Detection of Acute HIV Infection: We Can’t Close the Window

(See the article by Rosenberg et al, on pages 528–34.) Acute human immunodeficiency virus (HIV) infection poses a dilemma for diagnosis, clinical management, and public health. It has been variously defined as the transient symptomatic illness associated with high-titer viral replication [1], the period from initial infection to complete seroconversion [2], and the phase between the appearance of detectable HIV RNA and detectable antibodies to HIV [3]. In the context of therapeutic trials, primary HIV infection includes the acute infection interval and the first 6 months after seroconversion, during which viral set point is established [4, 5]. “Detecting acute infection” has also been used synonymously with “closing the window,” the period during which tests for HIV are negative in persons who are infected. In this issue of the journal, Rosenberg et al [6] evaluated how well the Determine HIV-1/2 Ag/Ab Combo rapid test (Combo RT) compared with an algorithm of parallel antibody tests followed by RNA polymerase chain reaction. The Combo RT, designed for use at the point of care with finger-stick whole blood to detect HIV antibody and HIV antigen (and distinguish the two) in 20 minutes, did not fare very well in this comparison. The Combo RT is a lateral flow test (ie, an immunoassay that incorporates all necessary reagents into a single test strip). The test result is read visually, without instrumentation, after the specimen flows across the strip. Lateral flow tests have revolutionized HIV testing because they are easy to perform, require minimal training, have long-term stability, require no dedicated instrumentation, and can use whole blood specimens. In the United States, lateral flow HIV tests are used extensively in nonclinical settings. However, high hopes for point-of-care rapid tests able to identify acute infection, predicated on detecting p24 antigen, may be unrealistic. Recently developed fourth-generation HIV antigen-antibody (Ag/Ab) combination assays, designed for use in the laboratory, detect both p24 antigen and antibodies against HIV but do not distinguish between the two. They exhibit an analytic sensitivity for p24 antigen of 11–18 pg/mL [7], equivalent to approximately 30 000–50 000 copies/mL of HIV RNA [8]. In contrast, lateral flow assays using colloidal gold have a lower limit of detection in the range of 1 ng/mL [9], ∼100-fold higher than the p24 antigen concentration usually present in plasma [8]. With the ultrasensitive procedure that includes heat dissociation of p24 antigen–antibody immune complexes, signal amplification, and instrumentation, the limit of detection for p24 (without additional pretreatment of the specimen to disrupt viral particles) can be ≥250 000 RNA copies/mL [10]. The findings of Rosenberg et al substantiate this: ultrasensitive methods were used to test 7 of the 8 acute specimens and failed to detect p24 antigen in 2 that had RNA concentrations of 45 000 and ≥750 000 copies/mL. In contrast, the ability of the Combo RT to establish the presence of antibodies to HIV was excellent. The Combo RT was positive for antibody in 162 (99.4%) of the 163 persons with established infection and in 2 of the 8 patients with acute HIV infection who were negative on other antibody tests. This latter finding suggests that the Combo RT might be able to identify many of the earliest infections. As shown in Table 1, antibody tests differ in their sensitivity to detect HIV during seroconversion [11]. Most lateral flow rapid tests in current use (and also Western blot analysis) are based on second-generation immunoassay principles that detect only immunoglobulin (Ig) G–class antibodies and become positive, on average, 20–25 days after RNA appears [12]. The antibody component of the Combo RT detects both IgM- and IgG-class antibodies. Several studies suggest the Combo RT becomes positive at or just before sensitive third-generation laboratory antibody assays, approximately 10–14 days after RNA [13, 14]. Thus, the Combo RT might miss acute infections during the 10-day interval when only RNA can be detected but identify those during the subsequent period (also ∼10 days) when results of other rapid antibody tests are still negative [15]. This number might be substantial. In the United States, use of RNA assays to test pools of specimens that were negative on different antibody tests has identified a prevalence of acute HIV infection ranging from 0.5 cases/1000 persons tested in North Carolina to 4.0 cases/1000 in San Francisco [16, 17]. This prevalence might seem relatively low, but acute infections accounted for 5%–20% of all cases of HIV infection among persons seeking testing [16–18]. However, 49%–86% of these acute infections could also be detected by a third-generation antibody test with better sensitivity during seroconversion [18, 19]. Unfortunately, it is not possible to assess what proportion such patients might represent in the population tested by Rosenberg et al, because a negative Determine antibody test result was used as part of the reference standard that defined acute HIV infection. Human Immunodeficiency Virus (HIV) Testing Assays and Their Window Periods Abbreviations: EIA, enzyme immunoassay; Ig, immunoglobulin; PCR, polymerase chain reaction. Human Immunodeficiency Virus (HIV) Testing Assays and Their Window Periods Abbreviations: EIA, enzyme immunoassay; Ig, immunoglobulin; PCR, polymerase chain reaction. This study (together with the recent introduction of laboratory-based, fourth-generation Ag/Ab combination assays for HIV screening in the United States and their widespread availability in other countries) has several ramifications. First, substantiating the results of positive screening tests will require supplemental testing that can confirm the presence of HIV antibodies and either HIV proteins or HIV nucleic acid. Second, classifying infections as acute will depend on the supplemental testing algorithm. The algorithm that has been proposed to replace the traditional algorithm based on Western blot analysis (Figure 1) [20] is analogous to the detuned strategy initially used to identify recent infections [21]. After a reactive third-or fourth-generation screening test, a second-generation IgG antibody test is performed. If this is also reactive, the patient is considered to have established HIV infection. If this second antibody test is negative, the specimen is assayed for HIV RNA to determine whether acute HIV infection is present. Although other supplemental testing strategies can be used [3], testing only for RNA is not sufficient: HIV RNA assay results are negative in 3%–5% of patients with repeatedly reactive HIV antibody immunoassays and positive Western blot results [11, 18]. Proposed testing algorithm for diagnosis of human immunodeficiency virus type 1 (HIV-1) and type 2 (HIV-2) [20]. Asterisk indicates use of antigen-antibody (Ag/Ab) combination test. Determining whether HIV infection was recent in a patient with a positive HIV test result has clinical utility for differentiating whether symptoms of a viral illness might be due to acute HIV infection. The implications for clinical management are less clear. Observational studies suggest that antiretroviral therapy during acute infection might modify the long-term course of disease [5], but the optimal duration and long-term benefits of therapy have not yet been established [22]. On the other hand, immunopathogenesis studies suggest the die is cast early on, during viral replication in the lymphoreticular system and before any HIV markers appear in the peripheral circulation [23]. Public health benefits are also anticipated from better identification of acute HIV infection [23, 24]. Models and phylogenetic studies incriminate this acute, hyperinfectious phase as the culprit that, despite its short duration, drives some HIV epidemics. [25, 26] Identifying acute HIV can improve case finding substantially. In one study, persons with acute HIV infection named 2.5 times as many partners and nearly twice as many partners with undiagnosed HIV infection as did persons with long-standing HIV infection [27]. However, no evidence yet exists for a specific intervention to demonstrate whether knowing that one has acute HIV infection specifically translates into more effective prevention than simply knowing that one is infected [28, 29]. No test can detect infection in persons immediately after they acquire HIV, so a negative RNA test result might offer false reassurance. The consequences of delivering a negative HIV test result to a person whose infection is at its most infectious stage are obvious. In addition to having extremely high levels of virus in genital secretions, simian immunodeficiency virus models suggest that the viral strains that succeed in establishing new infections might be up to 700 times more infectious than those that predominate during established infection [30]. Closing the window alone will not stop the relentless flow of new infections if, to stretch the analogy, the problem is a leaky roof. The duration of the eclipse phase, when infection is present but undetectable, averages 10 days—approximately the same as the acute infection interval as defined by detectable RNA before antibodies appear [12, 31]. Therefore, the number of acute infections detected by more sensitive testing strategies is probably proportional to the number of acute infections that are still being missed. Additional strategies are necessary to identify HIV-infected persons as soon as possible after they acquire HIV. The first is recognition of the acute retroviral syndrome. At least half of persons who acquire HIV develop a nonspecific syndrome in the first 2–4 weeks after infection for which they seek medical care. It may be even more important simply to recommend more frequent testing for certain populations. The days when we recommended testing 6 months after an exposure are over. We can’t close the window completely—but we can certainly lower it. Alere has provided Determine HIV-1/2 Ag/Ab Combo test kits to J. D. S. as part of a separate National Institutes of Health–funded research project. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.