The multiple dose pharmacokinetics of pegylated liposomal doxorubicin (PL-DOX), known as DOXIL (US) and CAELYX (EU), was characterized in dogs and a pharmacokinetic/pharmacodynamic model to identify a relationship between drug exposure and the probability of observing treatment-related palmar-plantar erythrodysesthesia (PPE) was developed. Twenty dogs were assigned to PL-DOX groups (2/sex/ group) that received intravenous PL-DOX doses of 0.5 mg/kg ql, 2, or 4 weeks; 1.0 mg/kg q2weeks; or 1.5 mg/kg q4weeks for 12 weeks. Blood was collected for HPLC analysis of doxorubicin concentration pre-dose and periodically up to 120 h after dosing three times during treatment. Plasma drug concentration was modeled using iterative 2-stage analysis. Dermal lesions (PPE) were scored twice weekly for six regions of each dog using a 0-6 severity scale; maximum severity was 36. PPE score data were modeled using an approach in which the % probability of PPE was related to a hypothetical effect site by a series of Hill-type functions. Pharmacokinetics were best modeled as a one-compartment open model. Vss (ml/kg), CLt (ml/hr/kg) and half-life (h) were 44.1, 1.39 and 23.1, respectively. Cmax increased linearly with dose. CLt decreased with repeated doses. 5 A two-compartment pharmacodynamic model, which correctly predicted 97% of the observed lesion severity, was developed to establish the relationship of lesion severity to dose intensity (a measure of drug exposure incorporating the effect of both dose level and dosing frequency, which can be expressed in units of mg/kg/week). The model demonstrated that maximal PPE was positively correlated with dose intensity, the major factor that affects the incidence and severity of dermal lesions. 6 The model can be used to predict acceptable dose intensities in humans utilizing body surface area conversion factors and comparative AUCs for dogs and humans. It predicts that a dose intensity of 10-12.5 mg/m2 of PL-DOX will be well tolerated in patients. The results of recent clinical studies are consistent with this prediction.