Secondary Logo

Share this article on:

Effects of CCR532 and CCR2-64I alleles on HIV-1 disease progression: the protection varies with duration of infection

Mulherin, Stephanie Aa,b; O'Brien, Thomas Ra; Ioannidis, John PAa,c; Goedert, James Ja; Buchbinder, Susan Pd; Coutinho, Roel Ae; Jamieson, Beth Df; Meyer, Laurenceg; Michael, Nelson Lh; Pantaleo, Giuseppei; Rizzardi, G Paoloi; Schuitemaker, Hannekej; Sheppard, Haynes Wk; Theodorou, Ioannis Dl; Vlahov, Davidm; Rosenberg, Philip Safor the International Meta-Analysis of HIV Host Genetics


Objective: To examine temporal variation in the effects of CCR5-Δ32 and CCR2-64I chemokine receptor gene polymorphisms on HIV-1 disease progression.

Design: Pooled analysis of individual patient data from 10 cohorts of HIV-1 seroconverters from the United States, Europe, and Australia.

Methods: We studied HIV-1 seroconverters of European (n = 1635) or African (n = 215) ancestry who had been genotyped for CCR5-Δ32 and CCR2-64I. We used Cox proportional hazards models with time-varying coefficients to determine whether the genetic protection against AIDS (1987 case definition) and death varied with time since seroconversion.

Results: Protection against AIDS conferred by CCR5-Δ32 held constant at a 31% (RH 0.69, 95% CI 0.54, 0.88) reduction in risk over the course of HIV-1 infection, whereas protection against death held constant at a 39% reduction in risk (RH 0.61, 95% CI 0.45, 0.88). When the period from AIDS to death was isolated, the survival benefit of CCR5-Δ32 diminished 2 years after AIDS. Protection against AIDS conferred by CCR2-64I was greatest early in the disease course. Compared with individuals without CCR5-Δ32 or CCR2-64I, individuals with one or two copies of CCR2-64I had a 58% lower risk of AIDS during the first 4 years after seroconversion (RH 0.42, 95% CI 0.23, 0.76), a 19% lower risk during the subsequent 4 years (RH 0.81, 95% CI 0.59, 1.12), and no significant protection thereafter.

Conclusion: The protection against AIDS provided by CCR5-Δ32 is continuous during the course of infection. In contrast, the protection provided by CCR2-64I is greatest early in the course of infection.

From the aDivision of Cancer Epidemiology and Genetics, NCI, NIH, DHHS, Rockville, MD 20852, USA; bDepartment of Epidemiology, University of North Carolina School of Public Health, Chapel Hill, NC 27599, USA; cDepartment of Hygiene and Epidemiology, University of Ioannina School of Medicine and Ioannina Biomedical Research Institute, Ioannina 45110, Greece; dHIV Research Section, San Francisco Department of Public Health, San Francisco, CA 94102-6033, USA; eMunicipal Health Service Amsterdam, Department of Public Health and Environment, Amsterdam, the Netherlands; fDepartment of Medicine, Division of Hematology/Oncology, University of California at Los Angeles School of Medicine, Los Angeles, CA 90095-1745, USA; gService d'Epidemiologie – INSERM U292, Hopital de Bicetre, 94276 Le Kremlin-Bicetre Cedex, France; hDivision of Retrovirology, Walter Reed Army Institute of Research, Rockville, MD 20850, USA; iLaboratory of AIDS Immunopathogenesis, Centre Hospitalier Universitaire Vaudois, 1011 Lausanne, Switzerland; jDepartment of Clinical Viro-Immunology, Sanquin Research at CLB and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands; kViral and Rickettsial Disease Laboratory, California Department of Health Services, Berkeley, CA 94704, USA; lLaboratoire Central d'Immunologic Cellulaire et Tissulaire, Hopital Pitie Salpetriere et CNRS UMR 7627, 75013 Paris, France; mCenter for Urban Epidemiologic Studies, New York Academy of Medicine, New York, NY 10029.

Correspondence and requests for reprints to: Philip S. Rosenberg, Ph.D., Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd, Room 7006, Rockville MD 20852. Email: tel: 301-435-3996; fax: 301-402-0081.

Received: 10 May 2002; revised: 20 September 2002; accepted: 8 October 2002.

Back to Top | Article Outline


HIV-1-infected individuals who carry CCR532 or CCR2-64I chemokine receptor gene polymorphisms progress significantly more slowly to clinical AIDS and death. Many studies have investigated these polymorphisms in individual cohorts [1–23]. A recent meta-analysis of these cohorts found that in the absence of potent antiretroviral therapy, both CCR532 and CCR2-64I carriers progressed to AIDS at a 25% slower rate than individuals who lacked either of these protective alleles [24]. They also progressed more slowly to death, by approximately 35% and 25%, respectively.

CCR532 is a deletion mutation that renders the CCR5 cell surface receptor non-functional. CCR5 is the major co-receptor for the R5 HIV-1 variants (non-syncytium-inducing strains) that dominate in early infection. The deficit of HIV-1 co-receptors appears to impede viral replication in vitro [14,25,26]; and circulating levels of HIV-1 RNA in serum or plasma are lower during early chronic infection among CCR5-Δ32 heterozygotes than in individuals with wild-type CCR5 [7,24]. In contrast, the CCR2-64I single nucleotide polymorphism codes for a conservative substitution and a mechanism for its effect remains to be elucidated. CCR2-64I is quite common in individuals of African and European descent, suggesting that it is a relatively old polymorphism that has been conserved over time [18]. In our population, the CCR2-64I allele occurs with a frequency of 0.10 and 0.14 in individuals of European and African descent, respectively. The CCR5-Δ32 allele has a frequency of 0.09 in individuals of European descent.

Most studies have assumed that these gene variants restrict the rate of HIV-1 disease progression by a constant factor over time. However, Meyer et al. [7] reported that the survival benefit for CCR5-Δ32 was limited to the first 7 years after seroconversion among subjects in the SEROCO cohort. A time-restricted effect was also observed in the Copenhagen seroprevalent cohort [23]. In the Copenhagen cohort, CCR5-Δ32 carriers actually had poorer survival after the onset of AIDS [27]. Time-dependency was also reported for a CCR5 promoter polymorphism; disease progression was faster for CCR5-P1 homozygotes, but only for the first 5 years after seroconversion [28].

Characterizing the temporal restriction of a gene's effect may shed light on its underlying biological mechanisms. However, the ‘time-averaged’ effects of these chemokine genes are relatively subtle, and individual HIV-1 cohort studies have limited power to dissect variation in the effects of these genes over time. In this paper, we analyse time-trends in the effects of the CCR5-Δ32 and CCR2-64I polymorphisms using the largest available database, the International Meta-Analysis of HIV Host Genetics [24]. By pooling information from many studies, we increased our ability to characterize temporal patterns.

Back to Top | Article Outline


Data were contributed by 19 groups of investigators from the United States, Europe, and Australia [24]. Only cohorts of HIV-1 seroconverters were included in this analysis, provided that the subjects had been genotyped for both CCR5-Δ32 and CCR2-64I and prospectively followed from seroconversion to AIDS (1987 definition) and death. These cohorts have typically enrolled and genotyped consecutive HIV-1 seroconverters, and thus there is no strong selection bias in favor of slow or rapid progressors.

Because time-averaged results between seroconverter cohorts were not significantly heterogeneous in the meta-analysis, we studied all of the seroconverters who had been included in that report [24], plus 18 additional individuals from five cohorts who were ineligible to be in the previous analysis because they belonged to subgroups with fewer than 20 individuals. We also included 80 additional individuals from the Multicenter Hemophilia Cohort Study who did not previously have genotype results. As in the previous analysis, we censored follow-up after 1 January 1996, to minimize the effects of potent antiretroviral therapy. In all, these participants contributed an additional 846.3 person-years of follow-up to AIDS, 38 AIDS events, 889.7 person-years of follow-up to death, and 43 deaths to the analysis. In total, we analysed 13 514.5 person-years of follow-up, 619 AIDS events and 535 deaths (Appendix 1).

The CCR5 and CCR2 chemokine receptor genes are tightly linked on chromosome 3p21–22. Therefore, we analysed compound CCR5-CCR2 genotypes to compare individuals who carried CCR5-Δ32 or CCR2-64I with individuals who were homozygous for the wild-type allele of both genes (compound wild-type homozygotes). CCR5-Δ32 homozygotes are highly resistant to infection and none were included in our analysis. CCR5-Δ32 is restricted to individuals of European descent, whereas CCR2-64I is prevalent among individuals of both European and African ancestry. Therefore, our analysis of CCR5-Δ32 was limited to individuals of European ancestry, whereas our analysis of CCR2-64I included individuals of either European or African ancestry. All models were adjusted for age at seroconversion, and stratified by cohort. CCR2-64I models were further stratified by racial ancestry. Neither CD4 T-cell count nor HIV-RNA load were included in our analyses, because these markers may be on the causal pathway through which these genetic variants affect the likelihood of AIDS or death.

The standard Cox proportional hazards model assumes that the effect of a covariate on the hazard rate for an event is constant over time. In our analysis, the covariate is compound CCR5-CCR2 genotype and the outcome events are AIDS and death. This standard model provides a ‘time-averaged’ gene effect, and its statistical significance can be assessed using the likelihood ratio test. However, we hypothesized that CCR5-Δ32 or CCR2-64I may not confer a constant protective effect over time, but instead may provide more protection early in the course of infection. To characterize potential time-variation in the effects of CCR5-Δ32 and CCR2-64I, we relaxed the proportional hazards assumption using several prespecified models of gene-by-time interactions. These interaction models were based on categorical, cubic polynomial, and cubic spline functions of time (technical details are presented in Appendix 2).

For our primary analyses, a priori we tested the significance of categorical gene-by-time interactions based on 4-year time intervals (0–4, 4–8, 8–12, and 12+ years since seroconversion). However, when the number of events, or follow-up time, was sparse, we merged the first or last time intervals, respectively, with immediately adjacent intervals. In addition, previous studies reported that the protective effect of CCR5-Δ32 is substantially diminished 7 years after infection [7]. Therefore, we also fitted a categorical gene-by-time interaction model with a single cutpoint at 7 years. Finally, in an exploratory analysis, we examined models that allow the relative hazard (RH) to vary continuously over time using cubic polynomial or cubic regression splines. However, for formal hypothesis tests, we relied on the categorical models, because these provide a limited number of easily interpretable parameters that characterize the RH over the entire time course of the disease.

We also fitted categorical gene-by-time interaction models to study survival after diagnosis of AIDS. For this period, we fitted a time-invariant model and a categorical interaction model with a single cutpoint 2 years after AIDS diagnosis.

We selected our final models using the following procedure. If none of the interaction models was statistically significant at the α = 0.05 level, then the standard time-invariant model was selected. If one or more gene-by-time interaction model fit significantly better, then we selected the most parsimonious model using the Akaike Information Criterion (AIC) value (Appendix 2).

Finally, we contrasted the rate ratios for CCR5-Δ32 and CCR2-64I using a direct approach based on a Poisson regression model of the log rates. This method tabulated the numbers of events and person-years of follow-up by cohort, genotype, and quadrennial periods since seroconversion; contrasting rate ratios were compared using a Wald test.

Back to Top | Article Outline


Categorical gene-by-time interaction models

CCR532 heterozygotes progressed significantly more slowly to AIDS than compound wild-type homozygotes (Wald chi-squared, P = 0.003) and death (Wald chi-squared, P < 0.001) based on the standard time-invariant Cox model. Moreover, no significant time-dependency was observed for the effect of CCR5-Δ32 on the risk of AIDS or death, regardless of the model used (Table 1). CCR5-Δ32 heterozygotes had approximately a 30% lower risk of AIDS and a 40% lower risk of death than compound wild-type homozygotes. None of the time-trends were statistically significant, although a trend towards subtle time-dependency was observed for the model with a single cutpoint at 7 years (Table 1), consistent with the original observation in the SEROCO cohort.

Table 1

Table 1

The CCR532 effect showed significant time-dependency when the period from AIDS diagnosis to death was isolated (Table 1). Compared with compound wild-type homozygotes, CCR5-Δ32 heterozygotes had a statistically significant 43% lower risk of death during the 2 years after an AIDS diagnosis, but no significant protection thereafter.

Carriers of CCR2-64I were also protected against AIDS (Wald chi-squared, P = 0.032) and death (Wald chi-squared, P = 0.034) based on the standard time-invariant Cox model. Significant time-dependency was observed for the effect of CCR2-64I on the risk of AIDS (Table 2). However, the time-dependency was not significant for death or the period from AIDS to death. Compared with compound wild-type homozygotes, individuals with the CCR2-64I allele (homozygotes and heterozygotes combined) had a 58% lower risk of AIDS in the first 4 years after seroconversion. This declined to a 19% lower risk of AIDS during the subsequent 4 years, and no significant protection thereafter. Contrasting the two protective polymorphisms using a Poisson regression model of the AIDS hazard rate by cohort, quadrennial time period, and genotype showed that CCR2-64I carriers had a significantly lower rate of AIDS than CCR5-Δ32 heterozygotes during the first 4 years after seroconversion (Wald chi-squared, P = 0.04).

Table 2

Table 2

Back to Top | Article Outline

Smooth gene-by-time interaction models

Effect of CCR5-Δ32 on AIDS incubation period and survival

Candidate gene-by-time interaction models are summarized in Table 3. To obtain a smooth estimate of the CCR5-Δ32 effect over time, we fitted a cubic polynomial gene-by-time interaction model and five cubic regression spline models with knots at time 3 and 6 years, 4 and 8 years, and 5, 6, and 7 years. The cubic polynomial model had a superior AIC value to any of the regression splines, but it did not fit significantly better than the time-invariant model. A closer examination of the time-specific RH of AIDS estimated by the cubic polynomial model demonstrated that CCR5-Δ32 heterozygotes are significantly protected from AIDS from approximately 2 to 5 years post-seroconversion (Fig. 1a). The RH for this period vary from 0.55 (95% CI 0.30, 0.99) to 0.71 (95% CI 0.51, 0.99). Estimates of the CCR5-Δ32 effect during the first 2 years post seroconversion are imprecise because of the paucity of AIDS cases among CCR5-Δ32 heterozygotes in this time period (2 events).

Fig. 1.

Fig. 1.

Table 3

Table 3

Similarly, the effect of CCR5-Δ32 on survival did not demonstrate significant variation over time when seven cubic regression spline models with knots at time 3 and 8 years, 3 and 9 years, 4 and 10 years, and 5, 7, 8, and 10 years were evaluated. However, the graph of the cubic polynomial model demonstrates a protective benefit for CCR5-Δ32 from approximately 3.3 to 8.2 years post-seroconversion, and a modest but not significant protective effect thereafter (Fig. 1b). The RH for the former period ranged from 0.49 (95% CI 0.31, 0.77) to 0.69 (95% CI 0.48, 0.99), indicating a 30–50% reduction in the risk of death. Infrequent fatalities among CCR5-Δ32 heterozygotes before 4 years post-seroconversion resulted in extremely wide 95% confidence intervals during this period of follow-up.

Back to Top | Article Outline

Effect of CCR2-64I on AIDS incubation period and survival

CCR2-64I showed a strong, time-restricted effect in both categorical and smooth gene-by-time interaction models. The models consistently indicated an early protective effect for this polymorphism. Cubic polynomial and cubic spline gene-by-time interaction models with knots at 3 and 6 years, 4 and 8 years, and 4, 5, 6, and 7 years were investigated. Although none of the smoothed gene-by-time models reached statistical significance at the α = 0.05 level, the cubic polynomial model, and hazard spline model with knots at 4 and 8 years, performed better than the time-invariant model based on the AIC. CCR2-64I provided substantial protection from AIDS during the first 5 years after seroconversion. The risk of AIDS was reduced by approximately 80% in the second year post-seroconversion, 50–70% in the third year, 40–50% in the fourth year, and 30–40% in the fifth year post-seroconversion. (Fig. 1c).

CCR2-64I also had a time-restricted effect on survival. Smoothed estimates were calculated using cubic polynomial and cubic spline gene-by-time interaction models with knots at 3 and 8 years, 3 and 9 years, 4 and 10 years, and 5, 7, 8, and 10 years. A cubic spline with a knot at 8 years performed significantly better than the time-invariant model, and had the best AIC value (Fig. 1d). The effect of CCR2-64I on survival was seen later in the disease course than the effect on AIDS risk. Early estimates are imprecise because of the paucity of deaths among CCR2-64I carriers before 4 years post-seroconversion, and estimates beyond 10 years are also imprecise.

Back to Top | Article Outline


It has frequently been assumed that the inherited resistance against progression to AIDS and death conferred by the CCR5-Δ32 and CCR2-64I genetic polymorphisms is constant over the course of the disease. In reality, the relationships may be more complex. The large amount of data available from the International Meta-Analysis of HIV Host Genetics [24] allowed us to fit flexible gene-by-time interaction models for the effects of CCR5-Δ32 and CCR2-64I. We implemented a vigorous and systematic approach to model selection to avoid over-interpreting results. The findings of time-dependence were largely consistent, regardless of the modeling approach being employed. The models that best characterize the effects of these polymorphisms are presented in Table 4.

Table 4

Table 4

We observed no significant time-dependency for the effect of CCR5-Δ32 on progression from seroconversion to AIDS or from seroconversion to death. A previous report found a suggestion of time-dependency in the CCR5-Δ32 effect on progression to AIDS, but it did not reach statistical significance [7]. Our data suggest that any temporal effects of the CCR5-Δ32 polymorphism on progression to AIDS are likely to be modest. However, we did observe significant time-dependency for the period from AIDS diagnosis to death. CCR5-Δ32 heterozygotes remain protected from death for approximately 2 years after an AIDS diagnosis, but thereafter have no protection, and possibly an increased risk. By 2 years after AIDS, the surviving group may be enriched with individuals whose circulating virus can use the CXCR4 co-receptor [26]. However, X4 viruses often emerge before AIDS [29,30]; and even after they emerge, X4 viruses may not be the only strains circulating in an individual [1]. Therefore, this explanation for the waning CCR5-Δ32 effect requires further study.

In contrast to CCR5-Δ32, the effect of the CCR2-64I polymorphism on progression to AIDS varied over time. During the first 4 years after seroconversion, the protective effect of this variant is larger than previously recognized; moreover, carriers of CCR2-64I are significantly more protected from an AIDS event than carriers of CCR5-Δ32 during this period. Our previous meta-analysis of this database showed that in a ‘time-averaged’ sense, CCR2-64I is associated with a 24% lower risk of AIDS. However, using time-dependent models, we found CCR2-64I carriers had approximately a 60% lower risk of AIDS in the first 4 years after seroconversion, a 20% lower risk of AIDS during the subsequent 4 years, and no significant protection thereafter.

A mechanism for the CCR2-64I effect remains to be elucidated. As CCR2 is a minor co-receptor for HIV-1, it is unlikely that the profound reduction in the risk of AIDS that we observed in the first 4 years after seroconversion directly reflects the role of CCR2 as a viral co-receptor. Instead, a mechanism involving CCR5 or CXCR4, the major HIV-1 co-receptors, is more likely. It has been proposed that the CCR2-64I variant may be in linkage disequilibrium with a functional mutation in CCR5 that travels on the same haplotype [5], or that the protein produced by the CCR2-64I variant may interact with CCR5 or CXCR4 [31,32] to decrease the expression of these co-receptors. Although the 64I variant has not been associated with altered baseline expression rates of CCR5 [31,33], it has recently been reported that the CD4 T-lymphocytes in individuals who carry the CCR2-64I variant may be slower to re-express CCR5 after internalization, which could result in less CCR5 available on the cell surface [34]. Other studies have shown that CCR2-64I carriers develop X4 HIV-1 strains more rapidly than wild-type individuals [35–37], and that the CCR2-64I protection is lost after X4 strains emerge [36]. Putting these data together with our findings of an early effect for CCR2-64I, one could speculate that the allele may provide early protection through an interaction with CCR5, but that this protection is lost after X4 strains emerge in these patients. Further epidemiological studies of CCR2-64I and viral phenotype in early chronic infection are needed to examine this hypothesis.

The potential limitations of our study must be considered. First, not all subjects in the cohorts were genotyped. Our results might be subject to bias if the clinical course and genotype frequency differ in those who were not genotyped. However, the most plausible direction of this potential bias strengthens our main conclusions because rapid progressors may be less likely to have samples available for genotyping (because of shorter survival) and more likely to be wild-type for the CCR2-64I and CCR5-Δ32 alleles (because of the observed protective effects of the mutant alleles). Second, although our statistical methods are sensitive and our database is the largest available, the genetic effects are subtle. As a consequence, our estimates are somewhat imprecise, as reflected by rather broad confidence intervals. Finally, survival in some cohorts might be better than others because of differences in access to healthcare or socioeconomic status. As allele frequencies also differ by cohort (Appendix 1), one might be concerned that our results could be affected by confounding. Fortunately, the cohort-stratified survival analysis we used protects us from this potential bias: it controls very closely for the effects of cohort on disease progression by allowing each cohort to have a different baseline hazard function.

The course of HIV-1 infection is governed by multiple genetic factors, which may operate at different time intervals. Our finding that the effect of CCR2-64I is limited to early infection may shed light on the mechanism by which this allele improves prognosis in HIV-1 infection, which is heretofore unexplained. It also suggests that similar analyses for other candidate genetic polymorphisms that are being proposed as regulators of the rate of HIV-1 disease progression may be of interest. Large-scale analyses, similar to this pooled analysis of individual patient data, would be required to probe such time-dependent associations.

Sponsorship: This work was supported by National Institutes of Health, National Research Service Award (5 T32 A107001) administered by the University of North Carolina, Chapel Hill, Division of Infectious Diseases, and by the Cancer Research Training Award, National Cancer Institute, Division of Cancer Epidemiology and Genetics, Viral Epidemiology Branch.

Back to Top | Article Outline

References Roda Husman AM, Koot M, Cornelissen M, Keet IP, Brouwer M, Broersen SM, et al. Association between CCR5 genotype and the clinical course of HIV-1 infection. Ann Intern Med 1997, 127:882–890.
2.Dean M, Carrington M, Winkler C, Huttley GA, Smith MW, Allikmets R, et al. Genetic restriction of HIV-1 infection and progression to AIDS by a deletion allele of the CKR5 structural gene. Hemophilia Growth and Development Study, Multicenter AIDS Cohort Study, Multicenter Hemophilia Cohort Study, San Francisco City Cohort, ALIVE Study. Science 1996, 273: 1856–1862.
3.Hendel H, Henon N, Lebuanec H, Lachgar A, Poncelet H, Caillat-Zucman S, et al. Distinctive effects of CCR5, CCR2, and SDF1 genetic polymorphisms in AIDS progression. J Acquir Immune Defic Syndr Hum Retrovirol 1998, 19:381–386.
4.Katzenstein TL, Eugen-Olsen J, Hofmann B, Benfield T, Pedersen C, Iversen AK, et al. HIV-infected individuals with the CCR delta32/CCR5 genotype have lower HIV RNA levels and higher CD4 cell counts in the early years of the infection than do patients with the wild type. Copenhagen AIDS Cohort Study Group. J Acquir Immune Defic Syndr Hum Retrovirol 1997, 16:10–14.
5.Kostrikis LG, Huang Y, Moore JP, Wolinsky SM, Zhang L, Guo Y, et al. A chemokine receptor CCR2 allele delays HIV-1 disease progression and is associated with a CCR5 promoter mutation. Nat Med 1998, 4:350–353.
6.Liu R, Paxton WA, Choe S, Ceradini D, Martin SR, Horuk R, et al. Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection. Cell 1996, 86:367–377.
7.Meyer L, Magierowska M, Hubert JB, Rouzioux C, Deveau C, Sanson F, et al. Early protective effect of CCR-5 delta 32 heterozygosity on HIV-1 disease progression: relationship with viral load. The SEROCO Study Group. AIDS 1997, 11:F73–F78.
8.Michael NL, Chang G, Louie LG, Mascola JR, Dondero D, Birx DL, et al. The role of viral phenotype and CCR-5 gene defects in HIV-1 transmission and disease progression. Nat Med 1997, 3:338–340.
9.Michael NL, Louie LG, Rohrbaugh AL, Schultz KA, Dayhoff DE, Wang CE, et al. The role of CCR5 and CCR2 polymorphisms in HIV-1 transmission and disease progression. Nat Med 1997, 3:1160–1162.
10.Morawetz RA, Rizzardi GP, Glauser D, Rutschmann O, Hirschel B, Perrin L, et al. Genetic polymorphism of CCR5 gene and HIV disease: the heterozygous (CCR5/delta ccr5) genotype is neither essential nor sufficient for protection against disease progression. Swiss HIV Cohort. Eur J Immunol 1997, 27:3223–3227.
11.Mummidi S, Ahuja SS, Gonzalez E, Anderson SA, Santiago EN, Stephan KT, et al. Genealogy of the CCR5 locus and chemokine system gene variants associated with altered rates of HIV-1 disease progression. Nat Med 1998, 4:786–793.
12.O'Brien TR, Winkler C, Dean M, Nelson JA, Carrington M, Michael NL, et al. HIV-1 infection in a man homozygous for CCR5 delta 32. Lancet 1997, 349:1219.
13.Rizzardi GP, Morawetz RA, Vicenzi E, Ghezzi S, Poli G, Lazzarin A, et al. CCR2 polymorphism and HIV disease. Swiss HIV Cohort. Nat Med 1998, 4:252–253.
14.Samson M, Libert F, Doranz BJ, Rucker J, Liesnard C, Farber CM, et al. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature 1996, 382:722–725.
15.Schinkel J, Langendam MW, Coutinho RA, Krol A, Brouwer M, Schuitemaker H. No evidence for an effect of the CCR5 delta32/+ and CCR2b 64I/+ mutations on human immunodeficiency virus (HIV)-1 disease progression among HIV-1-infected injecting drug users. J Infect Dis 1999, 179:825–831.
16.Smith MW, Carrington M, Winkler C, Lomb D, Dean M, Huttley G, et al. CCR2 chemokine receptor and AIDS progression. Nat Med 1997, 3:1052–1053.
16.Smith MW, Dean M, Carrington M, Huttley GA, O'Brien SJ. CCR5-delta 32 gene deletion in HIV-1 infected patients. Lancet 1997, 350:741; discussion 742.
18.Smith MW, Dean M, Carrington M, Winkler C, Huttley GA, Lomb DA, et al. Contrasting genetic influence of CCR2 and CCR5 variants on HIV-1 infection and disease progression. Hemophilia Growth and Development Study (HGDS), Multicenter AIDS Cohort Study (MACS), Multicenter Hemophilia Cohort Study (MHCS), San Francisco City Cohort (SFCC), ALIVE Study. Science 1997, 277:959–965.
19.Stewart GJ, Ashton LJ, Biti RA, Ffrench RA, Bennetts BH, Newcombe NR, et al. Increased frequency of CCR-5 delta 32 heterozygotes among long-term non-progressors with HIV-1 infection. The Australian Long-Term Non-Progressor Study Group. AIDS 1997, 11:1833–1838.
20.Wilkinson DA, Operskalski EA, Busch MP, Mosley JW, Koup RA. A 32-bp deletion within the CCR5 locus protects against transmission of parenterally acquired human immunodeficiency virus but does not affect progression to AIDS-defining illness. J Infect Dis 1998, 178:1163–1166.
21.Zimmerman PA, Buckler-White A, Alkhatib G, Spalding T, Kubofcik J, Combadiere C, et al. Inherited resistance to HIV-1 conferred by an inactivating mutation in CC chemokine receptor 5: studies in populations with contrasting clinical phenotypes, defined racial background, and quantified risk. Mol Med 1997, 3:23–36.
22.Eugen-Olsen J, Iversen AK, Benfield TL, Koppelhus U, Garred P. Chemokine receptor CCR2b 64I polymorphism and its relation to CD4 T-cell counts and disease progression in a Danish cohort of HIV-infected individuals. Copenhagen AIDS Cohort. J Acquir Immune Defic Syndr Hum Retrovirol 1998, 18:110–116.
23.Eugen-Olsen J, Iversen AK, Garred P, Koppelhus U, Pedersen C, Benfield TL, et al. Heterozygosity for a deletion in the CKR-5 gene leads to prolonged AIDS-free survival and slower CD4 T-cell decline in a cohort of HIV-seropositive individuals. AIDS 1997, 11:305–310.
24.Ioannidis JP, Rosenberg PS, Goedert JJ, Ashton LJ, Benfield TL, Buchbinder SP, et al. Effects of CCR5-Delta32, CCR2-64I, and SDF-1 3'A alleles on HIV-1 disease progression: an international meta-analysis of individual-patient data. Ann Intern Med 2001, 135:782–795.
25.Benkirane M, Jin DY, Chun RF. Dating the origin of the CCR5-Delta 32 AIDS-resistance allele by the coalescence of haplotypes. Am J Hum Genet 1998, 62:1507–1515.
26.Berger EA, Murphy PM, Farber JM. Chemokine receptors as HIV-1 coreceptors: roles in viral entry, tropism, and disease. Annu Rev Immunol 1999, 17:657–700.
27.Garred P, Eugen-Olsen J, Iversen AK, Benfield TL, Svejgaard A, Hofmann B. Dual effect of CCR5 delta 32 gene deletion in HIV-1-infected patients. Copenhagen AIDS Study Group. Lancet 1997, 349:1884.
28.Martin MP, Dean M, Smith MW, Winkler C, Gerrard B, Michael NL, et al. Genetic acceleration of AIDS progression by a promoter variant of CCR5. Science 1998, 282:1907–1911.
29.Koot M, Keet IP, Vos AH, de Goede RE, Roos MT, Coutinho RA, et al. Prognostic value of HIV-1 syncytium-inducing phenotype for rate of CD4+ cell depletion and progression to AIDS. Ann Intern Med 1993, 118:681–688.
30.Shankarappa R, Margolick JB, Gange SJ, Rodrigo AG, Upchurch D, Farzadegan H, et al. Consistent viral evolutionary changes associated with the progression of human immunodeficiency virus type 1 infection. J Virol 1999, 73:10489–10502.
31.Lee B, Doranz BJ, Rana S, Yi Y, Mellado M, Frade JM, et al. Influence of the CCR2-V64I polymorphism on human immunodeficiency virus type 1 coreceptor activity and on chemokine receptor function of CCR2b, CCR3, CCR5, and CXCR4. J Virol 1998, 72:7450–7458.
32.Mellado M, Rodriguez-Frade JM, Vila-Coro AJ, de Ana AM, Martinez-AC. Chemokine control of HIV-1 infection. Nature 1999, 400:723–724.
33.Mariani R, Wong S, Mulder LC, Wilkinson DA, Reinhart AL, LaRosa G, et al. CCR2-64I polymorphism is not associated with altered CCR5 expression or coreceptor function. J Virol 1999, 73:2450–2459.
34.Sabbe R, Picchio GR, Pastore C, Chaloin O, Hartley O, Offord R, et al. Donor- and ligand-dependent differences in C-C chemokine receptor 5 reexpression. J Virol 2001, 75:661–671.
35.Nakayama EE, Hoshino Y, Xin X, Liu H, Goto M, Watanabe N, et al. Polymorphism in the interleukin-4 promoter affects acquisition of human immunodeficiency virus type 1 syncytium-inducing phenotype. J Virol 2000, 74:5452–5459.
36.van Rij RP, de Roda Husman AM, Brouwer M, Goudsmit J, Coutinho RA, Schuitemaker H. Role of CCR2 genotype in the clinical course of syncytium-inducing (SI) or non-SI human immunodeficiency virus type 1 infection and in the time to conversion to SI virus variants. J Infect Dis 1998, 178: 1806–1811.
37.Vicenzi E, Ghezzi S, Brambilla A, Sheppard HW, Lazzarin A, Poli G, et al. CCR2-64I polymorphism, syncytium-inducing human immunodeficiency virus strains, and disease progression. J Infect Dis 2000, 182:1579–1580.
38.Akaike H. Information theory and an extension of the maximum likelihood principle. In: Proceedings of the Second International Symposium on Information Theory. Petrov BN, Csaki F (editors). Budapest: Akademia Kiado; 1973. pp. 267–281.
    Back to Top | Article Outline

    Appendix 1

    The following table describes the cohorts included in this pooled analysis of HIV-1 seroconverters. Two subjects of African descent in the AIDS Link to the Intravenous Experience cohort who had been included in the previous meta-analysis [24] were excluded here because they were missing CCR5-Δ32 genotype data. The analysis excludes seroprevalent cohorts because the dates of seroconversion are unknown, and as a consequence, it would be difficult to interpret time trends in the effects of a gene.

    Back to Top | Article Outline

    Appendix 2

    Let t be the years-since-seroconversion, I{.} be the indicator function equal to 1 if a logical condition in brackets is true, 0 otherwise, and (x)+ = max(0,x) be the ‘truncation’ function. Let GENE = `+’ if wild-type, ‘−’ if variant, AGE be the age-at-seroconversion, ANCESTRY = 1 if African, 0 if European and h c 0(t) be an unspecified baseline hazard function that is specific for cohort c. The baseline hazard function specifies the event rate (of AIDS or death) that will occur in an infinitesimal time interval [t,t + Δ) among susceptible individuals in cohort c who are still at risk at time t and who have baseline covariate values for GENE, AGE, and ANCESTRY. (Models for CCR2-64I include ANCESTRY whereas models for CCR5-Δ32 do not.)

    The standard time-invariant Cox proportional hazards model is

    This model implies that the log relative hazard (log RH(t)) contrasting individuals with the variant allele with individuals with the wild-type allele is constant over time, log RH(t) = β.

    An extended Cox model with a categorical time interaction at 5 years is

    This model implies that

    An extended Cox model with a cubic polynomial time interaction is

    which implies that

    The following equation describes an extended Cox model with a gene-by-time interaction specified using a cubic regression spline with knots at 4 and 8 years:

    For this model, log RH(t) is a cubic regression spline with knots at 4 and 8 years:

    This equation specifies three cubic polynomials for the intervals (0,4], (4,8], and (8,t max); the pieces join together smoothly so that the function and its first two derivatives are continuous.

    Each of the extended Cox models contains the time-invariant model as a special case; statistical significance of parameters describing gene-by-time interactions is assessed using the standard likelihood ratio test. In the models described above, there are 1, 3, and 5 degrees of freedom, respectively, for interaction.

    In our analyses of CCR5-Δ32 and CCR2-64I, we fitted several categorical- and cubic-spline-based models of gene-by-time interaction in order to determine which model provides the best description of the genes’ effects. Model selection was performed using a two-step procedure. First, we ruled out interaction models that were not statistically significant at the α = 0.05 level using the likelihood ratio test. Second, if more than one model was significant, we selected the model with the lowest value of the AIC [38]. The AIC value is defined as −2 times the log likelihood ratio of the model, plus two times the number of degrees of freedom. AIC is widely used to balance goodness-of-fit and parsimony.


    AIDS; alleles; CCR5; CCR2; chemokine receptors; HIV-1

    © 2003 Lippincott Williams & Wilkins, Inc.