HIV Diagnostics: New Developments and Challenges
Grosvenor Resort, Orlando, Florida
February 28 – March 1, 2005
The CDC/APHL-sponsored, HIV diagnostics meeting was attended by over 200 researchers, laboratorians, and industry representatives to share the latest information on testing technologies and alternative methods to increase the uptake of testing and diagnosis of persons with HIV infection, and to discuss HIV incidence and drug-resistance surveillance. The two-day meeting included over 50 poster and 39 oral presentations on the following seven topics: (1) performance characteristics and use of recently approved HIV enzyme immunoassays; (2) nucleic acid amplification testing (NAAT) for acute HIV infection; (3) rapid HIV testing in the United States and international settings; (4) options for confirmatory testing in different settings; (5) dried blood spot (DBS) testing; (6) tests for recent HIV infection; and (7) FDA requirements for use of tests for recent infection for purposes other than surveillance. A brief summary of the content and themes for each of these topics are presented below.
1. New HIV Enzyme Immunassays (EIA)
This session included presentations on (1) the performance characteristics of the bioMerieux Vironostika HIV-1 Plus O and Bio-Rad Genetic System’s HIV-1/2 Plus O, (2) an evaluation of Bio-Rad’s new EIA assay by the Florida Bureau of Laboratories and by the Maryland Department of Health, and (3) a survey on HIV assays currently in use in public health laboratories. Both of the two new HIV EIAs incorporate 3 (Vironostika) or 4 (Bio-Rad) antigens including a group O specific antigen for increased detection of HIV-1 Group O. One of the antigens in the Bio-Rad assay is a HIV-2 specific antigen that allows detection of HIV-2. Both assays incorporate reagents that allow detection of IgG and IgM antibodies. In the case of the Bio-Rad assay, IgG and IgM are detected by utilizing the antigen-sandwich approach, while the Vironostika Plus O assay uses a goat anti-human immunoglobulin that binds both IgG and IgM. The Bio-Rad HIV-1/2 Plus O has been available since 2003 and replaces the Genetic Systems HIV-1/HIV-2 peptide assay. bioMeriuex’s Vironostika HIV-1 Plus O should be available in June, 2005 and will replace the Vironostika HIV-1 Microelisa system. Compared with their predecessors, both assays have greater sensitivity to detect (1) lower antibody titers, (2) persons earlier during seroconversion, (3) Group O infection, and (4) HIV-2 infection (Bio-Rad only). However, laboratories that have adopted the Bio-Rad HIV-1/2 Plus O assay have observed an increase in the number of repeatedly reactive EIAs with negative or indeterminate Western blots. The cause of these repeatedly reactive EIAs with negative or indeterminate blots could not be confirmed for all specimens. They might be caused by the increased sensitivity resulting in identification of samples that would be missed by less sensitive Western blots or indirect immunofluorescence assays, or by cross contamination, cross-reaction, or non-specific binding in the testing procedure. The Maryland Department of Health Laboratory, for example, found decreased specificity of Bio-Rad’s Plus O assay when testing cadaveric samples when compared to the Bio-Rad HIV 1/2 peptide assay. The Florida Bureau of Laboratories recommended that laboratories that use Bio-Rad’s Plus O assay increase monitoring and adopt procedures to prevent cross contamination which may produce higher false-positive rates. In a survey of 56 public-health laboratories, 63% of responders currently use the Vironostika HIV-1 EIA, (the assay least sensitive for early seroconversion); 10% use the Bio-Rad HIV1/2 Plus O, and 22% reported that they intend to change EIAs in 2005.
2. Nucleic Acid Amplification Tests (NAAT) for Acute HIV Infection
This session included presentations on the application of NAAT for donor screening and for publicly funded HIV testing in domestic and international settings, and an overview on CDC’s new primary HIV infection (PHI) study. The feasibility of screening for acute HIV infection using NAAT on pooled seronegative specimens has been repeatedly demonstrated in different settings and in multiple countries. However, the cost-effectiveness of detecting acute HIV infection through NAAT varies considerably by setting and is expected to change with adoption of third-generation, highly sensitive EIAs and use of rapid HIV tests that obviate serum collection for those who test negative. Over the first three years of NAAT screening for U.S. blood donors, for example, only 12 acute HIV infections were identified in 37 million donations, corresponding to >$1,000,000 cost per quality-adjusted life year. In Maryland, NAAT screening (in-house RT-PCR) of over 15,000 pooled specimens seronegative by the 2nd-generation Genetic Systems HIV-1/HIV-2 assay has not identified any acute infections, possibly because of increased availability and uptake of oral fluid and rapid HIV testing by high-risk persons. However, NAAT screening of clients tested in public-health clinics in San Francisco and North Carolina who were antibody negative by the 1st-generation Vironostika identified 10% and 4%, respectively, more HIV infections than would have been identified by using the Vironostika assay alone. Identification of acutely infected persons and referral of their partners may be a particularly important prevention strategy if, as suggested by some studies, a high proportion of HIV transmission is attributed to persons with PHI. CDC’s new PHI study is expected to begin in the summer of 2005 and will be conducted by health departments in Florida and Los Angeles, in collaboration with the Florida Bureau of Laboratories and the New York State Wadsworth Laboratory (respectively). The primary objective of this study is to determine the yield and cost effectiveness of pooled NAAT in addition to antibody screening, coupled with partner notification and referral efforts to detect HIV infections that would have remained undiagnosed with conventional 2nd and 3rdgeneration EIA screening.
3. Rapid HIV Testing in the United States and International Settings
The session on domestic testing included presentations on implementation barriers and outcomes of rapid testing by several state and local health departments. The implementation of rapid HIV tests in public-health settings has varied considerably by state and local health department due to both state regulatory requirements and system-development needs. State regulations that require phlebotomy training for personnel performing finger-stick rapid-tests (California) or that CLIA-waived laboratories have lab directors with a PhD (Pennsylvania) were barriers for implementing CLIA-waived rapid tests in community-based organizations. However, system development needs such as revisions to state and local laboratory forms, operator training, and establishing effective quality assurance programs were, by far, the most important implementation barriers. Addressing these needs requires considerable planning and resources, and as a consequence, roll out of rapid testing in the United States has taken time but has been successfully implemented in several states. The number of rapid tests performed by the Florida Department of Health (FDOH), for example, increased from approximately 4,000 in 2003 to approximately 24,000 in 2004. In 2005, FDOH expects to conduct between 40,000 and 50,000 rapid HIV tests. Data presented from New York State and Florida, and from CDC’s OraQuick post-marketing surveillance project suggest that (1) most persons seeking HIV testing services prefer rapid over conventional tests; (2) nearly all persons who are tested with rapid HIV tests receive their results; (3) most health-department rapid HIV test programs have instituted effective quality assurance programs; and (4) rapid tests have performed as expected with very few invalid and false-positive tests reported. As a consequence of the successful roll out of rapid testing, however, some laboratories have had a reduction in the number of conventional HIV tests performed. The potential reduced need for public health laboratory HIV testing and corresponding loss in funding should be monitored closely to mitigate any impact on requisite infrastructure.
Presentations on rapid HIV testing in international settings described the implementation and performance of rapid testing in resource poor countries. Rapid testing in CDC supported Global AIDS Program countries has expanded quickly from a total of less than 250,000 tests conducted in approximately 250 sites in 2002 to approximately 2,000,000 tests conducted in 1800 sites in 2004. Of the 15 targeted President’s Emergency Plan for AIDS Relief (PEPFAR) countries, 8 have implemented rapid testing in over 600 sites. In these 8 countries, a sequential rapid testing algorithm is currently used in 5 and a parallel testing algorithm is used in 3. In 4 countries with quality assurance programs that involved re-testing in reference laboratories, concordance between on-site rapid testing and reference laboratory re-testing ranged from 95.7% to 99.5%. However, roll out of rapid testing in PEPFAR and other resource poor countries has identified considerable laboratory infrastructure and training needs to develop and sustain effective quality-assurance programs. For example, only one of the PEPFAR countries has implemented an external proficiency-testing program for rapid test operators. In 6 external assessment exercises conducted in 29 Asian and Pacific countries by the national reference laboratory in Australia, rapid tests produced higher error rates than EIAs. Follow-up visits to participating laboratories indicated insufficient documentation of laboratory procedures, insufficient use of external controls, and a lack of understanding and documentation of quality-assurance principles and practices.
4. Options for Confirmatory Testing
This session included presentations on potential alternative HIV testing algorithms that (1) reduce the use of the Western blot (WB) by using NAAT and highly sensitive EIAs for supplemental confirmatory testing, (2) eliminate the WB by using rapid tests in combination, and (3) use a combination of the Multispot rapid test and HIV-1 and HIV-2 EIA/WB assays to detect HIV-2 infection. Two studies (CDC and American Red Cross) indicated that including NAAT within an HIV testing algorithm increases sensitivity by detecting acute HIV infection and reduces the need for the WB and the number of indeterminate WB results given to non-infected persons. However, because some HIV-infected persons have non-detectable viral loads, both studies suggested that NAAT cannot completely replace the WB as a supplemental confirmatory test. The CDC study, for example, found that as many as 3.3 % of serology positive individuals were NAAT negative. Of 12.4 million American Red Cross blood donations from 1989 through 1999, 11,080 were EIA repeatedly reactive (RR), and of these, 7.1% were WB positive, 46.6% were WB indeterminate, and 46.3% were WB negative. The high-rate of indeterminate and negative WB results is attributed, in part, to the FDA-required interpretation for blood donation screening (any WB background discoloration or band must be reported, as opposed to only viral bands). Because of the high rate of WB negative and indeterminate results, an alternative algorithm for blood donors has been proposed to the FDA. The algorithm involves an initial EIA screen; individual NAAT on donations that are EIA-RR; and pooled NAAT on all seronegative donations with discriminatory NAAT on deconstructed pools to identify RNA-positive donors. Donations that are EIA-RR/NAAT positive are considered infected (WB is not performed). EIA-RR/NAAT-negative donations are tested with a second EIA using a different antigen. If the second EIA is negative, the patient is considered to be non-infected (WB is not performed). Western blots are only used to resolve donations that are EIA-RR by the screening EIA, NAAT-negative, and EIA-RR by the second EIA. The new algorithm would improve sensitivity by identifying donors with early infection, and eliminate 98.5% of WB tests and 98.6% of indeterminate results (and unnecessary anxiety) given to non-infected donors.
A prospective evaluation of rapid tests conducted by the CDC and the Los Angeles Department of Health Services demonstrated the sensitivity and specificity of individual rapid tests. On the basis of these results, simulated algorithms were developed using up to 3 different rapid tests. Each was as sensitive and specific as the conventional EIA/WB algorithm used for comparison. In this study, persons who tested negative on the first rapid test were considered to be noninfected (no further tests performed). If the first rapid test was positive, a second different rapid test was performed. If both tests were positive, the patient was considered to be infected (no further tests performed). Finally, a third rapid test (different from the first two) was used if only the first two rapid test results were discordant. For these discordant cases, persons who tested positive on 2 of 3 rapid tests were considered to be infected; persons who tested negative on 2 of the 3 rapid tests were considered to be noninfected. The algorithm was evaluated using 10 combinations of up to 3 different rapid tests (2 evaluated using whole blood only, 1 using whole blood and oral fluid). Compared with the conventional EIA/WB algorithm, all 10 combinations of the rapid-test algorithm yielded sensitivities > 98% and all were 100% specific (no false positives).
The final presentation was on a new HIV testing algorithm used by the NYC Public Health Laboratory (PHL), which was instituted because of increasing concerns about HIV-2 infection. The average annual number of HIV-2 infections by the NYC-PHL was 5 from 1988 through 1993 (mean 158,959 specimens tested/year), 12 from 1994 through 1997 (mean 156,020 specimens tested/year), and 38 from 1998 through 2004 (mean 120,200 specimens tested/year). Instituted in April 2004, the new HIV-1/2 diagnostic algorithm involves (1) screening all specimens with the Bio-Rad EIA HIV-1/2 Plus O followed by an HIV-1 WB on EIA-RR specimens; (2) using the MultiSpot rapid test and HIV-2 EIA/WB algorithm on HIV-1 WB positive specimens, and (3) using a second different HIV-1 EIA (Bio-Rad rLav) and HIV-2 EIA/WB algorithm on WB negative or indeterminate specimens.
5. Dried Blood Spot (DBS) Testing
This session included presentations on the use of DBS testing for (1) the Newborn Screening Quality Assurance Program (NSQAP), (2) HIV surveillance and research in Canada and internationally, (3) the diagnosis of HIV infection in infants through detection of HIV DNA and RNA, (4) BED-CEIA-based detection of recent infection, and (5) HIV subtype and drug-resistance testing. Since 1989, the NSQAP has conducted routine evaluations of new filter paper lots to ensure uniformity of testing, and produces, distributes, and evaluates DBS quality control and proficiency testing materials. Currently 49 laboratories in 18 countries participate in the NSQAP quality assurance program for HIV testing. Multiple studies presented suggest that, provided they are adequately collected, allowed to dry, and subsequently protected from heat and humidity, dried blood spots are stable and can be used for multiple applications for HIV surveillance (prevalence, BED-CEIA-based incidence estimation, and subtype determination) and clinical management (viral load and drug resistance). Because of their functional versatility, low expense, and ease of use, transport, and storage, dried blood spots offer considerable advantages over conventional serum or plasma collections. In Canada, for example, DBS is used in multiple studies assessing HIV prevalence, incidence, drug resistance, and subtype infection among injection drug users and men who have sex with men. DBS may also play an important future role in the detection and clinical management of HIV-infected infants in Africa and Asia. CDC has developed a magnetic bead-based, real-time reverse transcriptase, polymerase chain reaction (PCR) assay used to detect both HIV-1 DNA and RNA from dried blood spots obtained from infants. Test results of the real-time PCR were 99% concordant with conventional NAAT (Amplicor). The per test cost of the real-time PCR is $5 (U.S.), considerably less than commercially-available NAAT. Technology transfer of the test is currently underway in Uganda, Kenya, Thailand, and South Africa.
6. Status of Tests for Recent Infection
This session included presentations on (1) preliminary results from a less sensitive (LS) incidence-assay approach for a simple and inexpensive particle-agglutination assay, (2) a proposed weighting mechanism for assay-based HIV incidence estimation, (3) an evaluation of the comparability between two Vironostika LS assays, and (4) the calibration of the BED incidence-assay for multiple HIV subtypes. The Serodia HIV-1/HIV-2 particle-agglutination (PA) assay was modified to act as an LS assay. At a dilution of 1:58,000, 134/152 (88%) of samples from recently infected persons were correctly identified by the PA assay. The study demonstrated the possibility of modifying an inexpensive, commercially available rapid test in resource poor countries to classify persons with recent HIV infection and for incidence surveillance. However, use of routinely collected sera for incidence estimation is dependent, in part, on the completeness of testing from all HIV positive persons. In a study conducted in the Caribbean, from 36 to 75% of all HIV-positive samples were not available for incidence testing. A model constructed by the Institute of Human Virology, Baltimore MD, suggests that incidence estimates are subject to a significant downward bias when > 50% of positive specimens are unavailable for incidence testing. To adjust for missing incidence test results, a simple weighting procedure was proposed that adjusts the denominator of the incidence estimator by the proportion of positive specimens available for testing (i.e., # EIA-negative specimens multiplied by the proportion of available positive specimens). The weighting mechanism is consistent with that in the CDC’s protocol for incidence calculations. When reporting assay-based incidence estimates, however, researchers should describe the number of HIV-positive specimens that were unavailable for incidence testing and the potential downward bias of reported estimates if many positive specimens were unavailable. Because the Vironostika Plus O (VPlusO) assay is expected to replace the Vironostika MicroElisa System in 2005, CDC evaluated the comparability of the less-sensitive versions of both assays (VPlusO-LS vs. V-LS) using 1009 specimens obtained through the HIVNET Infected Participant Cohort study. In this study, VPlusO was 90.2% concordant with V-LS (472 specimens classified as incident by VPlusO-LS vs. 459 specimens classified as incident by V-LS). Both assays yielded similar window periods for detecting recent infection (212 days for V-LS; 196 days for VPlusO). Finally, the BED capture EIA (BED-CEIA) incidence assay has been calibrated for several non-B HIV-1 subtypes using seroconversion panels obtained from Thailand, Kenya, Ethiopia, and Zimbabwe. These evaluations suggest small differences in the window period with different subtypes with overlapping confidence intervals, such that a consensus window period of 153 days based on an optical density cutoff of 0.8 can be used for incidence estimations. In three validation studies, BED-CEIA incidence estimates were comparable to observed incidence in 3 prospective US and international cohorts.
7. FDA Requirements for Use of Tests for Recent Infection
This session included discussion on FDA requirements for tests used for recent infection for research purposes and for public-health purposes other than surveillance. Research purposes, for example, would include assays used to identify persons eligible for the National Institutes of Health Acute Infection and Early Disease Research Program. Public-health purposes would include assays used to concentrate partner notification efforts on persons who were identified as recently infected. Results of assays used for these above purposes involve clinical management and reporting of results to individuals and providers, and thus are required to have an FDA investigational new drug (IND) or investigation device exemption (IDE) designation. The current BED-CEIA incidence assay manufactured by Calypte is not subject to IND/IDE requirements when used for population-based surveillance purposes. The current version of the Vironostika LS-EIA could be suitable for identifying individuals with acute infection, especially if lower cutoffs are used to avoid misclassification of individuals who are beyond the acute stage. Further validation would be needed to calibrate the Vironostika Plus-O assay for such a purpose. During the session, both Calypte Biomedical and bioMerieux indicated they would consider applying to the FDA for an IDE to permit these other uses of the assays.