Use of 18F-FDG PET to Predict Response to First-Line Tuberculostatics in HIV-Associated Tuberculosis

Abstract

This prospective pilot study examined the relationship between the severity and extent of tuberculosis as assessed by (18)F-FDG PET at the time of diagnosis and response to treatment or treatment failure at 4 mo. Twenty-four consecutive HIV patients with newly diagnosed tuberculosis were prospectively included in the study after providing written informed consent. Seventeen patients had pulmonary tuberculosis, and 7 patients had extrapulmonary tuberculosis. All patients underwent whole-body (18)F-FDG PET; none were receiving tuberculostatics at the time of the PET investigation. After undergoing (18)F-FDG PET, the patients were given tuberculosis treatment (the classic triad: isoniazid, rifampicin, and ethambutol) and reevaluated for treatment response: monthly assessment of sputum, smears, and cultures in patients who proved positive at the time of diagnosis, and clinical and radiologic (when relevant) assessment 4 mo after treatment instigation in all patients. Quantitative (18)F-FDG PET results (averaged (18)F-FDG maximum standardized uptake value [SUVmax] derived from early and delayed imaging), percentage change in SUVmax, and number of involved lymph node bastions were related to treatment response or failure. Age, sex, viral load, CD4 status, duration of HIV treatment, SUVmax of lung and splenic lesions (early and delayed), and percentage change in SUVmax of lymph nodes were not significantly different between responders and nonresponders (P ≥ 0.3). In contrast, SUVmax of involved lymph node bastions (both early and delayed) and number of involved lymph node bastions were significantly higher in nonresponders than in responders (respective P values were 0.03, 0.04, and 0.002). Using a cutoff of 5 or more involved lymph node bastions, responders could be separated from nonresponders with a sensitivity, specificity, and positive and negative predictive value of, respectively, 88%, 81%, 70%, and 93%. Using a cutoff of 8.15 for early SUVmax of lymph node bastions and of 10 for late SUVmax of lymph node bastions, a comparable sensitivity of 88% came at the cost of a lower specificity: 73% and 67%, respectively. In this pilot study, a cutoff of 5 or more involved lymph node bastions allowed for separation of tuberculostatic responsive and nonresponsive tuberculosis-infected HIV patients with a sensitivity of 88%, a specificity of 81%, and a negative predictive value of 93%. These findings warrant confirmation by additional studies on larger cohorts of patients.