Q J Med 1999; 92: 609-617
© 1999 Association of Physicians
Commentary papers |
Study requirements for investigating HLA-associated progression of HIV disease, and review
From the MRC Biostatistics Unit, Cambridge, and 1 MRC-BIAS and 2 Regional Infectious Diseases Unit, Edinburgh, UK
Dr S.M. Gore, MRC Biostatistics Unit, Robinson Way, Cambridge CB2 2SR
Introduction
Human Leukocyte Antigens (HLA), also known as transplantation antigens, determine the immune response. We give a brief account of them, and of more complex immunology, to explain why HLA phenotype may be specifically important in HIV disease, as it is in other infectious diseases.1 Study requirements for investigating HLA-associated progression of HIV disease are illustrated with Scottish data; susceptibility to HIV infection is considered only briefly.
Systematic review of published and other diverse data on three HLA associations with HIV progression is attempted, and the difficulties illustrated, for: the ancestral phenotype HLA-A1, B8, DR3 (which is associated with a range of autoimmune diseases); the allele B35 (which is common in Africans, and has been implicated empirically in HIV progression); and B27 (because Ohno2 established a theoretical basis for its association with slow HIV progression, which McNeil et al.3 corroborated).
The systematic review suggests that major HLA effects on HIV progression are equivalent to at least 20 additional years of age. We also present a partial validation of Kaslow's HLA profile score4 which models intermediate effects also, but conclude that more rigorous analysis still requires an international multi-cohort analysis workshop to be convened to map HLA effects on HIV disease progression.5,6
Human leukocyte antigens (HLA)
The most studied genes of the major histocompatibility complex (MHC) encode for the polymorphic Human Leukocyte Antigens (HLA). These are molecules on the surface of nucleated cells which represent the major mechanism by which the T lymphocyte immune system recognizes what is `self' from `non-self'. There are three classes of MHC genes: class I, encoding HLA-A, B and C, which are found on all nucleated cells; class II, encoding HLA-DR, DP and DQ, which are located on immune cells, namely: B lymphocytes, activated T lymphocytes and macrophages; and class III, of which more later, encoding the complement system and tumour necrosis factor (TNF).
HLA, often called the transplantation antigens, determine the binding and presentation of peptide antigens to T lymphocytes. Through natural selection and otherwise, linkage disequilibrium has evolved such that HLA and other genes in the major histocompatibility complex occur together non-randomly. There is thus geographical and ethnic variation in individual gene frequency, and also in so-called ancestral haplotypes, two of which are: A1, B8 ... DR3 ... DQ2 and A2, B35 ... DR5 ... DQ1.
A child who inherits the paternal haplotype A2, B35 ... DR1 ... DQ2 and the haplotype A1, B8 ... DR3 ... DQ2 from its mother has the HLA phenotype: A1, A2; B8, B35; ...; DR1, DR3; ...; DQ2, and is thus DQ2 homozygous (having inherited the same DQ antigen from father and mother) but is heterozygous at the other loci. Siblings share zero, one or both haplotypes. The non-inherited maternal antigens, to which the child is tolerant in utero, may be `acceptable mismatches' in transplantation.7 Whether they have a role in HIV progression is unknown.
Serological assays type HLA molecules by determining the protein products for which certain HLA antigens encode. More recent DNA-based techniques (using restriction fragment length polymorphism, polymerase chain reaction, and direct sequencing) study the genome directly and adopt a different but related notation, see Keet et al.8 We use serological notation in this paper.
Complex immunology: influence of HLA on HIV disease progression?
HIV replicates preferentially in activated CD4 lymphocytes, and HLA determine the immune response. Susceptibility to, or progression of, other diseases, including infectious diseases, has been found to be HLA-associated.
There was early evidence9 that HLA-A1, B8, DR3 influenced the clinical course of HIV infection. HLA antigens might, however, be more strongly associated with the rate of immunological deterioration, than with clinical manifestations of HIV disease, some of which—such as Kaposi's sarcoma, tuberculosis and Reiter's syndrome—are themselves HLA-linked. If HLA phenotype identifies a subset of individuals who are immunologically vulnerable to HIV disease progression (see McNeil et al.3), then such patients may merit special consideration for early treatment or for randomization in novel controlled trials.
HLA have greatly interested molecular immunologists working on HIV disease. Different HLA class I molecules select distinct epitopes derived from HIV proteins to stimulate cytotoxic T lymphocyte (CTL) responses. Phillips et al.10 noted less mutation in the HLA-B27-restricted epitope compared to HLA-B8-restricted peptides, and suggested that the resulting effect on CTL recognition could explain the association of HLA-A1, B8, DR3 with poor outcome in HIV infection. They also remarked that the effect of HLA-B27 on HIV progression was then unknown. Ohno2 subsequently established a molecular basis for an association between HLA-B27 and slow progression of HIV disease, namely that the nonapeptide derived from gp120 has virtually `optimal' binding characteristics for HLA-B27, and is therefore likely to be `tolerogenic' in that it resembles the preferred self peptides presented to T lymphocytes on this molecule.
Immunology does not lack for complex theories on how HLA influences HIV progression. Firstly, as above, there is the varying efficiency of HIV antigen presentation by different HLA molecules: viral diversity may enable HIV ultimately to evade immune responses. Secondly, there is genetic predisposition to aberrant immune responses, and the notion that HIV acts as an environmental trigger for these, for example in individuals with the DR3 ancestral haplotype who anyway have exaggerated immune responsiveness. Third is the idea of molecular mimicry between HIV and HLA. Several studies provide evidence of homology between class I and II HLA and specific amino acids of HIV-1 gp 120. In theory, the greater the homology, the more rapid is HIV progression. There are two schools of thought on molecular mimicry—alloimmunity and autoimmunity. Alloimmunity protects and is mediated, for example, by disparity between host and donor HLA. If this theory were correct, patients with rare HLA phenotypes might enjoy slower HIV progression. Autoimmunity, by contrast, sees AIDS as essentially an autoimmune disease mediated by molecular mimicry between HIV-1 and host cellular antigens. This results in selective depletion of CD4+ T cells. One version of this theory is that inherent susceptibility to autoimmunity is triggered by HIV. AIDS could result in two ways from autoimmune response directed against CD4: either via autoreactive T cells against CD4, as normally present, being stimulated in the presence of HIV, or HLA molecules may differentially stimulate activation of T suppressor cells (whose function is to prevent autoimmune responses).
Low-immune-responsive or rare HLA phenotypes thus merit attention, but—for elucidating study requirements, our primary aim in this paper—it is sufficient for us to concentrate on HLA-A1, B8, DR3; HLA-B35; and HLA-B27.
Study requirements and biostatistical notes
Susceptibility
The best design for studying HLA-associated susceptibility to HIV infection requires HLA-typing of HIV-infected pregnant women whose diagnosis was made before or during pregnancy11 together with HLA-typing of the infant, who is then followed-up to 18 months to determine HIV status. This design allows the influence of non-inherited maternal antigens to be assessed; see Luscher12 and also MacDonald et al.13 who reported that the risk of perinatal HIV-1 transmission increased with class I HLA concordance between mother and infant, and suggested that discordant HLA may thus protect infants against HIV because of allogeneic infant anti-maternal MHC immune responses.
Highly HIV-exposed individuals who have escaped HIV infection, such as Gambian prostitutes,14 have been hypothesis-generating about HLA-associated susceptibility. Determination of a suitable HLA-typed control group is difficult, however. Partner studies have also been considered, but adjusting adequately for partners' differential sexual (or injecting) exposure is again difficult.15
Progression
Prospective follow-up of patients with narrowly defined seroconversion intervals is the basic requirement for studies of HLA-associated HIV progression. In addition, several important time intervals need to be documented for proper inferences to be drawn. These are discussed below, and practical difficulties illustrated, with reference to two Scottish cohorts of HIV-infected individuals: the Edinburgh City Hospital (ECH) HIV Cohort3 and the Glenochil HIV Cohort, see Taylor et al.16
Interval 1
Interval 1 is from `infection focus' to `recruitment into HIV cohort'. A long Interval 1 risks that early HIV-related deaths are missed from the cohort; HIV infection may even be undiagnosed in these cases, particularly if infections occurred at the start of the global HIV epidemic when clinical experience was limited, see Gore et al.17 Later in the HIV epidemic, acute retroviral syndrome, itself associated with rapid progression and occurring in some 14% of infections, is more likely to lead to early diagnosis, so that later-established cohorts may be enriched for patients whose diagnosis was made on the basis of acute retroviral syndrome.
Edinburgh's injecting drug users were mainly infected with HIV in 1983, recruitment to the ECH HIV Cohort began in late 1985; and by 24 March 1994, the ECH Cohort included nearly 700 patients, of whom 68% were injection drug users.3 Only 13 seroconversion intervals were defined on the basis of acute retroviral syndrome, however.
During January to June 1993, injector-inmates were infected with HIV in Glenochil Prison with recruitment to the Glenochil HIV Cohort in July 1993 after an infection control exercise16 in which 14 HIV seroconversions in injectors were recognized, 13 with the same virus.18 Three infected prisoners had seroconversion illnesses. Willing anonymous salivary HIV surveillance at Glenochil Prison in July 199419 estimated that over 20 men had been infected in the prison during January to June 1993, not all of whom had participated in the infection control exercise. One prisoner-patient in the Glenochil HIV Cohort developed AIDS in August 1993, within 2 months of his seroconversion illness: we suggested that he was likely to be HLA-A1, B8, DR3 and so it proved. He died in May 1994. A second prisoner-patient, who had had a seroconversion illness and whose HLA phenotype included A1, B8, DR1 (not DR3), developed AIDS in January 1995. In addition, two other HLA-A1, B8, DR3 prisoner-patients, one of whom had had a seroconversion illness, became severely HIV-immunodeficient (=CD200 case, as defined by CD4 Collaborative Group20) by 31 December 1995. There were no CD200 cases among the eight other HLA-typed prisoner-patients, two of whom were B35, and one B27. One of the above three HLA-A1, B8, DR3 prisoners also had B35 in his phenotype (personal communication, Dr J. McMenamin).
Because hepatitis B as well as HIV infection was transmitted among injectors in Glenochil Prison in early 1993, as also in Edinburgh a decade earlier in 1983–84, see Bath et al.21, an epidemiological study17 was initiated in 1996 in Edinburgh to discover (and it did) undiagnosed HIV infections in individuals, particularly injectors, who had been tested for hepatitis B surface antigen in 1983–84 at the Regional Virus Laboratory in Edinburgh but had died by December 1986.
Interval 2
Interval 2 is from `recruitment into HIV cohort' to `start of HLA typing'. Long Interval 2, for funding or other reason, combined with serological typing risks missing HLA phenotypes for patients who were recruited but died early from rapid progression of their HIV disease.
For example, funded serological typing of the ECH Cohort only began in 1989; and concentrated on patients for whom a narrow seroconversion interval had been established by retrospective HIV testing of stored sera.22 DNA-based techniques are preferable for newly established cohorts, and were used for the Glenochil HIV Cohort: even so, HLA phenotype has been established for only 12 of Glenochil's 14 diagnosed HIV infections. As already noted, 20 men are estimated to have been infected, others of whom may be recruited when disease progression makes their diagnosis inevitable. Epidemiological surveillance23—using soundex of prisoners' surname and date of birth—was initiated in 1996 to facilitate future recognition of such cases within Scotland.
Interval 3
Interval 3 is the duration of HLA typing. Interval 3 may be protracted because of logistical constraints imposed by the laboratory undertaking serological typing, and the need for doctors to schedule requests to patients accordingly, or—with DNA-based techniques—because of the time needed to locate appropriately stored samples.
By 24 March 1994, only 155 (50%) of the 313 ECH patients with narrow interval estimates of seroconversion had been HLA-typed. Moreover, serological HLA-typing was initiated in the sickest patients so that information could be obtained whilst they were still alive.
Interval 4
Interval 4 is the epoch of follow-up. If Interval 4 is defined, say, as from 3 to 13 years after seroconversion, it misses out on rapid progressions to AIDS or death within the first 3 years after seroconversion as occurred in both Glenochil's and Edinburgh's injectors.17 Also, if proportionality is assumed, as commonly it is in applications of Cox regression, the relative hazard associated with a rapid-progression HLA phenotype such as HLA-A1, B8, DR3 will be under-estimated if the epoch of follow-up is from 3 to 13 years as compared to from 3 to 9 years after seroconversion. Mallal,24 however, conjectured that HLA influences on opportunistic infections become apparent only in later disease.
In Table 1
, we illustrate overall versus epoch-specific analyses for 248 injecting drug users in the ECH seroconversion cohort with follow-up to 31 December 1996 and median time from seroconversion to recruitment of 1467 days (inter-quartile range 990–2084 days). Unusually, for the reasons given by McNeil3 about probable completeness of ascertainment, we have not implemented left truncation when analysing the sub-cohort of injectors with narrow seroconversion intervals. Notice that our `other' category, mainly of untyped patients, reveals—as expected—excess mortality in the early epoch: some of these patients are untyped because they had died before HLA-typing was initiated. Notice also that, as expected, rapid progression to clinical events is associated with HLA-A1, B8, DR3 more strongly in the first than in the second epoch of follow-up. The sum of the log likelihoods for the two epoch-specific analyses is superior to the overall log likelihood, which supports our non-proportionality thesis.
|
Besides checking for non-proportionality of hazards, there is a further analysis consideration: relative hazard cannot be estimated for a slow-progression antigen, such as B27, if no endpoint, AIDS or death, has yet occurred as was the case when McNeil3 reported on HLA-B27's association with slow progression in the ECH HIV cohort. Instead, using a random effects model, McNeil estimated individual rates of decline in percent CD4 cells, a measure of immunological progression, and regressed these slopes on HLA and other cofactors. By so doing, the regression coefficients for HLA-A1, B8, DR3 (-0.70, SE=0.24; n=21 patients) and for HLA-B27 (1.15, SE=0.35; n=9 patients) were found to be opposite in sign, both significant, and roughly equal in magnitude.
Results
The results are in two parts: systematic review of published and other progression data which were available to us before 1996; followed by a subsection, entitled `Other data sources', which reviews a major 1996 publication by Kaslow et al.4 and presents analytic insights on unconventional or eccentric data sources which do not conform to the ideal that analysis should be restricted to seroconverter cohorts and principally be reported as relative hazard of AIDS.
Systematic review essayed: HLA-A1, B8, DR3; B35; B27
Table 2 summarizes published progression data which were available before 1996, together with published and unpublished results from the ECH HIV cohort.3 In practice, different endpoints than the relative hazard of AIDS have had to suffice in Table 2. Also, not all cohorts measured time from seroconversion, covariate adjustment for baseline CD4 count being often adopted as a surrogate for unknown seroconversion times. Relative hazard was not reported by all investigators, and so it has been approximated crudely in Table 2—as the relative `percent with AIDS'—since this could usually be discovered, whereas times at risk since seroconversion were seldom cited. Kaplan-Meier plots were also rare.
|
Unusually, times at risk since seroconversion were reported in the early paper by Steel9 which concerned 18 patients with haemophilia, eight of them HLA-A1, B8, DR3, who were HIV infected during March to May 1983 by a single batch of locally-produced factor VIII. By the end of 1987, seven of the eight patients with HLA-A1, B8, DR3 had developed CDC category IV disease (estimated cumulative time at risk=236 months from April 1983) but only two had of the other 11 patients (estimated cumulative time at risk=400 months from April 1983). Relative hazard was 5.9 but relative `percent with AIDS' was 4.4 from which ln(relative risk) of 1.48, as shown in Table 2
, was derived. The approximation used in Table 2 is thus not ideal, but was conservative in the one study for which comparison was possible.
Relative risks for non-significant regression effects are commonly not reported. This reporting bias is particularly acute for HLA effects because of their multiplicity and a generalized editorial insistence on brevity. Table 2 includes two illustrations, the first of which is our previously unpublished B35 regression effect for ECH HIV Cohort. Also, Kaplan25 reported an estimate of the relative hazard for HLA-A1, B8, DR3 but not for B35: the latter was not formally statistically significant in their data, and therefore was not documented. Such problems will bedevil any literature-based, supposedly systematic review of HLA associations with disease progression.
The pooled estimates and statistical error shown in Table 2 were derived straightforwardly, as described by Gore et al.31 in another context. They should, however, be interpreted cautiously against an overlay of extra uncertainty due to a) methodological disparity, b) reporting bias, and c) possible incompleteness in ascertainment of relevant published cohorts. Nonetheless, the broad conclusion from Table 2 is that HLA-A1, B8, DR3 and B35 may have similar effects on HIV progression (ln RR about 0.9, SE=0.2) which are opposite in sign but comparable in magnitude to that anticipated for B27—based on McNeil's comparison3 between HLA-A1, B8, DR3 and B27 in respect of immunological progression.
It is important to note that these major HLA effects have an impact on HIV progression which is the equivalent of 20 additional years of age. This claim rests upon the pooled ln RR per decade of age being around 0.40 (SE 0.08) based on Rosenberg et al.,32 Carre et al.33 and the Multicohort Analysis Project Workshop;34 or, based on the quality-assured dataset analysed by the Collaborative Group on AIDS Incubation and HIV Survival,6 was 0.31 (SE 0.04) when Kaposi's sarcoma was excluded as first AIDS indicator disease, but was 0.39 (SE 0.04) per decade of age in respect of mortality.
Major HLA effects, unlike immunological senescence, apply for only a proportion of patients, however: for example, whereas about a fifth of Scots have HLA-A1, B8, DR3 in their phenotype, only 8% have B27.
Other data sources
Additional support for the above quantitative inferences comes from diverse small datasets, important work by Kaslow et al.,4 and our inter-cohort validation thereof (see Table 3).
|
The diverse small datasets include: the Glenochil cohort, described above; the six patients with haemophilia originally reported on by Phillips et al.,10 three of whom were B27 and three HLA-A1, B8, DR3; and 25 long-term `non-progressors' whose HLA phenotypes were fully reported by Cao35 and by Pantaleo36 who remarked: `No particular patterns of HLA haplotypes were observed'.
Expected deaths in the six Oxford patients—based on ECH's HLA-A1, B8, DR3 death rate at that time of 13/21, see McNeil,3 and assumed `equal but opposite' for B27—were 1.9 and 0.2, respectively. In fact, two A1, B8, DR3 and zero B27 patients had died by end August 1995 (personal communication, Dr Sara Marshall). Of the 25 non-progressors reported on by Cao35 and Pantaleo,36 one was HLA-A1, B8, DR3; one was B35; and three were B27. Since the frequency of B35 in Whites in the USA is roughly twice that of B27, the Cao and Pantaleo data are consistent with a 6-fold relative enhancement in the representation of B27 among long-term non-progressors.
Detailed statistical analysis by Kaslow et al.4 generated an HLA profile that predicted time from HIV infection to onset of AIDS. This HLA profile took into account interactions with other class III MHC genes which have biological plausibility, specifically transporters associated with antigen processing or TAP genes.
The HLA profile was developed in a cohort of 139 male seroconverters from the Multicenter AIDS Cohort Study (MACS) and evaluated in a second unrelated but demographically and epidemiologically comparable group of 102 seroconverters from the DC Gay (DCG) cohort. The two North American cohorts were typed serologically for HLA class I antigens, and molecularly for HLA-DR, DQ and TAP genes. Kaslow estimated from the MACS group the relative hazards associated with individual alleles or with combinations of class I and class II antigens and alleles of the TAP genes. These alleles or combinations were scored on an arbitrary scale according to the associated relative hazards, and this scale was then used to assign high or low HLA-associated risk to individuals in the DCG cohort. In the evaluation cohort, the profile discriminated a six-fold difference between groups with the shortest and longest times to AIDS. Hill37 outlined the attractions and dangers of Kaslow's approach, and called for further investigation to retest the predictive power of the algorithm. We present one such evaluation, but see also Mann et al.38
At 31 December 1996, the ECH HIV cohort comprised 310 patients with narrowly defined seroconversion intervals, of whom 165 were HLA-typed. The present analysis differs from that of McNeil3 in having longer follow-up, a few extra HLA types, and by our having deselected cases whose date of first HIV-antibody-positive test was based on patient report. It also differs from Kaslow4 in that we are unable to address the association of TAP genes (for which patients in the ECH cohort were not typed) with class I antigens, other than by reliance on linkage disequilibrium. That said, 126 (76%) of the 165 fully or partially typed patients out of 310 in the ECH seroconverter cohort carried at least one of Kaslow's contributing markers (slow progression: A25, A26, A32, B18, B27, B51, B57; fast progression: A23, A24, A28, A29, B8, B35, B37, B49, B60). For each subject studied, the algebraic sum of the scored markers constituted their HLA profile which was assigned to one of five HLA score categories. Table 3 shows the hazard of AIDS relative to the zero score category for ECH and DCG cohorts and also the highly significant and consistent trend in ln RR (ECH: -0.39, se 0.11) over the five HLA score categories.
Because Cao and Pantaleo reported HLA phenotype (but not TAP genes) for their 25 long-term non-progressors, we have also scored HLA profile for them on the same basis as in ECH. Scores were: -2 (1 cases), -1 (6 cases), 0 (6 cases), +1 (8 cases), +2 or more (4 cases) with, as expected, significantly more of the long-term non-progressors having scored positively (48%: 12/25) than of ECH's fully or partially HLA-typed seroconverters (13%: 22/165).
Finally, B27 was present for 17/241 patients (7%) in the MACS and DCG cohorts and—in the grouping B27 or B51 or B57—it was associated with approximate RR of 0.32, or ln RR of -1.1 (se 0.38) in MACS. Table 4 brings our review of HLA-A1, B8, DR3; B35; and B27 up to date by inclusion of these other data.
|
Discussion
We have concentrated on design, analysis and reporting requirements for studying HLA influences on clinical and immunological progression of HIV disease.
The polymorphism of HLA, and geographical and ethnic variation in gene frequency, make consistent reporting between studies an unlikely achievement even by those authors who limit themselves to studying only those alleles with 5% or more relative frequency in their patient population. Relative hazards which are not formally statistically significant often go unreported on grounds of parsimony, but this jeopardizes systematic review of the literature.
Inferences about HLA effects on HIV progression depend not only upon prospective follow-up of patients with narrowly-defined seroconversion intervals but also upon appreciation of four crucial intervals—from focus of infection to recruitment into cohort, from recruitment to initiation of HLA typing, time period for HLA typing, and the epoch of follow-up over which relative hazards are estimated. Seldom is there complete reporting on these four intervals, let alone consideration of the ascertainment biasses that they give rise to.
Inter-cohort validation of HLA risk scores for HIV progression was first reported by Kaslow et al.,4 and has been extended herein to the ECH HIV cohort. However, robust and rapid progress in the mapping of HLA influences on HIV progression will only be made if a Multicohort Analysis Workshop can be convened for this specific purpose, at which complex immunological theories on how HLA influences HIV progression can be tested rigorously and potential biases dealt with appropriately by careful attention to statistical method.
Despite the limitations inherent in systematic review of the HLA literature, even for restricted phenotypes such as we have attempted, and the eccentricity of locating relevant small data-sets, some of which we have illustrated, there is a strong case, both a priori1,2 and empirically, for a workshop to map HLA influences on HIV progression. The empirical justification is that the apparent HLA profile effects are the equivalent of a decade of age, and the major HLA effects the equivalent of two decades of immunological senescence, as we have indicated in Table 4. Further specific insights to HIV pathogenesis and therapeutic implications of HLA profile are thus a realistic prospect from such a workshop.
Acknowledgments
We thank the Medical Research Council for funding (SPG-9116497 and SPG-9102632) and our co-grantholders Dr Peng Lee Yap, Dr David J. Goldberg and Dr A. Graham Bird for scientific discussions on HLA and on HIV immunology. We are grateful to Dr J. McMenamin for information on the Glenochil HIV cohort.
References
1. Hill AV. The immunogenetics of human infectious diseases. Ann Rev Immunol 1998; 16:593–617.[Web of Science][Medline]
2.
Ohno S. How cytotoxic T cells manage to discriminate nonself from self at the nonapeptide level. Proc Natl Acad Sci USA 1992; 89:4643–7.
3. McNeil AJ, Yap PL, Gore SM, Brettle RP, McColl M, Wyld R, Davidson S, Weightman W, Richardson AM, Robertson JR. Association of HLA types A1-B8-DR3 and B27 with rapid and slow progression of HIV disease.Q J Med 1996; 89:177–85.[Abstract]
4. Kaslow RA, Carrington M, Apple R, Park L, Munoz A, Saah AJ, Goedert JJ, Winkler C, O'Brien SJ, Rinaldo C, Detels R, Blattner W, Phair J, Erlich H, Mann DL. Influence of combinations of human major histocompatibility complex genes on the course of HIV-1 infection. Nature Med 1996; 2:405–11.[Web of Science][Medline]
5. Ioannidis JPA, O'Brien TR, Rosenberg PS, Contopoulos-Ioannidis DG, Goedert JJ. Genetic effects on HIV disease progression. Nature Med 1998; 4:536.[Web of Science][Medline]
6. Collaborative Group on AIDS Incubation and HIV Survival, including the CASCADE EU Concerted Action. Time from HIV-1 seroconversion to AIDS and death in 13 000 individuals before the widespread use of highly active anti-retroviral therapy: effect of age at seroconversion and exposure category. Lancet 1999; submitted for publication.
7.
Birmingham WJ, Grailer AP, Heisey DM, Claas FHJ, Norman D, Mohanakumar T, Brennan DC, de Fijter H, van Gelder T, Pirsch JD, Sollinger HW, Bean MA. The effect of tolerance to noninherited maternal HLA antigens on the survival of renal transplants from sibling donors. N Engl J Med 1999; 339:1657–64.
8. Keet IPM, Klein MR, Just JJ, Kaslow RA. The role of host genetics in the natural history of HIV-1 infection: the needles in the haystack. AIDS 1996; 10 (suppl A):S59–67.
9. Steel CM, Ludlam CA, Beatson D, Peutherer JF, Cuthbert RJG, Simmonds P, Morrison H, Jones M. HLA haplotype A1 B8 DR3 as a risk factor for HIV-related disease. Lancet 1988; 1:1185–8.[Web of Science][Medline]
10. Phillips RE, Rowland-Jones S, Nixon DF, Gotch FM, Edwards JP, Ogunlesi AO, Elvin JG, Rothbard JA, Bangham CRM, Rizza CR, McMichael AJ. Human immunodeficiency virus genetic variation that can escape cytotoxic T cell recognition. Nature 1991; 354:453–9.[Medline]
11. Gibb DM, MacDonagh SE, Tookey PA, Duong T, Nicoll A, Goldberg DJ, Hudson CN, Peckham CS, Ades AE. Uptake of interventions to reduce mother-to-child transmission of HIV in the United Kingdom and Ireland. AIDS 1997; 11:F53–8.[Web of Science][Medline]
12. Luscher MA, Choy G, Embree J, Nagelkerke NJ, Bwayo JJ, Njenga S, Plummer FA, Barber BH, MacDonald KS. Anti-HLA alloantibody is found in children but does not correlate with a lack of HIV type 1 transmission from infected mothers. AIDS Res Hum Retroviruses 1998; 14:99–107.[Web of Science][Medline]
13. MacDonald KS, Embree J, Njenga S, Nagelkerke NJ, Ngatia I, Mohammed Z, Barber BH, Ndinya-Achola J, Bwayo J, Plummer FA. Mother-child class I HLA concordance increases perinatal human immunodeficiency virus type 1 transmission. J Infect Dis 1998; 177:551–6.[Web of Science][Medline]
14. Rowland-Jones S, Sutton J, Ariyoshi K, Dong T, Gotch F, McAdam S, Whitby D, Sabally S, Gallimore A, Corrah T, Takiguchi M, Schultz T, McMichael A, Whittle H. HIV-specific cytotoxic T-cells in HIV-exposed but uninfected Gambian women. Nature Med 1995; 1:59–64.[Web of Science][Medline]
15. Fielding KL, Brettle RP, Gore SM, O'Brien F, Wyld R, Robertson R, Weightman R. Heterosexual transmission of HIV analysed by generalized estimating equations. Statistics Med 1995; 14:1365–78.
16.
Taylor A, Goldberg D, Emslie J, Wrench J, Gruer L, Cameron S, Black J, Davis B, McGregor J, Follett E, Harvey J, Basson J, McGavigan J. Outbreak of HIV infection in a Scottish prison. Br Med J 1995; 310:289–92.
17. Gore SM, Brettle RP, Burns SM, Lewis SC. Early mortality of undiagnosed but prevalent (in 1983–84) HIV infection in Lothian injectors who tested Hepatitis B surface antigen positive (group A) or negative but were high risk for blood-borne virus transmission (group B) in 1983–84. J Infect 1998; 37:166–72.[Web of Science][Medline]
18.
Yirrell DL, Robertson P, Goldberg DJ, McMenamin J, Cameron S, Leigh Brown AJ. Molecular investigation into outbreak of HIV in a Scottish prison. Br Med J 1997; 314:1446–50.
19.
Gore SM, Bird AG, Burns SM, Goldberg DJ, Ross AJ, Macgregor J. Drug injection and HIV prevalence in inmates of Glenochil prison. Br Med J 1995; 310:293–6.
20. CD4 Collaborative Group. CD4 surveillance in Scotland: perspectives on severe HIV-related immunodeficiency. AIDS 1997; 11:1509–17.[Medline]
21. Bath GE, Scott TG, Sibbald CJ, Upton PA, Ramsay CN, Willocks LJ. Acute Hepatitis B in Edinburgh 1975–92: a retrospective study in a population where human immunodeficiency virus is highly prevalent. Epidemiol Infect 1997; 119:85–9.[Medline]
22. Burns SM, Brettle RP, Gore SM, Peutherer JF, Robertson JR. The epidemiology of HIV infection in Edinburgh related to the injecting of drugs: an historical perspective and new insights regarding the past incidence of HIV infection derived from retrospective HIV antibody testing of stored samples of serum. J Infect 1996; 32:53–62.[Web of Science][Medline]
23. Hutchinson SJ, Gore SM, Goldberg DJ, Yirrell DL, McGregor J, Bird AG, Leigh-Brown AJ. Methodology used to follow-up previously undiagnosed infections in the HIV outbreak at Glenochil prison. J Epidemiol Infect 1999; in Press.
24. Mallal SA. The Western Australia HIV Cohort Study, Perth, Australia. J Acquired Immun Defic Syndrome Hum Retrovirol 1998; 17:s23–7.
25. Kaplan C, Muller JY, Doinel C, Lefrere JJ, Paquez F, Rouger P, Salmon D, Salmon C. HLA-associated susceptibility to Acquired Immune Deficiency Syndrome in HIV-1-seropositive subjects. Hum Hered 1990; 40:290–8.[Web of Science][Medline]
26. Kaslow RA, Duquesnoy R, van Raden M, Kingsley L, Marrari M, Friedman H, Su S, Saah AJ, Detels R, Phair J, Rinaldo C. A1, Cw7, B8, DR3 HLA antigen combination associated with rapid decline of T-helper lymphocytes in HIV-1 infection. A report from the Multicenter AIDS Cohort Study. Lancet 1990; 335:927–30.[Web of Science][Medline]
27. Cameron PU, Mallal SA, French MAH, Dawkins RL. Major Histocompatibility Complex genes influence the outcome of HIV infection. Ancestral haplotypes with C4 null alleles explain diverse HLA associations. Hum Immunol 1990; 29:282–95.[Web of Science][Medline]
28. Klein MR, Keet IPM, d'Amaro J, Bende RJ, Hekman A, Mesman B, Koot M, de Waal LP, Coutinho RA, Miedema F. Association between HLA frequencies and pathogenic features of Human Immunodeficiency Virus type 1 infection in seroconverers from the Amsterdam cohort of homosexual men. J Infect Dis 1994; 169:1244–9.[Web of Science][Medline]
29. Itescu S, Mathur-Wagh U, Skovron ML, Brancato LJ, Marmor M, Zeleniuch-Jacquotte A, Winchester R. HLA-B35 is associated with accelerated progression to AIDS. J Acquired Immun Defic Syndrome 1991; 5:37–45.
30. Sahmoud T, Laurian Y, Gazengel C, Sultan Y, Gautreau C, Costagliola D. Progression to AIDS in French haemophiliacs: association with HLA-B35. AIDS 1993; 7:497–500.[Web of Science][Medline]
31. Gore SM, Fontaine O, Pierce NF. Impact of rice based oral rehydration solution on stool output and duration of diarrhoea: meta-analysis of 13 clinical trials. Br Med J 1992; 304:287–92.
32. Rosenberg PS, Goedert JJ, Biggar RJ for the Multicenter Hemophilia Cohort Study and the International Registry of Seroconverers. Effect of age at seroconversion on the natural AIDS incubation distribution. AIDS 1994; 8:803–10.[Web of Science][Medline]
33. Carre N, Deveau C, Belanger F, Boufassa F, Persoz A, Jadand C, Rouzioux C, Defraisssy J-F, Bucquet D and the SEROCO Study Group. (1994) Effect of age and exposure group on the onset of AIDS in heterosexual and homosexual HIV-infected patients. AIDS 1994; 8:797–802.[Web of Science][Medline]
34. Multicohort Analysis Project Workshop. Immunologic markers of AIDS progression: consistency across five HIV-infected cohorts. AIDS 1994; 8:911–21.[Web of Science][Medline]
35.
Cao Y, Qin L, Zhang L, Safrit J, Ho DD. Virologic and immunologic characterization of long-term survivors of Human Immunodeficiency Virus type 1 infection. N Engl J Med 1995; 332:201–8.
36.
Pantaleo G, Menzo S, Vaccarezza M, Graziosi C, Cohen OJ, Demarest JF, Montefiori D, Orenstein JM, Fox C, Schrager LK, Margolick JB, Buchbinder S, Giorgi JV, Fauci AS. Studies in subjects with long-term nonprogressive Human Immunodeficiency Virus infection. N Engl J Med 1995; 332:209–16.
37. Hill AVS. HIV and HLA: confusion or complexity? Nature Med 1996; 2:395–6.[Web of Science][Medline]
38. Mann DL, Garner RP, Dayhoff DE, Cao K, Fernandez-Vina MA, Davis C, Aronson N, Ruiz N, Birx DL, Michael NL. Major histocompatibility complex genotype is associated with disease progression and virus load levels in a cohort of human immunodeficiency virus type 1-infected Caucasians and African Americans. J Infect Dis 1998; 178:1799–802.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||