HIV Antibody Assays Testing
HIV Antibody Assays
HIV InSite Knowledge Base Chapter
May 2006
Niel Constantine, PhD, University of Maryland School of Medicine
Introduction
An arsenal of laboratory methods is available to screen blood, diagnose infection, and monitor disease progression in individuals infected by HIV. These tests can be classified into those that: 1) detect antibody, 2) identify antigen, 3) detect or monitor viral nucleic acids, and 4) provide an estimate of T-lymphocyte numbers (cell phenotyping). The focus of this discussion is on antibody detection, the most widely used and, in most situations, most effective way to identify HIV infection.
General Considerations
Tests to detect antibody to HIV can be further classified as: 1) screening assays, which are designed to detect all infected individuals, or 2) confirmatory (supplemental) assays, which are designed to identify individuals who are not infected but who have reactive screening test results. Accordingly, screening tests possess a high degree of sensitivity, whereas confirmatory assays have a high specificity. Tests with high sensitivity produce few false-negative results, whereas tests with high specificity produce few false-positive results. These classes of assays, performed in tandem, produce results that are highly accurate, reliable, and appropriate to protect the blood supply or assist in the diagnosis of HIV infection. Technical errors do occur, however, and there are biologic factors that can limit the accuracy of HIV tests. Therefore, along with the testing process, there is the requirement for an extraordinary and dedicated quality assurance program.(1) Regardless of the results, because laboratory tests are not perfect, they are meant to be a supplement for clinical diagnosis.
Early Detection and the Window Period
Specific antibody to HIV is produced shortly after infection, but the exact time depends on several factors, including host and viral characteristics. Importantly, antibody may be present at low levels during early infection but not at the detection limit of some assays. Using the early-generation tests, antibody could be detected in most individuals by 6 to 12 weeks after infection. Newer-generation assays, including the third-generation antigen sandwich assays, can detect antibody at about 3-4 weeks after infection.(2) This window period before the detection of antibody can be shortened by several days using antigen tests, and by several more days using nucleic acid detection methods.(3) Therefore, in most individuals, the window period may be only 2-3 weeks if an all-inclusive testing strategy is used. Most antibody tests currently on the market have near perfect and equivalent degrees of sensitivity for detecting most individuals who are infected with HIV (epidemiologic sensitivity), but they vary in their ability to detect low levels of antibody (analytical sensitivity), such as those occurring before complete seroconversion.(2) Although tests are available to detect specific HIV immunoglobulin M (IgM) antibody, these tests have shown little utility in identifying early infection because IgM responses to HIV are not produced consistently during early infection.(4) The ability of some tests (eg, third-generation tests) to detect IgM antibody simultaneously with immunoglobulin G (IgG) detection, however, may be responsible for their higher analytical sensitivity.
Tests to Screen for HIV Infection
For the laboratory diagnosis of HIV, the mere presence of specific antibodies signals that infection has occurred. For the diagnosis to be correct, however, detection depends on the use of tests that are effective in identifying HIV antibodies, and not antibodies directed to other infectious agents that may be antigenically similar. Antigens used in HIV diagnostic tests must be appropriately specific, and usually are purified antigens from viral lysates, or antigens produced through recombinant or synthetic peptide technology. The use of such antigens allows HIV screening tests to possess both sensitivity (to detect infection) and specificity (to detect noninfection). In the United States, screening tests for HIV must be licensed by the Food and Drug Administration (FDA), regardless of whether they are used for screening blood, diagnosis, or monitoring disease.
Considerations in Choosing a Screening Test Methodology
Regardless of the particular screening test used, serum or plasma samples first are tested (screened) using a test with high sensitivity, most often an enzyme-linked immunosorbent assay (ELISA), "rapid test," or "simple method" (described below). ELISA is the screening method used most commonly, with the other 2 methodologies offering more rapid results with simple procedures applicable for use in point-of-care testing and in developing countries. With the advent of new therapies to treat HIV infection and the recommendation to institute therapy as soon as possible (but no later than 72 hours) after exposure,(5) rapid assays may be the most appropriate for testing the source patient after exposure (eg, needlestick injuries). More recently, tests have been developed using fluids that can be obtained conveniently outside the clinical laboratory. Whole blood from fingerstick and oral fluid (saliva) has been shown to be as effective as serum or plasma for detecting antibodies to HIV.
Reactive Results
Regardless of the screening method, a sample producing a reactive result must be screened again in duplicate, with at least 2 of the 3 results being repeatedly reactive before verifying infection with confirmatory assays. The most common reason for nonrepeatable results by screening tests is technical error.
Samples that produce repeatedly reactive results by screening tests must be further tested using confirmatory tests, or other confirmatory strategies (see below). Although screening tests are exquisitely sensitive, they lack an adequate degree of specificity. An example is their low predictive values when testing a population having a low prevalence of infection. When testing a population of 100 individuals, a test having a specificity of 99% can be expected to produce 1 false-positive result. If 1 individual in that same population is truly infected, the test will produce 2 positive results (1 from the infected individual, and 1 false positive). Therefore, if a positive result is produced when testing these 100 individuals, there is only a 50% chance that it represents an accurate result. Consequently, additional testing is required to differentiate between true- and false-positive results. A complete review of screening assays and a description of the use of test indexes has been published.(1)
HIV Screening Assays
Enzyme-Linked Immunosorbent Assays/Enzyme Immunoassays
ELISA is the most commonly used type of test to screen for HIV infection because of its relatively simple methodology, inherent high sensitivity, and suitability for testing large numbers of samples, particularly in blood testing centers. More than 40 different ELISA test kits are available, but only about 10 are licensed by the FDA for use in the United States.(1)
A common feature of all varieties of ELISA is the use of enzyme conjugates that bind to specific HIV antibody, and substrates/chromogens that produce color in a reaction catalyzed by the bound enzyme conjugate. The most popular ELISA involves an indirect method in which HIV antigen is attached to a well of a 96-well microtiter plate. Antibody in the sample is allowed to react with the antigen-coated solid support, usually for 30 minutes at 37º C or 40º C. After a wash step to remove unbound serum components, addition of a conjugate (an antihuman immunoglobulin with a bound enzyme) binds to the specific antibody that is attached to the antigens on the solid phase. Following another wash, addition of an appropriate substrate results in color development that is detected by a spectrophotometer and is proportional to specific HIV antibody concentration in the sample. Optical density (OD) values are produced as the colored solution absorbs transmitted light, and provide an indication of the amount of color, which is proportional to the amount of antibody bound (ie, antibody concentration). A mathematical calculation, usually based on the OD of the negative controls multiplied by a factor, produces a cutoff value on which the OD of the sample is compared to determine the antibody status; samples with OD cutoff values >1.0 (in an indirect ELISA) are considered antibody reactive (positive). Several indirect ELISA tests incorporate polyvalent conjugates (anti-IgG and anti-IgM) and antigen-sandwich configurations in order to increase sensitivity for detecting early infection (during seroconversion).
Alternate ELISA methodologies include a competitive format in which specific HIV antibody in the sample competes with an enzyme-bound antibody reagent for antigen sites on the solid phase. In this method, color development is inversely proportional to specific HIV antibody concentration.
A more recent addition to ELISA technology is the antigen sandwich method in which an enzyme (alkaline phosphatase or horseradish peroxidase) is conjugated to an HIV antigen (similar to the immobilized antigen on the solid phase). The antibody in the sample is "sandwiched" between 2 antigen molecules, 1 immobilized on the solid phase and 1 containing the enzyme. Subsequently, the addition of substrate results in color development in proportion to antibody concentration. The antigen sandwich ELISA is considered the most sensitive screening method, given its ability to detect all isotypes of antibody (including IgM).(2) One disadvantage of this method is the relatively large volume (150 µL) of sample required, which may make repeat testing and testing of samples from infants difficult.
Antibody Testing Strategies for Identifying Early HIV Infection and Estimating Incidence: Sensitive/Less-Sensitive ("Detuned") Assays
In contrast to the prevalence of HIV infection (ie, the number of persons infected), the incidence of HIV infection is defined as the change in prevalence of infection over time (ie, the number of new infections occurring). Incidence estimates most often are calculated by testing a cohort of individuals at 2 different time periods and observing the number of new infections. As is easily understood, this strategy is difficult due to the need to locate individuals for follow-up testing. However, incidence estimates are important, not only for identifying specific populations where educational endeavors can have the most benefit or where changes in infection patterns are occurring, but also for targeting these populations for therapeutic intervention or vaccines.
Detection of individuals in early infection provides several benefits. Identifying infections within the previous 4 months facilitates tracking of intravenous drug and/or sexual contacts, as only contacts in a defined, recent time period require tracing. Further, because the high viral loads of early infection are associated with increased transmission risk, identification of high-incidence populations may assist in effectively targeting prevention interventions. Monitoring areas of high incidence of HIV infection has clinical and therapeutic implications for neonatal diagnosis and for the early initiation of antiretroviral treatment, and also can provide information for prognosis, identify communities most likely to benefit from preventive vaccines, and assist in the enrollment of recently infected individuals in studies of pathophysiology or pharmacotherapy.
Laboratory-based strategies that can distinguish recently infected individuals from those with established infection have been devised. In these methods, the procedures of conventional ELISA or rapid assays have been modified to allow discrimination of antibody titer or antibody avidity. These modified assays have been called "detuned" assays or "sensitive/less-sensitive" (S/LS) assays.
During acute HIV infection, prior to the appearance of antibody (window period or pre-seroconversion), HIV infection can be confirmed only by the demonstration of circulating p24 antigen, or by the presence of viral RNA or DNA. Although highly sensitive antibody assays exist to detect very low levels of HIV antibody in blood, the window period prior to appearance of antibody rarely can be shortened to less than 3 weeks. Once antibody has appeared, titers progressively increase during 3-5 months until levels peak, at which time they remain fairly constant throughout the remainder of infection. Also, antibodies during early infection usually are of low avidity, but avidity increases as infection progresses. Therefore, HIV infection can be divided into categories of recent or established infection, depending on the quantity of antibody present or their avidities. These parameters can be exploited as tools in order to estimate the relative time that HIV infection occurred. For example, if antibody titers or antibody avidity is low, it is likely that infection occurred within the past 4 months; conversely, high-titer or high-avidity antibodies signal an established infection that has been present for longer than 4 months. Several epidemiological studies have used the S/LS testing strategy to predict incidence in San Francisco and in Rio de Janeiro, Brazil.(6-8)
The first S/LS strategy, then, is based on the principle that antibody titer increases with time and that recent infection can be assumed if test results become nonreactive following dilution of the individual's serum. In such a case, an initially reactive sample, when tested with the routine (sensitive), becomes nonreactive when diluted in a modified (less-sensitive) assay. Conversely, the serum of an individual with established HIV infection would remain reactive following dilution in the less-sensitive assay due to high levels of antibody. This strategy is used only on individuals who are confirmed positive using the Centers for Disease Control and Prevention (CDC) interpretive criteria via Western blot, as persons who are negative for antibody would not be candidates for determining the time of infection. This system, also known as the Serologic Testing Algorithm for Determining Recent HIV Seroconversion (STARHS), has been developed and adopted by the CDC. The initial assay system that was modified by dilution and validated using persons of known seroconversion or infection times was the FDA-licensed, first-generation Abbott HIV-1 Viral Lysate ELISA (3A11).(9) Specific modifications in the procedure of the 3A11 ELISA were made to 4 parameters in order to decrease the sensitivity (for the less-sensitive assay). To construct the less-sensitive test, the sample dilution was increased to 1:20,000, the sample incubation time was reduced to 30 minutes, the conjugate incubation time was reduced to 30 minutes, and the OD cutoff value was adjusted. In order to compare results obtained with the less-sensitive assay, OD readings for individual samples are standardized by calculating standardized OD (SOD) values based on the formula: SOD = (sample OD value - negative control OD value)/positive control OD value). A cutoff SOD (0.75) has been determined statistically and nonreactive samples have an SOD less than the cutoff. When such a sample shows this reversion by the S/LS test, the time interval from seroconversion was calculated to be 129 days or about 4 months (95% confidence interval: 109-149 days). Several studies have validated the S/LS algorithm by analyzing individuals with known early infection as determined by clinical evaluation, recent seroconversion, high-risk behavior, and antigen and nucleic acid analyses. A limitation of the S/LS test strategy may be the detection of individuals with long-standing infection (0.4%) and late-stage AIDS (2%). Thus, CD4 cell counts and clinical information may be required to support results obtained by the S/LS test algorithm for maximum accuracy. The S/LS strategy is inexpensive, reproducible, and can give a fairly accurate estimate of the time of infection. Because the 3A11 ELISA no longer is available, other test kits (Vironostika, bioMérieux) have been substituted, and have been considered to be equivalent in performance. More recently, another quantitative ELISA method has been introduced, and reportedly performs effectively with samples from persons who are infected with non-B HIV clades. This assay, the BED assay (Calypte; Lake Oswego, OR), incorporates synthetic peptide antigens and can classify infections for clades B, E, C, and A/D.(10)
The second method to identify the time of infection for incidence estimation is based on antibody avidity and has been developed using a third-generation ELISA. This method is known as the Avidity Index Protocol. Avidity describes the collective interactions between antibodies and a multivalent antigen. Avidity measurements are used with a variety of infectious diseases to offer confirmatory evidence of acute infection, to distinguish reactivation from primary infections, and to permit diagnosis of acute infection from a single sample. An individual's differential binding or avidity index (AI) correlates with the estimated length of time from the initial infection by HIV. Thus, the strength of the interaction between antigen and the antibody present in early infection is weak because low-avidity HIV-1 antibody comprises the majority of antibodies found in early infection. The relative avidity of antibody is stronger in established infection and can be estimated serologically based on resistance of the antigen-antibody complex to chaotropic agents. Chaotropic agents are dissociating reagents such as urea (at concentration of 4, 6, and 8 M), potassium thiocyanate (KSCN; 1-3 M), magnesium chloride (2 and 4 M), diethylamine (0.025, 0.05, and 0.1 M), and guanidine HCl (3 and 6 M).
The most widely recognized AI test is a recombinant viral lysate enzyme immunoassay (EIA) from Bio-Rad Laboratories (Hercules, CA), modified by the incorporation of a dissociation and wash step. The chaotropic agent that demonstrated the ability to dissociate low avidity HIV antibody molecules most effectively was 2.5 M KSCN. Procedurally, duplicate wells of a diluted sample are incubated with HIV antigen. Antibodies to HIV bind to the antigen, and following a wash step, a solution of dissociating reagent is added to one of the wells (test) while wash solution is added to the other well (control). Results are interpreted based on a calculation of the AI from a percentage of the ratio of the OD of the KSCN-treated specimen to that of the nontreated control. Samples demonstrating an AI <80% are taken to represent early infection and are associated with the 3- to 4-month (120-day) time period following seroconversion. This method has been validated with samples from seroconversion panels and samples from individuals with clinically established HIV infection.
Our laboratory has developed a rapid S/LS method using the Uni-Gold HIV test (Trinity Biotech; Wicklow, Ireland), a 10-minute, visually read, rapid test. This method, based on a dilution of serum for the LS mode, has shown excellent results in comparison with the Abbott 3A11 assay and when assessed using samples from individuals with known seroconversion dates. In addition, we obtained preliminary results using an HIV saliva test, SalivaCard (Trinity Biotech), that shows utility as an S/LS tool.(11) More recently, we have developed a simple and low-cost particle agglutination assay as an S/LS assay and shown it to be 97% accurate (unpublished observation). The advantage of rapid and simple S/LS assays is that they are portable and can be used to identify high-incidence populations in remote areas where ELISA instrumentation cannot be supported. Further, even in developed countries, they can be adapted easily for use in mobile testing centers to identify recently infected individuals so that they can be counseled appropriately to find contact persons within the past several months or to immediately direct individuals to appropriate treatment centers. Finally, the noninvasiveness of saliva-based rapid assays may increase testing participation.
Fourth-Generation Assays for the Simultaneous Detection of HIV Antigen and Antibody
Antibody can be detected in a majority of individuals within 6-12 weeks after infection using the earlier generations of assays, but may be detected within 3-4 weeks when using the newer third-generation antigen sandwich assays.(2) The window period can be shortened to about 2 weeks using p24 antigen assays or reduced to 1 week with the implementation of nucleic acid detection assays.(12) Consequently, the window period between infection and detection of infection may be <2 weeks if a comprehensive testing approach is utilized. The detection of p24 antigen by ELISA is a simple cost-effective technique to demonstrate viral capsid (core) p24 protein in blood during acute infection due to the initial burst of virus replication after infection. In order to maximize the detection of all infected individuals, including those in early infection, antibody, antigen, and viral RNA tests should be used. However, viral RNA tests are expensive, time consuming, and unavailable in many laboratories. Laboratories that possess ELISA capability can increase the ability to detect most infections by testing for both HIV antibody and p24 antigen. During the late 1990s, assays in an ELISA format that have the capability to detect both HIV antibody and HIV p24 antigen simultaneously were developed, thereby eliminating the need to perform separate assays.
The new generation of combination ELISAs that simultaneously detect both antigen and antibody has been developed and marketed, and offers advantages for decreasing the time, personnel, and costs necessary to perform each assay individually. These assays have demonstrated a high analytical sensitivity of detection that is most likely attributed to the combination of a third-generation format (antigen sandwich) for antibody detection and the ability to simultaneously detect HIV p24 antigen. To date, there are 8 commercial, combination antibody and antigen assays that have been developed and evaluated.(13-31) These fourth-generation assays include the VIDAS HIV DUO Ultra (bioMérieux; Marcy l'Etoile, France), Enzymun-Test-HIV-Combi (Boehringer; Mannheim, Germany), Vironostika HIV Uni-Form II Ag/AB (Organon Teknika; Boxtel, Netherlands), AxSYM-HIV Ag/AB (Abbott Laboratories; Abbott Park, IL), Enzygnost HIV Integral (Dade Behring; Marburg, Germany), Genescreen Plus HIV Ag-AB (Bio-Rad), and COBAS Core HIV Combi (Roche Diagnostics; Mannheim, Germany). The eighth assay is an 18-minute, double-antigen sandwich combination assay called the Elecsys-HIV Combi (Boehringer) that has been reported to have a specificity of 99.8% when challenged with a cohort of hospitalized patients.(16) This rapid assay is based on electrochemiluminescence and is reported to reduce the window period by 5 days over antibody tests. A ninth, unidentified assay is a lineal immunoenzymatic assay evaluated to have a sensitivity of only 99.5% and a specificity of 94.8%.
The benefits of testing for both antibody and antigen are justifiable due to the need to identify individuals with both established and early HIV infection not only for the blood donor population but also for some clinical applications. Early detection of infection via antigen testing promotes the prompt referral of infected individuals for the initiation of treatment, counseling, and prevention interventions to reduce the risk of transmission. Due to their ability to detect p24 antigen, the fourth-generation ELISAs will be of value in detecting early infection. These assays are highly applicable for the diagnosis of early and established HIV infection by hospital and private clinical laboratories and other laboratory settings. In these settings, individuals to be screened for infection are of higher risk groups than the blood donor population, and thus require the use of testing methodologies with high levels of analytical sensitivity to detect primary infection. Of significance, the high level of analytical and epidemiological sensitivity demonstrated by most of the fourth-generation assays with seroconversion and clade panels, as well as a variety of patient populations, makes them ideal for use in a variety of testing situations for the diagnosis of early and established infection. In routine laboratory settings, HIV-infected samples that are identified via antigen detection would not have been identified by the usual screening antibody assays, because antigen testing of patients is not performed commonly as a screening tool outside blood banks. The detection of early infection has been shown to be beneficial for the prompt initiation of appropriate antiretroviral therapy in a clinically relevant time frame. Additionally, early detection will help in the timely implementation of interventions such as the counseling of patients, prevention of transmission, and management of infection.
Rapid Tests
Rapid assays for detecting specific HIV antibody were developed in the late 1980s, and are defined as tests that can yield results in <30 minutes. These tests gained popularity in the early 1990s, and as technology became refined, proved to be as accurate as the ELISA when performed carefully by experienced personnel. Technical errors are common with these assays, however, because users become careless with these simple procedures. For example, pipettes are not always held in a vertical position as recommended, resulting in an incorrect delivery of reagent volumes. In addition, many laboratory workers attempt to test multiple samples simultaneously, resulting in inaccuracies in the timing of steps.
When performed correctly, rapid HIV assays are accurate and have wide utility in a number of testing situations. Application includes emergency rooms, physicians' offices, point-of-care testing, autopsy rooms, funeral homes, small blood banks, and situations involving stat HIV testing (where immediate treatment is recommended for exposures). Rapid HIV assays have proven particularly useful for testing pregnant women in labor who have not received prenatal care (ie, of unknown HIV status). It has been shown that the institution of antiretroviral therapy (eg, zidovudine) is effective in reducing transmission of HIV, and that this should be provided as early as possible to the mother and subsequently to the newborn. Rapid HIV testing of the mother who is near delivery allows treatment to be initiated prior to delivery if a positive serostatus is determined.(32) Importantly, these rapid assays are easy to perform and have utility in developing countries, where facilities may not be optimal, stable electricity may be unavailable, and formal education programs for laboratorians are absent.
One class of rapid tests is the "dot blot" or "immunoblot"; they produce a well-circumscribed colored dot on the solid phase surface if the test is positive. Most of these rapid assays now incorporate a built-in control to indicate that the test was performed correctly. This control is an antihuman immunoglobulin that binds any immunoglobulin in the sample and produces a separate indicator when all reagents are added appropriately. In addition, several varieties are available that include 2 "dots," which allows the differentiation of HIV-1 and HIV-2 infection. The procedures for the dot-blot assays are similar regardless of the exact format of the test. Most require drop-wise additions of reagents in the following sequence: buffer, sample, wash buffer, conjugate, wash buffer, substrate, and stop solution. Some assays substitute an IgG binding dye (protein A gold reagent) for the antiimmunoglobulin conjugate, thereby decreasing the procedure by a step.
The newer 1-step rapid assays, also known as immunochromatographic assays, are convenient, self-contained tools for HIV serologic testing, consisting of a flat cartridge device, usually plastic or paper. Whole blood, oral fluid, or serum is placed at the tip of the device and allowed to diffuse along a strip that is impregnated with reagents (often protein A colloidal gold) that bind and permit visual detection of HIV antibodies; some use third-generation (antigen sandwich) technology. These tests can be completed in <10 minutes (some within 2 minutes), require little or no addition of reagents, and contain a built-in quality-control reagent to control for technical errors. Some tests can be stored at a wide range of temperatures (from 15º C to 30º C), and are transported easily. For example, one type (Determine; Abbott) comes in "cards" of 10 tests each, making it possible to carry 100 tests in a shirt pocket; the cards require no reagents, just addition of serum or plasma. The test can also be performed on whole blood, or blood collected via fingerstick (this requires 1 buffer addition). These types of rapid HIV tests are gaining in popularity because of their simplicity, ease of interpretation, and robustness.(33,34) In particular, the fingerstick collection method is taught easily to health care personnel in outreach situations or mobile vans. The use of fingerstick specimens also may prevent unnecessary collection and discarding of full units of donated blood (where blood is collected prior to testing at a remote laboratory and held until results become available). Another variety of lateral flow devices allows for the use of saliva, plasma, whole blood, or fingerstick specimens, thereby adding flexibility in sample type (see "Alternatives to Classic Tests and Testing Strategies" below).
Other rapid test formats include dipsticks, in which antigen is attached on the "teeth" of comblike devices; several of these rapid tests have the ability to differentiate HIV-1 and HIV-2. Disadvantages include a subjective interpretation, difficulty in reading if the laboratorian is color-blind, and a higher cost than that of the ELISA. Currently, 4 rapid HIV tests are approved for use in the United States.
Simple Tests
This type of HIV test requires longer than 30 minutes for results, but consists of procedures that can be performed easily without instrumentation. Within this class of tests are agglutination assays in which antigen-coated particles (red blood cells, latex particles, or gelatin particles) are allowed to react with serum antibodies to form visible clumping (agglutination). If red blood cells are used, the technique is termed passive hemagglutination; with the use of latex particles, it is known as latex agglutination. In East Asia, an HIV gelatin particle agglutination test is popular, offering good sensitivity, low cost, and ease of performance. It incorporates a quality control system to detect nonspecific antibodies directed toward the gelatin particles themselves, and results can be obtained within 2 hours with minimal hands-on time. Although appropriate for use in facilities with limited testing capabilities, this test must be performed under temperature-controlled conditions.
Tests to Confirm HIV Infection
Most testing algorithms require the use of very specific assays, such as the Western blot, indirect fluorescent antibody (IFA) assay, or the radioimmunoprecipitation assay (RIPA), to verify reactive screening test results. If performed and interpreted correctly, these extremely specific tests should not produce biologic false-positive results. They are, however, more laborious and more expensive than screening assays.
These confirmatory tests do not have to be FDA licensed in the United States when used for purposes other than testing blood donors. For blood donors, a licensed confirmatory test is used for purposes of donor reentry, for which the results must be negative. The primary purpose of confirmatory tests is to ensure that uninfected individuals who test reactive by screening assays are not identified incorrectly as being HIV infected.
Western Blot Test
Methodology
The Western blot probably is the most widely accepted confirmatory assay for the detection of antibodies to the retroviruses. Most authorities consider it the gold standard for validation of HIV results. It is based on using an electrophoretic technique to separate HIV antigens derived from a lysate of virus grown in culture. This technique denatures the viral components, imparts a negative charge to the antigens, and separates them primarily on the basis of their molecular weights. The separation of antigens in the technique allows for the identification of specific antibodies to each of the viral antigens in a subsequent set of steps similar to the ELISA methodology.
A purified HIV antigen mixture is layered onto a sodium dodecyl sulphate (SDS) polyacrylamide gel slab and then electrophoresed. The viral proteins (HIV antigens) migrate through the molecular pores of the gel at rates determined by electrical charge and molecular weight. The proteins with higher molecular weight migrate less and form bands closer to the starting point. The proteins on the gel are then transferred ("blotted") to nitrocellulose paper by another electrophoretic procedure. This paper is cut into thin strips, each with the full distribution of viral protein antigen bands. A single test strip is incubated with a 1:50 or 1:100 dilution of a test sample or a control and then washed and incubated with a labeled (tagged) antihuman globulin. At this point, the procedure is similar to any other indirect immunoassay. The label usually is an enzyme (horseradish peroxidase or alkaline phosphatase) that will react with a specific colorless substrate to produce an insoluble colored band on the strip wherever there is an antigen-antibody complex. Reaction with a positive serum sample produces a pattern of bands on the strip that is characteristic of HIV. Many of these bands have been identified as specific viral gene products.
The HIV-1 viral antigens are separated as follows (from top to bottom): gp160, gp120, p66, p55, p51, gp41, p31, p24, p17, and p15 (Figure 1). The "gp" designation refers to glycoproteins; "p" indicates proteins. The numeric values (x100) indicate molecular weights. It is important to remember that nonviral proteins derived from the host cells in which the virus was grown also are present on the nitrocellulose strip. They can form bands in many places, but often are near the middle molecular weight (40,000 to 60,000) region. These nonviral protein bands may produce difficulty in interpretation of results by producing nonspecific reactions.
Interpretation of Results
Depending on the particular antibodies in the sample, reactivities with the separated antigenic components result in band profiles. The type of profile (the combination and intensity of bands that are present) determines whether the individual is considered positive for antibodies to HIV. The classification of Western blot results is determined by certain criteria. Most institutions now follow the CDC guidelines, which require reactivity to at least 2 of the following antigens: p24, gp41, gp120/160 for a positive classification. It is now universally accepted that a negative result is the absence of all bands. Two organizations, however, including the World Health Organization (WHO), suggest that results also can be reported as negative if there is only a very weak p17 band. Indeterminate classifications occur when there is reactivity to 1 or more antigens, but not fulfilling the criteria for positivity. Figure 1 depicts examples of positive, negative, and indeterminate Western blot results.
Unfortunately, sera from some noninfected individuals show some reactivity to 1 or more antigens if tested by Western blot. This reactivity may occur in as many as 15% of normal noninfected persons, and many times occurs in persons who are nonreactive by screening assays. Therefore, if ELISA-nonreactive sera are tested by Western blot, many will result in an indeterminate profile. Most indeterminate results show only weak reactions to the Gag proteins (mostly p17, p24 and/or p55); other patterns occur but are less frequent. Any Western blot reactivity that does not meet the requirements for being positive or negative must be considered indeterminate.
Some individuals who exhibit indeterminate results (eg, reactivity to p24 and p55) later seroconvert, demonstrating that a p24 and p55 profile can indicate early infection. Conversely, other individuals may have the identical profile for long periods of time (years) and never seroconvert (ie, they are not infected). In fact, most indeterminate Western blot results from noninfected individuals exhibit the p24 and/or p55 profile. Therefore, an indeterminate Western blot result cannot predict early infection.
Most authorities suggest that persons with indeterminate results should be retested after several months, although seroconversion may be detected in a shorter period of time. If at all possible, the retesting of an individual at a later time should be performed in parallel with reassay of the initial sample on the same run with the same kit lot numbers and the same assay conditions to ensure that the samples can be compared directly. The WHO recommends retesting persons after 2 weeks if highly suggestive Western blot profiles are produced, although other organizations suggest waiting 1-6 months before retesting. If an individual is retested over a period of 6 months and becomes negative or the band profiles do not progress, infection with HIV generally can be ruled out. For poorly understood reasons, many individuals continue to exhibit indeterminate results for years but are not infected. If an individual does progress serologically (more bands or greater intensity of bands) or converts to positive (seroconversion) during retesting, the individual probably was infected at the time of the first test (early infection). It should be noted that individuals who have received vaccination for HIV (eg, subunit gp160) may be misidentified as positive based on reactions to the envelope antigens alone.
The significance of an indeterminate Western blot result varies depending on the risk factors, clinical status of the patient, and the Western blot profile produced. For example, individuals with a history of high-risk behavior are more likely to be the ones who later seroconvert, because the chances of their being infected are high. In addition, some Western blot profiles are more suggestive of early infection (eg, p24, p31, and p55) than are others (eg, p17 only). Many initially indeterminate results that subsequently become negative or remain indeterminate probably are a result of nonspecific reactions, hypergammaglobulinemia, the presence of cross-reactive antibodies, infection by HIV-2, or infection by an unknown, but related retrovirus. There have been a few reports where autoimmune diseases (eg, systemic lupus erythematosus) can cause false-positive HIV tests, including Western blot.(35) Also, it is known that some individuals with AIDS may lose reactivity to p24, and perhaps other antibodies, later in disease, so that even AIDS patients may have indeterminate Western blot results by some criteria. Ancillary tests, such as polymerase chain reaction (PCR) and viral culture may be helpful in resolving these indeterminate results if the diagnosis is in question.
Indirect Immunofluorescent Antibody Assay
In this technique, cells (usually lymphocytes) are infected with HIV and are fixed to a microscope slide. Serum containing HIV antibodies is added and reacts with the intracellular HIV. The slide is washed and then allowed to react with antiimmunoglobulin antibodies with a covalently bound fluorescence label attached. The reaction is visualized using a fluorescent microscope. This technique has the advantage of sometimes providing definitive diagnosis of samples that have yielded indeterminate results by Western blot analysis. Disadvantages to its use include the requirement of an expensive microscope and a subjective interpretation, thus necessitating well-trained individuals.
Modified Western Blot
Western blot assays that have the ability to identify and differentiate infections by HIV-1 and HIV-2 have been developed. Most incorporate the use of viral lysates from HIV-1 and synthetic peptides artificially applied from HIV-2 on the same nitrocellulose strip (a modified or augmented Western blot). In this case, multiple HIV-1 antigens and 1 HIV-2-specific band (gp36 or gp41) are present on the strip. Criteria established by manufacturers include reactions to 1 gene product from each of the 3 major groups (Gag, Pol, and Env) for positivity for HIV-1. To be considered positive for HIV-2, the test must show reactions to the HIV-2-specific antigen plus a reaction to HIV-1-specific antigens, which alone do not meet the criteria for positivity for HIV-1.
Line Immunoassay
Another alternative to the classic Western blot and IFA confirmatory tests is the line immunoassay (LIA). In this assay, recombinant or synthetic peptide antigens are applied on a nitrocellulose strip, rather than electrophoresed as in the Western blot. This use of "artificial" antigens decreases the presence of contaminating substances derived from cell culture that can cause interference and sometimes false reactions. The use of LIA is popular in Europe, but these tests have not been licensed for use in the United States. A number of reports have verified that the accuracy is equivalent to the Western blot.
HIV-2 Tests
HIV-2 is endemic primarily in areas of West Africa, but the increased prevalence and distribution of HIV-2 infections necessitate the use of tests that can detect HIV-2 infection. In the United States, about 80 cases of HIV-2 have been verified, most of which have been linked to West Africa. Biologically, HIV-2 is very similar to HIV-1: the HIV-2 genome exhibits about 60% homology in the more conserved gag and pol genes, and 30 to 40% homology in the other viral genes and long terminal repeat (LTR) sequences. The antigens of HIV-2 are similar to those of HIV-1, but the molecular weights may vary slightly. For example: the Gag proteins of HIV-2 have designations of p56, p26, and p16; the Pol proteins p68 and p34; and the envelope glycoproteins gp36 (or gp41), gp140, and gp105. As with HIV-1 screening tests, a variety of test formats are available to detect antibodies to HIV-2, including ELISA beads, ELISA microtiter, and rapid/simple assays.
Diagnostically, HIV-2 infections can present problems. Screening tests designed to detect infection by HIV-1 do not always detect infection by HIV-2 and vice versa. Most cross reactions represent antibody induced by the core (p26) and/or Pol antigens (p68, p34), because these are highly conserved between the two different viruses. A lack of reactivity with heterologous viruses, however, dictates the need for an extra measure of vigilance to identify infections that might not be readily apparent using some HIV-1 assays. By HIV-1 ELISA, the OD readings of HIV-2-positive specimens may be high negative; by Western blot, the results may be indeterminate. Therefore, it is important to recognize slightly high negative readings and suggestive indeterminate results by HIV-1 tests, and consider evaluating the serum using HIV-2 tests.
To address this issue, commercially available HIV-1/2 "combination tests," which incorporate antigens from both viruses, can be used to screen sera in an attempt to identify either infection. The subsequent differentiation of HIV-1 and HIV-2 infections necessitates the use of highly specific ELISA (eg, synthetic peptide-based), Western blot, radio-immunoprecipitation assays, or PCR.
In late 1991, the FDA licensed the first combination HIV-1/HIV-2 screening test and recommended that blood banks start screening for HIV-2 by mid-1992. Blood banks in the United States can use either the licensed HIV-2 ELISA screening test together with the HIV-1 ELISA, or one of the licensed HIV-1/2 combination tests. Samples that test positive by the combination test are tested by an HIV-1 Western blot. If the result is negative or indeterminate by this HIV-1 Western blot, 1 or more specific HIV-2 tests are used to further analyze the sample. Combination tests are considered to be equivalent to their predecessors in terms of sensitivity (ie, near perfect).
HIV-2 confirmatory tests include the Western blot and the RIPA. In addition, EIA tests and some rapid tests that use chemically synthesized peptides corresponding to a unique immunogenic region within the respective transmembrane glycoproteins exhibit good correlation with the Western blot and the RIPA for identifying and differentiating HIV-1 and HIV-2 antibodies. Furthermore, these tests are valuable for differentiating samples that produce reactions to both viruses (dual reactors).
For HIV-2 confirmation, most organizations that have created criteria for positive HIV-2 Western blot agree on the necessity for reactivity to the envelope antigens. The WHO requires reactivity to at least 2 HIV-2 envelope antigens, whereas other organizations require reactivity to p26 (Gag) and to gp34 or gp105 (Env). If a specimen is tested by both HIV-1 and HIV-2 Western blot, the blot exhibiting the strongest reactivity to envelope antigens usually indicates which infection is present.
Alternatives to Classic Tests and Testing Strategies
As technology evolves, alternatives to the classic tests and testing strategies arise. Each offers 1 or more attractive features that may simplify collection, testing, or interpretation of results.
Oral Fluid ("Saliva") HIV Tests
Noninvasively collected specimens, such as oral fluids, have been used for HIV testing as a more convenient alternative to blood samples. Although generally referred to as "saliva," the fluid used for testing is actually crevicular fluid from capillaries beneath the tooth-gum margin, which is a transudate of blood and therefore similar to the samples used in serum-based tests. The concentration of antibodies in oral fluids is about 1/400 of that in plasma, however, because of the dilutional effect of fluids from the salivary glands (true saliva),(36) necessitating extremely sensitive tests that are able to detect small quantities of antibody. The testing technology to detect these low quantities is now available, and oral fluid tests, both ELISA and rapid tests, are accurate.(37,38)
A complete review of oral fluid testing for HIV has been published.(36) The use of oral fluids for testing offers advantages, such as ease of collection, group collections, collection from persons in whom blood is difficult to obtain, and an increase in collection adherence.(1,36)
Rapid oral HIV tests were introduced in the mid-1990s.(37) As with ELISA, the sensitivity of these tests to detect HIV in oral fluid needed to be boosted because of the low level of antibody in oral fluid, which was compounded by the dilutional effect of pure saliva.(36) In 2004, a rapid HIV test was licensed by the FDA for use with oral fluid. This test, the OraQuick Advance (OraSure Technologies; Bethlehem, PA), is a combination collection and testing device. Consisting of an absorbent (porous) pad on a stick coupled to a lateral flow testing device, it is swabbed once around the gums, and then placed in a vial of buffer solution. Following a 20-minute incubation, the results are read like other lateral flow rapid tests (a control line is included also). The manufacturer claims 100% sensitivity and specificity equivalent to that of ELISA HIV tests. Therefore, a positive result must be considered preliminary until confirmed by a more specific test, such as Western blot. This device also can be used for testing plasma, whole venipuncture blood, or blood collected via fingerstick, thereby giving flexibility for different testing situations. As of March 2006, rapid oral HIV testing is approved for use only by clinical laboratories and Clinical Laboratory Improvement Amendments-waived laboratories, but licensing for home use remains under consideration.
Urine Tests
Intact IgG antibodies are found in urine, but their exact origin is unknown. The collection of urine is simple, noninvasive, and inexpensive, and the sample can be stored at room temperature for extended periods of time. The use of urine for testing is appropriate for physicians' offices, health clinics, and in developing countries where health care personnel may not be trained professionally or where clean needles for drawing blood may not be available. The major disadvantage is that there is not an approved urine-based confirmatory assay, necessitating the collection of blood when results are reactive. The FDA has approved an ELISA and Western blot for use to test urine for antibodies to HIV-1.
Although urine testing for HIV has not gained in popularity as much as would be expected, companies are interested in modifying their serum-based rapid assays to offer rapid tests that can use urine samples. The market for these tests is the same as that for rapid serum tests, ie, occupational exposure cases, pregnant women without known HIV status, and public health clinics that provide results during the initial visit (to prevent loss to follow-up when patients do not return for their results). Although it would seem that serum-based tests could be modified easily to accept urine samples, this is not the case. There are a number of factors that influence rapid tests differently from the way they influence ELISA-type tests. For example, because urine is much less viscous and contains less protein than serum, flow rates through these rapid devices are increased dramatically. Consequently, this leaves less time for antigen-antibody reactions to occur. Also, the variability in the pH of urine appears to affect reaction time (since antigen-antibody reactions are pH dependent); the pH of urine varies considerably from individual to individual. However, our laboratory has been successful in modifying one manufacturer's serum-based test (only 1 of 6 manufacturers' tests could be modified successfully). Nevertheless, this shows proof of principle that rapid urine tests can be developed.
Home Collection for Testing
As of this writing, home collection, but not home testing, is approved by the FDA. These collection devices are filter paper for the collection of whole blood via fingerstick. The samples are mailed to a laboratory, eluted, and screened with ELISA tests. Results and counseling are made available by telephone. More recently, the FDA is considering the use of over-the-counter (OTC) rapid tests, particularly oral fluid tests for home use, in order to increase the prevalence of HIV testing. However, how to address needs for HIV test counseling (which traditionally includes discussion of risk reduction and education on the implications of test results) in the setting of home testing is unclear.
Alternative Confirmatory Strategies Using Screening Tests
In most industrialized countries, confirmation of HIV infection is accomplished using Western blot or IFA technologies. In developing countries, these assays may be available in reference laboratories, but it is common to find alternative confirmatory strategies for cost savings because funds to purchase expensive confirmatory tests or equipment may be unavailable. Several investigators have verified that similar predictive values can be obtained by using 2 screening assays in tandem. This method can result in up to 80% cost savings.(39) It is important to select appropriate tests, with the most sensitive tests used in the initial testing. These strategies recommend initial screening using ELISA or a rapid/simple assay, followed by a second ELISA or rapid/simple assay; the initial and second tests must be of different principle (bead vs microtiter) and/or use a different antigen source (lysate vs recombinant or synthetic peptide).
Another recent advance that makes use of prior technology, but in a novel format, includes a rapid confirmatory assay that incorporates several different HIV antigens on 1 rapid test device (similar to combination HIV-1 and HIV-2 rapid tests). These rapid, flow-through tests are performed in an identical manner to rapid screening testing (addition of several reagents in drop-wise fashion) and produce "reaction profiles" similar to those of the Western blot test and LIA. A thorough evaluation of one of these rapid confirmatory tests has produced excellent results.(40) Several companies are introducing these assays to address the issue of expensive and cumbersome Western blot confirmatory assays and the associated need for significant laboratory infrastructure.
HIV Diagnostic Dilemmas
The major difficulties with the laboratory diagnosis of HIV infections include: 1) indeterminate Western blot results; 2) minimally reactive confirmatory test results from noninfected individuals; 3) inconsistent results when repeating specimens or testing follow-up specimens; 4) the occurrence of technical errors; 5) false-negative results due to HIV Group O viruses; and 6) laboratory diagnosis of HIV infection in the newborn.
Indeterminate Western Blot Results
In reference to samples that show inconclusive results (eg, indeterminate Western blot results), a follow-up specimen in 1-3 months is the most effective means for resolution. At this interval of time, serum from almost all individuals who are infected will show an increase in reactivity by serologic assays (eg, an increase in the OD by ELISA or more bands by Western blot) or will seroconvert. It is important to test both samples on the same run to obtain a clear indication of changes in reactivity (ie, to ensure that intertest variations do not contribute to small differences in reactivity). Alternatively, IFA, PCR, viral culture, or antigen assays may be helpful.
Minimally Reactive Western Blot Results
These results occur occasionally, perhaps due to early infection (seroconversion) when antibody levels have not yet peaked, and on rare occasions for unknown reasons in individuals who are later found not to be infected with HIV. In the latter case, reactions to p24 usually are noted, as are weak reactions to gp41 or gp 120/160. In these cases, it is important to note on the report form that "on rare occasions, this profile has been found in persons who are not infected, and submission of a new specimen in several weeks is recommended."
Inconsistent Results
Inconsistent results when repeating specimens or testing new specimens from the same individual are rare, but real occurrences. Explanations include mislabeling of specimens, technical errors in the laboratory, the use of different test systems, or problems with components of the test system. If an individual is seroconverting, repeat testing by the same assay on the same specimen can produce results that fluctuate around the cutoff value. Alternatively, wide variations in values usually are a sign of technical error and should be investigated thoroughly through quality assurance monitoring. Inconsistencies with follow-up specimens can be due to seroprogression in truly infected individuals, seroreversion in persons who are not infected, or mislabeling or technical errors.
Technical Errors
Technical errors do occur, and although they cannot be eliminated totally, they can be minimized through the institution of a thorough quality assurance program and documented preventive measures. Clerical errors are the most common, and can be addressed effectively through a dedicated supervisory review mechanism. Several essential quality assurance measures are outlined subsequently.
False-Negative Results for HIV Group O
False-negative results by HIV serologic assays have been verified when testing some individuals infected by HIV Group O viruses.(41) This group of viruses, found primarily in Cameroon and Gabon, also has been reported in Europe and the United States.(42) Several "acceptable" routine HIV screening assays have been documented to produce false-negative results in up to 20% of sera from individuals infected with Group O viruses.(43) Although it is difficult to recommend measures to prevent this misdiagnosis, manufacturers of test kits are addressing this problem by incorporating antigens from Group O viruses.(44) Health care providers can be vigilant by inquiring as to the geographic origin of persons tested, or their contact with persons from these areas of Africa. The same is true for HIV-2 infections, when HIV-1-only assays are used (see above).
Diagnosis in the Newborn
The laboratory diagnosis of HIV in the neonate has been difficult since the first tests were developed, principally because of the omnipresence of maternal antibody up until 1 year after birth, at which time the infant may serorevert. Subsequently, it may be several more months until the infected infant's immune system is competent enough to produce antibody (seroconversion). Antigen assays can be of help, as can PCR, to detect HIV DNA or RNA in the infant. At present, however, definitive diagnosis in the newborn is still difficult, particularly before 6 months of age.
Monitoring the Testing Process: Quality Assurance, Quality Control, and Quality Assessment
The validity of diagnostic test results depends on the quality of a number of measures used before, during, and after the test is performed.(1) To ensure the quality of test results, a program consisting of quality assurance, quality control, and quality assessment is necessary. In most developed countries, regulatory agencies provide guidelines for such a program. In contrast, formal programs and recommendations generally are unavailable in developing countries. Brief descriptions and examples of quality assurance programs follow.
Quality assurance encompasses all measures, from receipt of specimens through final reporting, to ensure that the final results are as accurate as the assays allow. Specimens must be inspected upon arrival for suitability; logging, processing, and review of all accompanying paperwork must be performed and monitored carefully. Also included are an organized record keeping system, standard operating procedure manuals that act as references, a continuing education program, supervisory review of results, a system for evaluation of laboratory personnel, use of the most appropriate tests/strategies, a mechanism for timely reporting, compliance with regulatory requirements, storage of specimens for follow-up testing, appropriate reporting forms, variance reporting for errors/inconsistencies, a good management system, and, of course, a good quality control and quality assessment program.
Quality control refers to those specific measures that ensure the test is performing as expected. Such measures include careful inspection of internal (kit) control values that validate the test, monitoring of physical parameters (temperatures, functioning of equipment), validation of new reagents (different kit lots), and use of extraordinary measures such as external controls to verify the manufacturer's claims (the use of external controls is recommended but not required). A detailed description of quality control measures has been published.(1)
Quality assessment is a means to challenge the overall performance of the laboratory. This process usually consists of the testing of a panel of samples with known reactivity provided by an external source. Such assessment, usually performed quarterly, yields some information about the overall quality of the laboratory's performance. Results from each laboratory are compiled and feedback is provided. Other measures of assessment include internal (self-inspections of the laboratory and testing process), specimens provided by the laboratory supervisor for blinded testing by personnel, and review of the total operation by an external agency. The ultimate challenge in totally assessing the ability of a laboratory to produce accurate results is to provide these panels of specimens in a blinded manner so that personnel are unaware that they are being monitored.
