Molecular Automation

Photo by Susan Urmy

As medical science advances upon the molecular drivers of different cancers, and as cancer treatment consequently becomes more targeted and precise, clinical testing for molecular alterations has come online over the past decade to help clinicians match patients to these newer treatments.

“We have pills that allow us to target certain genomic alterations in tumors, usually with very few toxicities,” said thoracic oncologist and informaticist Travis Osterman, DO, MS, assistant professor of Biomedical Informatics and Medicine and director of Cancer Clinical Informatics at Vanderbilt-Ingram Cancer Center. “Oftentimes we’re able to maintain patients on that treatment for months or years before needing to think about changing to our more traditional approaches like chemotherapy.”

Like many other clinical tests, these molecular tests produce structured data, that is, discrete values that can be clearly represented in tabular format in information systems. And unlike information that may be buried in text or images, structured data can readily be put to work by information systems for things like automated clinical decision support, or for electronic searches that clinicians might undertake to find health care outcome patterns among their patients.

Today, however, many treatment centers post these actionable test results into the electronic health record (EHR), not as structured data, but as uploaded text files, as PDFs. Osterman is leading the implementation of a new, dedicated genomics module at Vanderbilt University Medical Center (VUMC) that will make these test results much easier to access.

VUMC’s EHR software comes from health information technology behemoth Epic Systems Corp., based in Verona, Wisconsin. Osterman is working to turn on Epic’s new genomics module at VUMC by summer 2021.

In this vital question of which cancer patients should get which targeted therapies, Osterman highlights four molecular profiling modalities that loom large.

• Gene panel: This involves full genomic sequencing of a few hundred selected genes in a tumor sample and a control sample (blood or saliva), searching for DNA alterations in the tumor that give rise to any of the known and treatable protein alterations in cancer.
• FISH, or fluorescence in situ hybridization: fusion proteins, some of which are implicated in cancer and its treatment, arise through the joining of two or more different genes, and to search for evidence of these, FISH maps selected genes or portions of genes.
• RNA sequencing (RNA-Seq): the principal role of RNA (ribonucleic acid) is carrying instructions from DNA for the synthesis of proteins; sequencing of this messenger molecule alerts clinicians to the overexpression of amplified genes, useful for cancer prognosis and diagnosis and as a marker of drug resistance.
• Immunohistochemistry: in cancer, this microscopy-based technique is used to examine cellular abnormalities, including those giving evidence of treatment options such as immunotherapy.

The upcoming change to structured results for these tests will be a boon for population health management.

“For the clinician who’s deciding what treatment to use in a given patient’s case, the PDF works fine,” he said. “But what happens when a new targeted drug suddenly becomes available? When we have structured versions of all of these test data, we can sift through our entire patient population and contact oncologists about any specific patients of theirs who might stand to benefit.”

With Epic’s genomics module turned on, clinician orders will be transmitted electronically to outside laboratories that conduct these tests, and VUMC’s lab information system will be able to track fulfillment automatically instead of manually. As new results are returned, alerts will pop up in the EHR message inboxes of clinicians, and, for the first time, they will be able to filter their patients’ records by genomic test results. Structured results for these molecular tests will also allow automatic alerts to fire whenever a result appears to match a patient for recruitment to a given clinical trial. Crucially, the new genomics module will also integrate up to 10 years of backlogged structured data into the current EHR, providing a comprehensive basis for prospective automated clinical decision support around experimental treatments and newly approved drugs.

“We have new drug approvals in the oncology space that sometimes come out at an alarming rate,” Osterman said, “and it’s hard for clinicians to keep up on that. The same thing is true of clinical trial openings.”

He says the practice of oncology at VICC can also benefit significantly from the new ability of individual clinicians to search through their own patients by molecular test result. “There’s still so much of medicine and oncology that falls in the gaps between evidence-based medicine guidelines, where you have to make your best judgment on what the best treatments are,” he said.

With a new case at hand, the clinician’s ability to search for similar test results can help inform these judgments. “We want to make that easy for providers,” he said.