Incidence of amikacin ototoxicity: a sigmoid function of total drug exposure independent of plasma levels.

A sigmoid curve was found to closely describe the relationship between the incidence of amikacin ototoxicity (greater than or equal to 15 dB hearing loss at a given frequency) and either (1) total dose, or (2) the area under the curve (AUC) describing plasma drug concentration v time over the total period of amikacin administration (total AUC) in continuously infused guinea pigs. Total dose or total AUC estimates of the drug exposure required to produce ototoxicity in 50% of the animals (ED50s) were not significantly different over an eight-fold range of dosing rates or plasma concentrations. A theoretical explanation for this result is that ototoxicity occurs only when a critical amount of drug is accumulated at the ototoxic site by an essentially unidirectional process with a rate that is slow and linearly related to the extracellular drug concentration. The sigmoid relationships for pooled data were parallel in slope for all hearing frequencies from 2 to 32 kHz, and the ED50s showed a strong negative linear relationship to the log of the hearing frequency over this range. The magnitude of ototoxicity expressed as the number of octaves (frequency ratios of 2) for which hearing loss damage was continuous from 32 kHz downward, was correlated to both total dose (r = .605) and total AUC (r = 0.703). No relationship between ototoxicity and plasma level or dosing rate was found. The extreme steepness of the dose-effect curve for the incidence of ototoxicity greatly amplified the variability between individuals and offers an explanation for the unpredictability of aminoglycoside ototoxicity in human patients. The results indicate that either total dose or total AUC (in cases of highly unpredictable blood levels), and not peak or trough serum levels, should be used as an index of ototoxic risk and that the safety limits of drug exposure should be set conservatively.