Bayesian modeling for analyzing heterogeneous response in preclinical mouse tumor models.

Area of Science:

• Oncology
• Biostatistics
• Pharmacology

Background:

• Assessing anticancer treatment efficacy in mouse models is crucial for clinical trial progression.
• Tumor volume data analysis is complicated by response heterogeneity and data loss.
• Traditional statistical methods lack robustness for heterogeneous responses.

Purpose of the Study:

• To develop a statistical method for analyzing heterogeneous in vivo tumor responses.
• To create a translatable assessment comparable to clinical RECIST criteria.
• To improve the accuracy and sensitivity of preclinical treatment efficacy evaluation.

Main Methods:

• Developed INSPECT (IN vivo reSPonsE Classification of Tumors) using Bayesian modeling.
• Classified individual tumor behaviors into nonresponder, modest responder, stable responder, and regressing responder categories.
• Validated the method using published and simulated tumor growth data.

Main Results:

• INSPECT demonstrated higher accuracy and sensitivity than existing methods.
• The method effectively balanced false-negative and false-positive rates.
• A case study highlighted INSPECT's value in drug development and translation.

Conclusions:

• INSPECT provides a robust and translatable method for analyzing heterogeneous in vivo tumor responses.
• The INSPECT methodology enhances the assessment of anticancer drug efficacy in preclinical studies.
• The "INSPECTumours" package offers a user-friendly web interface for analysis and reporting.