197: Evaluation of the Impact of Physiological Changes of Pregnancy & Cancer on Oncology Drug Pharmacokinetics Towards Implementing a “Cancer-Pregnancy Physiologically Based Pharmacokinetic (PBPK) Population Model”

Statement of Purpose: The treatment of cancer during pregnancy remains challenging with knowledge gaps in drug’s dose, safety and efficacy. Many efforts have focused on physiologically based pharmacokinetic (PBPK) models that separately account for pregnancy or cancer-induced physiological changes to predict drug pharmacokinetics (PK). However, investigation of the joint impact of cancer and pregnancy is lacking. Our research aims to develop a cancer-pregnancy (CP) population model and provide more relevant pharmacokinetics characterization.

Description of Methods & Materials: We conducted literature review of the physiological changes in pregnant and cancer populations which can impact drug PK. Paclitaxel (a CYP3A4 and CYP2C8 substrate, bound to albumin) and docetaxel (a CYP3A substrate, bound to albumin and alpha glycoprotein (AGP)) were selected based on availability of clinical data in pregnant cancer patients. PBPK analyses were conducted in SimCYP (v21), using the software’s library model of paclitaxel and a new model developed for docetaxel. Initially, drug models were verified against clinical PK data in nonpregnant cancer patients, using the cancer population model. Predictions were then compared between healthy volunteers (HV) and cancer population model to identify any key cancer-related physiological changes for inclusion in a CP model. Simulations were conducted using the default healthy pregnancy and CP models, accounting for CYP abundance and binding protein levels. Predictive performance among different population models was evaluated by visual comparison of predicted and observed concentration-time profiles and predicted to observed ratios of maximum plasma concentration (C_max), area under the concentration-time curve (AUC) and clearance (CL).

Data & Results: The cancer population model accurately predicted AUC, C_max and CL values within ±30% of observed values in cancer patients for both paclitaxel and docetaxel. The HV model resulted in similar predictive performance for both drugs. Despite incorporating changes of binding protein levels in cancer population model, with 15% reduction in albumin levels resulting in 17% increase in paclitaxel unbound fraction, and a 2-fold increase in AGP levels resulting in 15% decrease in docetaxel unbound fraction, there was no major impact ( < 15% change) on the predicted CL for both drugs. The healthy pregnancy model appeared to overestimate paclitaxel exposure (40% higher AUC) in pregnant cancer patients. The modified model with 2-fold increase of CYP2C8 expression predicted AUC and CL values within ±25% of observed values. The healthy pregnancy model accurately predicted the PK of docetaxel in pregnant cancer patients, with values within ±25% of observed data.

Interpretation, Conclusion or Significance: In vitro data indicated an increase in CYP2C8 expression during pregnancy, which was tested to improve prediction accuracy of paclitaxel PK in pregnant cancer patients. Changes in binding protein levels in cancer patients had minimal impact on the predicted clearance of paclitaxel and docetaxel. A CP population model has the potential to evaluate the combined effect of cancer and pregnancy on the PK of oncology drugs in pregnant individuals. Additional model evaluation with other drugs and information regarding binding protein levels, CYP expression and clinical PK data in this population is necessary to continually improve our understanding of CP pharmacokinetics.

Disclosures: This project was supported by a research fellowship for Xinxin Yang from the Oak Ridge Inst for Science & Education through an interagency agreement between the Department of Energy and the US Food & Drug Administration.