Predicting clinical trial success for Clostridium difficile infections based on preclinical data

Area of Science:

• Translational science
• Drug discovery
• Infectious disease research

Background:

• Preclinical models are crucial for drug discovery but their clinical translation success is poorly understood.
• Limited success in translating findings from animal models to human patients hinders drug development.

Purpose of the Study:

• To investigate the translational success of treatments for Clostridium difficile infection from preclinical models to human patients.
• To develop a predictive system for enhancing preclinical to clinical translation efficiency.

Main Methods:

• Analysis of preclinical and clinical experiment pairs for Clostridium difficile infection treatments.
• Univariate and multivariate random forest modeling to identify predictors of translation success.
• Development of a recommendation system using key features like dosage and endpoints.

Main Results:

• Only 36% of preclinical-to-clinical experiment pairs showed successful translation.
• Sustained response endpoint and subject age were identified as negative predictors of translation success.
• The developed recommendation system achieved 0.76 accuracy and 0.71 F1 score, boosting translational efficiency by 25% using only 7 features.

Conclusions:

• The study highlights significant challenges in preclinical to clinical translation for Clostridium difficile infection.
• A novel recommendation system can improve translational efficiency and de-risk clinical outcomes.
• The proposed method is adaptable for various diseases, serving as a decision support system for drug discovery.