To further substantiate this finding and to examine the biological principle underlying the protection offered by CCR5 delta32 heterozygosity, we analyzed the role of CCR5 genotype alone and in relation to established progression markers in the clinical course of HIV-1 infection in participants from the Amsterdam Cohort Studies.

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample

Between October 1984 and March 1986, 961 asymptomatic men who were living in the Amsterdam area and who reported having had at least two homosexual contacts in the preceding 6 months were enrolled in a prospective study on the prevalence and incidence of HIV-1 infection and risk factors for AIDS [25]. In the first serum sample taken, 728 men tested negative for HIV-1 antibodies; 131 of these men underwent seroconversion during the study. The remaining 238 men were positive for HIV antibodies; 5 of these men refused to participate further. Enrollment of seropositive persons was stopped after 6 months (in April 1985).

Epidemiologic studies on the incidence of HIV-1 infection [26] showed that infection in seroprevalent homosexual men must have occurred an average of 1.5 years before entry into the Amsterdam Cohort Studies. Therefore, the time of seroconversion for seroprevalent men was set at 1.5 years before study entry. No differences in AIDS-free survival were found between persons who underwent seroconversion during the study and seroprevalent persons by using Kaplan-Meier (P > 0.2) and Cox proportional-hazard analyses in which the development of AIDS was the end point criterion (relative hazard, 1.17 for persons who had seroconversion compared with seroprevalent persons [95% CI, 0.84 to 1.63]). This result suggests a good estimation of the seroconversion date in the latter group. When we restricted our analyses to persons who had seroconversion, relative hazards were similar but less precise than estimates for the group as a whole. Therefore, we used 131 persons who had seroconversion and 233 seroprevalent persons as one study sample. Every 3 months, clinical and epidemiologic data were collected and serum and peripheral blood mononuclear cells were cryopreserved.

Most seropositive men (n = 242 [66%]) did not receive early treatment. The remaining 122 men (34%) received zidovudine (70 [19%]), didanosine 10 [3%]), or other antiretroviral therapy (42 [12%]) before AIDS was diagnosed. None of the men received a combination of more than two antiretroviral drugs during our study. The mean age of participants at the time of seroconversion was 34.5 years (range, 19.5 to 57.7 years). By 1 January 1996 (the censor date), 189 men had developed AIDS according to the 1987 definition of AIDS [27] (median follow-up, 5.9 years [range, 0.6 to 12.3 years]), 94 men had not developed AIDS (median follow-up, 10.1 years [range, 0.3 to 13.7 years]), and 81 men were lost to follow-up (median follow-up, 2.0 years [range, 0.6 to 12.5 years]).

A nested case–control study done using the same group of participants from the Amsterdam Cohort Studies was designed to identify factors that may be correlated with long-term survival. Long-term survivors (n = 23) remained free of clinical diseases for at least 9 years, with a mean CD4⁺ T-lymphocyte count of more than 400 cells/mm³ in the eighth and ninth year of HIV-1-positive follow-up (median follow-up, 10.8 years [range, 9.1 to 11.1 years]; mean CD4⁺ T-lymphocyte counts in the ninth year of follow-up, 534 cells/mm³ [range, 408 to 953 cells/mm³]). Each long-term survivor was matched with two progressors (men who developed AIDS after 2 to 7 years of HIV-1-positive follow-up). Matching was based on mean CD4⁺ T-lymphocyte count (± 250 cells/mm³) in year 2 of HIV-positive follow-up, HIV-1 serostatus at entry in the cohort study, and age (± 10 years).

Use of Polymerase Chain Reaction for CCR5 Genotyping

Samples of DNA were available for CCR5 genotyping for 343 of 364 men (94%). Genomic DNA was isolated from cryopreserved peripheral blood mononuclear cells (Qiagen blood kit, Qiagen, Hilden, Germany) and 100 mg of DNA was analyzed by using polymerase chain reaction (PCR) with primers (sense, position 612 to 635 in CCR5, 5'-GATAGGTACCTGGCTGTCGTCCAT-3'; antisense, position 829 to 850 in CCR5, 5'-AGATAGTCATCTTGGGGCTGGT-3') flanking the described 32-nucleotide deletion in the CCR5 gene [17, 18]. Samples were amplified with 1 unit of Taq polymerase (Promega, Madison, Wisconsin) in the provided buffer with a final MgCl₂ concentration of 3 mmol/L. Conditions of PCR comprised 5 minutes of denaturation at 95°C; 30 cycles of 1 minute at 95°C, 1 minute at 56°C, and 2 minutes at 72°C; and 5 minutes of elongation at 72°C in a Perkin Elmer Cetus DNA thermal cycler 480 (Perkin Elmer, Foster City, California). Products of PCR were analyzed by using 2% agarose gel electrophoresis and ethidium bromide staining. Five randomly chosen samples with a reduced product size revealed the described 32-base pair deletion on automatic DNA sequencing (data not shown) [17, 18].

Virologic Assays

Cocultivation of HIV-1-positive peripheral blood mononuclear cells with MT2 cells was performed every 3 months to detect syncytium-inducing HIV-1 variants [28, 29]. Serum viral load was measured by using a quantitative HIV-1 RNA nucleic acid-based sequence amplification (Organon Teknika, Boxtel, the Netherlands) with electrochemiluminescent labeled probes [30]. Serum samples obtained approximately 2 years after seroconversion (1 year after seroconversion; mean time point, 2.3 years [range, 1.5 to 3.0 years]) were available for measurement of HIV-1 RNA viral load for 335 of 364 participants (92%). Serum levels of HIV-1 RNA were analyzed after log¹⁰ transformation. Numbers of RNA copies that were below the test threshold of quantification were arbitrarily set at 10 (³).0 copies/mL.

Immunologic Assays

Antibodies to HIV-1 were detected in serum by using a commercial recombinant HIV-1/-2 enzyme immunoassay (Abbott, Chicago, Illinois) and were confirmed with an HIV-1 Western blot IgG assay (version 1.2, Diagnostic Biotechnology Ltd., Singapore, Thailand). Enumeration of CD4⁺ and CD8⁺ T lymphocytes was done by using flow cytofluorometry. For seroprevalent persons for whom we estimated the time of seroconversion to have been 18 months before entry into the cohort study, CD4⁺ T-lymphocyte count was first measured 18 months after the estimated time of seroconversion. Beginning in January 1988, reactivity of T lymphocytes in response to stimulation with CD3 monoclonal antibodies in vitro was routinely determined in whole-blood cultures [31]. The proliferative response measured after 4 days of culture by incorporation of [³H] thymidine was expressed as a percentage of the median values of the responses measured in two to five healthy controls tested on the same day.

Statistical Analysis

The Fisher exact test was used to compare HIV-1-seronegative participants with HIV-1-seropositive participants for CCR5 genotype distributions. In the case–control study, conditional logistic regression was performed to estimate the chance that a CCR5 delta32 heterozygote would be a long-term survivor. The Mann-Whitney U test was used to compare CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes.

For each participant, the slope of the decrease in CD4⁺ T lymphocytes was determined separately by fitting a simple regression line to his CD4⁺ T-lymphocyte count. At least three CD4⁺ T-lymphocyte counts had to be available for analysis; this was the case for 66 (97%) of the 68 CCR5 delta32 heterozygotes and 250 (91%) of the 275 CCR5 wild-type homozygotes.

A Kaplan-Meier analysis was used to estimate the cumulative incidence of conversion to syncytium-inducing HIV-1 variants in relation to CCR5 genotype. We also estimated the duration of AIDS-free survival in relation to CCR5 genotype for the period during which only non-syncytium-inducing variants were present (conversion to syncytium-inducing HIV-1 was used as a censor criterion) or for the period after conversion to syncytium-inducing HIV-1 variants. A Kaplan-Meier analysis and a Cox proportional-hazards analysis were used to study the predictive value of CCR5 genotype alone or in combination with serum viral RNA load, CD4⁺ T-lymphocyte count, T-lymphocyte function, and syncytium-inducing phenotype. We evaluated the predictive value of the markers by fitting separate Cox models at 2, 4, 6, and 8 years after seroconversion. Participants were at risk from each specific time point; this method excluded participants who had previously developed AIDS. Because data on HIV-1 RNA load were available approximately 2 years after seroconversion only, data on viral load were not included in the models at 4, 6, and 8 years after seroconversion. All markers were also analyzed as time-dependent covariates. Participants who did not have AIDS were censored at 1 January 1996. Significance in survival analyses was determined by the log-rank, Wilcoxon, and likelihood ratio tests.

Optimal cut-off points for serum viral RNA load, CD4⁺ T-lymphocyte counts, and T-lymphocyte reactivity were determined on the basis of the sensitivity and specificity in predicting the development of AIDS by using a receiver-operating characteristic curve [32]. Analysis of the curve allows the choice of a value that maximizes sensitivity in relation to specificity or vice versa. At 2, 4, 6, and 8 years after seroconversion, the receiver-operating characteristic values for CD4⁺ T-lymphocyte counts were less than 500 cells/mm³, less than 400 cells/mm³, less than 400 cells/mm³, and less than 300 cells/mm³, respectively; T-lymphocyte function was less than 60%, less than 40%, less than 30%, and less than 20%, respectively, of that of HIV-negative controls. Two years after seroconversion, the receiver-operating characteristic value for serum RNA load was greater than 10⁴.5 copies/mL. Number Crunching Statistical Systems (NCSS, version 6.0, Klaysville, Utah) was used for statistical analyses.

Results

Top Methods Results Discussion Author & Article Info References

CCR5 Genotype Distribution

In our study sample, 80% of HIV-1 seropositive participants were homozygous for the CCR5 wild-type genotype, and 20% were heterozygous for CCR5 delta32. No HIV-1-infected participant was homozygous for CCR5 delta32. This distribution did not significantly differ from the distribution of CCR5 genotype as described for a random population of white persons from a similar geographic region (Belgium and France) [17].

CCR5 Genotype and Clinical Course of HIV-1 Infection

In the nested case–control study, the proportion of CCR5 delta32 heterozygotes was substantially higher among long-term survivors (48%) than among matched controls (9%). The crude chance of long-term survival was higher among CCR5 delta32 heterozygotes, as determined by conditional logistic regression (odds ratio, 6.9 [CI, 1.9 to 24.8]; P < 0.001).

Kaplan-Meier analysis of all 343 HIV-1-seropositive participants with a known CCR5 genotype showed a highly significant prolonged duration of AIDS-free survival (P < 0.001) in carriers of the heterozygous genotype compared with carriers of the wild-type genotype (Figure 1, left). Univariate Cox proportional-hazard analysis done at the time of seroconversion showed a relative hazard for disease progression of 2.6 (CI, 1.7 to 3.8) for CCR5 wild-type homozygotes (Table 1). Kaplan-Meier analysis also showed a significantly prolonged time to death (P < 0.001) for CCR5 delta32 heterozygotes compared with CCR5 wild-type carriers (Figure 1, right). Heterozygosity for CCR5 delta32 was not associated with prolonged survival after AIDS was diagnosed (data not shown). Censoring participants at the moment when therapy was initiated did not change the outcome of our analysis (data not shown).

View larger version (10K): [in this window] [in a new window]   Figure 1. C-C chemokine receptor 5 (CCR5) genotype and disease progression. Kaplan-Meier plots for time to development of AIDS (left) or death (right) for homozygotes (thin lines) and CCR5 delta32 heterozygotes (thick lines). The numbers of men at risk are indicated for each time point.

CCR5 Genotype and Kinetics of CD4⁺ T-Lymphocyte Counts

A significant difference was seen in the median decrease in CD4⁺ T-lymphocyte counts between CCR5 wild-type homozygotes (decrease of 70 cells/mm³ per year) and CCR5 delta32 heterozygotes (decrease of 51 CD4⁺ cells/mm³ per year) (P = 0.01). Moreover, as shown in Figure 2, the mean CD4⁺ T-lymphocyte count in CCR5 delta32 heterozygotes from 6 months after seroconversion was approximately 100 cells/mm³ higher than that in CCR5 wild-type homozygotes. This effect was consistent over time: Mean CD4⁺ T-lymphocyte counts were 530 cells/mm³ and 420 cells/mm³ 5 years after seroconversion and 340 cells/mm³ and 270 cells/mm³ 10 years after seroconversion for CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes, respectively. The observed difference was not caused by a higher CD4⁺ T-lymphocyte count around the time of seroconversion (3 months after seroconversion, means of 688 cells/mm³ and 684 cells/mm³ for CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes, respectively).

View larger version (25K): [in this window] [in a new window]   Figure 2. C-C chemokine receptor 5 (CCR5) genotype and mean CD4+ T-lymphocyte counts from the time of seroconversion. White circles represent CD4+ T-lymphocyte counts for CCR5 wild-type homozygotes; black circles represent counts for CCR5 delta32 heterozygotes. Error bars represent SEs of the mean.

CCR5 Genotype and Serum Viral Load

We observed previously that from 12 months after seroconversion, participants with HIV-1 RNA levels greater than 10⁴.6 copies/mL had more rapid progression to AIDS than did participants with RNA levels of less than 10⁴.0 copies/mL [2]. Serum viral RNA levels were measured at a mean of 2.3 years (range, 1.5 to 3.0 years) after seroconversion. The median viral load in participants with the CCR5 wild-type genotype was 10 (⁴).3 copies/mL (range, 10³.0 to 10⁶.0 copies/mL); this level was 2.6-fold higher than the median load in CCR5 delta32 heterozygotes (10³.9 copies/mL [range, 10³.0 to 10⁵.6 copies/mL]). This difference was significant (P = 0.01) and graphically evident by a shift to the left in the distribution of viral loads in heterozygotes (Figure 3).

View larger version (15K): [in this window] [in a new window]   Figure 3. C-C chemokine receptor 5 (CCR5) genotype and serum viral load. The probability distributions of serum viremia levels 1.5 to 3.0 years after seroconversion in CCR5 wild-type homozygotes (white bars) and CCR5 delta32 heterozygotes (striped bars) are shown.

Genotype and HIV-1 Biological Phenotype

At the end of the study, no substantial difference was found in the total frequency of syncytium-inducing HIV-1 in CCR5 delta32 heterozygotes (21 of 59 participants [36%]) compared with CCR5 wild-type homozygotes (81 of 219 participants [37%]). To analyze a possible difference in the rate of evolution to syncytium-inducing HIV-1, Kaplan-Meier survival analysis of CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes was performed with emergence of syncytium-inducing HIV-1 as an end point criterion. Although evolution to syncytium-inducing HIV-1 infection tended to be more rapid in CCR5 wild-type homozygotes (25% of conversions occurred 5.4 years after seroconversion) than in CCR5 delta32 heterozygotes (25% of conversions occurred 7.6 years after seroconversion; relative hazard, 1.52 [CI, 0.93 to 2.48]), this difference was not significant (Figure 4, top).

View larger version (16K): [in this window] [in a new window]   Figure 4. C-C chemokine receptor 5 (CCR5) genotype and HIV-1 biological phenotype. Kaplan-Meier plots for CCR5 delta32 heterozygotes (thick lines) and CCR5 wild-type homozygotes (thin lines) for time to emergence of syncytium-inducing HIV-1 (top), time to AIDS for the period during which only non-sancytium-inducing variants were present (middle), and time to AIDS for the period after the emergence of syncytium-inducing HIV-1 (bottom). Curves are cut off when fewer than 10 men are at risk.

Because non-syncytium-inducing viruses can use only CCR5 as a coreceptor, we compared duration of AIDS-free survival in CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes for the period during which only non-syncytium-inducing variants were present. The duration of AIDS-free survival in the presence of only non-syncytium-inducing variants was significantly longer in CCR5 delta32 heterozygotes than in CCR5 wild-type homozygotes (P < 0.001) (Figure 4, middle). This protective effect of CCR5 delta32 heterozygosity in persons with only non-syncytium-inducing variants was stronger (relative hazard, 3.96 [CI, 2.14 to 7.36] when compared with the total group (relative hazard, 2.55 [CI, 1.73 to 3.75]) (Figure 1, left) and with the group that was analyzed for the period after the emergence of syncytium-inducing HIV-1 variants (relative hazard, 2.33 [CI, 1.08 to 5.03]) (Figure 4, bottom). Even in the latter group, however, CCR5 delta32 heterozygosity was still significantly associated with prolonged AIDS-free survival (P = 0.038) (Figure 4, bottom).

Cox Proportional Hazard Analysis

Univariate (Table 1) and multivariate (Table 2) Cox proportional-hazards analyses were done to determine the predictive value of CCR5 genotype compared with HIV-1 phenotype and optimally predictive values of CD4⁺ T-lymphocyte counts, T-lymphocyte function, and viral RNA load. At 2, 4, 6, and 8 years after seroconversion, low CD4⁺ T-lymphocyte counts, presence of syncytium-inducing HIV-1 variants, and CCR5 wild-type homozygosity were predictive for subsequent progression to AIDS. Data on the serum viral RNA load were only available approximately 2 years after seroconversion; high levels were predictive of disease progression. Data on T-lymphocyte function (response to stimulation with CD3 monoclonal antibodies) became predictive at 6 years after seroconversion; this finding may be explained by the few data points in the early phase of the study (the assay was not available until 1988).

Multivariate analysis done at 2 years after seroconversion that excluded T-lymphocyte function as a covariate because of insignificance in univariate analysis indicated that a CD4⁺ T-lymphocyte count less than 500 cells/mm³, a serum viral RNA load greater than 10⁴.5 copies/mL, the presence of syncytium-inducing variants, and CCR5 wild-type genotype were independent predictors for progression to AIDS. Multivariate analysis performed 6 years after seroconversion revealed that T-cell function less than 30% of the function in HIV-1 seronegative controls, presence of syncytium-inducing HIV-1 variants, and CCR5 wild-type genotype were independent predictors of disease progression; data on serum HIV-1 RNA load were not available for this time point. Time-dependent analysis that included CCR5 genotype, CD4⁺ T-lymphocyte counts, and HIV-1 biological phenotype showed these markers to be independent. We did not find significant interaction terms between CCR5 and other markers (data not shown).

Cumulative Incidence of AIDS

Using the Kaplan-Meier product-limit method, we analyzed, at 2 and 6 years after seroconversion, whether clinical progression to AIDS increased with an accumulating number of independent progression markers fulfilled. At year 2, CCR5 genotype, CD4⁺ T-lymphocyte counts less than 500 cells/mm³, and HIV-1 RNA load exceeding 10⁴.3 copies/mL were included; at year 6, CCR5 genotype, T-lymphocyte function less than 30% of the function of HIV-negative controls, and presence of syncytium-inducing HIV-1 were included. In the group that met none of the progression criteria 2 years after seroconversion, as many as 87% of participants remained AIDS-free after a subsequent 4-year follow-up. Any additional progression marker increased the risk for AIDS; the incidence of AIDS in the group with CCR5 wild-type homozygous genotype, low CD4⁺ T-lymphocyte counts, and high levels of serum RNA at 2 years was 44% within the same 4-year period (Figure 5, top). At 6 years after seroconversion, CCR5 delta32 heterozygosity, preserved T-lymphocyte reactivity, and presence of only non-syncytium-inducing variants was associated with disease progression in only 7% of case-patients in the following 2 years. In contrast, progression to AIDS in this same period from year 6 until year 8 after seroconversion occurred in 35% of patients who had CCR5 wild-type genotype, low T-lymphocyte function, and syncytium-inducing HIV-1 variants (Figure 5, bottom).

View larger version (15K): [in this window] [in a new window]   Figure 5. Relation between C-C chemokine receptor 5 (CCR5) genotype and virologic and immunologic prognostic markers for progression to AIDS at 2 years (top) and 6 years after seroconversion (bottom). Kaplan-Meier plots show cumulative progression to AIDS. Participants were stratified according to the number of criteria that were fulfilled (CCR5 wild-type genotype, CD4+ T-lymphocyte count <500 cells/mm3, and viral RNA level >104.5 copies/mL at 2 years after seroconversion; CCR5 wild-type genotype, T-lymphocyte function less than 30% than the function of HIV-negative controls, and syncytium-inducing phenotype at 6 years after seroconversion). The dotted line indicates that none of the 3 criteria was met, the thick line indicates that 1 of the 3 criteria was met, the intermediate line indicates that 2 of the 3 criteria were met, and the thin line indicates that all 3 criteria were met. The numbers of men at risk are indicated for each time point. Curves are cut off when fewer than 10 men are at risk.

Discussion

Top Methods Results Discussion Author & Article Info References

We observed a strong correlation between heterozygosity for CCR5 delta32 and prolonged AIDS-free survival. In 343 HIV-1-seropositive participants from the Amsterdam Cohort Studies, a Cox proportional-hazard analysis on CCR5 delta32 heterozygosity yielded a relative hazard of 0.39 for disease progression. This finding was in agreement with the relative hazards reported by Dean and colleagues [21] (0.61) and Zimmerman and coworkers [24] (0.53) but not the results of a study by Huang and associates [33]. The latter group did not find substantially delayed disease progression in CCR5 delta32 heterozygotes. We found that CCR5 genotype predicted disease progression independently of viral RNA load, CD4⁺ T-lymphocyte counts, T-lymphocyte function, and biological phenotype; further study is warranted on the value of CCR5 genotyping for better estimation of the clinical course of HIV-1 infection.

The biological mechanism by which CCR5 delta32 heterozygosity provides protection from disease progression is not known. Liu and colleagues [18] found reduced in vitro viral replication in cells that were heterozygous for the CCR5 deletion compared with replication in wild-type CCR5 cells. We found a 2.6-fold decrease in serum viral load approximately 2.5 years after seroconversion in CCR5 delta32 heterozygotes compared with CCR5 wild-type homozygotes. This result is similar to those of Huang and associates [33]. Mellors and coworkers [1] described a relative hazard for death of 1.55 for every threefold increase in baseline HIV-1 RNA concentration. This indicates that the 2.6-fold lower RNA levels that we observed in CCR5 delta32 heterozygotes may be relevant to delayed disease progression.

The finding that CCR5 genotype, HIV-1 RNA load, CD4⁺ T-lymphocyte count, and T-lymphocyte function are independent predictors of disease progression does not mean that a pathway other than virus-driven loss of CD4⁺ T lymphocytes is exclusively responsible for progression to AIDS. These results only indicate that at 2 years after seroconversion, CCR5 wild-type homozygotes compared with CCR5 delta32 heterozygotes have a higher risk for more rapid progression to AIDS, even if CD4⁺ T-lymphocyte count and viral load are the same for both groups at that time point [34].

We observed a slower rate of decrease in CD4⁺ T-lymphocyte count in CCR5 delta32 heterozygotes compared with CCR5 wild-type homozygotes. Because we did not consider missing CD4⁺ T-lymphocyte counts caused by attrition, measurement errors, and biological variability [35, 36], our estimates may be biased. However, around the time of seroconversion, no differences were observed between CD4⁺ T-lymphocyte counts in CCR5 wild-type homozygotes and those in CCR5 delta32 heterozygotes. The mean CD4⁺ T-lymphocyte count was consistently higher in CCR5 delta32 heterozygotes from 6 months after seroconversion. A higher immunologic setpoint or CD4⁺ T-lymphocyte count seems to be established in CCR5 delta32 heterozygotes during the initial phase of infection. However, we found no correlation between CCR5 genotype and clinical severity of acute infection (data not shown).

At the end of the study, CCR5 wild-type homozygotes and CCR5 delta32 heterozygotes had the same frequency of syncytium-inducing HIV-1 variants, although conversion to these variants tended to be delayed in CCR5 delta32 heterozygotes. The reduced viral replication in heterozygotes may prolong the time during which the mutations required for conversion to a syncytium-inducing phenotype accumulate and are selected [37-39]. Of interest, Kaplan-Meier analysis for progression to AIDS in CCR5 delta32 heterozygotes and CCR5 wild-type homozygotes who carry only non-syncytium-inducing variants showed that protection from AIDS by CCR5 delta32 heterozygosity is stronger in but is not restricted to these persons. This stronger protection may be explained by the fact that non-syncytium-inducing variants (in contrast to syncytium-inducing variants) seem to be restricted to the use of CC-chemokine receptors [16] (Unpublished data). However, AIDS-free survival in the presence of syncytium-inducing HIV-1 variants was also significantly delayed in CCR5 delta32 heterozygotes. This finding concurs with the observation that primary syncytium-inducing variants can also use CCR5 [16]. In addition, these findings may indicate that non-syncytium-inducing variants that remain present even after the emergence of syncytium-inducing HIV-1 variants may significantly contribute to the pathogenesis of AIDS [40, 41]. In a recent report by Michael and colleagues [23], delayed disease progression in CCR5 heterozygotes was restricted to carriers of non-syncytium-inducing HIV-1. That study sample was much smaller, and a single-virus phenotype determination was used to stratify persons for Kaplan-Meier analysis. As a result, the study by Michael and colleagues included both the period during which only non-syncytium-inducing variants were present in persons who carried syncytium-inducing variants and the period during which syncytium-inducing variants were present in the non-syncytium-inducing curve for persons in whom syncytium-inducing conversion occurred after virus phenotype was determined. This may explain the discrepancy between their results and ours.

Twelve of the 23 long-term survivors in our cohort did not carry CCR5 delta32. Although reports have shown that the clinical course of HIV-1 infection is determined by many factors, we are currently investigating whether other genetic changes cause CCR5 dysfunction in these persons or whether a more benign clinical course is independent of coreceptor function.

Drs. Cornelissen, de Wolf, and Goudsmit and Ms. Bakker: Academic Medical Center, Department of Human Retrovirology, Meibergdreef 15, 1105 AZ Amsterdam, the Netherlands.

Drs. Keet and Coutinho and Ms. Prins: Municipal Health Service Amsterdam, Department of Public Health and Environment, Nieuwe Achtergracht 100, 1018 WT Amsterdam, the Netherlands.