Table 2.

Analysis for IL10 polymorphic sites (STRs and SNPs) and haplotype association^* with HIV-1 infection and/or AIDS progression^†

IL10 locus	Allele (model)	Distance from AUG, kb	HIV-1 infection‡		Progression to AIDS outcomes
			No. of individuals	P value	No. of individuals	CD4<200		AIDS-1993		AIDS-1987		Death
						RH	P	RH	P	RH	P	RH	P
STR−3975-R	283  (rec)	4.0	507	0.43	379	1.47	0.007	1.46	0.003	1.45	0.009	1.46	0.02
STR−1140-G	169  (dom)	1.1	506	0.03	379	1.25	0.26	1.18	0.36	1.44	0.06	1.72	0.009
SNP−1082	G  (rec)§	1.1	879	0.50	419	0.89	0.18	0.85	0.03	0.78	0.008	0.77	0.39
SNP−592-5′A	5′A  (dom)¶	0.6	1008	0.48‡	514	1.27	0.05	1.44	0.0009	1.51	0.001	1.40	0.02

^*Genotypes from the STR loci were tested for infection and progression to AIDS considering each locus as a family of tests. The P values for STR IL10-R(−3975) were significant after Bonferroni correction for the number of alleles at each of these STR loci. Significant associations indicated by boldface were observed for HIV-1 infection (STR−1140-G) and for progression to AIDS. The association of the STR IL10-R(−3975) with disease progression is the result of a strong linkage disequilibrium between IL10-R(−3975) allele 283 (Fig. 2A) and the IL10-5′A promoter allele. Thus, a haplotype survey of 1,698 human chromosomes (Caucasian), which excluded ambiguous double heterozygotes, showed that 94% of IL10-5′A-bearing chromosomes carry an IL10-R(−3975)283 allele, a significant departure from random expectation for association of included alleles in that haplotype (68%; P < 0.0001). The STR-G(−1140) is also in strong linkage disequilibrium (P ≤ 0.0001) with IL10-5′A and tracks its effects as well. That the significant signals with STR−3975-R, STR−1140-G, and IL10−592-5′A follow different genetic models, recessive and dominant, is likely because of incomplete linkage disequilibrium between these polymorphisms. 

^† The AIDS-delaying influence of IL10(−1082)-G SNP was apparent by considering a recessive model where IL10-(−1082)-G/G homozygotes were compared with other genotypes in a Cox analysis. 

^‡ The AIDS-delaying influence of IL10-(−592)-5′A was observed by considering a dominant model where IL10-(−592)-+/5′A and 5′A/5′A were compared to IL10(−592)-+/+ genotypes (see Fig. 2 B–D). 

^§ The two SNP loci, IL10(−1082) and IL10(−592), both show an effect on AIDS progression and are in strong linkage disequilibrium with each other. Thus, in a sample of 3,626 Caucasian chromosomes, three [−1082 −592] genotype combinations were never observed: [G.+]/[G.5′A], [+.G]/[G.5′A], and [G.5′A]/[G.5′A], because of the complete absence of the [G.5′A] haplotype, as previously reported (42–45). To determine whether AIDS protection was determined by recessive protection of IL10−1082-G or by the dominant susceptible influence of IL10−592-5′A, we compared three haplotype groups in Cox relative hazard model analyses: (i) those who retained one or two copies of IL10−592-5′A (ii) those who were homozygous for IL10−1082-G; and (iii) other patients who contained neither IL10−592-5′A nor IL10−1082G/G genotypes (referent group). The significant epidemiologic signals were observed with the IL10−592-5′A-bearing genotypes (group i; RH = 1.22–1.48, P = 0.24–0.01), but not with the IL10−1082-G/G homozygotes who lack IL10−592-5′A (group ii; RH = 0.66–0.92, P = 0.71–0.07) when compared to the referent group. The above three-haplotype-stratified analysis is similar to those used in very complex haplotype analyses (46) but much simpler and implicates IL10−592-5′A as the operative SNP in the epidemiologic effects on AIDS progression. 

^¶ The failure to reveal an infection effect of IL10-5′A in the face of association with a smaller group of rigorously defined (47) high-risk exposed uninfected patients (n = 72; see text), could be because of the difficulty of assessing the extent of HIV-1 exposure in the larger group of HIV-1 antibody negative study participants (n = 631 in this table).