Prognostic Value of HIV-1 Syncytium-Inducing Phenotype for Rate of CD4+ Cell Depletion and Progression to AIDS

Methods

In October 1984, a study on the natural course of HIV-1 infection was started in Amsterdam using a cohort of 1182 asymptomatic homosexual men [27]. All HIV-1-seropositive men in this cohort were clinically examined at least every 3 months, and 3-month serum and peripheral blood mononuclear cells (PBC) were cryopreserved for virologic and immunologic studies. Informed consent for these studies was obtained from all men. Study participants were staged according to the Centers of Disease Control and Prevention (CDC) classification [28].

In the present study, all HIV-1-seropositive men who were still asymptomatic (CDC II or III) in May 1988 (n = 225) were followed prospectively for their HIV-1 biological phenotype and progression to AIDS with a maximum follow-up period of 30 months (mean follow-up, 24.8 months). From this group, 33 men received zidovudine from the start of the study and 4 started zidovudine treatment in the course of the study. The remaining 188 men were not prescribed any other treatment, with the exception of 28 men who received prophylaxis for Pneumocystis carinii pneumonia (PCP) during the last 1 to 12 months of the study period.

In addition to this prospective study, the actual interval between the appearance of SI isolates and progression to AIDS was determined in 45 men who did not receive antiviral treatment. In these patients a phenotypic switch from NSI to SI isolates was deleted by sequential virus isolation during the prospective study described above (n = 22) or during previous studies in the same cohort during the period 1985 to 1988 (n = 23) [6, 18]. For the analysis of the temporal relation between phenotype conversion and changes in the rate of CD4+ cell decline, these 45 men were compared with 45 controls. The controls were randomly selected from the group of seropositive nontreated cohort men in whom SI isolates were never detected and were pairwise matched for CD4+ cell count and serum HIV-1 p24-antigen status at the time of phenotype conversion, as well as pairwise matched for age and length of seropositive period. The time of seroconversion was known for 12 of 45 men shifting to SI variants. These 12 men were matched with 12 men who had a known moment of seroconversion and had persistent NSI isolates. The remaining 33 men were seropositive at their entry into the cohort study. Previous epidemiologic studies, however, indicated that most of the cohort participants who were seropositive at entry became infected within 1 to 1.5 years before the start of the cohort study [29]. Therefore, we matched men pairwise who were seropositive at entry into the cohort study with controls who were seropositive at entry. Because most of these men (74%) were seropositive for at least 3 years before the phenotype switch occurred, it is unlikely that this method of matching would result in an important difference in duration of infection between the case and the control groups.

Virus Phenotyping

HIV-1 was isolated from 0.5 10⁶ fresh or 1 10⁶ cryopreserved PBC from infected men by cocultivation with 1.0 10⁶ MT-2 cells (MRC AIDS Reagent Project, Hertfordshire, United Kingdom), as previously described [23, 30]. Cultures were kept for 3 weeks. Virus replication was detected by observation of syncytium formation and detection of HIV-1 p24 in the culture supernatant [23]. Cocultivation of patient PBC with MT-2 cells is a sensitive and specific method for the detection of SI isolates [23, 26], a subset of HIV ability to induce syncytia in phytohemagglutinin-stimulated peripheral blood lymphocyte cultures and by their tropism for the H9 T-cell line [17, 31]. NSI isolates, detectable by cocultivation of patient PBC with phytohemagglutinin-stimulated peripheral blood lymphocytes, do not replicate in MT-2 cells [23]. The intra-assay variation of this test was less than 2% in 108 repeatedly tested men.

In the prospective study, virus isolation was done on PBC from all study participants during the last 6 months of follow-up. Virus isolation was done on cryopreserved PBC from earlier time points for all men with SI variants at the end of follow-up. The moment of the phenotype conversion (defined as the intermediate period between the last SI-negative and the first SI-positive time point) was further defined by virus culture from sequential PBC samples obtained in the course of the study. Men with SI variants at entry invariably were positive in the MT-2 test at later time points. Similarly, men developing SI variants during the study always tested positive for SI variants at subsequent time points (mean number of time points tested, 3.1; range, 1 to 6). Peripheral blood mononuclear cells cryopreserved at the start of the study from 20 men without SI variants at the end of follow-up were all negative for SI variants. In some of the men studied previously [6, 18], sequential HIV-1 isolates were recovered by cocultivation with phytohemagglutinin-stimulated peripheral blood lymphocytes and were phenotyped as previously described [17].

Sera were tested for HIV-1 p24 in a commercial solid-phase immunoassay (Abbott Laboratories, Chicago, Illinois) according to the manufacturer's instructions.

CD4+ Cell Counts

CD4+ cell counts were determined at 3-month intervals by flow cytofluorometry. The reference value for CD4+ cell counts obtained in our laboratory for seronegative homosexual men (n = 41) during this study period was 0.71 10⁹/L (CI, 0.34 to 1.18).

Statistical Analysis

Number Crunching Statistical Systems (NCSS, version 5.0) was used for statistical analysis. The Kaplan-Meier product-limit method was used to estimate the cumulative incidence of clinical disease and for cumulative HIV-1 phenotype conversion. Difference in survival was analyzed by the log-rank and Wilcoxon rank-sum tests. Stepwise forward Cox proportional-hazard analysis was used to calculate the significant contribution of covariates on AIDS diagnosis. The Mann-Whitney test and the paired samples t-test were used to compare CD4+ cell count and rate of CD4+ cell decline between groups and within groups, respectively. The chi-square test was used to analyze the frequency of PCP prophylaxis in men with and without SI variants and to compare the groups with and without SI variants for serum HIV-1 antigen status.

Results

Prospective Study

Baseline Characteristics

We investigated changes in HIV-1 biological phenotype and clinical progression during a 30-month follow-up period in a prospectively studied group of 225 asymptomatic seropositive men. The relevant characteristics of the study participants, divided in three groups according to their viral phenotype during the study, are summarized in Table 1. The prevalence of HIV-1 p24 antigenemia at entry was not different (P = 0.08) among the three groups and was not different when the zidovudine-treated men (n = 37) were excluded (not shown). No differences in CD4+ cell counts at entry were observed between participants with only NSI variants during the entire study period and those converting from NSI to SI during the study (P = 0.2). However, the baseline mean CD4+ cell counts in persons with SI variants at entry were lower than in persons with NSI variants at entry (P = 0.001). Zidovudine treatment had been initiated more frequently in men with SI HIV-1 variants at entry (Table 1). Because zidovudine treatment was likely to influence the natural course of HIV-1 infection, the 37 men who received zidovudine during the study were excluded from further analyses.

Prevalence and Incidence of HIV-1 Variants with SI Phenotype

In 22 of 188 nonzidovudine-treated men, SI isolates were already present at the start of the study. In addition, a conversion from NSI to SI variants occurred in the course of the study in 22 of the men without SI variants at entry (n = 166). The moment of the phenotype conversion in these 22 men (defined as the intermediate between the last SI-negative and the first SI-positive time point) was further defined by virus culture from sequential PBC samples obtained in the course of the study. The cumulative incidence of phenotype conversion, as determined by the Kaplan-Meier method, was 7.3% after 1 year of follow-up and 15.4% during the whole study period.

HIV-1 Phenotype and Clinical Progression

The cumulative incidence of progression to AIDS for men without zidovudine treatment, with or without SI variants at the start of the study, was determined according to the Kaplan-Meier method (Figure 1). In the course of the study, 36 men developed AIDS. The median time for progression to AIDS for the men with SI isolates at entry was 16.6 months. After 30 months, the probability of progression to AIDS for the group of men with SI isolates at start was 70.8%. In contrast, during the same period, a 15.8% estimated progression was observed for the group of men without SI isolates at study entry (difference between both survivals: P < 0.0001 in both the log-rank and the Wilcoxon test), including the 22 men who developed SI variants in the course of the study. Of these 22 men with HIV-1 phenotype conversion, 5 developed AIDS during follow-up.

Laboratory markers were evaluated as univariate predictors for progression to AIDS using Cox proportional-hazard modeling, at the start of the prospective study (Table 2). CD4+ cell counts less than 0.5 10⁹/L or less than 0.3 10⁹/L, HIV-1-antigenemia, and SI phenotype at the start of the study were all predictive for progression to AIDS. Multivariable Cox proportional-hazard modeling was used to determine which laboratory markers were independent predictors of progression (Table 3). In this study cohort, the isolation of SI HIV-1 variants at entry was associated with a higher relative hazard (6.7) compared with antigenemia (4.9) and with CD4+ cell counts less than 0.3 10⁹/L (4.3) at entry. Zidovudine treatment was started more frequently in men with SI variants (see Table 1). However, when zidovudine-treated men (n = 37) were also included in the Cox analyses, SI variants at entry were associated with the highest relative risk for progression to AIDS (5.1; CI, 3.0 to 8.8), compared with antigenemia (2.1; CI, 1.2 to 3.6) and with CD4+ cell counts less than 0.3 10⁹/L (2.9; CI, 1.7 to 5.0).

Starting in January 1990, PCP prophylaxis was given for men with CD4+ cell counts repeatedly less than 0.2 10⁹/L. Of the 145 men still participating in the study as of January 1990, 28 men were eligible for PCP prophylaxis before the end of follow-up. During this period, PCP prophylaxis was started more frequently among men who had SI isolates (12 of 30) than among men without SI isolates (16 of 115) (P = 0.005, chi-square test). In those who received PCP prophylaxis, progression to AIDS occurred in 3 of 12 men with SI isolates and in 2 of 16 men without SI isolates.

Lag Time between HIV-1 Phenotype Conversion and Progression to AIDS

Because the moment of conversion from NSI or SI isolates was not known for the 22 men with SI isolates at the start of the study, only a minimum estimate of the delay between phenotype conversion and AIDS diagnosis could be made from the prospective study. We identified the median lag time between the first appearance of SI variants and progression to AIDS using the following method. We analyzed the clinical progression of all cohort men (n = 45) in whom a conversion from NSI to SI phenotype had been defined during the prospective study or previous studies [6, 18]. The cumulative incidence of progression to AIDS for these 45 men, after the emergence of SI isolates (time zero), is shown in Figure 2. Based on these data, we found a median interval between NSI to switch and AIDS diagnosis of 23.1 months, compatible with the minimal median lag time of 16.6 months observed in the prospective study.

HIV-1 Biological Phenotype and CD4+ Cell Decline

Prospective Study

For the analysis of CD4+ cell decline, the 188 non zidovudine-treated men were divided into three groups according to the evolution of their viral phenotype during the study. In agreement with previous observations [24], the highest rate of CD4+ cell decline was observed for men with SI variants already at entry (13.2 10⁷/L per year) (Figure 3). In men in whom SI variants were not detected until the end of follow-up, the rate of CD4+ cell decline was 3.7 times lower (3.6 10⁷/L per year). An intermediate rate of decline (7.2 10⁷/L per year) was found for the group of men converting to SI viruses in the course of the study.

Matched-Control Study

To study more accurately the temporal relation between changes in CD4+ cell decline and NSI to SI phenotype conversion, we analyzed CD4+ cell decline in the 45 men with a known time of viral phenotype conversion described above, after aligning them for the time of NSI to SI conversion (time zero) (Figure 4). In the 45 months before NSI to SI conversion, a moderate rate of CD4+ cell counts was observed; however, this rate increased threefold after NSI to SI phenotype conversion (median CD4+ cell decline: 6.0 compared with 18.0 10⁷/L per year, P = 0.003, paired sample t-test). The mean CD4+ cell count at the moment of phenotype conversion was in the low-to-normal range: 0.48 10⁹/L (CI, 0.42 to 0.54).

To investigate to what extent the observed increase in rate of CD4+ cell decline was specifically associated with the emergence of SI variants, we compared these 45 men who had a NSI to SI phenotype conversion with 45 randomly selected cohort men from whom SI isolates were never obtained. These men were pairwise matched for CD4+ cell count and serum HIV-1 p24-antigen status at the moment of phenotype conversion as well as for age and duration of the seropositive period. In the group without SI isolates, an essentially constant rate of CD4+ cell decline was observed during the entire study period (see Figure 4). Before time zero (that is, the moment both groups were matched), the rate of CD4+ cell decline between men developing SI variants and controls was similar (median CD4+ cell decline: 6.0 and 7.2 10⁷/L per year, respectively). In contrast, after time zero the decline of CD4+ cell counts in the group with SI isolates was 2.7 times greater than that of the control group (18.0 compared with 6.6 10⁷/L per year, P < 0.0001, Mann-Whitney test). A comparable difference in rate of CD4+ cell decline after phenotype conversion was observed when the analysis was limited to the 12 men with a defined time of seroconversion and their matched controls (data not shown).

CD4+ Cell Decline before AIDS Diagnosis

In contrast to the group of men with SI isolates, only few members of the control group in the comparative study progressed to AIDS before the end of follow-up (19 of 45 and 7 of 45, respectively). Therefore, this study did not formally exclude the possibility that directly before AIDS the rate of CD4+ cell decline was similar in men with and without SI isolates. We therefore compared 3-monthly CD4+ cell counts of all men in the prospective study who progressed to AIDS during follow-up and of whom the viral phenotype at the time of diagnosis was known (n = 34). Of these men, 19 progressed to AIDS with SI isolates and 15 progressed to AIDS without SI isolates. For this comparison, men were aligned for the moment of AIDS diagnosis (time zero) (Figure 5). Also directly preceding AIDS diagnoses, the rate of CD4+ cell decline in men with SI isolates was greater compared with men without SI isolates (25.0 10⁷/L per year compared with 8.5 10⁷/L per year, P = 0.001, t-test). Accordingly, a difference was observed in absolute CD4+ cell counts at the time of AIDS diagnosis between the two groups (0.08 10⁹/L; CI, 0.05 to 0.12 and 0.25 10⁹/L; CI, 0.15 to 0.41) for the SI and NSI groups, respectively (P = 0.0035, Mann-Whitney test).

Discussion

We analyzed the relation between the appearance of SI variants, the rate of CD4+ cell decline, and the clinical progression of HIV-1 infection. Viral phenotype was determined with a convenient and reliable cocultivation assay suitable for screening of large groups of patients. This binary assay defined two groups of patients, those with and without SI variants, at the start of a prospective study. After 30 months, the probability of progression to AIDS for the group of men with SI isolates at start was 70.8%. In contrast, during the same period, a 15.8% estimated progression was observed for the group of men without SI isolates at the start of the study. Using Cox analysis, the presence of SI variants at entry was the covariate with the greatest relative risk (6.7); it also had the greatest relative risk after statistical adjustment for other risk factors such as decreased CD4+ cell counts and HIV-1 antigenemia (relative risk, 4.1 and 5.2, respectively). Men without SI variants had a 15.4% probability of developing SI variants during a 30-month follow-up period.

Using PCP prophylaxis during the last phase of the prospective study might have biased the results. The proportion of men with SI variants receiving prophylaxis, however, was greater than the proportion of men without SI variants who received prophylaxis (40% compared with 14%). Therefore, it is unlikely that omission of the PCP prophylaxis would have resulted in less of a statistical difference in AIDS incidence between men with and without SI variants.

The analysis of 45 men for whom the moment of phenotype conversion was documented allowed for an accurate estimate of the duration of time between the first observation of SI variants and the development of AIDS. Generally, SI variants were first observed a long time before AIDS diagnosis (median interval, 23.1 months).

Additionally, the appearance of SI variants is prognostic for a rapid decline of CD4+ cell counts. In the period before HIV-1 phenotype conversion, the rates of CD4+ cell decline of the two groups were comparable with matched controls. After phenotype conversion, however, the rate of CD4+ cell depletion was 2.7 times greater than that of the control group.

This difference in rate of CD4+ cell decline in the period before AIDS diagnosis was reflected in statistically lower CD4+ cell counts at AIDS diagnosis for men with SI isolates, compared with AIDS patients with NSI isolates. This result suggests that AIDS patients with SI isolates are more severely immunocompromised than patients without SI isolates at the time of AIDS diagnosis. The risk for developing^{AIDS-associated} symptoms increases with decreasing CD4+ cell counts [32, 33]. The relatively high CD4+ cell count at which participants without SI variants developed AIDS probably is the consequence of the slow rate of CD4+ cell decline in these patients; this puts them at risk, for comparatively longer periods, for contracting opportunistic infections or developing Kaposi sarcoma. In a preliminary study, we obtained evidence for differences in the spectrum of clinical symptoms between patients with and without SI variants at the moment of AIDS diagnosis, which may be related to these differences in rate of CD4+ cell decline. Also, the survival after AIDS diagnoses appeared to be reduced for patients with SI variants [24].

It has been debated whether SI isolates arise as a consequence of the loss of CD4+ cells or precede it, in which case they might be causally involved in CD4+ cell depletion [6, 13, 23, 34, 35]. Syncytium-inducing variants typically emerge in patients with moderately reduced CD4+ cell counts (mean, 0.48 10⁹/L in this study), suggesting that a certain degree of immune dysfunction is required before overt replication of these variants is allowed [8]. On the other hand, emergence of SI variants coincided with a threefold increased rate of CD4+ cell loss, in contrast with a continuous slow CD4+ cell depletion in men with only NSI variants. This suggests that these viruses may be directly or indirectly involved in CD4+ cell depletion. This hypothesis is further supported by our incidental observations in patients with extremely rapid progressive clinical courses [19, 20, 36]. The indication that a viral biological phenotype may be directly linked to AIDS pathogenesis further increases the potential value of viral phenotype as a prognostic marker.

In a preliminary double-blind study of zidovudine treatment of asymptomatic HIV-1-infected patients, we observed a statistically significant delay in clinical progression for patients without SI isolates. In contrast, clinical progression was not significantly delayed in patients with SI isolates (Koot and colleagues. Unpublished observations). This phenotype-associated differential efficacy of zidovudine suggests that virus phenotype should be considered in the evaluation of controlled drug trials. Also, the possibility of using the evolution of viral phenotype as a marker for the efficacy of antiviral therapy should be investigated.