicor assay. "We have done a variety of in-house PCR tests for HCV," Dr. Gretch says, "and the Roche assay has evolved to where it is doing as well as any of our in-house tests."
His laboratory uses Bayer’s bDNA assay for determining HCV viral load to stage patients for probability of response to therapy. (Above about 2 million copies/mL the chance of a sustained response is significantly reduced.) "Most laboratories in the U.S. use the Roche Monitor assay because it is more sensitive," Dr. Gretch says. "But we don’t like it because at high viral load, above about 500,000 copies/mL, where most patients’ HCV RNA values fall, it is not linear." Dr. Gretch also uses bDNA for natural history studies. Bayer will soon introduce a third-generation bDNA assay that will detect as few as 2,000 copies/mL and be linear up to 50 million copies/mL.
David Hillyard, MD, director of molecular pathology at ARUP, Salt Lake City, identifies a move toward reporting HCV PCR results in international units (IUs), following a WHO recommendation, so that outputs can be standardized across platforms. Last summer tests became available that have been standardized to a reference pool of HCV virions. "Because this change can be confusing to clinicians, we are currently reporting out both IUs/mL and copies/mL until clinicians get familiar with both systems," Dr. Hillyard says. Because molecular testing for HCV RNA is also done with in-house assays, Dr. Hillyard notes, "it is a tremendous challenge to standardize. Proficiency testing is not nearly as organized as it needs to be." (Dr. Versalovic, a member of the CAPMicrobiology Resource and Molecular Pathology committees, says the CAPplans to consolidate molecular HCV proficiency testing in the HCVN Survey in 2002.)
Sequencing of the HCV genome is also used to guide therapy. "Clinicians more often require that we do HCV genotyping now," says Stephen Dumler, MD, associate professor of pathology and director of the division of medical microbiology, Johns Hopkins University. Type 1 HCV strains have about a 30 percent response rate to interferon plus ribavirin, significantly lower than the 65 to 70 percent response with non-type 1 strains. (Unfortunately, about 70 percent of U.S. patients are infected with type 1 virus.) A reliable commercial HCV genotyping assay is available (Inno-Lipa, Innogenetics). The Hopkins laboratory developed an in-house assay in which phylogenetically informative segments of the HCV genome are amplified, then sequenced using an automated ABI 377 instrument and compared to a database using built-in software.
Even patients who don’t have a sustained response to interferon therapy show slowed disease progression, and data suggesting this led to the NIH-sponsored HALT-C trial (hepatitis C antiviral long-term treatment against cirrhosis trial). In HALT-C, approximately 1,350 patients with chronic HCV and advanced hepatic fibrosis (Ishak stage 3-6) are treated with pegylated interferon plus ribavirin for 24 weeks. Patients who remain HCV-RNA positive (expected to be about two-thirds of subjects) are randomly assigned either to continue treatment with low-dose pegylated interferon alone for an additional 42 months or to have treatment discontinued. Liver biopsy is performed at baseline and after two and four years of treatment. Dr. Gretch’s laboratory is acting as the core virology laboratory for the trial.
HALT-C seeks to determine whether long-term, low-dose interferon therapy can reduce progression to cirrhosis, incidence of hepatocellular carcinoma, and need for hepatic transplantation among patients who do not achieve virologic response to optimal therapy. With the death rate from chronic HCV predicted to rise threefold in the next decade, any amelioration of the clinical impact of this disease would be a public health benefit.
In HIV diagnostics, says Angela Caliendo, MD, PhD, medical director of the microbiology and molecular diagnostics laboratories in the Department of Pathology and Laboratory Medicine, Emory University School of Medicine, "the biggest advance is that resistance testing has become part of the standard of care for managing HIV-infected patients." Measuring HIV-1 RNA in plasma (viral load) is still the primary approach to managing HIV-infected persons. "But when viral load gives you an indication of lack of response or loss of response," Dr. Caliendo says, "then resistance testing becomes the next step." Results of resistance testing are used with viral load, CD4 cell count, and the patient’s drug history to select a new regimen.
Detecting resistance to antiretroviral drugs is also recommended in HIV-infected pregnant women. And, says ARUP’s Dr. Hillyard, it "makes good theoretical sense" to search for resistance in selected cases of primary HIV-1 infection. "Up to 10 percent of newly infected patients have some evidence for resistance in their newly acquired virus," he says.
Two major methods are used to search for resistance in HIV-1, genotypic and phenotypic. In a genotypic assay, the protease and reverse transcriptase (RT) regions of the HIV genome are amplified and sequenced. Any mutations detected are analyzed by a set of rules, formulated by companies or expert committees, that predict sensitivity or resistance to antiretroviral drugs. "We have only a rudimentary and scanty understanding of how powerful or how useful that information is," Dr. Hillyard cautions. "So far a couple of studies have demonstrated that clinicians who have genotypic information can bring viral load about a half log lower." Further studies are ongoing.
In phenotypic resistance testing, the protease and RT genes from the patient’s viral isolate are cloned into a standard HIV-1 vector. The recombinant vector is introduced into cells, which are exposed to anti-HIV drugs. "This assay is very reminiscent of a microbial drug resistance assay," Dr. Hillyard notes. Because of the complexity of the cloning and recombination steps, phenotypic assays won’t even be done in most reference laboratories, Dr. Hillyard says. Samples have to be sent to the companies that devised the assays, ViroLogic and Virco. (Virco, a Belgian company, has an agreement with commercial labs to do its assay in the United States.)
Virco also offers what it calls a "virtual phenotype." "Basically," Dr. Hillyard explains, "you derive a sequence as in genotypic testing, then send it to Virco." The company queries a databank to see whether it has done a phenotypic assay on a similar or identical sequence, and, if it has, extrapolates the phenotypic resistance profile to your isolate. "In theory," Dr. Hillyard notes, this process "goes beyond rules-based analysis but has not been adequately evaluated by clinical studies."
Unfortunately, Dr. Caliendo says, "How [genotypic and phenotypic assays] compare to each other is not fully understood. Results from the two assays don’t always correlate. Which more accurately reflects how a patient will respond to therapy is unknown." Nor have there been trials on patients who fail their first regimen; studies reported to date have been in advanced patients. "There is a lot of work to be done on the clinical utility of resistance testing," Dr. Caliendo says.
She does genotypic resistance testing in her laboratory using the Visible Genetics system. A "challenge" is how she describes her lab’s experience setting up the genotypic assay. "It was the most complex assay that we have done in a clinical molecular laboratory," she adds. "You must have extensive experience in molecular technology before setting up this assay.
"One of the reasons we brought in genotyping," she continues, "was to be able to monitor closely the quality of testing. I am very interested in getting proper proficiency testing and quality control." At present there is neither adequate control material nor adequate proficiency testing for HIV-1 resistance testing. A private corporation, AcroMetrix, has started offering PT for a fee. It will send two five-specimen panels each year containing mixtures of wild-type HIV-1 isolates with strains bearing reverse transcriptase and protease mutations. Subscribers can get results and analysis of responses on a Web site. The CAP will also provide a challenge this year and plans to expand its PT program for HIV-1 genotyping in 2002. "The CAP plans to consolidate PTfor HIV viral load testing and HIV-1 genotyping in the HIVSurvey in 2002," Dr. Versalovic says.
Adding nucleic acid-based HPV testing to cervical cancer screening with the Pap test enhances the value of screening to patients, says Nancy Kiviat, MD, chief of pathology at Harborview Medical Center and director of cytopathology at University of Washington Associated Hospitals in Seattle. Under Dr. Kiviat’s leadership, all University of Washington Physicians Network (UWPN) clinics recently adopted routine testing of ASCUS smears for HPV. When obtaining a cervical specimen, a provider discusses with the patient whether she wants "HPV if ASCUS" or "HPV regardless" and marks that on the order form. Says Dr. Kiviat: "HPV testing has been adopted as the default policy for all UWPN clinics. Unless clinicians indicate otherwise, this is what happens. In practice," she adds, "it is rare for a patient to decline HPV testing."
If the HPV test is positive, the patient is scheduled for immediate colposcopy and biopsy. If it is negative, the patient returns to the regular screening schedule and has her next Pap test one year later. "One of the most important advantages of this approach," Dr. Kiviat says, "is that it more rapidly resolves the situation of a woman walking around thinking something is wrong but not knowing for sure." Women are immediately informed about the ASCUS finding, and a final report with the ASCUS diagnosis and HPV result goes out in 10 days to two weeks.
Dr. Kiviat estimates that 40 percent of ASCUS specimens are HPV positive. "So 60 percent of women [with ASCUS] are told they had mild changes on Pap smear that are not of clinical consequence and that they had no evidence of exposure to the agent that has been shown to play a central role in cervical malignancy." These women are spared anxiety, a return visit, and possible colposcopy.
Dr. Kiviat set the basis for this protocol four years ago when she introduced Cytyc’s ThinPrep liquid collection system. Clinicians place the cervical sample into a liquid-filled vial. Part is used for the Pap slide and the remainder stored. If an HPV test is indicated, it is done on the stored material. "We strongly discourage non-liquid samples now," Dr. Kiviat says. About six months ago she instituted routine HPV testing using Digene’s hybrid capture assay. A few months ago the current overall screening protocol was endorsed by the heads of the departments of obstetrics/gynecology, family practice, and medicine and the head of UWPN clinics. Dr. Kiviat’s ongoing communication with clinicians and her presentation of data supporting use of HPV testing facilitated acceptance.
Dr. Kiviat and outcomes researchers are analyzing how incorporating HPV testing into cervical cancer screening affects primary care practice. Outcome measures include time to resolution of abnormal Pap tests and cost-effectiveness. "We are assuming it will eliminate all return clinic visits for ASCUS," Dr. Kiviat says.
Other institutions are still in the early stages of introducing HPV testing. "I think it’s a good thing," Dr. Caliendo agrees. "We are preparing to evaluate the whole process." She finds gynecologists more interested in both liquid-based collection and HPV testing than when she arrived at Emory two years ago. "At that time we were just beginning to talk about it," she says. "Now we are beginning to implement it."
At Johns Hopkins, Dr. Dumler says, "HPV testing is blossoming slowly but surely. Getting samples took much longer than we anticipated. We want to be able to do HPV testing not as a standalone assay but in close collaboration with gynecologists and cytologists." Dr. Dumler’s vision is to provide clinicians with an integrated report containing both cytological and microbiological results, much like the system Dr. Kiviat and her colleagues have established. "We think this will be a superior format," he says.
Ultimately, Dr. Kiviat says, "Our goal is to assess HPV testing as a primary screen for cervical cancer control." It would replace cytology, in other words. Studies she and others have conducted have shown that using HPV testing as a primary screen may result in too many women being referred for colposcopy in certain age groups. While cytology is specific but not sensitive, HPV testing is sensitive but not specific. "The test as it is now may or may not be useful for primary screening," Dr. Kiviat says. "If it is useful, it will be most helpful in women over 35 years of age, most of whom have already acquired and resolved HPV infection. To make this assay useful for screening of all women will require further research," she concludes.
New treatment options for hepatitis B virus are causing a surge in demand for molecular testing. "Similar to HIV, nucleic acid testing is becoming the test of choice for quantitating the therapeutic response of HBV," Dr. Gretch says. As some anti-HIV drugs are found to work against HBV, viral load measurement can be used to maximize their effect. "There is no answer to how best to treat hepatitis B infection right now," Dr. Gretch says. "But nucleic acid testing is one tool to approach the question."
Putting molecular technology into clinical perspective, Dr. Wilson says, "There is no question that PCR revolutionized biomedical research. But the difference with diagnostic molecular biology is that, at the end, a physician has to interpret the results and use the results to treat the patient." He cites a remark by Barth Reller, MD, to the effect that, "We can sequence the human genome, but we still don’t know how to interpret the presence of coagulase-negative staphylococci in blood cultures." Nonetheless, Dr. Wilson believes that molecular diagnostics will play an increasingly important role in the clinical microbiology laboratory. Current applications, he says, represent "the front end of the coming wave of nucleic acid amplification."
William Check is a freelance medical writer in Wilmette, Ill.