Predictive value of TNF-α, IFN-γ, and IL-10 for tuberculosis among recently exposed contacts in the United States and Canada
M. R. Reichler, Christina S. Hirsch, Yan Yuan, Awal Khan, Susan E. Dorman, Neil W. Schluger, Timothy R. Sterling, I Bakhtawar, Catherine Ledoux, James McAuley, J Beison, M Fitzgerald, Monika Naus, Motohiro Nakajima, Neil W. Schluger, Yael Hirsch‐Moverman, Jennifer L. Moran, Henry M. Blumberg, J Tapia, L Singha, E Hershfeld, B. Roche, Bonita T. Mangura, Alberto Fernández de Sevilla, Tina M. Sterling, T Chavez-Lindell, Fernanda Maruri, Susan E. Dorman, Wendy Cronin, Elizabeth Munk, Abdul Qadir Khan, Ye Yuan, Bo Chen, Fenglian Yan, Y Shen, Hui Zhao, Hui Zhang, Patricia A. Bessler, M Fagley, M. R. Reichler, Tina M. Sterling, J Tapia, Christina S. Hirsch, Cheng Luo
Abstract
BACKGROUND: We examined cytokine immune response profiles among contacts to tuberculosis patients to identify immunologic and epidemiologic correlates of tuberculosis. METHODS: We prospectively enrolled 1272 contacts of culture-confirmed pulmonary tuberculosis patients at 9 United States and Canadian sites. Epidemiologic characteristics were recorded. Blood was collected and stimulated with Mycobacterium tuberculosis culture filtrate protein, and tumor necrosis factor (TNF-α), interferon gamma (IFN-γ), and interleukin 10 (IL-10) concentrations were determined using immunoassays. RESULTS: Of 1272 contacts, 41 (3.2%) were diagnosed with tuberculosis before or < 30 days after blood collection (co-prevalent tuberculosis) and 19 (1.5%) during subsequent four-year follow-up (incident tuberculosis). Compared with contacts without tuberculosis, those with co-prevalent tuberculosis had higher median baseline TNF-α and IFN-γ concentrations (in pg/mL, TNF-α 129 versus 71, P < .01; IFN-γ 231 versus 27, P < .001), and those who subsequently developed incident tuberculosis had higher median baseline TNF-α concentrations (in pg/mL, 257 vs. 71, P < .05). In multivariate analysis, contact age < 15 years, US/Canadian birth, and IFN or TNF concentrations > the median were associated with co-prevalent tuberculosis (P < .01 for each); female sex (P = .03) and smoking (P < .01) were associated with incident tuberculosis. In algorithms combining young age, positive skin test results, and elevated CFPS TNF-α, IFN-γ, and IL-10 responses, the positive predictive values for co-prevalent and incident tuberculosis were 40 and 25%, respectively. CONCLUSIONS: Cytokine concentrations and epidemiologic factors at the time of contact investigation may predict co-prevalent and incident tuberculosis.