Artificial Intelligence and Amikacin Exposures Predictive of Outcomes in Multidrug-Resistant Tuberculosis Patients

Aminoglycosides such as amikacin continue to be part of the backbone of treatment of multidrug-resistant tuberculosis (MDR-TB). We measured amikacin concentrations in 28 MDR-TB patients in Botswana receiving amikacin therapy together with oral levofloxacin, ethionamide, cycloserine, and pyrazinamide and calculated areas under the concentration-time curves from 0 to 24 h (AUC 0–24 ). The patients were followed monthly for sputum culture conversion based on liquid cultures. The median duration of amikacin therapy was 184 (range, 28 to 866) days, at a median dose of 17.30 (range 11.11 to 19.23) mg/kg. Only 11 (39%) patients had sputum culture conversion during treatment; the rest failed. We utilized classification and regression tree analyses (CART) to examine all potential predictors of failure, including clinical and demographic features, comorbidities, and amikacin peak concentrations ( C max ), AUC 0–24 , and trough concentrations. The primary node for failure had two competing variables, C max of 0–24 of 41 kg was a secondary node with a score of 35% relative to the primary node. The area under the receiver operating characteristic curve for the CART model was an R 2 = 0.90 on posttest. In patients weighing >41 kg, sputum conversion was 3/3 (100%) in those with an amikacin C max of ≥67 mg/liter versus 3/15 (20%) in those with a C max of C max and AUC 0–24 below the threshold, 7/7 (100%) failed, compared to 7/15 (47%) of those who had these parameters above threshold (RR = 2.14; 95% CI, 1.25 to 43.68). These amikacin dose-schedule patterns and exposures are virtually the same as those identified in the hollow-fiber system model.