| Picking the right site |
 |
|
|
November 2006 Picking the right site for HIV testing A quiet laboratory, late at night. A laboratory director stares into a computer screen, poring over an Excel spreadsheet, weighing the factors for and against bringing HIV testing in-house—cost, reimbursement, volume, technical staff, training, turnaround time, quality control, choosing an assay, satisfying clinicians. Our perplexed laboratorian stands and paces, muttering. We can just make out the words: “To test or not to test: What a difficult dilemma. And so unnecessary. If our conscientious director had only attended the workshop on bringing HIV testing in-house at the CAP ’06 annual meeting! In addition to addressing technical issues regarding HIV quantitative testing and genotyping and the technical and medicolegal issues surrounding rapid HIV testing, the session informed attendees how to improve patient outcomes and reduce laboratory costs by collaborating with clinicians. “Laboratories need to make sure that clinicians have access to molecular HIV testing, both HIV viral load testing and genotyping to determine viral resistance,” says Kathleen G. Beavis, MD, chief of the Microbiology and Virology Laboratories at Stroger Hospital of Cook County (formerly Cook County Hospital), Chicago, and chair of the CAP’s Microbiology Resource Committee. “Whether a lab brings these tests in-house or sends them out will pretty much depend on the volume of tests and the level of expertise.” Dr. Beavis, who organized and chaired the workshop, is reluctant to set a threshold volume that justifies doing viral load testing in-house, because this number depends on workflow and the overall economics of the contract the laboratory has with its reference laboratory. “Prices vary all over the map,” she says. She notes that some institutions might even be able to get diagnostic or viral load testing done free at a state laboratory. For diagnostic testing, in addition to a standard EIA, Dr. Beavis recommends that laboratory directors consider implementing rapid HIV testing for employee exposures and for women who present in labor with an unknown HIV status. “Whether a lab implements rapid testing for routine patient care as well depends more on its workflow and volume,” she says. “If your workflow allows you to do a couple of EIA runs per day, it might be less critical to offer a rapid test.” Since Dr. Beavis’ laboratory performs two EIA runs per day for HIV diagnosis, needlesticks that occur Monday through Friday on first shift are done with the standard EIA. “We use the rapid test only when there would be a delay in getting the EIA performed,” she says. Though rapid tests are faster for single specimens, when doing many samples a standard EIA has the shorter turnaround time. Some microbiology laboratory directors hesitate to do HIV testing in-house because HIV tests, especially genotyping of the HIV genome for antiretroviral resistance, are more expensive than many other tests they do. “We think we’re expensive relative to clinical chemistry,” Dr. Beavis says. “However, even genotyping is a drop in the bucket compared to a course of HIV therapy.” She recommends that laboratories work with clinicians to support their request to implement HIV tests. Clinicians in the HIV/AIDS center at Stroger Hospital of Cook County supported her budget with letters to administration and justification for bringing tests in-house. With this support, Dr. Beavis was able to bring genotyping in-house. Of the alternative (send-out) method of determining antiretroviral drug resistance, phenotyping, Dr. Beavis says: “Intuitively, phenotyping seems like a good idea. It parallels what we do in bacteriology. However, there are no outcomes data to show that it is worth the extra cost and delay.” When thinking about bringing viral load testing in-house, “Labs should first consider volume,” agrees James Versalovic, MD, PhD, director of Microbiology Laboratories and of the Division of Molecular Pathology, Texas Children’s Hospital, and assistant professor of pathology, molecular virology, and microbiology and of molecular and human genetics, Baylor College of Medicine. This decision also depends on the laboratory’s comfort level with send-out testing. How compelling is it to do something on site, he asks, so that specimens are there to be retested if necessary? “Certainly below 100 tests per year is usually an insufficient volume,” says Dr. Versalovic, who previously served on the CAP Microbiology Resource Committee. “Anything above 100 tests per year would be open to consideration. It gets back to urgency, turnaround time, and cost—what the institution is willing to invest to set up testing. For institutions doing a low volume, it may take a while to recoup costs,” he cautions. Dr. Versalovic also agrees it can be important to have rapid HIV testing available for accidental needlesticks and to prevent vertical HIV transmission in women coming in for obstetrical services. “The landscape is changing a bit now that there is an FDA-approved test for HIV diagnosis using a molecular method,” says Dr. Versalovic, referring to Gen-Probe’s Aptima HIV-1 RNA Qualitative Assay, approved in early October. This method may make it possible for laboratories to combine ELISA with molecular testing for confirmation of HIV infection as well as viral load testing and monitoring, he suggests. “AIDS was just recognized as I was going to medical school,” said Dr. Beavis, in introducing the workshop. “This illness has really shaped our careers.” In the 25 years since it was recognized, HIV infection has gone from being a death sentence to being a treatable chronic disease, at least in the Western world, which can afford highly active antiretroviral therapy, or HAART. Dr. Beavis’ clinical colleague, David Barker, MD, quantified this impact by showing during the workshop the dramatic drop in mortality among patients in the CORE Center for the Prevention, Care and Research of Infectious Diseases of Chicago’s Cook County Bureau of Health Services. Mortality attributable to HIV infection declined 17-fold in the five years from 1995, when HAART was introduced, to 2000. Total mortality dropped 11-fold. “These declines reflect national trends due to better therapy,” said Dr. Barker, who is chief medical officer of the CORE Center. (Dr. Versalovic noted that the impact of therapy on children has been so great that his hospital has shifted its focus to international pediatric AIDS care.) The consequence, Dr. Barker noted, is a greatly increased prevalence of people living with HIV infection. “As a result, clinicians are managing many more people living with HIV infection,” he continued, “and they use lots of lab tests to do this.” The CORE Center now manages about 4,500 HIV-positive individuals. What clinicians routinely want from the laboratory, Dr. Barker said, are CD4 counts with a turnaround time of one to three days and viral load testing with a turnaround time of less than 10 days. Clinicians order a set of tests (viral load, CD4, etc.) at baseline, four to eight weeks after changing therapy, and routinely two to four times per year. Resistance testing has the lowest usage rate of all HIV tests, according to Dr. Barker. It is useful for patients frankly failing their regimen, a very low fraction of all HIV-positive people. He noted that 10 percent of patients coming to clinic have a high CD4 count and are not on therapy. Between 35 percent and 60 percent are doing well and do not need resistance testing. Another 10 percent to 15 percent have a viral load <1,000 copies/mL. “Genotyping is not entirely reliable with low loads,” he said. Twenty percent to 25 percent have poor adherence and look like they are failing but actually are not. That leaves five percent to 10 percent needing a resistance test each year. The cost of genotyping—from $200, plus the time of an experienced technologist—needs to be weighed against the $800 to $1,000 per month cost of HAART. Overall laboratory costs make up about five percent of the average annual cost of care for HIV-positive patients, which is about $15,000, Dr. Barker calculates. Medications make up about two-thirds. Inpatient cost has declined greatly since HAART was introduced. From 1994 to 2004 the average daily census of HIV-positive patients at Stroger Hospital of Cook County declined from 84 to 22, Dr. Barker said. This reduction saves $68 million per year. For the clinician, the most important feature of viral load testing is low-end repeatability. “Low-end sensitivity is not important if the lack of repeatability means that reported values cross back and forth across a threshold,” Dr. Barker said. Distinguishing 75 copies/mL from none is more important than distinguishing between 150,000 and 100,000 copies/mL. Interpretation of a viral load test has great impact on patients, Dr. Barker explained. To a patient “undetectable” means “You are fine.” “Detectable” means “You will progress and get sick and maybe eventually die [of your disease].” He suggested thinking about viral load testing in the context of monitoring for relapse of leukemia or breast cancer. Unfortunately, many clinicians don’t understand coefficients of variation and are uncomfortable with reporting of viral load results in log units. “Only clinical virologists think in logs. All others want whole numbers,” Dr. Barker said. Though Dr. Beavis thinks “it would be nice if clinicians were comfortable with logs,” she reports viral load test results in both logs and whole numbers, while encouraging clinicians to become familiar with logarithmic values. Presenting the laboratorian’s viewpoint at the workshop, Dr. Versalovic listed the most important HIV-related tests: diagnosis of HIV infection by antibody detection with ELISA followed by Western blot confirmation; viral load testing for following patients on therapy; genotyping for detecting mutations in reverse transcriptase (RT) or protease genes that confer resistance to HAART; and phenotyping to detect resistance. Viral load and CD4 counts at baseline are the main predictors of mortality, Dr. Versalovic noted (Taha TE, et al. AIDS. 2000;14:453–459). In this study, a one unit log10 increase in HIV RNA level increased the hazard of child mortality by more than twofold. Results of viral load testing before therapy is initiated show the patient’s viral load setpoint, which indicates the probable rate of progression to AIDS. During therapy, a rise in viral load may indicate noncompliance or drug resistance and treatment failure. For quantitative determination of HIV RNA, nothing is new, Dr. Versalovic said. For several years there have been three FDA-approved technologies and all are “basically equivalent” in performance, he said. Reverse transcription PCR (RT-PCR, Roche) and NASBA (BioMerieux) are based on target amplification; branched DNA (bDNA, Bayer) on signal amplification. All require an internal standard or an external calibration curve. Nucleic acid amplification methods have a higher sensitivity than antibody tests for detecting HIV infection, particularly for those who are highly contagious and acutely infected (Pilcher CD, et al. N Engl J Med. 2005;352:1873–1883). However, serology by EIA remains satisfactory for routine diagnosis. In the most recent adult treatment guidelines of the International AIDS Society-USA panel, the role of testing remains central (Hammer SM, et al. JAMA. 2006;296:827–843). The guideline says, “Initiation of therapy continues to be recommended in all symptomatic persons and in asymptomatic persons after the CD4 cell count falls below 350/microL and before it declines to 200/microL.” For patients on therapy, “The virologic target for patients with treatment failure is now a plasma HIV-1 RNA level below 50 copies/mL.” This makes low-end sensitivity in viral load testing “even more important,” Dr. Versalovic noted. Among viral load testing methods, RT-PCR performed with the Roche Amplicor Monitor on the Cobas automated platform is “the most widely used system,” Dr. Versalovic said. It was the first system to combine amplification and hybridization in a single microtiter well. He said it was “revolutionary” when it was introduced but added, “Frankly, its widespread application has led to some stagnation.” (In the 2006 CAP HIV viral load survey, about one-fifth of participants use bDNA and about two-thirds use PCR.) Dr. Versalovic advised laboratorians to watch for real-time methods for RT-PCR and NASBA. Real-time adaptations of amplification technology are easier to perform, require less labor, and carry reduced cost. Dr. Versalovic expects them to perform better with lower coefficients of variation and to be more accurate and reproducible. “Although we haven’t implemented them in the clinical lab yet, so that remains to be seen,” he cautioned. Dr. Beavis, whose laboratory has done about 13,000 viral load tests per year for the last several years, initially used RT-PCR for this test. However, she noted, it requires three technologists, two with molecular expertise, and is labor-intensive, requiring considerable hands-on time for extraction. She recently switched to the bDNA procedure, which requires only one technologist—a tech with a chemistry background. Now, she said, the three technologists formerly needed to run RT-PCR run HIV testing, quantitative HCV, HIV genotyping, and HCV genotyping. When figuring costs, she noted, “Personnel is as big a factor as reagent costs.” Another factor in her decision was that RT-PCR required three controls per nine patient samples at that time, more than with bDNA. “We had rows of Cobas instruments,” she said. To complicate matters, sometimes the cell pellet was lost, requiring a repeat sample. Also, about 10 percent of samples required dynamic range repeats. This is because the Roche Monitor assay doesn’t have an adequate lower limit of sensitivity to call patients “undetectable.” For samples below its lower limit, a second, more sensitive RT-PCR must be run. With bDNA, the lower limit of sensitivity is not as low as the more sensitive RT-PCR—75 copies/mL versus 50—but, Dr. Beavis told CAP TODAY, “Our clinicians don’t consider that a clinically relevant difference.” And a bDNA result can be achieved with one test. Turnaround time for viral load testing in Dr. Beavis’ laboratory is two to three days. Though Dr. Beavis admitted that “real-time methods will be the way to go” when they are approved, she noted, “They are not here yet. Right now those tests are not FDA cleared, and that presents another hurdle for the lab.” Dr. Barker cited a brief report from a reference laboratory that found a 79 percent coefficient of variation for RT-PCR viral load testing at the low end, but only 20 percent for bDNA (Peter JB, Blum RA. J Acquir Immune Defic Syndr. 2002;30:261–262). At the Stroger Hospital CORE Center, the fraction of undetectable viral load results increased slightly after switching to the bDNA method, according to Dr. Barker. “We’ve come to believe that PCR was telling us randomly that people were undetectable sometimes when they really weren’t,” he said. Speaking to CAP TODAY, Dr. Versalovic conceded that bDNA is simpler technically and requires only one assay for its full dynamic range. And, he agrees, “bDNA is easier to implement and may be the most user friendly right now.” The CAP HIV/HV2 survey generally shows higher coefficients of variation for RT-PCR. However, he reiterates that RT-PCR is likely to get a boost in performance and user friendliness with real-time assays. And he notes, “RT-PCR has a longer history and a larger user base, which provides some comfort level for labs.” Clinician comfort levels are an additional consideration. “If physicians support a particular approach, even if it is better for the lab, we need to think about that,” he says. Dr. Versalovic next reviewed resistance testing. The International AIDS Society-USA Panel recommended resistance testing for those with acute HIV-1 infection, anyone with HIV-1 infection within the previous 12 months, patients on treatment with a suboptimal HIV-1 response to therapy, at first or multiple regimen failure, and in a pregnant woman if she has detectable plasma HIV-1 RNA levels (Hirsch MS, et al. Clin Infect Dis. 2003;37:113–128). Performing resistance testing in all newly diagnosed patients was first recommended several years ago. The most recent treatment guidelines say, “Resistance assays may be of value in selecting the initial treatment regimen, because transmission of drug-resistant HIV strains leading to suboptimal virologic responses has been documented and there is evidence of increasing rates of drug resistance among newly diagnosed patients both in Europe and the United States.” Several randomized trials have shown that genotyping plus expert advice improves patient outcomes by facilitating drug selection. For instance, in the VIRADAPT study, conducted among heavily pretreated patients, those receiving genotyping had lower viral load over six months and more of them had less than 200 copies/mL (Durant J, et al. Lancet. 1999;353:2195–2199). Benefit was maintained over 12 months (Clevenbergh P, et al. Antivir Ther. 2000;5:65–70). Similar benefit was found in GENOPHAR (Bossi P, et al. HIV Med. 2004;5:352– 359) and the French Narval study (Vray M, et al. Antivir Ther. 2003;8:427–434). Two main methods for genotyping HIV-1 are reverse hybridization (Inno-Lipa, Bayer) and sequencing by electrophoresis (Trugene, Bayer; Viroseq, Abbott). Sensitivity of the Inno-Lipa method is lower, Dr. Versalovic said. By a short time after infection, all patients are infected with HIV-1 quasi-species (genetic mixtures). Sequencing detects 20 percent minority variants, while Inno-Lipa can detect variants present at less than 10 percent. Viral quasi-species below this level won’t be detected, but Dr. Versalovic questions whether such minority strains are clinically important. Interpreting results of genotyping is complicated and requires computer algorithms, which are widely available. “As pathologists we’re continually challenged with how we manage information,” Dr. Versalovic said. “Interpreting complex data will enhance the role of the [molecular] pathologist in the next few years.” Among 56 participating laboratories doing genotyping, proficiency testing by the CAP and Acrometrix found satisfactory performance by 55 percent of laboratories using Viroseq, 62 percent of those using Trugene, and 33 percent using a homebrew method. (“Satisfactory” requires identifying eight of nine mutations present in a 50:50 mixture.) The alternative to genotyping for detecting resistance is phenotyping (PhenoSense, ViroLogic). In this assay, the protease and RT genes from the patient’s virus are recombined into a standard HIV-1 strain, which is then cultured to obtain an IC50. At first PT looks like a standard antibiogram. “But,” Dr. Versalovic noted, “it is more complicated, requires send-out testing, and is very expensive.” One company translates genotyping results into a “virtual phenotype” through an extensive database of genotype/phenotype correlates. Costs of the three types of resistance testing are $800 to $1,000 for phenotyping, $200 to $400 for genotyping, and $400 to $500 for virtual phenotyping. While these prices may seem expensive, Dr. Versalovic notes, “The overall cost of treatment completely dwarfs the cost of a resistance test.” Because most patients will be treated for a long time, selecting an effective first regimen makes resistance testing a good cost-benefit proposition. In the CREST study, patients were assigned to genotyping plus an interpretive algorithm or to virtual phenotyping (Hales G, et al. PLoS Clin Tr. 2006;1:e18). After 48 weeks there was no significant difference between groups in HIV-1 viral load or CD4 counts. The authors concluded, “[R]esistance testing using genotyping linked to a reliable interpretive algorithm is adequate for the management of HIV infection.” In summary, Dr. Versalovic said:
As future trends to watch, Dr. Versalovic cites the push to “routinize” HIV screening and possibly remove the barrier of informed consent, resulting in an “opt-out” rather than “opt-in” system; and faster genotyping/sequencing with better informatics. Dr. Barker presented a clinician’s view of resistance testing. Most trials show a 13 percent to 28 percent advantage in achieving virologic control with the next regimen in patients managed with genotyping, he said. Havana, a European multicenter study, went a bit further: It compared genotyping plus an algorithm, drug recommendations from an expert advisory committee, the combination of both, or neither (Tural C, et al. AIDS. 2002;16:209–218). Viral load was measured at three and six months. “Factors associated with a higher probability of plasma HIV-1 RNA <400 copies/mL were HIV-1 genotyping and expert advice in patients failing a second-line antiretroviral therapy,” the investigators reported. Dr. Barker concluded, “Expert review of results, especially in conjunction with treatment history, can outperform genotyping results plus an algorithm.” At 24 weeks, 59 percent of patients receiving expert advice (with or without genotyping results) achieved viral load <400 copies/mL, significantly greater than the 41 percent among those who did not receive expert advice, even if their clinician got genotyping results. Algorithms for interpretation of genotyping are “not very helpful,” in Dr. Barker’s view, because they do not account for past mutations that are now lost from view but are reflected in the patient’s medication history. However, he noted, in patients failing their third regimen, nothing makes much difference, since no drugs are left. Resistance testing is especially important among newly diagnosed patients, Dr. Barker stressed, because 11 percent of them are infected with strains that bear resistance mutations in the RT or protease genes or both. “The implication is that those who know they are HIV-positive and have taken medications and developed resistant virus are infecting others,” Dr. Barker said. Detecting primary resistance matters because people with resistant virus fail initial therapy faster (Little SJ, et al. N Engl J Med. 2002;347:385–394). He offered these conclusions about resistance testing:
Rapid testing for HIV diagnosis is important for women in labor with unknown HIV status and for employee needlesticks, Dr. Beavis said. About 300 HIV-positive babies are born each year in the U.S. Simple peripartum antiretroviral prophylaxis can greatly reduce vertical transmission (Lallemant M, et al. N Engl J Med. 2004; 351:217–228; Moodley D, et al. J Infect Dis. 2003;187:725–735). Rapid point-of-care testing after 34 weeks of gestation or during active labor by obstetric nurses makes prophylaxis possible, as demonstrated by the national MIRIAD (Mother-Infant Rapid Intervention at Delivery) study, in which Cook County Hospital participated (Bulterys M, et al. JAMA. 2004;292:219–223). Both rapid test and EIA detected all 34 positives among the 4,849 women who were tested. Rapid test had four false-positive results, EIA 11. When MIRIAD was being conducted, explicit informed consent was required from pregnant women for testing. Eighty-five percent of women approached for testing in MIRIAD agreed. Since then many states have changed to an opt-out rule—a woman has to sign that she refuses testing. In Illinois, state law requires rapid HIV testing for pregnant women with unknown HIV status who come to labor and delivery unless they decline testing in writing. For employee needlesticks, too, anti-HIV therapy reduces transmission when begun soon after percutaneous exposure. Dr. Beavis offers rapid testing for employee exposures on all three shifts seven days per week. In Illinois the consent of the source patient is not needed to test, but this varies by state. Sixteen source patients were HIV-positive and 22 were HCV-positive in 2004 at her hospital. Rapid testing can also be used for routine patient screening. “We don’t use rapid tests for patient screening because our volume is so huge,” Dr. Beavis told CAP TODAY. Her laboratory performs 75 EIAs for HIV diagnosis each day. For EIA, the turnaround time is three hours, 30 minutes, including pipetting and results entry. Essentially this is a “next day” test, Dr. Beavis says. Each rapid test takes about 20 minutes, but they must be done one at a time, so they are not feasible for a large-volume setting. For a lab that does three specimens per day, on the other hand, rapid testing is a good option. Do the rapid test in-house and send out for confirmatory testing. Sometimes a clinician argues the rapid test is being used for employees and women in labor and not on his or her patients. Dr. Beavis stresses that they are seeking different goals in the different situations. “For employees and moms, we are trying to prevent transmission of disease, where hours matter,” she notes. “This is different from routine diagnostics.” In an attempt to reduce the number of undiagnosed HIV cases in the community, a federally funded pilot study was done in the emergency department at Stroger Hospital. Over 15 months, trained personnel randomly approached 5,290 ED patients and offered HIV counseling and rapid testing on site; 3,200 (60 percent) agreed. “I was really surprised by how many consented,” Dr. Beavis says. This approach found 82 new HIV-positive people. More than 75 percent of patients who tested positive reported for followup care. Among the 5,290 patients asked about testing, 87 percent were approached by counselors. The other 13 percent were referred to the counselors by ED physicians if they thought a patient had a high probability of being HIV-positive. “It’s no surprise,” Dr. Beavis says, “that those referred by doctors had a higher yield from testing.” The positivity rate among referred patients was 11.6 percent, compared with 1.2 percent among randomly selected patients. “Even though the yield was much higher among doctor-referred persons,” Dr. Beavis told CAP TODAY, “the randomly approached people represent a group who would never have been tested or diagnosed.” So both methods are valuable. Another element of the ED study was to compare accuracy of the rapid test to that of EIA. “Rapid testing was equivalent to EIA,” Dr. Beavis says, “but rapid test results really need to be confirmed, because there were a few false-positive findings.” So, if you do rapid testing, have provision for people who test positive to get confirmatory testing. This typically means getting a phlebotomy specimen. Also, Dr. Beavis advised, you will need medical and psychosocial support for those who test positive, and a plan for getting them a followup appointment. “You always need to think ahead,” she cautions. “Ask yourself, if I get a positive result, how will I deal with this patient?” In summary, Dr. Beavis said, for routine diagnostics, she does EIA; for women in labor with unknown HIV status, the nurses do a point-of-care rapid test; for employee needlesticks, EIA is the norm, with rapid test when EIA is unavailable. They have not extended the ED study, but if they did, Dr. Beavis would use a POC rapid test. Two important additional points about rapid testing during labor: A positive rapid test proceeds directly to Western blot confirmation, and bringing specimens to the laboratory adds more than one hour to the turnaround time. There are now four FDA-approved rapid tests:
Dr. Beavis offers this final thought: ”Understand your state law.” For example, Illinois law prohibits providing unconfirmed positive HIV test results from an EIA to patients or physicians, except for women in labor and employee needlesticks. However, giving unconfirmed results from rapid tests is allowed. She also notes that the Centers for Disease Control and Prevention recently recommended routine universal screening. However, laws about consent and confirmatory testing are state based. “For me personally,” Dr. Beavis says, “in some populations where you know HIV prevalence is higher, you might go to universal screening. Right now I think it would be more efficient to put our effort into these higher-risk populations, since resources are limited.” For a hospital-based laboratory, she notes, universal screening might mean testing everyone admitted, a daunting prospect. William Check, PhD is a medical writer in Wilmette, Il. |
