The Clinical Significance of CD4 Counts in Asian and Caucasian HIV-Infected Populations: Results from TAHOD and AHOD

Abstract

The significance of interethnic variation in CD4 counts between Asian and Caucasian populations is not known. Patients on combination antiretroviral therapy from Treat Asia and Australian HIV Observational Databases (TAHOD, predominantly Asian, n = 3356; and AHOD, predominantly Caucasian, n = 2312, respectively) were followed for 23 144 person-years for AIDS/death and all-cause mortality endpoints. We calculated incidence-rates and used adjusted Cox regression to test for the interaction between cohort (TAHOD/AHOD) and time-updated CD4 count category (lagged by 3 months) for each of the endpoints. There were 382 AIDS/death events in TAHOD (rate: 4.06, 95%CI: 3.68-4.50) and 305 in AHOD (rate: 2.39, 95%CI: 2.13-2.67), per 100 person-years. At any given CD4 count category, the incidence-rates of endpoints were found to be similar between TAHOD and AHOD (in the adjusted models, P > .05 for the interaction term between cohort type and latest CD4 counts). At any given CD4 count, risk of AIDS or death was not found to vary by ethnicity, suggesting that the CD4 count thresholds for predicting outcomes defined in Caucasian populations may be equally valid in Asian populations.

Figure 1

Incidence rates per 100 person years of (a) AIDS or death and (b) all-cause mortality by latest CD4 count per mm³ (lagged by 3 months) in TREAT Asia HIV Observational Database (TAHOD) and Australian HIV Observational Database (AHOD). Error bars represent 95% confidence intervals.

Figure 2

a, Rate of loss to follow-up, per 100 person years, by last known CD4 count per mm³ (lagged by 3 months), in TREAT Asia HIV Observational Database (TAHOD) and Australian HIV Observational Database (AHOD). Error bars represent 95% confidence intervals. b, Overall estimated all-cause mortality rate in TAHOD (with latest CD4 count <50 cells/mm³), after accounting for assumed mortality rates in those who were lost to follow-up (LTFU), and the comparison with observed rate in AHOD. The higher the assumed mortality in LTFU patients, the similar are the mortality rates between TAHOD and AHOD. Error bars represent 95% confidence intervals.

Figure 3

Incidence rate of AIDS or death (left vertical panel) and all-cause mortality (right vertical panel), by latest CD4 count (cells/mm³), in TREAT Asia HIV Observational Database (TAHOD) and Australian HIV Observational Database (AHOD). *P value is for the interaction term (product of cohort [TAHOD/AHOD] and categorized CD4 count lagged by 3 months), obtained in the multivariate models predicting respective endpoint, adjusted for same covariates as listed in Tables 2 and 3. The data are when (a) every second time-dependent CD4 count excluded in AHOD; (b) censoring tuberculosis-related AIDS diagnosis or deaths with tuberculosis as an underlying cause; and (c) using latest (time-updated) CD4 counts and HIV RNA levels, without lagging (ie, using the recent most measurements). The latest CD4 counts were lagged by 3 months in (a) and (b). Error bars represent 95% confidence intervals.