HIV-1 causes progressive immune deterioration and ultimately clinically manifests as AIDS in humans. But as with all etiologic agents, HIV must replicate in the milieu of a given host to produce disease, and it is the balance between the organism's virulence and the resistance of the host that determines the course and severity of disease. Accordingly, certain intrinsic host characteristics, such as older age and genetic makeup, may be true risk factors for rate of progression to AIDS or death. Other markers that change quickly after infection, such as CD4+ cell count, neopterin and β2-microglobulin levels, are consequences rather than determinants of disease progression. Early HIV-1 plasma viral load is both a consequence of events at inception of the infection and closely related causally to the rate of future disease progression . The number of viral particles in plasma is proportional to the total number of productively infected CD4+ lymphocytes in a given person , which in turn is related to the ability of the particular host to contain the infection.
Recent reports [2–4] on viral dynamics and the relationship between viral load and rate of progression  also show that plasma viral load is in the causal pathway for AIDS, leading to the quantitative loss of CD4+ lymphocytes. In addition, estimating the rate of progression of HIV-1 infection is most reliably accomplished by measuring the incubation time from HIV-1 seroconversion to AIDS or death.
Although replication of HIV-1 produces disease, host factors appear to strongly influence these rates of replication in different persons, in that they account for a wide range of incubation periods observed for AIDS . Kaslow et al.  showed that combinations of markers in the HLA region predict time to AIDS, and Pantaleo et al.  have focused on host-viral interaction during very early HIV-1 infection as a prime determinant of subsequent course of disease progression. Fig. 1 summarizes a likely sequence in the pathogenetic order of risk variables and outcomes that prompted this study. In this report, we corroborate that HLA profile is highly associated with rate of progression and suggest that the products of these genes exert their effect early in HIV-1 infection through the pathway that establishes the virus concentration (i.e., plasma viral load). This identifies at least one mechanism whereby events within 6–12 months following sero-conversion during HIV-1 replication determine the rate of progression.
The Multicenter AIDS Cohort Study (MACS), initiated in 1984, longitudinally evaluates the natural history of HIV-1 infection in homosexual men . Occurrence of AIDS and death in this cohort are continuously updated. AIDS is defined here by the 1987 surveillance definition . Participants are monitored for HIV infection and CD4 cell count every 6 months. Of the 3141 initially HIV-seronegative entrants, 487 seroconverted prior to May 1996. Of these 487 participants, 139 white men were selected for HLA typing on the basis of rate of disease progression and length of follow-up. By 17 November 1993, 71 men had developed AIDS and 68 had not. For this analysis, CD4+ cell count and plasma viral load were measured a median of 9 months (interquartile range, 7.9–9.7 months) after seroconversion, the midpoint between the last seronegative and first seropositive visit.
Kaslow et al.  identified combinations of HLA markers in these 139 participants that conferred either protective or adverse effects on the rate of HIV disease progression. An HLA scoring profile was developed by means of a detailed analysis using Cox proportional hazards models  to compute the relative hazards of AIDS following HIV-1 seroconversion in men carrying a given marker compared with all men not carrying that marker. The HLA scoring profile effectively discriminated 71 AIDS cases from the remaining 68 participants who were AIDS-free as of 17 November 1993. Table 1 shows the value (−1, 0, or +1) assigned to each marker combination. For each individual who carried any of those markers shown, HLA score profile was calculated as the algebraic sum of the values given to those markers. The predictive capacity of the score profile was independently confirmed in two different cohorts of homosexual men [6,11].
This analysis extends the follow-up of 139 participants to 21 February 1996. Taking advantage of the additional follow-up time, from 13 November 1993 to 21 February 1996, we corroborated the ability of the HLA scoring profile to predict subsequent AIDS amongst the 68 participants, referred to as the confirming subset, who were AIDS-free at the time of the initial analysis when the scoring profile was developed. Twenty-six of the 68 persons developed AIDS during the extended follow-up time period; the capability of the HLA scoring profile to discriminate these 26 AIDS cases from the other 42 would, therefore, not be vulnerable to a potential circular bias.
As described previously , class I typing was performed serologically by conventional microcytotoxicity assay and reagents. Molecular amplification methods were used on lymphocyte DNA for class II alleles (DRB1, DQA1, DQB1) and for polymorphic segments in the genes encoding transporter proteins associated with antigen processing (TAP1 and TAP2). Lymphocytes were phenotypically analyzed by two-color flow cytometry of either stained specimens of Ficoll-Hypaque-separated peripheral blood mononuclear cells or EDTA-anticoagulated whole blood stained with fluorescein- and phycoerythrin-conjugated antibodies, as previously described [12,13]. HIV-1 RNA (plasma viral load) was quantified in heparinized plasma samples stored at 70°C. A sensitive (version 2) branched DNA assay (Chiron Corporation, Emeryville, California, USA) was used, which has a lower quantification limit of 500 HIV-1 RNA copies/ml . Samples were tested from duplicate 1.0 ml specimens after viral particles were pelleted by centrifugation at 23 500 g for 1 h at 4°C.
Cox proportional hazard models were fit to risk variables with time to AIDS and duration of survival from incident HIV-1 infection as outcomes . Such variables included CD4+ lymphocyte count at the time HIV-1 viral load was measured, age at HIV-1 sero-conversion and HLA scoring profile in five categories (≤ −2, −1, 0, +1, ≥ +2), which represent the algebraic sum of positive and negative integers assigned to HLA markers associated with slower and more rapid progression of infection, respectively . For numerical stability, only three HLA scoring profile categories (< 0, n = 50; 0, n = 54; > 0, n = 35) were summarized in the Kaplan-Meier plots. In order to maintain comparability of the Kaplan-Meier plots, the CD4+ cell count and HIV-1 RNA viral load data were trichotomized into similarly sized groups as these three HLA scoring profile groups.
Linear regression was used to study the association of HLA scoring profile with viral load and CD4+ cell count, the association of viral load with CD4+ cell count, and the association of CD4+ cell count with viral load. Viral load data were log10-transformed. CD4+ cell count was analyzed in increments of 100 cells (×106/l).
In order to reassure ourselves that the observed associations did not simply reflect HLA-AIDS relationships that were purposefully maximized by our original analytic strategy , analyses were replicated both excluding and including the 71 cases of AIDS that occurred before 1994 on whose discrimination the HLA scoring profile was originally constructed.
The group of 139 men who were HLA typed were representative of all seroconverters as of 1995 in important risk factors for progression (Table 2). The confirming subset who remained AIDS-free until 1995 (n = 68), of whom 26 subsequently developed AIDS, had higher HLA scoring profiles, higher CD4+ cell counts, and lower viral load than those who developed AIDS earlier. The overall median follow-up of the 139 participants was 9.5 years and for the confirming subset of 68 was 10.7 years. Table 3 compares the 71 AIDS cases who were used to develop the HLA scoring profile with the 26 who subsequently developed AIDS and with the 42 who remained AIDS-free. The expected trend of baseline and early markers of infection was observed with the original 71 AIDS cases having the worst values and the 42 AIDS-free having the best. The HLA scoring profile was significantly lower in the 26 subsequent AIDS cases than in the 42 AIDS-free group (P = 0.05), reconfirming that the HLA scoring profile has predictive value for AIDS that occurred after its initial development.
Stratified Kaplan-Meier graphs (Fig. 2) show the relationships of the HLA scoring profile, HIV-1 RNA plasma viral load and CD4+ cell count to time to AIDS in the 139 seroconverters. Median times to AIDS were 5.2, 7.8 and 10.7 years for the three HLA scoring profile levels (< 0, 0 and > 0; Fig. 2A). The median times to AIDS in the graphs for HIV-1 RNA categories (Fig. 2B) and CD4+ cell count categories (Fig. 2C) were virtually identical to those of the HLA scoring profile. The median times to death by HLA scoring profile score category were 6.9, 9.8 and 11.4 years.
Cox proportional hazard models for the entire 139 participants are shown in Table 4 for risk of AIDS. HLA scoring profile, but not age, was significantly associated with the time to AIDS in both univariate and multivariate analyses. The relative hazard of AIDS for each incremental change in HLA scoring profile was 0.64 in the multivariate model, which included age. A proportional hazards model was also fit using the subset of 68 who were AIDS-free at the time the HLA scoring profile was developed. Within this subset, the relative hazard for time to AIDS per unit change (e.g., −1 to 0) in HLA scoring profile was 0.69 (P = 0.05) in the univariate model, only somewhat higher than the relative hazard of 0.52 for the entire group of 139. Early post-seroconversion risk factors, plasma viral load and CD4+ cell count, were both significantly associated with time to AIDS. The impact of these variables on time to death was similar (data not shown).
In both univariate and multivariate linear regression models (Table 5), the HLA scoring profile and CD4+ cell count, but not age, were statistically associated with HIV-1 RNA plasma viral load. Similar findings were made using non-parametric Spearman correlations. Table 5 shows in the multivariate model that for each unit increment in the HLA scoring profile the plasma viral load was 39% lower. There was a significant association between CD4+ cell count and plasma viral load. For every 100 cell (× 106/l) increment in CD4+ cell count, viral load was 13% lower after controlling for age and HLA scoring profile. HLA scoring profile, plasma viral load and age were also studied in univariate and multivariate linear regression models where CD4+ cell count, in increments of 100, was the dependent variable. For every unit increase in the HLA scoring profile in a multivariate model, the CD4+ cell count was higher by 58 × 106 cells/l (0.58 of the 100 cell increment; P = 0.03). For each log10 increase in the plasma viral load, the CD4+ cell count was lower by 116 × 106 cells/l (1.16 × 100 cell increment; P = 0.002). However, age had no effect on early CD4+ cell count.
HIV-1 replication is clearly an important determinant of disease progression in HIV-1 infection . Its strength as a predictor of subsequent disease progression probably derives from the CD4+ lymphocyte infection burden in a given host [2–4]. It is therefore instructive to consider the host factors that may be involved in determining the HIV-1 RNA viral load level shortly after HIV-1 seroconversion. Fig. 1 depicts a causal pathway where immunogenetic makeup, which dictates the immune response to HIV-1 through a number of potential mechanisms, is a primary factor. Correlations between HLA scoring profile, early plasma viral load, CD4+ cell count and time to AIDS or death support both an underlying genetic influence on HIV disease progression and an effect that takes place relatively soon after seroconversion. Such genetically mediated mechanisms controlling HIV-1 replication probably include antigen processing characteristics, β-chemokine production, expression and phenotype of chemokine receptors and endogenous cytokine production, amongst others [15–17]. The specific sequence of immunologic events through which the HLA genes represented in the HLA scoring profile act are still unclear and need further study. In addition, we have not considered HIV-1-specific factors related to either virulence or attenuation that may independently contribute to the rate of disease progression.
HLA scoring profile and viral load were stronger predictors of time to AIDS by the Wald χ2 statistic (Table 4) than CD4+ cell count, all of which were also independently significantly associated with time to AIDS. The strong associations of HLA scoring profile with viral load and CD4 cell counts at approximately 9 months after seroconversion may provide useful quantitative relationships (Table 5). For instance, multivariate linear regression allows for quantitative associations to be made between HLA scoring profile and viral load and also on the subsequent associations with CD4+ cell count. We noted that for every HLA scoring profile unit increase, the viral load was 0.39 lower than the value of HLA scoring profile immediately preceding it. These relationships may ultimately lead to understanding the relative contributions of host factors that affect HIV-1 replication and viral load, especially when specific allele combinations or chemokine receptor expression are found to be responsible for the bulk of the difference in viral load. The mechanism by which HLA scoring profile may mediate its effect is probably through cytotoxic T-lymphocyte and natural killer cellular responses [18,19]. Early data indicate that the chemokine receptor/ligand systems (encoded on different chromosomes and operating through a different mechanism) influences viral load and outcomes independently of HLA effects .
Age is clearly associated with shorter duration in time to AIDS in many populations, including the MACS [6,21], but in contrast to HLA scoring profile, age was not associated with early plasma viral load and CD4+ cell count in the MACS. While a type II error is possible, this finding suggests that the effect of age may not occur early after seroconversion, but may accumulate slowly over time or may be mediated through senescence of an immune system that is relatively unable to replenish itself, as opposed to a direct effect through early plasma viral load.
As previously noted by Kaslow et al. , some qualification of HLA scoring profile validity is needed because it was originally constructed in this cohort. However, this HLA scoring profile has been independently confirmed in other cohorts [6,11]. The validity of the HLA scoring profile is also demonstrated in Table 3, which shows that HLA scoring profile, viral load and CD4+ cell count maintained a relationship with each other that was consistent with their respective association with rate of progression to AIDS. In addition, HLA scoring profile has continued to predict onset of AIDS in this group beyond the original date of its construction both in Table 4 and in restricted analyses excluding the 71 AIDS cases occurring prior to 1994. The univariate relative hazard for the HLA scoring profile effect on AIDS was 0.69 in the confirming subset analysis, only somewhat higher than the relative hazard of 0.52 for the entire group of 139 participants. It is probably arguable that the HLA scoring profile had an ‘advantage’ over the other risk variables in the analysis of the entire group because of the way it was developed. However, the confirmatory subset of 68 participants is important because it supports the contention that circular bias did not play a significant role here and indicates the robust nature of the HLA scoring profile beyond the original separation of the 71 initial AIDS cases from the 68 others.
The HLA scoring profile predicted risk of HIV-1 disease progression dramatically and independently of HIV-1 RNA viral load and CD4+ cell count. If these findings are confirmed, HLA alleles or a refined HLA scoring profile may become useful clinically to judge a patient's need for antiretroviral treatment. Such information would be used in conjunction with viral load and CD4+ cell count, particularly among patients with early stage HIV-1 infection. Identifying the highly relevant HLA markers would make this task simpler and reduce costs. Given that molecular amplification technologies are similar, we would expect the cost of measuring relevant HLA markers to be comparable to that of measuring of HIV-1 RNA plasma viral load, and measuring HLA markers would need to be performed only once.
1. Mellors JW, Rinaldo CR Jr, Gupta P, White RM, Todd JA, Kingsley LA: Prognosis in HIV-1 infection predicted by the quantity of virus in plasma
2. Ho DD, Neumann AU, Perelson AS, Chen W, Leonard JM, Markowitz M: Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection
3. Wei X, Ghosh KS, Taylor ME, et al.
: Viral dynamics in human immunodeficiency virus type 1 infection
4. Perelson AS, Neumann AU, Markowitz M, Leonard JM, Ho DD: HIV-1 dynamicsin vivo
: virion clearance rate, infected cell life-span and viral generation time
5. Veugelers PJ, Page KF, Tindall P, et al.
: Determinants of HIV disease progression among homosexual men in the tricontinental seroconverter study
. Am J Epidemiol
6. Kaslow RA, Carrington M, Apple R, et al.
: Influence of combinations of human major histocompatibility complex genes on the course of HIV-1 infection
. Nat Med
7. Pantaleo G, Demarest JF, Schacker T, et al.
: The qualitative nature of the primary immune response to HIV infection is a prognosticator of disease progression independent of the initial level of plasma viremia
. Proc Natl Acad Sci U S A
8. Kaslow RA, Ostrow DG, Detels R, et al.
: The Multicenter AIDS Cohort Study: rationale, organization and selected characteristics of the participants
. Am J Epidemiol
9. Centers for Disease Control: Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome
10. SAS Institute: PHREG, SAS Release 6.09
. Cary: SAS Institute, Inc.; Year?.
11. Keet IPM, Tang J, Klein M, et al.
: Consistent associations between HLA and time to AIDS in three prospective seroconverters cohorts
. Fourth Conference on Retroviruses and Opportunistic Infections
. Washington, DC, January 1997 [abstract LB16].
12. Giorgi JV, Cheng HL, Margolick JB, et al.
: Quality control in the flow cytometric measurement of T-lymphocyte subsets: the multicenter AIDS cohort study experience
. Clin Immunol Immunopathol
13. Hoffman RA, Kung PC, Hansen WP, Goldstein G: Simple and rapid measurement of human T lymphocytes and their subclasses in peripheral blood
. Proc Natl Acad Sci USA
14. Kern D, Collins M, Fultz T, et al.
: An enhanced sensitivity branched DNA assay for quantification of human immuno-deficiency virus, type 1 RNA in plasma
. J Clin Microbiol
15. Cocchi F, DeVico A, Garzino-Demo A, Arya SK, Gallo RC, Lusso P: Identification of RANTES, MIP-1
βas the major HIV-suppressive factors produced by CD8+ T cells
16. Fauci AS: Host factors and the pathogenesis of HIV-induced disease
17. Moore JP: Coreceptors: implications for HIV pathogenesis and therapy
18. Walter JB, Brander C, Mammen M, et al.
: Stimulation of human cytotoxic T cells with HIV-1-derived peptides presented by recombinant HLA-A2 peptide complexes
. Int Immunol
19. Goulder PJ, Bunce M, Krausa P, et al.
: Novel, cross-restricted, conserved, and immunodominant cytotoxic T lymphocyte epitopes in slow progressors in HIV type 1 infection
. AIDS Res Hum Retroviruses
20. Kaslow R, Koup R, Zimmerman P, et al.
: HLA scoring profile (HSP) and CCR5 deletion heterozygosity as predictors of AIDS in seroconverters
. Fourth Conference on Retroviruses and Opportunistic Infections
. Washington, DC, January 1997 [abstract 22].
21. Hoover DR, Saah AJ, Guccione M, et al.
: Observed HIV-1 disease progression times in gay men with access to treatments
. XI International Conference on AIDS.
Vancouver, July 1996 [abstract TuC435].
The Multicenter AIDS Cohort Study
A. Saah, H. Armenian, H. Farzadegan, D. Hoover, N. Kass, C. Lyketsos, J. Margolick, J. McArthur, E. Taylor, S. Weng (The Johns Hopkins School of Hygiene and Public Health).
J. Phair, J. Chmiel, B. Cohen, M. O'Gorman, D. Variakojis, J. Wesch, S. Wolinsky (Howard Brown Health Center and Northwestern University Medical School).
R. Detels, B. Visscher, J. Fahey, J. Giorgi, A. Kaplan, M. Lee, O. Martinez-Maza, E. Miller, H. Morgenstern, P. Nishanian, J. Oishi, J. Taylor (University of California, UCLA Schools of Public Health and Medicine).
C. Rinaldo Jr, R. Anderson, J. Becker, R. Day, P. Gupta, M. Ho, L. Kingsley, J. Mellors, O. Ndimbe, T. Silvestre (University of Pittsburgh Graduate School of Public Health).
Data Coordinating Center:
A. Muñoz, C. Enger, S. Gange, L. Jacobson, R. Lyles, S. Piantadosi, S. Su (The Johns Hopkins School of Hygiene and Public Health).
National Institutes of Health:
L. Schrager (National Institutes of Allergy and Infectious Diseases), S. Melnick (National Cancer Institute).
Keywords:© 1998 Lippincott Williams & Wilkins, Inc.
AIDS; HIV-1; natural history; HLA; HIV-1 RNA plasma viral load