Objective The association between pre-antiretroviral (ART) inflammation and immune activation and risk for incident tuberculosis (TB) after ART initiation among adults is uncertain. Design Nested case-control study (n = 332) within ACTG PEARLS trial of three ART regimens among 1571 HIV-infected, treatment-naïve adults in 9 countries. We compared cases (participants with incident TB diagnosed by 96 weeks) to a random sample of controls (participants who did not develop TB, stratified by country and treatment arm). Methods We measured pre-ART C-reactive protein (CRP), EndoCab IgM, ferritin, interferon gamma (IFN-γ), interleukin 6 (IL-6), interferon gamma-inducible protein 10 (IP-10), lipopolysaccharide (LPS), soluble CD14 (sCD14), tumor necrosis factor alpha (TNF-α), and CD4/DR+/38+ and CD8/DR+/38+ T cells. Markers were defined according to established cutoff definitions when available, 75th percentile of measured values when not, and detectable versus undetectable for LPS. Using logistic regression, we measured associations between biomarkers and incident TB, adjusting for age, sex, study site, treatment arm, baseline CD4 and log10 viral load. We assessed the discriminatory value of biomarkers using receiver operating characteristic (ROC) analysis. Results Seventy-seven persons (4.9%) developed incident TB during follow-up. Elevated baseline CRP (aOR 3.25, 95% CI: 1.55-6.81) and IP-10 (aOR 1.89, 95% CI: 1.05-3.39), detectable plasma LPS (aOR 2.39, 95% CI: 1.13-5.06), and the established TB risk factors anemia and hypoalbuminemia were independently associated with incident TB. In ROC analysis, CRP, albumin, and LPS improved discrimination only modestly for TB risk when added to baseline routine patient characteristics including CD4 count, body mass index, and prior TB. Conclusion Incident TB occurs commonly after ART initiation. Although associated with higher post-ART TB risk, baseline CRP, IP-10, and LPS add limited value to routine patient characteristics in discriminating who develops active TB. Besides determining ideal cutoffs for these biomarkers, additional biomarkers should be sought that predict TB disease in ART initiators.
