VA Takes Lead Role in Using Informatics to Validate and Translate Biomarkers

Uncategorized

Clinical informatics – the use of data from the electronic medical record (EMR) to inform the development and improvement of evidence-based medicine, the dissemination and implementation of health outcomes research, and translational research initiatives – can facilitate the validation and translation of biomarkers, said Julie Lynch, PhD, RN, MBA, at the second annual Global Biomarkers Consortium conference.

She spoke of the Veterans Administration (VA) experience in using clinical informatics and how it is uniquely positioned to evaluate the real-time translation of biomarkers to the bedside.

Clinical informatics is different from bioinformatics in that clinical informatics contains data within the EMR. Translational information bridges bioinformatics with clinical information, explained Lynch, nurse scientist at the VA, and a research associate at Dana-Farber Cancer Institute, Boston, Massachusetts. Although structured data sets of sequenced genes do not yet exist in the EMR, “with the $29 billion investment in the HITECH act with a move toward enabling electronic medical records throughout the whole US healthcare system in community hospitals, the possibility of doing that is real,” she said.

Translational informatics requires increased use of standardized data in the EMR, such as tumor SNOMED codes, so that clinical data can inform research.

Delays in the translation of cancer genomics and underrepresentation of minorities in National Cancer Institute (NCI) clinical trials have significant negative consequences for patients and the entire healthcare industry. Lynch used the example of the epidermal growth factor receptor (EGFR) assay, which was commercialized in 2005. In 2010, only 7800 lung cancer patients were being tested for EGFR, and there was a strong indication that black patients were not being tested. “The perception at NCI cancer centers was that everybody was doing EGFR testing in 2010,” she said.

Current diagnostics and drugs are being developed based on analysis of tumor tissue from patients enrolled in clinical trials at NCI hospitals or academic medical centers, but minorities are not well represented at these hospitals. “We have to move the clinical trials to the community, even if it’s a registration trial,” said Lynch. Because the incidence of mutations is very small, “to get the data we need we have to get out into the community, but it’s most likely to take place post approval.”

The VA is uniquely positioned to make significant contributions to translational informatics because each of the 158 VA medical centers and 958 outpatient clinics uses a nationally integrated EMR. The amount of data available from the VA is massive, with 83.6 million outpatient visits and 703,500 inpatient admissions. Each year, 52,000 veterans are diagnosed with cancer, of whom 18.5% are minority cancer patients. The VA has also maintained a cancer registry since 1995. In 2011, this registry began incorporating molecular data, including ALK tyrosine kinase receptor and EGFR mutation status for patients with lung cancer and KRAS for patients with colon cancer.

“The rapid and explosive growth of biomarkers has created challenges for large healthcare systems to consistently and accurately integrate test orders and results into the EMR,” she said. “With genomics becoming an increasingly important component of high-quality cancer care, and with the increasing price of targeted oncology drugs, evaluating appropriate biomarkers and targeted treatments is a priority in the VA.”

A single integrated computerized patient record system is used throughout the VA in all healthcare settings. It delivers an integrated record that covers all aspects of patient care and treatment. “The system allows us to do a key word search of the veteran’s entire medical record. The research that can be conducted using key word search, and natural language processing is substantially better than structured data sets,” said Lynch.

A 2012 analysis of the VA’s structured data sets revealed that there were substantial inconsistencies in the ordering, use, and reporting of biomarker tests within the data sets. Yet key word searches of the clinical notes discovered that the tests were being done but not captured in the structured data sets.

Data from reference laboratories illustrated a 10-fold greater utilization of biomarker tests than what was identified in the VA administrative data.

Factors that impeded accurate reporting of genomics are: 1) tests are entering clinical practice faster than updates can be made to the EMR, 2) most genetic diagnostic tests have been billed using a stack of Current Procedural Terminology (CPT) codes, and 3) in some cases veterans undergo testing outside the VA.

VA national clinical leadership and VA Health Informatics supported the development of a translational informatics application to address this problem.

“We established collaborations with reference laboratories to develop ongoing electronic data interfaces that would capture biomarker test orders and results from all VA medical centers,” she said. A data set that identifies clinical genetic diagnostic test orders and results is being stored in the Veterans Informatics, Information, and Computing Infrastructure.

As well, structured data sets are being adapted to include the 101 newly published CPT codes for clinical genetic diagnostic tests.

Using natural language processing tools, the VA is developing algorithms that will automatically scan the EMR to identify the use of genomic applications not captured by structured data sets.

Related Articles