Table 3.

Survival analysis for progression to three AIDS outcomes among seroconverters as a function of IL10-5′A/5′A or +/5′A vs. IL10-+/+ genotype

	Time* interval, years	AIDS-1993					AIDS-1987					Death
		n/event	RHu	P value	adjRH†	P value	n/event	RHu	P value	adjRH†	P value	n/event	RHu	P value	adjRH†	P value
All races	All	766/462	1.28	0.01	1.37	0.002	769/332	1.43	0.002	1.44	0.002	769/253	1.32	0.03	1.32	0.03
	0–5		1.06	0.66	1.17	0.26		1.26	0.24	1.25	0.27		1.04	0.86	1.05	0.82
	>5		1.54	0.002	1.58	0.0009		1.52	0.003	1.54	0.002		1.49	0.01	1.48	0.02
Caucasians	All	511/336	1.44	0.0009	1.45	0.0008	514/265	1.51	0.001	1.51	0.001	514/213	1.40	0.02	1.41	0.02
	0–5		1.23	0.21	1.27	0.15		1.21	0.42	1.26	0.33		1.05	0.85	1.09	0.75
	>5		1.63	0.0009	1.61	0.001		1.65	0.0007	1.63	0.001		1.57	0.007	1.56	0.008
Multicenter AIDS Cohort Study	All	345/234	1.56	0.0008	1.57	0.0007	348/186	1.52	0.005	1.53	0.004	348/147	1.34	0.08	1.36	0.06
	0–5		1.28	0.17	1.30	0.15		1.19	0.49	1.22	0.43		1.05	0.87	1.07	0.81
	>5		1.96	0.0005	1.96	0.0005		1.74	0.003	1.74	0.003		1.55	0.04	1.57	0.03
Multicenter Hemophilia Cohort Study	All	157/100	1.34	0.17	1.39	0.13	157/78	1.49	0.10	1.56	0.10	157/65	1.46	0.15	1.45	0.17
	0–5		0.97	0.95	1.15	0.75		1.15	0.83	1.39	0.62		0.95	0.95	1.33	0.74
	>5		1.48	0.11	1.50	0.10		1.56	0.09	1.54	0.10		1.53	0.13	1.49	0.16
African Americans‡	All	220/102	0.92	0.69	1.22	0.42	220/53	1.08	0.81	1.12	0.73	220/31	1.01	0.81	1.07	0.87
	0–5		0.74	0.24	0.93	0.80		1.48	0.37	1.31	0.57		1.02	0.96	0.96	0.94
	>5		1.42	0.37	1.93	0.13		0.77	0.54	0.96	0.93		1.25	0.74	1.32	0.69

Statistical analysis. Cox models (15, 16) were used for calculating relative hazards and P values for separate and combined cohorts and racial groups, as we have previously reported (5, 6). Nonsignificant 2- and 3-year windows in the initial infection period were grouped together and later times considered separately in further Cox analyses of AIDS outcomes (14, 35). A strict and conservative Bonferroni correction of the 120 P values presented would mean that only those of P ≤ 0.00042 are significant, which is coincidentally equivalent to the smallest observed P value (P = 0.0005; Multicenter AIDS Cohort Study, AIDS-1993 >5-year portion of the analysis). Because the tests were not independent, the observed statistical associations seem robust. Eighteen of the AIDS progression P values are lower than 0.002 (some by an order of magnitude), and overall, 45 are P ≤ 0.05 and 30 are P ≤ 0.01. 

^*Time was not taken into account in all models and was included as an interaction term by separating 0–5 and >5 years in the models (except death, where 0–6 and >6 years was used). 

^† Relative hazards were shown as RHu from unadjusted models and as adjRH from adjusted models, in which the effects of the previously described CCR5-Δ32 and CCR2-64I genotypes were taken into account. The effects of IL10 and CCR (CCR5-Δ32 and CCR2-64I) appear additive, because, in the Cox analyses, interaction terms were not significant (P = 0.53–0.87) for the four outcomes in Caucasians. 

^‡ IL10-5′A AIDS accelerating effects were most apparent in Caucasian and mixed ethnic group analyses. The less significant P values in African Americans are likely a consequence of fewer patients (n = 220 seroconverters), fewer AIDS cases after 5 years of infection (15 AIDS-1987 and 5 AIDS deaths), and the relative recency of the recruitment of the AIDS Link to the Intravenous Experience cohort, which comprises 70% of the African American seroconverters (5, 27). This interpretation is supported by high relative hazards (at >5 years); RH = 3.41 and 1.93 for CD4 < 200 and AIDS-1993, respectively, the two earliest AIDS end points.