Amikacin Concentrations Predictive of Ototoxicity in Multidrug-Resistant Tuberculosis Patients

Abstract

Aminoglycosides, such as amikacin, are used to treat multidrug-resistant tuberculosis. However, ototoxicity is a common problem and is monitored using peak and trough amikacin concentrations based on World Health Organization recommendations. Our objective was to identify clinical factors predictive of ototoxicity using an agnostic machine learning method. We used classification and regression tree (CART) analyses to identify clinical factors, including amikacin concentration thresholds that predicted audiometry-confirmed ototoxicity among 28 multidrug-resistant pulmonary tuberculosis patients in Botswana. Amikacin concentrations were measured for all patients. The quantitative relationship between predictive factors and the probability of ototoxicity were then identified using probit analyses. The primary predictors of ototoxicity on CART analyses were cumulative days of therapy, followed by cumulative area under the concentration-time curve (AUC), which improved on the primary predictor by 87%. The area under the receiver operating curve was 0.97 on the test set. Peak and trough were not predictors in any tree. When algorithms were forced to pick peak and trough as primary predictors, the area under the receiver operating curve fell to 0.46. Probit analysis revealed that the probability of ototoxicity increased sharply starting after 6 months of therapy to near maximum at 9 months. A 10% probability of ototoxicity occurred with a threshold cumulative AUC of 87,232 days · mg · h/liter, while that of 20% occurred at 120,000 days · mg · h/liter. Thus, cumulative amikacin AUC and duration of therapy, and not peak and trough concentrations, should be used as the primary decision-making parameters to minimize the likelihood of ototoxicity in multidrug-resistant tuberculosis.

FIG 1

Observed concentrations versus model prediction. A two-compartment model (central compartment or serum versus peripheral compartment) described the amikacin pharmacokinetics in MDR-TB patients in Botswana well, with an r² of 0.997.

FIG 2

Distribution of amikacin concentrations in the 28 patients. (A) Peak concentrations were normally distributed based on the D'Agostino and Pearson omnibus normality test (P = 0.774). The ratio of the lowest to highest peak concentration was 3.49. (B) Trough concentrations were not normally distributed (P < 0.001). Fifty percent of patients had a trough below the limits of detection. (C) The AUCs were normally distributed (P = 0.223); the ratio of the lowest to highest AUC was 4.09. (D) The cumulative AUCs were not normally distributed (P < 0.001) and had an even wider range, with a ratio of highest to lowest cumulative AUC of 33.67.

FIG 3

Optimum classification and regression analysis tree to identify concentration thresholds. The tree identified cumulative AUC thresholds of 54,519 and 91,914 days · mg · h/liter.

FIG 4

Probability of ototoxicity as a function of three different parameters. The probability of ototoxicity as a function of cumulative AUC is sigmoidal, with the TC₅₀ of 87,232 days · mg · h/liter shown in the equation. This value is very close to the 91,914 days · mg · h/liter identified by CART. The more continuous relationship between concentration and toxicity can be used to calculate the cumulative AUC versus the probability and risk of toxicity that can be tolerated by clinicians and patients. (B) Probability of ototoxicity as a function of patient weight. Since this is a nonmodifiable patient factor at the time when a patient presents with MDR-TB, it would not factor much in the decision making. (C) Probability of ototoxicity as a function of duration of therapy. The risk increases at the inflection point of 6 months and is a near certainty at 9 months.