Applying interpretable machine learning workflow to evaluate exposure-response relationships for large-molecule oncology drugs

AI-Driven Applications in Clinical Pharmacology and Translational Science: Insights From the ASCPT 2024 AI Preconference

Artificial intelligence (AI) is driving innovation in clinical pharmacology and translational science with tools to advance drug development, clinical trials, and patient care. This review summarizes the key takeaways from the AI preconference at the American Society for Clinical Pharmacology and Therapeutics (ASCPT) 2024 Annual Meeting in Colorado Springs, where experts from academia, industry, and regulatory bodies discussed how AI is streamlining drug discovery, dosing strategies, outcome assessment, and patient care. The theme of the preconference was centered around how AI can empower clinical pharmacologists and translational researchers to make informed decisions and translate research findings into practice. The preconference also looked at the impact of large language models in biomedical research and how these tools are democratizing data analysis and empowering researchers. The application of explainable AI in predicting drug efficacy and safety, and the ethical considerations that should be applied when integrating AI into clinical and biomedical research were also touched upon. By sharing these diverse perspectives and real-world examples, this review shows how AI can be used in clinical pharmacology and translational science to bring efficiency and accelerate drug discovery and development to address patients' unmet clinical needs.

A Tutorial and Use Case Example of the eXtreme Gradient Boosting (XGBoost) Artificial Intelligence Algorithm for Drug Development Applications

Approaches to artificial intelligence and machine learning (AI/ML) continue to advance in the field of drug development. A sound understanding of the underlying concepts and guiding principles of AI/ML implementation is a prerequisite to identifying which AI/ML approach is most appropriate based on the context. This tutorial focuses on the concepts and implementation of the popular eXtreme gradient boosting (XGBoost) algorithm for classification and regression of simple clinical trial-like datasets. Emphasis is placed on relating the underlying concepts to the code implementation. In doing so, the aim is for the reader to gain knowledge about the underlying algorithm and become better versed with how to implement the algorithm functions for relevant clinical drug development questions. In turn, this will provide practical ML experience which can be applied to algorithms and problems beyond the scope of this tutorial.

Multi-output prediction of dose-response curves enables drug repositioning and biomarker discovery

Drug response prediction is hampered by uncertainty in the measures of response and selection of doses. In this study, we propose a probabilistic multi-output model to simultaneously predict all dose-responses and uncover their biomarkers. By describing the relationship between genomic features and chemical properties to every response at every dose, our multi-output Gaussian Process (MOGP) models enable assessment of drug efficacy using any dose-response metric. This approach was tested across two drug screening studies and ten cancer types. Kullback-leibler divergence measured the importance of each feature and identified EZH2 gene as a novel biomarker of BRAF inhibitor response. We demonstrate the effectiveness of our MOGP models in accurately predicting dose-responses in different cancer types and when there is a limited number of drug screening experiments for training. Our findings highlight the potential of MOGP models in enhancing drug development pipelines by reducing data requirements and improving precision in dose-response predictions.

Dose Optimization in Oncology Drug Development: An International Consortium for Innovation and Quality in Pharmaceutical Development White Paper

The landscape of oncology drug development has witnessed remarkable advancements over the last few decades, significantly improving clinical outcomes and quality of life for patients with cancer. Project Optimus, introduced by the U.S. Food and Drug Administration, stands as a groundbreaking endeavor to reform dose selection of oncology drugs, presenting both opportunities and challenges for the field. To address complex dose optimization challenges, an Oncology Dose Optimization IQ Working Group was created to characterize current practices, provide recommendations for improvement, develop a clinical toolkit, and engage Health Authorities. Historically, dose selection for cytotoxic chemotherapeutics has focused on the maximum tolerated dose, a paradigm that is less relevant for targeted therapies and new treatment modalities. A survey conducted by this group gathered insights from member companies regarding industry practices in oncology dose optimization. Given oncology drug development is a complex effort with multidimensional optimization and high failure rates due to lack of clinically relevant efficacy, this Working Group advocates for a case-by-case approach to inform the timing, specific quantitative targets, and strategies for dose optimization, depending on factors such as disease characteristics, patient population, mechanism of action, including associated resistance mechanisms, and therapeutic index. This white paper highlights the evolving nature of oncology dose optimization, the impact of Project Optimus, and the need for a tailored and evidence-based approach to optimize oncology drug dosing regimens effectively.

Realizing the promise of Project Optimus: Challenges and emerging opportunities for dose optimization in oncology drug development

Project Optimus is a US Food and Drug Administration Oncology Center of Excellence initiative aimed at reforming the dose selection and optimization paradigm in oncology drug development. This project seeks to bring together pharmaceutical companies, international regulatory agencies, academic institutions, patient advocates, and other stakeholders. Although there is much promise in this initiative, there are several challenges that need to be addressed, including multidimensionality of the dose optimization problem in oncology, the heterogeneity of cancer and patients, importance of evaluating long-term tolerability beyond dose-limiting toxicities, and the lack of reliable biomarkers for long-term efficacy. Through the lens of Totality of Evidence and with the mindset of model-informed drug development, we offer insights into dose optimization by building a quantitative knowledge base integrating diverse sources of data and leveraging quantitative modeling tools to build evidence for drug dosage considering exposure, disease biology, efficacy, toxicity, and patient factors. We believe that rational dose optimization can be achieved in oncology drug development, improving patient outcomes by maximizing therapeutic benefit while minimizing toxicity.

Machine Learning-Based Quantification of Patient Factors Impacting Remission in Patients With Ulcerative Colitis: Insights from Etrolizumab Phase III Clinical Trials

Etrolizumab, an investigational anti-β7 integrin monoclonal antibody, has undergone evaluation for safety and efficacy in phase III clinical trials on patients with moderate to severe ulcerative colitis (UC). Etrolizumab was terminated because mixed efficacy results were shown in the induction and maintenance phase in patients with UC. In this post hoc analysis, we characterized the impact of explanatory variables on the probability of remission using XGBoost machine learning (ML) models alongside with the SHapley Additive exPlanations framework for explainability. We used patient-level data encompassing demographics, physiology, disease history, clinical questionnaires, histology, serum biomarkers, and etrolizumab drug exposure to develop ML models aimed at predicting remission. Baseline covariates and early etrolizumab exposure at week 4 in the induction phase were utilized to develop an induction ML model, whereas covariates from the end of the induction phase and early etrolizumab exposure at week 4 in the maintenance phase were used to develop a maintenance ML model. Both the induction and maintenance ML models exhibited good predictive performance, achieving an area under the receiver operating characteristic curve (AUROC) of 0.74 ± 0.03 and 0.75 ± 0.06 (mean ± SD), respectively. Compared with placebo, the highest tertile of etrolizumab exposure contributed to 15.0% (95% confidence interval (CI): 9.7-19.9) and 17.0% (95% CI: 8.1-26.4) increases in remission probability in the induction and maintenance phases, respectively. Additionally, the key covariates that predicted remission were CRP, MAdCAM-1, and stool frequency for the induction phase and white blood cells, fecal calprotectin and age for the maintenance phase. These findings hold significant implications for establishing stratification factors in the design of future clinical trials.

Machine Learning for Exposure-Response Analysis: Methodological Considerations and Confirmation of Their Importance via Computational Experimentations

Exposure-response (E-R) is a key aspect of pharmacometrics analysis that supports drug dose selection. Currently, there is a lack of understanding of the technical considerations necessary for drawing unbiased estimates from data. Due to recent advances in machine learning (ML) explainability methods, ML has garnered significant interest for causal inference. To this end, we used simulated datasets with known E-R "ground truth" to generate a set of good practices for the development of ML models required to avoid introducing biases when performing causal inference. These practices include the use of causal diagrams to enable the careful consideration of model variables by which to obtain desired E-R relationship insights, keeping a strict separation of data for model-training and for inference generation to avoid biases, hyperparameter tuning to improve the reliability of models, and estimating proper confidence intervals around inferences using a bootstrap sampling with replacement strategy. We computationally confirm the benefits of the proposed ML workflow by using a simulated dataset with nonlinear and non-monotonic exposure-response relationships.