AI Bridges the Gap Between Cancer Clinical Trials and Real-World Patients

By leveraging real-world data and machine learning, scientists bridge the gap between clinical trials and real-world patients, empowering personalized cancer care like never before.

Research: Evaluating generalizability of oncology trial results to real-world patients using machine learning-based trial emulations. Image Credit: PeopleImages.com - Yuri A / Shutterstock

A new study led by Winship Cancer Institute of Emory University and Abramson Cancer Center of the University of Pennsylvania researchers demonstrates that a first-of-its-kind platform using artificial intelligence (AI) could help clinicians and patients assess whether and how much an individual patient may benefit from a particular therapy being tested in a clinical trial.

This AI platform can help make informed treatment decisions, understand the expected benefits of novel therapies, and plan future care.

The study, published in the journal Nature Medicine, was led by board-certified medical oncologist Ravi B. Parikh, MD, MPP, medical director of the Data and Technology Applications Shared Resource at Winship Cancer Institute of Emory University and associate professor in the Department of Hematology and Medical Oncology at Emory University School of Medicine, who develops and integrates AI applications to improve the care of patients with cancer.

Qi Long, PhD, a professor of Biostatistics and Computer and Information Science, founding director of the Center for Cancer Data Science at the University of Pennsylvania, and associate director for Quantitative Data Science of the Abramson Cancer Center of Penn Medicine, was the co-senior author.

The study's first author was Xavier Orcutt, MD, a trainee in Parikh's lab. Other study authors included Kan Chen, a PhD student training in Long's lab, and Ronac Mamtani, an associate professor of medicine at the University of Pennsylvania.

Parikh and his fellow researchers developed "TrialTranslator," a machine-learning framework for systematically evaluating clinical trial results in real-world populations. By emulating 11 landmark cancer clinical trials using electronic health records (EHR) from Flatiron Health, they recapitulated actual clinical trial findings, enabling them to identify distinct groups of patients who may respond well to treatments in a clinical trial and those who may not.

"This framework leverages machine-learning models to stratify patients into prognostic risk groups, offering tailored insights for clinicians," Parikh explains. "We hope that this AI platform will provide a framework to help doctors and patients decide if the results of a clinical trial can apply to individual patients. Furthermore, this study may help researchers identify subgroups in whom novel treatments do not work, spurring newer clinical trials for those high-risk groups."

"Our work demonstrates the enormous potential of leveraging AI/ML to harness the power of rich, yet complex real-world data to advance precision medicine at its best," adds Long.

The TrialTranslator workflow consists of two steps. Step I involves the development of prognostic models tailored to specific cancer types, designed to predict the risk of patient-level mortality from time of metastatic diagnosis. The GBM emerged as the top-performing model for all four cancers and was selected for the trial emulation phase. Step II encompasses the trial emulation process, which unfolds in three parts: (1) eligibility matching: real-world patients are selected who meet key eligibility criteria (EC) from landmark clinical trials; (2) prognostic phenotyping: within each trial, selected patients are stratified based on prognostic risk scores using the cancer-specific GBM; and (3) survival analysis: IPTW-adjusted Kaplan–Meier (KM) survival curves are calculated for each prognostic phenotype to evaluate treatment effect and compare to respective RCT results.

Limited Generalizability of Trial Results

Parikh explains that clinical trials of potential new treatments are limited because less than 10% of all patients with cancer participate in a clinical trial. This means clinical trials often do not represent all patients with that cancer. Even if a clinical trial shows a novel treatment strategy has better outcomes than the standard of care, "there are many patients in whom the novel treatment does not work," Parikh says.

"This framework and our open-source web-based calculators will allow patients and doctors to decide whether results from phase III clinical trials are applicable to individual patients with cancer," he says, adding that "this study offers a platform to analyze the real-world generalizability of other randomized trials, including trials that have had negative results."

How They Did Their Analysis

Parikh and colleagues used data from a large, nationwide database of EHR maintained by Flatiron Health, emulating 11 landmark randomized controlled trials (studies that compare the effects of different treatments by randomly assigning participants to groups) that investigated anticancer regimens considered standard of care for the four most prevalent advanced solid malignancies in the United States: advanced non-small cell lung cancer, metastatic breast cancer, metastatic prostate cancer, and metastatic colorectal cancer.

What They Found

Their analysis revealed that patients with low- and medium-risk phenotypes, defined using machine-learning models to assess prognosis, had survival times and treatment-associated survival benefits similar to those observed in the randomized controlled trials. In contrast, those with high-risk phenotypes showed significantly lower survival times and treatment-associated survival benefits compared to the randomized controlled trials.

Their findings suggest that machine learning can identify groups of real-world patients in whom randomized controlled trial results are less generalizable. This means, they add, that "real-world patients likely have more heterogeneous prognoses than randomized controlled trial participants."

Why This Is Important

The research team concludes that the study "suggests that patient prognosis, rather than eligibility criteria, better predicts survival and treatment benefit." They recommend that prospective trials "should consider more sophisticated ways of evaluating patients' prognosis upon entry, rather than relying solely on strict eligibility criteria."

Furthermore, they cite recommendations by the American Society of Clinical Oncology and Friends of Cancer Research that efforts should be made to improve the representation of high-risk subgroups in randomized controlled trials "considering that treatment effects for these individuals might differ from those of other participants."

The study also included rigorous robustness checks, such as sensitivity analyses and validation using semi-synthetic data simulations, to confirm the reliability of the findings.

As to the role of AI in studies such as this one, Parikh says, "Soon, with appropriate oversight and evidence, there will be an increasing tide of AI-based biomarkers that can analyze pathology, radiology, or electronic health record information to help predict whether patients will or will not respond to certain therapies, diagnose cancers earlier, or result in better prognoses for our patients."

This research was supported by grants from the National Institutes of Health: K08CA263541, P30CA016520, and U01CA274576.