Background. The diagnosis of human immunodeficiency virus type 1 (HIV-1) infection by standard tests relies on the formation of HIV-1-specific antibodies. Early treatment of acute HIV-1 infection may have unique immunologic effects on host cellular and humoral responses. Rare cases of HIV-1 seroreversion have been reported for patients with advanced or rapidly progressive disease. Here, we report seroreversion that occurred in subjects with acute HIV-1 infection who initiated early antiretroviral therapy.
Methods. A total of 150 patients with symptomatic acute or early onset HIV-1 infection that was treated with antiretroviral therapy were observed prospectively by means of monthly clinical and laboratory evaluation, which included serial HIV enzyme-linked immunosorbent assay and Western blots, until a fully evolved HIV-1 antibody response was documented.
Results. Three patients who initiated antiretroviral therapy a mean interval of 8 days (range, 1-16 days) after presentation and were observed for a mean duration of 50.2 months (range, 40.2-55.7 months) did not develop a fully evolved HIV-1 antibody response or demonstrated complete or partial HIV-1 seroreversion, despite maintenance of cytomegalovirus-specific humoral responses. Virologic suppression and seroreversion (complete or partial) occurred a mean duration of 4.1 months (range, 2.3-5.7 months) and 15.5 months (range, 6.7-26.3 months), respectively, after the initiation of therapy. All patients maintained complete virologic suppression while receiving therapy and had an undetectable HIV-1 RNA load at the time of seroreversion.
Conclusions. Early antiretroviral therapy associated with durable virologic suppression in acute HIV-1 infection may abrogate the formation or detection of HIV-1-specific antibodies. Ongoing antigenic stimulation may be required to maintain HIV-1-specific humoral responses. Incomplete evolution of the HIV-1 antibody response and/or presence of seroreversion (although infrequently observed) underscore the potential unique immunologic effect of early antiretroviral therapy in patients with primary HIV-1 infection.
A total of 150 patients with symptomatic acute HIV-1 infection (i.e., a detectable HIV-1 RNA/DNA load and a negative HIV-1 antibody test) or early onset HIV-1 infection (i.e., a ⩽12-month history of infection, based on results of detuned ELISA, or a previously documented negative HIV-1 antibody test) were enrolled in an institutional review board-approved observational study at our institution over the past 8 years. The patients were observed prospectively with serial clinical and immunological assessments, including serial HIV ELISA and Western blot testing, until a fully evolved HIV-1 antibody response was documented. Baseline demographic and clinical data were collected, and monthly follow-up visits were scheduled for the subjects. HIV-1 antibody testing was performed using an HIV-1/HIV-2 (HIV-1/2) ELISA (Abbott) and an HIV-1 Western blot (BioRad). The HIV-1 load was measured using commercially available tests (Roche Amplicor and Chiron branched DNA), CD4+ T lymphocyte counts were assessed by flow cytometry, and B lymphocyte function was assessed by measuring the IgG response to cytomegalovirus. The time to virologic suppression was defined as the interval between the start of therapy and the first virus load measurement that yielded undetectable levels (<50 copies/mL); the virus load was confirmed by a second measurement.
Patient 1. A 30-year-old man presented to the emergency department with diarrhea, fever, neutropenia, and elevated liver function test results. At presentation, an HIV-1/HIV-2 ELISA was negative, the HIV-1 branched DNA load was >500,000 copies/mL, and the CD4+ T lymphocyte count was 473 cells/mm3. Treatment with zidovudine, lamivudine, and efavirenz was initiated the next day but was switched to zidovudine, lamivudine, and nevirapine after 3 days because efavirenz was not tolerated. On day 12 after presentation, Western blotting showed a reactive gp160 band and a weakly reactive p24 band (table 1). The virus load became undetectable by day 69. After 6 months of therapy, the CD4+ T lymphocyte count was 355 cells/mm3, HIV-1/2 ELISA results remained weakly positive, and results of a Western blot were indeterminate, with weak reactivity at the gp160 band only. After 11 months of therapy, results of the HIV-1/2 ELISA and Western blot were negative. Surveillance antibody testing over a 56-month follow-up period showed fluctuations between weakly positive and negative HIV-1/2 ELISA results, with nonreactive Western blots.
Patient 2. A 45-year-old man presented with fever, fatigue, and back and calf pain 2-6 weeks after repeated unprotected sexual exposures with a new partner. At presentation, the HIV-1 RNA load was 807,500 copies/mL, and the CD4+ T lymphocyte count was 281 cells/mm3. The final negative HIV-1 antibody test was obtained at an outside institution 7 days before enrollment in the cohort. A second Western blot for detection of antibody, performed on day 7 after presentation, had positive results with 2 reactive bands, and treatment with stavudine, lamivudine, and efavirenz was initiated (table 1). By day 71 after presentation, the Western blot was reactive at all bands. After 5 months of therapy, the virus load was <50 copies/mL, and the CD4+ cell count was 526 cells/mm3. After 26 months of therapy, an ELISA was positive for HIV-1, but results of a Western blot were indeterminate, with reactivity at the p24 band only. One year later (after receiving therapy for ∼38 months), results of an ELISA and a Western blot were positive and negative, respectively.
Patient 3. A 47-year-old man presented with symptoms consistent with acute retroviral syndrome after sexual exposure to a new partner known to be infected with HIV. At presentation, the HIV-1 branched DNA load was 437,155 copies/mL. Results of antibody tests were positive 16 days later, with 4 bands present on a Western blot; the CD4+ T lymphocyte count was 468 cells/mm3. On day 100 after presentation, results of a detuned ELISA were nonreactive, suggesting that infection occurred ⩽130 days earlier . Patient 3 initiated antiretroviral therapy with zidovudine, lamivudine, and nevirapine. After 6 months of therapy, the HIV-1 load was <50 copies/mL, and the CD4+ T lymphocyte count had increased to 638 cells/mm3. After 13 months of therapy, the patient had a positive HIV-1/2 ELISA result but indeterminate results of a Western blot, with reactivity at band p24 only (table 1). At a follow-up visit 40 months after presentation, patient 3 continued to have a positive ELISA results and indeterminate Western blot results, with reactivity at 1 band only.
Three patients with symptoms consistent with acute retroviral syndrome initiated antiretroviral therapy at a mean interval of 8 days (range, 1-16 days) after presentation and were observed for a mean duration of 50.2 months (range, 40.2-55.7 months). Virologic suppression occurred a mean interval of 4.1 months (range, 2.3-5.7 months) after initiation of antiretroviral therapy, and complete or partial seroreversion developed a mean interval of 15.5 months (range, 6.7-26.3 months) after the initiation of therapy. All patients maintained complete virologic suppression while receiving therapy and had undetectable HIV-1 RNA loads at the time of seroreversion. By the end of the follow-up period, the mean duration of viral suppression was 45.7 months, and the mean duration of seroreversion was 34.6 months. Patient 1 (who initiated antiretroviral therapy within 24 h after presentation) had Western blot results showing that the evolution of the antibody response was interrupted, with reversion of 2 previously present bands. Patient 2 developed a fully reactive Western blot, the results of which subsequently reverted to negative or indeterminate (1 reactive band) during serial assessments. Patient 3 had a Western blot with strong reactivity at 4 bands early in the course of infection, followed by subsequent partial seroreversion coincident with virologic suppression. All patients had positive IgG responses to cytomegalovirus in plasma specimens tested after the time of HIV-1 seroreversion (table 1), which demonstrated the ability to maintain an appropriate antibody response against a different virus causing persistent latent infection.
Reports of early HIV-1 seroreversion appear predominantly in studies involving children and infants and describe uninfected infants who clear passively transferred maternal HIV-1 antibodies [2, 3]. Rare cases of positive results of successive PCR assays for HIV-1 DNA for infants who were born to HIV-infected mothers and who subsequently demonstrated seroreversion have been attributed to either elimination of HIV infection by strong host immune responses or to abortive infection by an attenuated or defective virus . Cases of seronegative HIV-1 disease have been reported in studies involving adults, and various potential explanations have been proposed (table 2). Some investigators suggest that the majority of suspected cases of seroreversion are actually due to test-related errors and that true HIV-1 seroreversion is an extremely rare phenomenon .
Negative results of successive HIV-1 antibody tests for patients who tested positive for p24 antigenemia and/or HIV-1 nucleic acid amplification have occurred in cases in which the course of disease is rapid and severe [6, 7] and have been attributed to an aggressive disease course and associated immunological dysfunction. In the majority of these cases, antibody testing was first performed late during the course of disease, and the loss of antibody cannot be distinguished from lack of antibody formation [6–9].
The hallmark of most previously reported cases of seronegative HIV-1 disease is the absence of past detection of HIV-1 antibody, which distinguishes seronegativity from true seroreversion. Zaaijer et al.  reported a case of temporary seronegativity in a man with esophageal candidiasis, Pneumocystis pneumonia, and progressive immunosuppression. Although results of the HIV-1/2 ELISA fluctuated from positive to negative, results of the Western blot gradually became positive, which is consistent with the expected evolution of antibody response in patients with primary HIV-1 infection (complicated by severe early immunosuppression). A recent report involving a patient who initiated antiretroviral therapy (comprised of 5 drugs) and mycophenolate mofetil (MMF; 2 g/day) after presenting with acute HIV-1 infection described transient loss of HIV-1 antibody after a documented pretreatment HIV-1 Western blot that was reactive at all bands . After 2 months of therapy and sustained virologic suppression, HIV-1 antibody was no longer detectable. However, after antiretroviral therapy and MMF were discontinued, the plasma HIV RNA load rebounded, and serum HIV-1 antibodies were again detectable ⩽1 month later.
This patient received unconventional therapy because he was enrolled in a MMF trial . MMF, an inhibitor of inosine monophosphate dehydrogenase, has cystostatic activity on lymphocytes, which have no alternative enzymatic pathway for the production of guanosine nucleotides. MMF may therefore have direct antiviral activity, by depleting levels of the substrate acted on by reverse transcriptase, as well as immunologic activity, by selectively inhibiting lymphocyte division and reducing activated CD4+ lymphocyte levels [12, 13]. In addition to its known inhibitory effects on cell-mediated immunity, MMF may exert inhibitory effects on the humoral immune response. Significant delay in the synthesis and maturation of cytomegalovirus-specific antibodies has been demonstrated after primary cytomegalovirus infection in solid organ transplant recipients treated with MMF [14–16]. Although MMF may have contributed to seroreversion, the other 3 MMF-treated patients in this study did not demonstrate seroreversion .
Our description of 3 individuals with acute HIV-1 infection who initiated early conventional combination antiretroviral therapy and demonstrated an incompletely evolved HIV-1 antibody response initially and/or complete or partial HIV-1 seroreversion subsequently underscores the potential unique immunologic effects of early therapy in patients with primary HIV-1 infection. Because these patients did not receive MMF, the findings suggest that initiation of effective antiretroviral therapy alone early during the course of disease may be adequate to abrogate the humoral response to HIV-1. The preservation of cytomegalovirus-specific antibody responses contrasts with the loss of HIV antibody, despite the high-level viremia associated with acute HIV-infection only months before seroreversion. Because of the time required for the maturation and affinity of virus-specific antibodies, one hypothesis is that this process can evolve adequately in patients with chronic HIV-1 infection but may be aborted with early therapy and rapid virologic suppression in patients with acute HIV-1 infection. There is additional evidence to suggest that treatment of acute or early onset HIV-1 infection may result in the loss of HIV-specific antibodies ; however, this is most likely an uncommon event.
Other possible explanations for seroreversion include waning antibody responses due to prolonged virologic suppression, irrespective of when treatment is initiated. It is possible that successive testing may show that chronically infected patients who have long-standing virologic suppression and are receiving antiretroviral therapy also lose HIV-specific antibodies, but serial testing is not routinely practiced, and data on the natural history of disease in such patients are therefore limited. Of particular interest from an immunologic perspective is patient 1, who never developed a fully reactive Western blot, which raises the question of whether very early therapy (in this case, within 1 day after presentation) can actually terminate antibody evolution in some instances. Although patient 1 demonstrated HLA B57 positivity (a marker that has been correlated with slower HIV-1 disease progression ), this polymorphism affects cellular immune mechanisms and is less likely to be correlated with antibody seroreversion.
The clinical significance of seroreversion is unclear. The patient described by Jurriaans et al.  formed HIV-specific antibodies again after antiretroviral therapy was stopped and viral rebound occurred, and this may well be an expected outcome in most cases. Cellular reservoirs are seeded early during HIV-1 infection [19, 20], and provirus could readily emerge from these reservoirs if therapy is discontinued or becomes ineffective, resulting in rebound viremia and induction of antibody reformation. At the time of writing, the 3 patients in our study continue to receive antiretroviral therapy, have maintained virologic suppression, and have not reformed HIV-specific antibody during several years of follow-up. Very early therapy may afford the opportunity to alter immunologic responses to HIV-1 infection in a more durable manner. The clinical correlates of HIV-1 antibody loss in cases of acute and early infection deserve further study.
Financial support. Infectious Diseases Society of America (Roche Laboratories Postdoctoral Award to S.K.), Harvard University (T-32 AI007433 HIV Clinical Research Training Program grant to S.K.), and National Institutes of Health (AI040873 and Acute Infection and Early Disease Research Program grant to E.S.R.).
Potential conflicts of interest. All authors: no conflicts.