The estimated impact of the CCR-5 Δ32 gene deletion on HIV... : AIDS (original) (raw)
Introduction
The CC-chemokine receptor CCR-5 has been shown to be a coreceptor for entry of HIV-1 into CD4+ cells [1]. Homozygosity for a 32 base-pair deletion (Δ32) in the gene encoding CCR-5 has been associated with resistance to HIV-1 [2]. In addition, being heterozygous for the deletion in CCR-5 (CCR-5 +/Δ32) has been suggested to be a protective factor against being HIV-infected [3].
Several studies on the impact of CCR-5 +/Δ32 on HIV disease progression have been carried out, although no conclusion has yet been reached [3–10].
The differing results may be real or they may be attributable to systematic errors caused by study design [11]. Biases may arise when the cofactor studied, such as CCR-5 genotype, is associated with both the risk of being HIV-infected and progression of disease.
In our study, the impact of CCR-5 +/Δ32 on the progression rate to AIDS and death was estimated with two different study designs: (i) a prospective study of HIV-positive subjects with follow-up time from study entry (prevalent cohort), and (ii) a prospective study of HIV-positive subjects with well-defined time of HIV seroconversion, with follow-up time from the retrospectively assessed date of HIV seroconversion (retrospective incident cohort).
Materials and methods
Study population
Prevalent cohort
The Oslo HIV Cohort Study, a collaborative effort between the three major infectious disease clinics in Oslo, Norway (Ullevål Hospital, Rikshospitalet and Olafiaklinikken) that was started in 1988, included 409 subjects from different HIV exposure categories [12]. Almost all subjects (98%) were recruited before 1992 and represented 70% of all individuals with an HIV-positive diagnosis in Oslo before 1992 [13]. At study entry, the time of HIV seroconversion was generally not known. For 338 subjects (83%), blood cells had been collected. CCR-5 genotype could be obtained for 310 subjects, although information on follow-up was missing for one of these subjects. A total of 278 subjects were AIDS-free at study entry.
Retrospective incident cohort
For 105 of the 309 subjects in the prevalent cohort, the time of HIV seroconversion could be assessed retrospectively on the basis of HIV antibody analyses of stored serum samples from within a 2-year interval (n = 65), or information in medical records on clinical and serological signs of acute HIV infection (n = 40). The mean time from estimated seroconversion to inclusion in the Oslo HIV Cohort Study was 3.5 years (range, 0.1–11 years).
Clinical follow-up
The physicians responsible for clinical follow-up were also responsible for the reporting of AIDS. In addition, linkage to the National AIDS Registry [13] and the Central Person Registry [14] was performed after the end of the follow-up periods (1 January 1995 for the prevalent cohort, 1 January 1997 for the retrospective incident cohort). The AIDS definition was the prevailing European AIDS definition at the time of diagnosis [15].
Characterization of CCR-5
The individual CCR-5 genotypes were established by PCR using the primers 5′-CAG ATC TCA AAA AGA AGG TCT TCA and 5′-GAC CAG CCC CAA GAT GAC TA. Products of 112 and 80 base pairs were taken to reflect wild-type and Δ32 genotypes, respectively. Heterozygotes had both bands.
Statistical analysis and follow-up times
In the prevalent cohort, the impact of CCR-5 +/Δ32 on progression to AIDS and death from all causes was studied with stratified Kaplan-Meier analysis, and the differences tested using log-rank test. In addition, a multivariate Cox regression model was applied, controlling for gender, age and exposure category. The follow-up time was from the date of entry into the Oslo HIV Cohort Study until AIDS, death or censoring.
In the retrospective incident cohort, the follow-up time was from the estimated date of seroconversion until AIDS, death or censoring. Hence, subjects with known time of seroconversion contributed with follow-up time from the date of study inclusion in the prevalent cohort and from the estimated date of HIV seroconversion in the retrospective incident cohort. In the retrospective incident cohort, progression rate to AIDS and death was studied with survival methods with staggered entry time, when applying both Kaplan-Meier analyses and Cox regression models [16]. Survival analysis with staggered entry is a method of controlling for a potential bias associated with selection of subjects still alive and AIDS-free at study entry.
Results
Progression to AIDS and death in the prevalent cohort
A total of 35% (98 out of 278) of AIDS-free subjects at study inclusion in the prevalent cohort developed AIDS during a mean follow-up period of 41 months (SD, 25): 26% (14 out of 53) of subjects with CCR-5 +/Δ32 and 37% (84 out of 225) of subjects with CCR-5 +/+ (P = 0.07, log-rank test). The adjusted relative risk of CCR-5 +/Δ32 on progression to AIDS was 0.6 (95% confidence interval, 0.3–1.1), when controlling for age at study entry, gender and HIV exposure group (Table 1).
. Crude and adjusted relative risk (RR) with 95% confidence intervals (CI) of progression to AIDS for CCR-5 genotype, age, gender and HIV exposure group in a prevalent cohort and in a retrospective incident cohort.
Amongst the 98 subjects who developed AIDS, 75 (77%) died during a mean follow-up time of 16.7 months (SD, 13) after AIDS diagnosis. There was no significant difference in AIDS survival according to CCR-5 genotype (P = 0.89, log-rank test).
When studying the progression rate to death from any cause, all 309 subjects with obtainable CCR-5 genotype and follow-up time were included. A total of 22% (69 out of 309) had the CCR-5 +/Δ32 genotype; 121 subjects (39%) died during a mean follow-up period of 50 months (SD, 19), including 16 pre-AIDS deaths; 21% (12 out of 57) with CCR-5 +/Δ32 and 43% (109 out of 252) with CCR-5 +/+ died during follow-up (P < 0.01, log-rank test).
Progression to AIDS and death in the retrospective incident cohort
A total of 40% (42 out of 105) of subjects in the retro-spective incident cohort developed AIDS during a mean follow-up of 113 months (SD, 39): 33% (six out of 18) of subjects with CCR-5 +/Δ32 and 41% (36 out of 87) of subjects with CCR-5 +/+ (P = 0.96, log-rank test). In the Cox regression model with staggered entry time, and controlling for HIV exposure group, age at seroconversion and gender, the estimated relative risk of CCR-5 +/Δ32 on progression to AIDS was 1.0 (95% confidence interval, 0.4–2.5; Table 1).
A total of 37% (39 out of 105) of subjects died during a mean follow-up of 121 months (SD, 34). There was no significant difference in the progression rate to death from any cause between subjects with CCR-5 +/Δ32 (six out of 18) and those with CCR-5 +/+ (33 out of 87; P = 0.78, log-rank test).
Discussion
In this study, CCR-5 +/Δ32 was associated with slow progression to AIDS and death in the prevalent cohort, but not in the retrospective incident cohort. The differing findings regarding the impact of CCR-5 +/Δ32 genotype within this study and between previous studies may be explained by systematic errors imposed by the choice of HIV-infected patients to study.
If CCR-5 +/Δ32 is associated with decreased risk of being HIV-infected, subjects with this genotype have, on average, been HIV-infected for a shorter time when the risk populations become saturated with HIV infection. When these HIV-infected subjects are included in a prevalent cohort, a larger proportion of subjects with CCR-5 +/+ will develop AIDS during follow-up, because of the increasing AIDS risk by time (onset confounding) [11,17].
In our study and in other prevalent cohort studies, the estimated slower progression to AIDS in subjects with CCR-5 +/Δ32 may be explained by onset confounding [5,9,10]. In the Chicago Multicentre AIDS Cohort Study, statistically significant association between CCR- 5 genotype and progression to AIDS or death could not be reached either in the prevalent cohort or among seroconversion cases [3]. The direction of the estimates, however, was the same as in our study, with a protective impact of CCR-5 +/Δ32 in the prevalent cohort, but no impact in the seroconverter cohort. When the time of HIV seroconversion is known and used in the analysis, as in true incident cohorts or in cohorts with retrospectively assessed time of HIV seroconversion, onset confounding is eliminated as a source of error.
Slow HIV disease progressors are overrepresented in prevalent cohorts, mainly because the participants will have lived long enough to reach study inclusion. If CCR-5 +/Δ32 is associated with slow progression to AIDS, subjects with this genotype are likely to be over-represented in prevalent cohorts. An overrepresentation of subjects with the CCR-5 +/Δ32 genotype leads to an underestimated difference between the two CCR-5 genotype groups in prevalent cohorts [11]. In addition, in retrospective incident cohorts, selection of slow progressors occurs, unless the time from HIV seroconversion to study entry is short. In true incident cohorts no such bias occurs.
In the study by Dean et al. [4], which included sero-converters from several US HIV cohort studies, CCR- 5 +/Δ32 was significantly associated with slow progression to AIDS. However, their report gave limited information on the assessment of HIV seroconversion times. Considering the length of follow-up, it is likely that, for a large proportion of subjects, the time of HIV seroconversion was prior to study entry. If so, the estimated protective impact of CCR-5 +/Δ32 in this study may be underestimated due to selection of slow progressors [11]. Other selection biases in inclusion may also have occurred by including only subjects with long-term data available. The direction and magnitude of these potential biases was not discussed in the report. Retrospective information on HIV seroconversion may not be obtainable at random. For instance, subjects with acute HIV infection may be overrepresented. In our study, acute HIV infection was not associated with CCR-5 genotype or AIDS progression, and was thus not a confounder. However, other factors associated with both obtainable information on HIV seroconversion and disease progression may exist.
Different findings regarding the impact of CCR-5 genotype on the risk of HIV infection and disease progression may be real and caused by factors such as variability in the requirement of HIV for CCR-5 [18]. However, biases associated with study design and selection of study population may be significant, as illustrated in our study.
Acknowledgements
The authors thank Mette Sannes (Ullevål Hospital) and Astrid Vik (Olaflaklinikken) for valuable help in data collection, and Hilde Sorhaug for excellent technical assistance.
References
1. Deng H, Lui R, Ellmeier W, et al.: Identification of a major coreceptor for primary isolates of HIV-1. Nature 1996, 381:661–666.
2. Dragic T, Litwin V, Allaway GP, et al.: HIV-1 entry into the CD4+ cells is mediated by the chemokine receptor CC-CKR5. Nature 1996, 381:667–673.
3. Huang Y, Paxton W, Wolinsky S, et al.: The role of a mutant CCR5 allele in HIV-1 transmission and disease progression. Nat Med 1996, 2:1240–1243.
4. Dean M, Carrington, Winkler C, et al.: Genetic restriction of HIV-1 infection and the progression to AIDS by a deletion allele of the CCR5 structural gene. Science 1996, 273:1856–1862.
5. Eugen-Olsen J, Iversen AKN, Garred P, et al.: Heterozygosity for a deletion in the CKR5 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.
6. Garred P, Eugen-Olsen J, Iversen AKN, et al.: Dual effect of CCR5 Δ32 gene deletion in HIV-1-infected patients. Copenhagen AIDS Study Group [letter]. Lancet 1997, 349:1884.
7. Smith MW, Dean M, Carrington M, Hutteley GA, O'Brien SJ: CCR5-Δ32 gene deletion in HIV-1 infected patients [letter]. Lancet 1997, 350:741.
8. Stewart GJ, Ashton LJ, Biti RA, et al.: Increased frequency of CCR-5 Δ32 heterozygotes among long-term non-progressors with HIV-1 infection. AIDS 1997, 11:1833–1838.
9. Michael NL, Chang G, Louie LG, et al.: The role of viral phenotype and the CCR-5 gene defects in HIV-1 transmission and disease progression. Nat Med 1997, 3:338–340.
10. Garred P: Chemokine-receptor polymorphisms: clarity or confusion for HIV-1 prognosis? Lancet 1998, 351:2–3.
11. Brookmeyer R, Gail MH, Polk F: The prevalent cohort study and the acquired immunodeficiency syndrome. Am J Epidemiol 1987, 126:14–24.
12. Eskild A, Magnus P, Brekke T, et al.: The impact of exposure group on the progression rate to acquired immunodeficiency syndrome. Scand J Infect Dis 1997, 29:103–109.
13. Nilsen Ø, Hasseltvedt V, Lystad A: Number of HIV and AIDS Cases in Norway [in Norwegian]. Notification System for Infectious Diseases in Norway (MSIS-Rapport). Oslo: National Institute of Public Health; 1993.
14. Central Bureau of Statistics: Causes of Death (Health Statistics). Oslo: Central Bureau of Statistics; 1995.
15. Ancelle-Park R: Expanded European AIDS case definition. Lancet 1993, 341:441.
16. Cox DF, Oakes D: Analysis of Survival Data. London: Chapman and Hall; 1984:177–178.
17. Alcabes P, Munoz A, Vlahov D, Friedland GH: Incubation period of human immunodeficiency virus. Epidemiol Rev 1993, 15:303–318.
18. Doms RW, Peiper SC: Unwelcomed guests with master keys: how HIV uses chemokine receptors for cellular entry. Virology 1997, 235:179–190.
Appendix
Members of the Oslo HIV Cohort Study Group
Leiv S. Bakketeig (Group Leader, Department of Population Health Sciences, National Institute of Public Health), Johan N. Bruun (Department of Infectious Diseases, Ullevål University Hospital), Stig S. Frøland (Department A of Internal Medicine, Rikshospitalet), Harald Moi (Olaflaklinikken, Ullevål University Hospital), Torbjørn Berntsen (PLUSS, the National Organisation of HIV-positives), Helvi Holm Samdal (Department of Virology, National, Institute of Public Health) and Martinus Løvik (Department of Environmental Medicine, National Institute of Public Health).
Keywords:
HIV; AIDS; risk factors, CCR-5 genotype; progression; epidemiology; cohort
© 1998 Lippincott Williams & Wilkins, Inc.