Background:  Microbial translocation (MT) has been proposed as a mechanism for immune activation in AIDS when measured in a US population, but there is disagreement about its role in HIV in international settings. To date, covariates of MT have not been evaluated. We hypothesized that soluble CD14 (sCD14), an MT marker proposed to predict mortality in HIV, would differ based on the tested population.

Methods:  Nested case-cohort study within ACTG PEARLS, a multi-country HAART randomized trial of 1575 HIV-infected treatment-naive adults in 8 countries; 270 individuals were randomly selected. Samples were derived from sites in multiple countries in equal proportions: Brazil, Haiti, India, Malawi, Peru, South Africa, Thailand, US, and Zimbabwe, although there were disproportionately more missing samples from India. sCD14 was assessed by ELISA. Kruskal-Wallis tests compared sCD14 levels among groups. Baseline sCD14 results were available for 254 (94%).

Results:  Median (IQR) age was 35 years (29 to 41), and 122 (48%) were women. Persons of African origin were the most common race (53%), followed by Asian (23%), Caucasian (11%), and other (13%). Baseline median (IQR) CD4⁺ T cell count was 179 cells/mm³ (90 to 229), and median (IQR) plasma HIV-1 RNA was 5.1 log₁₀ copies/mL (4.6 to5.5). Median (IQR) plasma sCD14 before HAART was 2012 ng/mL (487 to 2,698). Persons with CD4⁺ T cell counts <200 cells/mm³ had higher sCD14 levels than persons with CD4⁺ T cell counts ≥200 cells/mm³, and differences between persons according to baseline CD4⁺ T cell count were significant (p = 0.04); baseline HIV-1 RNA was not associated with sCD14. Striking differences in sCD14 concentration were observed among countries (p <0.001, Table). Sex and race were also significantly associated with baseline sCD14 (p <0.001 and p = 0.02, respectively), and there was a trend towards significantly higher sCD14 levels in persons with TB history (p = 0.07). In a multi-covariate linear regression model, country remained significantly associated with plasma sCD14, even after controlling for race, sex, CD4, and TB history (p <0.001).

Conclusions:  Pre-HAART plasma sCD14 concentration differed by country after adjustment for CD4⁺ T cell count, race, sex, and TB history. Our findings may explain the variability of MT found in previous studies of HIV-infected persons. Variable baseline sCD14 levels in persons with HIV may in part be due to underlying biologic differences in gut integrity or immune activation.