Big Data May Guide Future Treatment Decisions

Virginia Powers, PhD
Published Online: Friday, Mar 17, 2017

Dr. Clifford A. Hudis

Clifford A. Hudis, MD, FACP

The paucity of patients participating in clinical trials makes data extrapolation and application complex, calling for a need to explore health technology solutions that tap the potential of real-world data, according to a talk by Clifford Hudis, MD, at the 15th St Gallen Breast Cancer Conference.

“Several factors compromise our ability to make evidence-based recommendations for standard treatments across diverse patient populations and limit our ability to accurately predict real-world benefits,” said Hudis, CEO of ASCO and Consultant in the Breast Medicine Service of Memorial Sloan Kettering Cancer Center. “The longstanding challenge is that we are not treating the patients that we are studying.”

In the United States, only approximately 3% of patients diagnosed with cancer are enrolled in clinical trials. Furthermore, substantial disparity exists in demographics between participants in clinical trials and the general population.

The graph for age distribution at diagnosis of breast cancer has a traditional bell shaped curve and includes ages ranging from 20 to 85 years and older, Hudis noted. However, the superimposed graph of the age of clinical trial participants represented a far younger cohort, with few persons aged 65 to 74 and none older than 74 years.

Although this age bias has recently begun to change, clinical trials of adjuvant therapies for breast cancer generally enroll younger and healthier patients, according to Hudis. “In contrast to the patients we treat, clinical trial participants have a better performance status, and few or no co-morbidities that require few or no co-administered drugs,” he said.

Steps toward obtaining data from a more representative populations were taken with the American Recovery and Reinvestment Act of 2009, which provided $19.2 billion in funding to help hospitals put health information technology systems in place. The act was responsible for an increase in electronic health record usage, primarily with the focus of tracking meaningful use.

Despite electronic record adoption, “one obstacle to curing cancer remained: patient data isn’t shared,” said Hudis. In fact, health information obtained under this act was not interoperable, representing a major hurdle. In general, the electronic health records were primarily maintained so that clinicians could adequately defend their billing practices for costly treatments and therapeutics upon audit. Despite this reality, the door had been opened to create a tool for patient data accumulation and sharing.

ASCO’s solution to this problem was a system called CancerLinQ, noted Hudis, who is the current chair of the big-data initiative’s board of directors. CancerLinQ compiles patient data, within confidentiality guidelines, and allows for data minining and sharing. “The primary purpose of CancerLinQ is to improve the quality of care and to enhance outcomes,” Hudis said.

For patients, potential services will include facilitated matching to clinical trials and safety monitoring during treatment, including side effects in real time and patients reported outcomes. This tool will provide clinical decision support to care providers with observational and guideline-driven second options delivered in real time. Moreover, CancerLinQ offers comparative data on the effectiveness of “off-label” use and may be hypothesis generating for researchers, Hudis noted.

By March 2017, nearly 2 million records had been incorporated in the system from 80 oncology care settings, which ranged from individual practices to large cancer centers. The system capitalizes on the data already being entered. In an average day, about 40% of a clinicians time is now spent on record entering, according to Hudis.

“This is time not spent with patients; the solution is the use of scribes, who follow the practitioner throughout the day typing relevant information,” he suggested. With this system, the clinician would be free to interact with, and treat more patients daily but with the drawback of the large expense of using a scribe.

The need for this data could continue to evolve in the near future, thanks in part to the 21st Century Cures Act. This bill, known as HR 34, would require the FDA to “evaluate the use of real-world evidence to support the approval of a new indication for a previously approved drug.” Additionally, the bill “enables the secure exchange of electronic health information ­ and allows for complete access, exchange and use of all electronically accessible health information.”

Hudis stressed that this does not undermine FDA approval of new drugs, which remains evidence-based, but allows for patient information to be used to inform off-label use of an existing drug. He explained that off label use is commonly employed in real-world practice, but now information regarding adverse events, dosage, and applicability to specific patients will be made available to guide this practice.

Hudis concluded: “Conventional research leads us forward but is limited by a narrow pool of subjects, and the expense and long time needed to establish an outcome. In the future, big data sets may augment and extend the high quality evidence from prospective research that includes older patients, comorbidities, concurrent medication and many other real-world implications of successful cancer treatment.”
Hudis C, Schilsky R, Miller R, et al. Extrapolating data from clinical trials to real life: Estimating magnitude of clinical benefit in early breast cancer. Presented at: 15th St. Gallen International Breast Cancer Conference; March 15-18, 2017. Presentation SL 1.1.


<<< View more from the St. Gallen's Breast Cancer Conference



View Conference Coverage
Online CME Activities
TitleExpiration DateCME Credits
Medical Crossfire®: Optimizing Treatment and Management of Soft Tissue Sarcoma in Community OncologyNov 30, 20171.5
Community Practice Connections™: 1st Annual European Congress on Immunotherapies in Cancer™Dec 13, 20171.5
Publication Bottom Border
Border Publication
x