TABLE 3.

Multivariable analysis of HLA class I supertypes in relation to HIV-1 outcomes among African Americans from the REACH (1996-2001) and HERS (1993-2000) cohorts^a

HLA class I supertype	Log10 VL				CD4 count
	Adjustedb		Adjustedc		Adjustedd		Adjustede
	ΔVL	P value	ΔVL	P value	ΔCD4 count	P value	ΔCD4 count	P value
A1s	+0.06	0.73	+0.001	0.99	+47	0.33	+67	0.17
A2s	−0.1	0.34	−0.14	0.18	−16	0.78	−9	0.87
A24s	+0.05	0.6	−0.01	0.86	+3	0.93	+28	0.4
A3s	−0.01	0.91	−0.006	0.96	−14	0.7	−27	0.45
B7s	+0.28	0.02	+0.22	0.05	−81	0.02	−52	0.13
B27s	−0.2	0.15	−0.25	0.05	+4	0.92	+24	0.49
B44s	+0.33	<0.01	+0.39	<0.001	−54	0.11	−50	0.13
B58s	−0.26	0.04	−0.17	0.17	+32	0.39	+39	0.29
B62s	−0.65	0.01	−0.49	0.03	+82	0.27	+53	0.44

^aEstimates correspond to the fixed effects of average log₁₀ VL and CD4 count changes within all patient visits. Both the estimates and P values were derived from mixed linear models, with imputed missing values for log₁₀ VL (24 observations) and CD4 count (38 observations). ΔVL and ΔCD4 count estimates correspond to the β coefficients from mixed linear models.

^bAdjusted for cohort, age, patient visit, and alleles associated with log₁₀ VL from uni- and bivariate analyses (A*32 and A*36 for A1s, A*74 for A3s, B*3501, B*53, and B*81 for B7s, B*14 and B*1510 for B27s, B*45 for B44s, and B*57 for B58s).

^cAdjusted for cohort, age, patient visit, and significant alleles within and outside the particular supertype, as applicable (A*32, A*36, A*74, B*3501, B*53, B*14, B*1510, B*45, B*57, and B*81).

^dAdjusted for patient visit, cohort, age, and alleles associated with CD4 from uni- and bivariate analyses (A*32 and A*36 for A1s, A*02 for A2s, A*74 and A*03 for A3s, B*3503, B*53, and B*81 for B7s, B*14 and B*27 for B27s, B*45 for B44s, and B*57 for B*58s).

^eAdjusted for patient visit, cohort, age, and significant alleles from within and outside the supertype, as applicable (A*02, A*03, A*32, A*36, A*74, B*14, B*27, B*3503, B*45, B*53, and B*57).