Tuberculosis case finding and mortality prediction: added value of the clinical TBscore and biomarker suPAR

SETTING: A suburban area of Bissau, the capital of Guinea-Bissau; the study was conducted among presumptive pulmonary tuberculosis (prePTB) patients seeking medical care for signs and symptoms suggestive of PTB.OBJECTIVE: To determine if a clinical TB score and a biomarker suPAR (soluble urokinase plasminogen activator receptor) have separate and composite ability to predict PTB diagnosis and mortality in prePTB patients.DESIGN: Observational prospective follow-up study conducted from August 2010 to August 2012.RESULTS: We included 1011 prePTB patients (mean age 34 years, 95%CI 33-35); 55% (n = 559) were female and 161 (16%) had human immunodeficiency virus (HIV) infection. Of all included patients, 10% (n = 101) were diagnosed with PTB. Mortality during follow-up was 5% (n = 48), with a mean survival time of 158 days (95%CI 27-289) in prePTB patients diagnosed with PTB vs. 144 days (95%CI 109-178) in those not diagnosed with PTB (P = 0.774). After adjusting for HIV status and age, the best separate predictor was suPAR 5 ng/ml, with a hazard ratio (HR) of 4.6 (95%CI 2.1-9.9) for mortality and 6.7 (95%CI 4.0-11.2) for TB diagnosis. All patients who died had a TBscore II + suPAR 7; the HR of the composite score for subsequent PTB diagnosis was 33.0 (95%CI 4.6-236.6).CONCLUSION: The proposed composite score of suPAR + TBscore II 7 can improve TB case finding and clinical monitoring.