Optimizing Phase III Trials in Pancreatic Cancer - ESMO Poster 2024
ESMO Showcases The Aurora Suite™ Approach
Our successful poster submission to the European Society of Medical Oncology (ESMO) Congress highlights how the Aurora Suite™ focuses on science, key data, and actionable insights to optimize trial strategy. By utilizing our technology, you can save time, reduce human error, gain improved visibility of risks, and move forward with a reliable trial strategy optimized for your goals.
Revolutionizing Clinical Trials: Harnessing Artificial Intelligence for Optimized Phase III Studies in Pancreatic Cancer.
M.-E. Bonneterre MD (1), B. Lekh (2), O. Gudmundsen (3)
(1) Medical Oncology, Endpoint, Marcq En Baroeul, France, (2) Trial Optimization and Innovation, Aurora Analytica AS, Oslo, Norway, (3) Strategy, Biomedical Visions AS, Oslo, Norway
Background
Pancreatic cancer (PC) constitutes over 3% of new cancer diagnoses in Europe, yet accounts for 7% of cancer deaths, ranking as the fourth leading cause of cancer mortality (1). PC remains lethal in over 80% of patients, with a 5-year relative survival rate of 13% (2,3). Although new therapies are explored, interventional clinical trials for PC represent merely 5.13% of global cancer trials. Addressing this high unmet need necessitates probing questions to enhance patient access to novel therapies and ensure robust trial design and execution.
We conducted research into key selection criteria for pancreatic cancer patients, including:
Incidence Areas: North America, parts of Europe, and highest incidence rates within Uruguay, Hungary, Japan, France, Austria, Czechia, Latvia, and Sweden.
Site Profile: University hospitals and cancer centres, or large private hospitals with oncology activity.
Investigators/KOLs: Oncologists specializing in digestive oncology, gastroenterologists and related fields.
Treatment: Most current first line therapy for BrCA mutated mPC is a platinum-based chemotherapy.
Figure 1 | Age-Standardised Incidence Rate per 100k population of Pancreatic Cancer. Source: WHO, Aurora Suite™, 2024.
Methods
Utilizing the Aurora Suite™, we generated data aggregation, visualization, and analysis. The Aurora Suite™ is an advanced web application that leverages machine learning and AI to support the design, planning, and execution of clinical trials. Our approach integrated public domain and primary data sources. The trial NCT02184195 (POLO)- served as a reference and was compared to Aurora Suite™ results for developing trial optimization recommendations (4).
We leveraged a range of data sources to input into the technology platform. Data was aggregated, transformed and analysed using the Aurora Suite™ to create a model and evaluate potential optimizations based on the completed trial and published report.
Aurora Analytica Trial Optimization Methodology for ESMO Abstract Poster Session, 2024.
Results
Using the Aurora Suite® and advanced AI capabilities in trial optimization, we compared the phase III PC reference trial 'POLO' with our results. We identified that POLO's country distribution includes a mix of favorable and non-favorable countries: Tier 1 (most favorable), Tier 2 (moderately favorable), Tier 3 (minimally favorable). The Aurora Suite® identified additional countries and sites that could have been used to optimize POLO’s trial strategy and enrollment. POLO enrolled 154 participants across 103 sites over 49 months, achieving a 0.03 patients/site/month (p/s/m) enrollment rate. Our analysis indicates an industry median enrollment of 0.12 p/s/m. The Aurora Suite® could have enabled POLO to enroll more than 200% higher, whilst reducing sites required and enrollment time by 30%.
Using the interactive analytics report we implemented the Aurora model and compared to our chosen reference trial.
Figure 2 | Utilization of The Aurora Suite™ to Conduct Trial Optimization. A: Enhanced Knowledge Operator (EKO™) Algorithm Alignment Profile. Competition: 11%, Experience: 47%, Operations: 5%, Other: 0%, Patients: 37%. B: Country Ranking - Global Map. Dark Purple: Tier 1 (most favourable), Purple: Tier 2 (favourable), Light Purple: Tier 3 (least favourable). C: Country Ranking - Regional Chart. Dark Purple: Tier 1 (most favourable), Purple: Tier 2 (favourable), Light Purple: Tier 3 (least favourable). D: Geo Matrix™ Data Map. High Risk Zones Identified: 8. Source: Geo Matrix™ Decision Engine, 2024.
Figure 3 | Enrolment Rate Benchmark Analysis (Patients/Site/Month) in Phase III. Pancreatic Cancer (only) - Total Trials: 91, Trials Used: 54 (Results: 23, No Results: 31), Indication Enrolment Median: 0.14 p/s/m (Phase III: 0.12). Pancreatic Cancer First Line (only) - Total Trials: 14, Trials Used: 3 (Results: 2, No Results: 1), First Line Enrolment Median: 0.12 p/s/m (Phase III: 0.03). Source: Aurora Consulting™, 2024.
Figure 4 | Operational Strategy For Reference Study: Original - POLO, & Optimized - Aurora Model. POLO; ClinicalTrials.gov ID: NCT02184195, Countries: 12, Enrolment: 154 patients, Sites Activated: 103, Enrolment Months: 49, Enrolment Rate Achieved: 0.03 p/s/m, Primary Completion Date: 2019-01-15. Aurora Model; Countries: 8, Enrolment Target: 154 patients, Optimized Total Sites: 55, Optimized Enrolment Months: 36, Optimized Enrolment Rate: 0.09 p/s/m. T1: Tier 1 (most favourable), T2: Tier 2 (favourable), T3: Tier 3 (least favourable). Source: Aurora Suite™, 2024.
Figure 5 | Site Activation and Enrolment Simulation of POLO and Aurora Model (Optimized). POLO; Start Date: Month 1, First Patient In: Month 8, Last Patient In: Month 56, First Site Activated: Month 8, Last Site Activated: Month 14. Aurora Model; Start Date: Month 1, First Patient In: Month 8, Last Patient In: Month 43, First Site Activated: Month 8, Last Site Activated: Month 11. Source: Simulation-L™ Decision Engine, 2024.
Discussion
The optimized Aurora model is expected to yield positive results (Figure 5) by improving overall study timelines and key milestones. The model is designed to achieve faster site activation and enrolment through a reduction in the complexity of the POLO (4) strategy—specifically, by optimizing and thereby reducing the number of countries and sites involved (Figure 4), and by conducting comprehensive enrolment calibration (Figure 3). These simplifications are anticipated to improve engagement among investigators, operational teams, patients, and other key stakeholders, as well as ease operational constraints (e.g., logistics, project management). The reduction in complexity is planned to be achieved by enhancing the visibility of strategic risks, such as regulatory start-up timelines, patient access (incidence, Figure 1), access to experienced investigators, and competition (Figure 2, 4). The Aurora Suite™ can be customized with any dataset, allowing users to monitor risks relevant to their investigation. By using this approach, we are enabled to identify potential mitigation strategies, such as the use of back-up countries (Figure 2) and strategic site distribution to address the specific competitive landscape.
However, potential limitations of this model should be acknowledged. These limitations may stem from the data used during the planning phase, as the model’s effectiveness in practice may be influenced by real-world factors that are difficult to predict at this stage. While the model has been optimized based on current data and assumptions, various practical factors—such as site-level interest, regulatory conditions, and unforeseen operational challenges—could impact the overall outcome if the strategy were to be implemented. To enhance the model’s predictive accuracy, future planning could incorporate additional data, such as electronic health records for pancreatic cancer patients, site-level interest metrics, site enrolment rate estimations, and more granular regulatory timeline data.
Conclusions
Innovative data-driven decision tools offer potential for expediting PC clinical trials, reducing trial duration, and providing enhanced efficiency leading to accelerated market access for new therapies and amplified patient benefits.
Accelerated Site Activation: Achieved 21% faster site activation, reducing the start-up period by 3 months through optimized country-site selection.
Faster Last Patient In (LPI): LPI reached 27% faster, reducing the timeline by 13 months, thereby enhancing stakeholder engagement by shortening study timelines.
Operational Simplification: Reduced operational complexity by limiting the number of countries involved and optimizing site selection.
Risk Mitigation: Successfully lowered the overall strategic/operational risk and improved the visibility of high-risk zones.
Enrolment Rate Optimization: Improved enrolment rates by 200%, targeting a baseline median of 0.09 patients per site per month (p/s/m).
Strategy Development Efficiency: Achieved over 80% time savings in strategy development compared to current methods.
Cost Savings: The combined optimizations and utilization of this technology offer significant cost-saving opportunities when compared with other approaches (e.g. manual data wrangling, other commercially available analytics platforms etc).
In conclusion, leveraging advanced technology platforms, supported by AI, such as The Aurora Suite™, can significantly improve the efficiency and optimization of clinical trials. These innovations enhance the probability of success for new treatments being approved for patients, ultimately accelerating the delivery of life-saving therapies.
Future research could focus on further refining site selection optimization to ensure efficient enrolment and improve patient access, thereby maximizing the potential impact of clinical trials.
References
Santucci et al (2024). European cancer mortality predictions for the year 2024 with focus on colorectal cancer. Annals of Oncology, 35(3). https://doi.org/January 28, 2024
Rojas, L.A et al. Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer. Nature 618, 144–150 (2023). https://doi.org/10.1038/s41586-023-06063-y
(n.d.). Surveillance & Health Equity Science. America Cancer Society. https://www.cancer.org/research/surveillance-and-health-equity-science.html
Kindler, H. L. et al(2022). Overall Survival Results From the POLO Trial: A Phase III Study of Active Maintenance Olaparib Versus Placebo for Germline BRCA-Mutated Metastatic Pancreatic Cancer. Journal of Clinical Oncology, 40(34). https://doi.org/July 14 2022
Conroy et al (2023). Pancreatic cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Annals of Oncology, 34(11). https://doi.org/5th September 2023
Mittal, A.; Moore, S.; Navani, V.; Jiang, D.M.; Stewart, D.J.; Liu, G.; Wheatley-Price, P. What Is Ailing Oncology Clinical Trials? Can We Fix Them? Curr. Oncol. 2024, 31, 3738–3751. https://doi.org/10.3390/ curroncol31070275
