Most recently infected individuals mount a vigorous antibody response directed against autologous HIV. During this time, HIV typically evolves rapidly in response to this neutralizing antibody response. As a consequence, at any time during early HIV disease, antibody responses are more likely to recognize earlier autologous viruses than contemporaneous virus (2, 18, 36, 44, 51). The well-documented emergence of antibody escape during early HIV infection argues against a protective role of neutralizing antibodies in the setting of chronic infection. However, several issues remain unresolved. First, the degree to which neutralizing antibody escape evolution persists indefinitely has not been well defined. Theoretically, HIV may be constrained in its ability to continuously and fully escape neutralizing antibody responses over a period of years. Second, the rapid emergence of neutralizing escape mutations in the setting of primary HIV infection does not preclude the possibility that a small subset of patients may develop and maintain neutralizing antibody responses that effectively control HIV replication. Although some studies suggest that potent neutralizing antibody responses contribute to the control of HIV in patients with nonprogressive HIV infection (i.e., long-term nonprogressors) (9, 40, 41), other studies have failed to detect effective neutralizing responses in these patients (5, 20, 30). Finally, the presence of viral escape from neutralizing antibodies does not rule out the possibility that partially effective responses might persist. The latter concept is supported by recent observations indicating that residual antiretroviral drug pressure often persists in the presence of high-level drug resistance, suggesting that there are limits in the ability of HIV to completely evade some antiviral responses (4, 8, 13).

Antiretroviral therapy dramatically affects the complex relationship that exists between the virus and the host response. Although most studies have shown that anti-HIV neutralizing antibody responses decline after the introduction of therapy (presumably due to a decrease in antigenic stimulation), serial structured treatment interruptions, which are designed to enhance HIV-specific T-cell responses, have been associated with enhanced neutralizing activity against autologous virus (35). Similar observations have been reported among patients with intermittent viremia (“blips”) (6). Finally, several studies have reported potent responses against autologous virus in the setting of partially effective antiretroviral therapy (6, 37, 38). The latter observations are consistent with a series of recent studies indicating that the emergence of drug-resistant HIV is associated with a decrease in relative virulence of HIV in vivo (12, 28) and that a significant subset of patients with low-level drug-resistant viremia exhibit heightened HIV-specific CD4⁺ T-cell and CD8⁺ T-cell responses (1, 14, 16, 42, 48).

Most earlier studies addressed the role of neutralizing antibody activity either by using lab-adapted strains (which could not provide insights about antibody responses against autologous virus) or by using labor-intensive approaches in which autologous virus was first cultured and passaged in vitro (which may have altered the virus). Because of these technical and practical limitations, novel techniques in which autologous virus can be assayed against patient-derived plasma in a high-throughput manner have been developed (11, 15, 44). In these techniques, pseudotyped vectors containing patient-derived env gene sequences undergo single rounds of replication in the presence or absence of patient-derived serum. These assays generally correlate well with the more established but labor-intensive peripheral blood mononuclear cell assays and have been recommended for use with studies assessing vaccine efficacy (7, 31, 34).

Given the number of unresolved questions regarding the role of neutralizing antibodies in the setting of established HIV infection and the recent development of assays able to quantify autologous responses in a high-throughput manner, we analyzed cross-sectional and longitudinal samples from a diverse group of recently and chronically infected adults. Our primary objective was to determine whether neutralizing activity directed against autologous virus has any associated impact on levels of viremia. We focused on four well-characterized groups of patients, including those with (i) acute infection, (ii) chronic infection with poorly controlled HIV replication (virologic noncontrollers), (iii) chronic infection with long-term control of HIV replication (virologic controllers), and (iv) chronic infection with long-term control of drug-resistant HIV (partial controllers on antiretroviral therapy [PCAT]). The latter two groups were chosen because evidence from our group and others indicates that they retain high levels of HIV-specific CD4⁺ T cells, a subpopulation that is likely important for the generation and differentiation of a neutralizing antibody response (16). As a secondary objective, we analyzed the level of cross-neutralizing activity against a large panel of heterologous primary viruses.

Study design. We assessed participants enrolled in two ongoing prospective cohort studies aimed at defining the immunologic and virologic characteristics associated with disease outcome in recently infected individuals (Options cohort) or in chronically infected individuals (SCOPE cohort). All subjects in each cohort provided written informed consent.

Characteristics of the UCSF Options project have been described previously (21). Enrolled subjects must have evidence of acute or recent HIV infection as defined by (i) negative or weakly reactive HIV antibody test, enzyme immunoassay, and Western blot analysis results, with detectable plasma HIV RNA levels, (ii) documentation of a negative HIV antibody test result within the previous 12 months and a positive antibody test at screen, or (iii) a history compatible with recent HIV infection, with laboratory evidence of recent antibody seroconversion by use of a sensitive/less-sensitive enzyme immunoassay testing strategy (23). From this cohort, we selected individuals who presented with acute/recent HIV infection and who had plasma samples available prior to the introduction of antiretroviral therapy.

The UCSF SCOPE cohort has enrolled over 600 chronically infected adults. All subjects are observed every 4 months (14, 22). From this cohort, we selected two subsets of chronically infected individuals who were controlling HIV replication in the absence of fully effective antiretroviral therapy: (i) virologic controllers, defined as antiretroviral drug-untreated individuals who have a steady-state plasma HIV RNA level below 2,000 copies/ml, and (ii) PCAT, defined as antiretroviral drug-treated individuals with multidrug-resistant HIV and a steady-state plasma HIV RNA level between 500 and 10,000 copies/ml. As a comparator group, we selected a group of virologic noncontrollers, defined as antiretroviral drug-untreated individuals with plasma HIV RNA levels of >2,000 copies/ml.

Longitudinal analysis was performed with 32 subjects. This included five individuals with acute HIV infection who initiated and subsequently discontinued antiretroviral therapy. This interruption period was studied in detail, as this may mimic the period of untreated early HIV infection. We also studied chronically infected individuals who remained untreated (five controllers and three noncontrollers) or who remained on a stable partial suppressive regimen (14 PCAT subjects). Finally, we studied seven virologic noncontrollers who initiated and responded virologically to combination antiretroviral therapy. In each case, longitudinal neutralizing antibody responses were measured against the baseline autologous virus. Longitudinal neutralizing antibody responses against contemporaneous autologous viruses and against laboratory-adapted strains were also measured for most subjects.

Cross-neutralizing activity against heterologous primary and laboratory viruses. The capacity of patient-derived plasma to neutralize heterologous isolates from other members of the cohort was assessed using two separate matrices. Eight participants were included in the first experiment and 17 subjects in the second experiment. In each experiment, baseline plasma samples were tested against the baseline autologous virus from each of the other subjects. No subject contributed data to both experiments. For all subjects, we also tested the ability of plasma to neutralize heterologous laboratory stains (NL4-3 and JR-CSF), as outlined below.

Neutralizing antibody assay. The antibody neutralization assay is a modification of the Monogram Biosciences HIV coreceptor tropism assay (44). Briefly, HIV genomic RNA was isolated from virus stocks or plasma. DNA spanning the open reading frame of gp160 was amplified by reverse transcription PCR using forward and reverse primers located immediately upstream and downstream of the env initiation and termination codons, respectively. env PCR products were digested and ligated to compatible ends in an expression vector. These ligation products were then introduced into competent Escherichia coli cells by transformation, and plasmid DNA was purified from bacterial cultures. Virus particles containing patient virus Env proteins were produced by cotransfecting HEK293 cells with env libraries plus an env HIV genomic vector that contains a firefly luciferase indicator gene. Recombinant viruses pseudotyped with patient-derived virus Env proteins were harvested 48 h posttransfection and incubated for 1 h at 37°C with serial fourfold dilutions of heat-inactivated patient plasma samples. U87 cells that express CD4 plus the CCR5 and CXCR4 coreceptors were inoculated with virus-plasma dilutions. Virus infectivity was determined 72 h postinoculation by measuring the amount of luciferase activity expressed in infected cells. Neutralizing activity was displayed as the percent inhibition of viral replication (luciferase activity) at each antibody dilution compared to an antibody-negative control: % inhibition = [1 − (luciferase with antibody/luciferase without antibody)] × 100. Titers were calculated as the reciprocal of the plasma dilution conferring 50% inhibition.

Recombinant viruses psuedotyped with Env proteins from amphotropic murine leukemia virus (aMLV) were used as a negative control. These Env proteins are able to mediate virus entry in U87/CD4/CCR5/CXCR4 cells but are not inhibited by anti-HIV Env antibodies. Recombinant viruses pseudotyped with Env proteins from HIV laboratory isolates known to be relatively neutralization sensitive (e.g., NL4-3 and SF162) or neutralization resistant (e.g., JR-CSF) were also studied in each experiment.

For samples derived from antiretroviral drug-treated patients, modifications to this assay were needed to exclude the potential inhibitory activity of drugs in patient-derived serum. HIV expression vectors were therefore constructed with multidrug-resistant reverse transcriptase and protease in experiments involving antiretroviral drug-treated patients. All pseudotyped virions (e.g., NL4-3, JR-CSF, aMLV, and autologous virus) contained the same reverse transcriptase/protease in these studies.

T-cell activation. T-cell activation (as defined by the coexpression of CD38 and HLA-DR) was measured for a subset of chronically infected individuals. As previously described (22), these studies were performed using freshly collected, EDTA-anticoagulated whole blood and analyzed by four-color flow cytometry with a Beckman Coulter Epics XL flow cytometer. Activated T cells were identified using fluorescein isothiocyanate-conjugated anti-HLA-DR, phycoerythrin-conjugated anti-CD38, phycoerythrin-cyanin red 5.1-conjugated anti-CD3, and phycoerythrin-Texas Red-conjugated anti-CD4 or -CD8. To exclude monocytes and natural killer cells, expression of CD3 was included in the definition of both CD4⁺ and CD8⁺ T cells. Cell subpopulations expressing the activation markers CD38 and/or HLA-DR were gated from the total CD3⁺ CD4⁺ or CD3⁺ CD8⁺ populations on a two-dimensional dot plot where quadrant gates, defined by pertinent isotype control antibodies, were used to delineate positive and negative populations.

Statistical analysis. Nonparametric tests were used for all analyses. Differences in continuous variables between patient groups were analyzed with the Mann-Whitney U test. All correlations were determined using Spearman's rank correlation. Changes in neutralizing antibody titers were assessed over time by using generalized estimating equations, treating the month of observation as a categorical predictor and log transforming the neutralizing antibody titer to meet model assumptions. Changes in neutralizing antibody titers over time were assessed with nonlinear tests of trend.

Subject characteristics. A total of 65 individuals were included in this study (Table 1). Twenty-four subjects had evidence of recent HIV infection (median estimated duration of infection, 2.4 months; interquartile range [IQR], 2.2 to 2.4). Eight subjects were virologic controllers and had maintained low-level viremia in the absence of therapy. Nineteen subjects were virologic noncontrollers. Fourteen subjects had maintained partial control of viral replication on antiretroviral therapy despite the presence of drug-resistant HIV.

TABLE 1. Cohort characteristics by treatment group

Neutralizing responses against autologous virus. The degrees to which antibodies found in patient plasma neutralize autologous virus from the same time point were assessed at the first time point for all patients. Low-level responses with titers in the range of 1:20 to 1:100 were observed with most chronically infected subjects (Fig. 1A). When all chronically infected subjects were considered, there was a consistently higher response to the contemporaneous autologous virus than to the non-HIV envelope control virus (murine leukemia virus [MLV]), suggesting that low-level autologous activity may exist in these individuals (P of <0.001 for the response to autologous virus versus MLV). In contrast, a neutralizing antibody response against autologous virus was generally absent in individuals with acute HIV infection and was much lower than that observed in individuals with chronic infection (P < 0.001) (Fig. 1A).

FIG. 1. (A) Neutralizing antibody titers against contemporaneous autologous virus for subjects with acute HIV infection (n = 19) and for subjects with chronic infection (n = 41). (B) Relationship between HIV RNA levels and the neutralizing titers (more ...)

There was a negative correlation between the neutralizing titer against autologous virus and the plasma HIV RNA level among all subjects (ρ = −0.29, P = 0.03, n = 57) but no association between the neutralizing antibody response against autologous virus and the CD4⁺ T-cell count. Since this analysis may have been confounded by rapidly changing HIV RNA levels in the setting of acute infection as well as by the complicated impact which treatment may have had on virus-host interactions, we analyzed untreated chronically infected subjects separately. A negative association between the neutralizing antibody response against autologous virus and the steady-state level of viremia was also observed within this subset (ρ = −0.40, P = 0.05) (Fig. 1B). Collectively, these observations provide indirect evidence supporting a persistent and partially effective neutralizing antibody response against autologous virus.

Cross-neutralizing activity against heterologous viruses. The capacity of antibodies in patient-derived plasma to neutralize primary isolates from other members of the cohort was assessed using two separate matrices. In the first experiment, plasma samples from five virologic controllers and three virologic noncontrollers were tested against the baseline virus from each of these subjects (Table 2). In the second experiment, plasma samples from three virologic controllers, eight virologic noncontrollers, and six PCAT were tested against the baseline virus from each of these subjects (Tables 3 and 4).

TABLE 2. Cross-neutralizing activity against heterologous virus, determined by using plasma from antiretroviral-untreated subjectsa

TABLE 3. Cross-neutralizing activity against heterologous virus, determined by using plasma from antiretroviral-untreated subjectsa

TABLE 4. Cross-neutralizing activity against heterologous virus, determined by using plasma from antiretroviral-treated subjects with partial viral suppressiona

The magnitude of the cross-neutralizing response was defined based on the average neutralizing titer against all heterologous viruses. The average magnitude of the neutralizing antibody against heterologous viruses was consistently higher than the response observed against contemporaneous autologous virus (P = 0.02; Wilcoxon rank test). This trend was observed within each of the three subsets of subjects studied: virologic controllers, virologic noncontrollers, and PCAT.

The breadth of the cross-neutralizing response was defined based on the number of viruses that were effectively neutralized by any given patient-derived plasma sample (where a positive response was defined by the presence of a >3-fold-higher response against the heterologous virus than against the aMLV control). The breadths of the responses varied but were generally higher for virologic noncontrollers than for virologic controllers: plasma from the noncontrollers neutralized a median of 71% of other viruses (IQR, 48% to 88%), while plasma from the controllers neutralized a median of 43% of other viruses (IQR, 25% to 54%). The PCAT subjects had very low levels of cross-neutralizing activity (plasma from these individuals neutralized a median of 4% of other viruses [IQR, 0% to 19%]).

When all subjects were considered, there was a positive association between the magnitude of the neutralizing responses against heterologous virus and the level of viremia (ρ= 0.48, P = 0.01, n = 25). This effect was also seen when we limited our analysis to the chronically untreated subset (ρ = 0.53, P = 0.02, n = 19) (Fig. 1C). There was also a similar association between the breadth of the neutralizing response against heterologous virus (in terms of the proportion of viruses which tested positive) and the level of viremia (ρ = 0.50, P = 0.01, n = 25).

The degree to which antibodies in patient plasma samples could neutralize heterologous viruses was also assessed by measuring responses to three well-characterized laboratory strains (NL4-3, SF162, and JR-CSF). Neutralizing responses to these viruses were much higher than those observed to occur against contemporaneous autologous virus (P of <0.001 for each pairwise comparison). Also, as seen with autologous responses, the response to these viruses was much higher for patients with chronic infection than for patients with acute infection (P of <0.001 for each pairwise comparison of acute versus chronic infection).

Viral evolution is evident in recent but not chronic infection. Previous studies have shown that among recently infected patients, viral evolution in response to neutralizing antibody activity is so rapid that titers against autologous virus obtained at the concurrent time point are low to undetectable, while titers against virus sampled several months earlier are often high (44, 51). Here, we assessed the degree to which neutralizing antibody titers against autologous virus emerged over time for the five acutely infected subjects who initiated and responded virologically to a combination regimen and who subsequently discontinued therapy (Fig. 2A). The neutralizing antibody response against autologous virus and laboratory strains was low prior to the introduction of therapy. Interruption of therapy was associated with a variable increase in HIV RNA levels but a consistent increase in the ability of patient-derived plasma to neutralize NL4-3, SF162, and JR-CSF (P of <0.001 for NL4-3, P of <0.0001 for SF162, and P of 0.03 for JR-CSF; test-of-trend analysis). Autologous responses against the baseline virus also increased over time (P = 0.002; test-of-trend analysis). Of note, although the response to contemporaneous virus was low for most subjects, it increased with time, suggesting residual activity of these responses at later time points (P = 0.04; test-of-trend analysis). These data among acutely infected individuals interrupting effective antiretroviral therapy are consistent with previous observations made in the setting of untreated primary infection (44) and support the concept that HIV is able to evolve rapidly in response to active antibody-mediated immunologic pressure. However, the observation that the response directed against contemporaneous virus increases with time suggests that the ability of the virus to completely evade neutralizing responses may be limited.

FIG. 2. Change in response to baseline autologous virus and to laboratory viruses over time in recent (A) and chronic (B) HIV disease. (A) Five acutely infected subjects initiated and later interrupted antiretroviral therapy. Viral rebound was associated with (more ...)

We also studied longitudinal changes in the setting of chronic infection (Fig. 2B). A total of 23 individuals were studied (five virologic controllers, four virologic noncontrollers, and 14 partial controllers on antiretroviral therapy). In each case, the response to the baseline virus was assessed every 16 weeks over a period of 1 year (where baseline was defined as the first sample available for this analysis). In contrast to observations with early HIV disease, neutralizing activity against baseline autologous virus did not increase over time (P> 0.30; test-of-trend analysis). Responses to laboratory strains also remained stable over time (P of >0.30 for each virus). Hence, viral evolution was not detected with patients who had been infected for several years or more.

Antiretroviral therapy rapidly decreases the level of neutralizing antibodies. Seven antiretroviral drug-untreated individuals initiated and responded to a combination antiretroviral regimen (Fig. 3, top). Neutralizing titers against autologous virus were low for these virologic noncontrollers prior to the introduction of antiretroviral therapy and did not change significantly during the first 1 to 2 years of treatment (Fig. 3, middle). In contrast, neutralizing activity against NL4-3 was high prior to therapy and declined dramatically within the first 4 months of observation (Fig. 3, bottom). Two patients experienced intermittent viremia during therapy. These isolated periods of detectable viremia were temporally associated with transient increases in the response against autologous virus but not against NL4-3 (Fig. 3, middle). These observations indicate that HIV antigen drives the production of neutralizing antibodies in the setting of advanced HIV disease. The rapid decrease in NL4-3 responses, with no change in autologous responses, provides further evidence that most of the anti-HIV specific neutralizing activity in these chronically infected individuals is ineffective.

FIG. 3. Change in response to autologous virus and NL4-3 after initiation of an effective antiretroviral treatment regimen. (Top) Seven antiretroviral drug-untreated subjects with chronic HIV disease initiated and responded virologically to combination antiretroviral (more ...)

Immune activation and heterologous responses. Several mechanisms may account for the observation that neutralizing responses against heterologous viruses, including antigen-driven stimulation of ineffective HIV-specific antibodies (i.e., “original antigenic sin”) and/or nonspecific immune activation, increased during chronic HIV infection. Although we did not measure B-cell function directly in this study, we did study the proportions of CD4⁺ and CD8⁺ T cells coexpressing both CD38 and HLA-DR. Among chronically infected individuals, there was a consistent significant direct association between CD4⁺ T-cell activation and the neutralizing antibody response against a panel of heterologous viruses (ρ = 0.52, P = 0.01, n= 22) and between CD4⁺ T-cell activation and neutralizing antibody response against NL4-3 (ρ = 0.36, P = 0.36, n = 37). There were similar associations with CD8⁺ T-cell activation (ρ of 0.42 and P of 0.05 for responses to heterologous primary viruses and ρ of 0.31 and P of 0.06 for responses to NL4-3).

Using a single-cycle recombinant virus assay to measure neutralizing antibodies against autologous and heterologous primary isolates, we made a number of observations relevant to the role of neutralizing antibodies in the settings of acute and chronic HIV infection. First, neutralizing activity against contemporaneous autologous virus was generally low, particularly with acutely infected individuals. However, among chronically infected individuals, there was a negative correlation between the response to autologous virus and plasma HIV RNA levels, suggesting that these antibodies may contribute to the control of HIV replication. Second, although the neutralizing antibody response against autologous virus was low during early HIV infection, it did increase with time, again suggesting that the ability of HIV to fully evade neutralizing antibodies is limited. Third, viral evolution (as defined by the ability of plasma to neutralize earlier virus more effectively than contemporaneous virus) was evident in early but not in chronic HIV infection. Fourth, many subjects, particularly those with advanced disease and higher levels of viremia, had high cross-neutralizing antibody responses directed against laboratory viruses and moderate cross-neutralizing titers directed against heterologous primary isolates. Finally, the initiation of highly active antiretroviral therapy resulted in a rapid decrease in the anti-NL4-3 neutralizing titer, with levels approaching a new steady state within 4 months of starting therapy. Collectively, these observations indicate that although HIV evolution is rapid in early HIV disease it does not persist indefinitely and that there may be constraints on the virus's ability to fully evade neutralizing antibody responses. Regardless of whether these neutralizing antibodies exert an anti-HIV effect in chronic disease, they appear to retain the ability to cross-neutralize viruses from other individuals, which may have clinical implications, as discussed below.

One of the central questions we sought to address is whether neutralizing antibodies determine, in part, the steady-state level of viral replication in the setting of either treated or untreated chronic HIV infection. Based on data indicating that an effective antigen-specific CD4⁺ T-cell response is critical for maintaining an effective B-cell response (10, 32), we selected from our cohort a subset of subjects who were known to have high-level CD4⁺ T-cell responses (i.e., virologic controllers and PCAT) (14). Although high-level neutralizing titers directed against contemporaneous autologous virus were not observed with most of these subjects, our observation that the autologous response was higher for patients with lower levels of viremia argues for a protective role (at least in the setting of chronic untreated HIV disease). Given the cross-sectional nature of this analysis, it is difficult to conclude whether detectable autologous neutralizing antibody responses are a cause or a consequence of low levels of viral replication. However, our observation that heterologous virus response increases with viremia argues against a mechanism whereby antibody production is generally spared in patients naturally controlling viral replication.

The well-documented ability of HIV to escape from neutralizing antibody responses is often cited as strong evidence against any protective role for these antibodies in chronic disease. In our current study, we extend these findings to include acutely infected individuals who initiated therapy at the time of their diagnosis and who subsequently interrupted antiretroviral therapy (hypothetically, the interruption phase may provide insights into host-virus dynamics during acute disease). As seen with untreated primary HIV disease (18, 36, 44, 51), neutralizing antibody responses against contemporaneous autologous viruses were generally lower than the responses against earlier viruses, suggesting that the immune system was not able to efficiently respond to a rapidly evolving quasi-species. There may be limits, however, to the ability of the virus to fully escape, as evidenced by the observation that responses to contemporaneous virus increased over time.

Although neutralizing activity against autologous virus was generally low in the setting of chronic infection, neutralizing activity against heterologous virus was often high. This was particularly true with regard to the response to NL4-3 and SF162 for chronically infected individuals with high-level viremia. NL4-3 is a lab-adapted strain that has presumably lost those virologic factors that protect against antibody-mediated neutralization in vivo (47). NL4-3-specific responses may therefore be a surrogate for the overall level of HIV-specific neutralizing titers. Our observation that neutralization titers to NL4-3 increased with advanced disease and then decreased rapidly in response to antiretroviral drug-mediated suppression of viral replication argues that the generation of these antibodies in the setting of advanced immunodeficiency is driven by antigenic stimulation. HIV-mediated production of antibodies may reflect nonspecific B-cell activation and/or an “original antigenic sin” phenomenon. With regard to the former point, our group and others have shown high levels of T-cell activation for patients with advanced disease and have determined that these levels decrease dramatically in response to treatment (22, 29, 50). Chronic polyclonal B-cell activation has also been reported in the setting of advanced HIV disease (27, 37, 45). Although we did not measure B-cell function directly in this study, we observed a consistent relationship between both CD4⁺ and CD8⁺ T-cell activation (as defined by expression of CD38 and HLA-DR) and the level of neutralizing antibody responses against heterologous virus, suggesting that immune activation drives some of the HIV-specific neutralizing antibodies in the setting of chronic disease. With regard to the second point (“original antigenic sin”), it is possible that B cells and other antibody-secreting cells that were primed by earlier viral strains continue to produce antibody in response to antigen-driven stimulation from related but neutralization-resistant strains. This phenomenon has been well described for other chronic viral infections (17, 25, 33).

Low to moderate cross-neutralizing activity was also observed when plasma from a disparate group of chronically infected subjects was tested against a panel of patient-derived viruses. Although the clinical relevance of these observations remains unclear, it is interesting to speculate that such neutralizing antibody responses against heterologous viruses might protect against HIV reinfection (or superinfection). Most well-documented cases of HIV superinfection have occurred in the setting of recent HIV infection (3, 24, 26, 43, 46, 52). Superinfection appears to be very uncommon in the setting of advanced HIV disease (19, 49). This potential relationship between the duration of HIV disease and susceptibility to HIV superinfection was tested empirically in a macaque model, where it was shown that the risk of superinfection was high during the acute stage but waned rapidly with time (39). Our observations that higher HIV RNA levels in chronic disease are associated with increased titers of response against heterologous viruses (in terms of both magnitude and breadth) suggest that humoral immunity might protect against reinfection in some individuals.

Our study has several limitations that deserve comment. Most importantly, our observation that the level of viremia is negatively correlated with the autologous responses is based on a cross-sectional analysis, making it difficult to discern cause and effect. Although low-level viremia may be a consequence of effective HIV immunity, it is also possible that low-level viremia may preserve immune function and the production of antibodies that have an insignificant impact on viral replication. This seems to be less likely given the observation that viral replication appears to drive the production of HIV-specific neutralizing antibodies, as shown by the consistent association between HIV RNA levels and responses to heterologous viruses, as well as by the observation that antiretroviral treatment decreases these responses. A second limitation pertains to our observations regarding viral evolution, which was evident for acutely infected individuals interrupting therapy but not evident for chronically infected individuals in steady state. A more appropriate comparison group for our chronically infected individuals would have been recently infected patients who did not receive antiretroviral therapy. However, a previous study of recently infected individuals using the same methods provided consistent evidence for viral evolution in early untreated HIV disease (44). More importantly, it should be emphasized that our conclusion regarding the lack of viral evolution is based on indirect inferences (i.e., a lack of increasing neutralizing titers against earlier viruses). A more definitive manner to test this issue is to sequence virus from plasma and peripheral blood mononuclear cells over time and then to directly compare rates of viral escape in early versus late HIV infection. Finally, although our conclusions suggest that HIV may not be able to fully evolve to escape all neutralizing pressures, it is also possible that the host may have limits in its capacity to continue to generate novel epitope-specific antibody responses.

In conclusion, low but measurable neutralizing antibody responses against autologous virus exist in some chronically infected individuals, suggesting a role of these responses in the durable control of HIV replication. Viral evolution, as defined based on evolving neutralizing responses over time, is clearly evident with recent infection but not with long-term chronic infection. The latter observation suggests that HIV might be constrained in its ability to evade neutralizing response indefinitely and/or that the immune system is eventually exhausted and unable to generate highly effective neutralizing antibody responses over time. Finally, chronic HIV infection drives the ongoing production of HIV-specific antibodies that effectively neutralize heterologous viruses. Whether this observation reflects antigen-driven expansion and/or abnormal B-cell activation remains to be determined.

This work was supported in part by grants from NIAID (AI052745, AI055273, AI47062, and AI44595), the California AIDS Research Center (CC99-SF, ID01-SF-049), the National Institutes of Health UCSF/Gladstone Institute of Virology & Immunology Center for AIDS Research (P30 MH59037, P30 AI27763), the Center for AIDS Prevention Studies (P30 MH62246), and the General Clinical Research Center at San Francisco General Hospital (5-MO1-RR00083-37).