Skip to main content
Premium Trial:

Request an Annual Quote

Vanderbilt's Precision Cancer Medicine App Brings Genomic Data to Point of Care


NEW YORK (GenomeWeb) – Enabling precision medicine at the point of care requires ready access to genomic information within the clinical workflow as well as tools to help clinicians make sense of the information presented to them.

As Electronic Health Records' vendors work to develop functionality that will enable the use of genomic data at the point of care, researchers from Vanderbilt University and elsewhere have developed a prototype of a clinico-genomic mobile application that provides some features that clinicians might use in interactions with patients. It also "demonstrates how to achieve end-to-end integration with a data warehouse operating in near-real time with the accompanying EHR system," the researchers wrote in a paper published earlier this year in the Journal of the American Medical Informatics Association that describes the app.

In its current iteration, the so-called Precision Cancer Medicine (PCM) app lets clinicians visualize mutation information from a single patient in the context of a wider pool of genomic data. Specifically, it displays somatic mutations within a disease-specific population context and includes links to resources as the Gene Wiki for gene level information, Vanderbilt's My Cancer Genome repository for gene-variant-disease information, and HemOnc for disease-specific and gene-specific treatment information. It shows clinicians and patients how common a given mutation is compared to others and how often the mutation shows up in the population compared to other alterations.

The app is designed to run on Apple devices and is based on the Substitutable Medical Applications and Reusable Technology (SMART) health IT platform. It uses an open access application programming interface and genomic extensions added to the Fast Healthcare Interoperability Resources (FHIR) standard from Health Level Seven International, which provide resources for "native representation of molecular profile data." The app can be deployed on different resources within the healthcare ecosystem — including multiple EHR systems — that use FHIR resources and extensions.

Jeremy Warner, an assistant professor of medicine and biomedical informatics at the Vanderbilt-Ingram Cancer Center and one of the PCM app's developers, discussed the tool during a presentation at the HL7 Genomics Policy Conference held last month in Washington DC. Warner is also the first author on the JAMIA paper that describes the app. He told GenomeWeb  that the idea for the PCM app grew out of work done at Vanderbilt on a precision cancer initiative and the development of the My Cancer Genome repository.

When Warner joined Vanderbilt in 2012, he became interested in ways to better report information gleaned from patients. "One of my major interests is data visualization, not just for clinicians but also for patients," he told GenomeWeb. At the time, reports from the school's cancer initiative were "fairly basic," offering general statistics on number of patients tested and observed mutations, for example, but none of this information was contextualized for patients or interactive, Warner said.  "So I wondered 'how can we repackage this in a way that is accessible to patients and clinicians and also has value added in that you could link through the display and get to other kinds of information that you might want?'" At the time, efforts to map the SMART technology to the FHIR standard were underway and "I was fortunate enough to get a small subcontract with the SMART project to basically create [the] app," he said.

The PCM app is one of two genomics-focused apps included in the SMART platform's gallery. It lets physicians visualize single mutations from patients identified by SNaPshot assays — which test for common somatic mutations across multiple cancer-associated genes — in the context of the broader pool of patients. The app generates pie charts that show, for example, a lung cancer patient with a KRAS mutation in the context of other lung cancer patients tested at Vanderbilt. Charts show things like the population distribution of mutated genes observed in lung cancer cases and observed variations in lung cancer patients with KRAS mutations.

On the backend, the app connects to the My Cancer Genome repository and to an FHIR server set up to hold data gleaned from Vanderbilt's electronic medical record, including patient demographics, primary cancer diagnosis, clinical progress notes, provider-patient communications, as well as data feeds from laboratory and billing systems.

To build the FHIR server, "we took data that had at one point been generated partly by the EMR and partly by the [laboratory information system] and staged it in a data warehouse," Warner said. "We were able to pull from that data warehouse and transform it into FHIR syntax and then have that on the FHIR server." Setting up the system this way let the developers sidestep the challenges of connecting to the live EMR and potentially crashing it, he said. Although the PMC app is separate from the EMR, Vanderbilt physicians use the same username and password that they use to log into their EMR to log into the PCM app.

Following its development, Warner and his colleagues piloted the PCM app with 14 users — nine fellows in an oncology training program and five practicing attending oncologists — and collected their feedback. Overall, clinicians seemed open to using the app at the point of care. "I see a potential use for clinicians who are seeing a wide variety of cancers," Warner said. It links them to the relevant knowledgebases where they can get to the information they need at the point of care without having to either memorize it or log onto a separate website to access it.

However, oncologists did ask for more decision support as well as for more links to external knowledgebases. For example, some clinicians who responded to the survey said the visualizations would be more valuable to them if they included links to additional information, such as available targeted agents, drug cost information, and patient outcomes data. Clinicians were also interested in tools to stratify patients by criteria such as age, gender, treatment exposure, or stage.

Some respondents also thought that the visualizations offered information that would be confusing for the average patient. Warner conceded that as it stands, patients may not get as much value from the app but patient engagement moving forward is going to become increasingly crucial. "As medicine gets more complicated, we've got to get patients more involved because they might end up being the experts on their almost unique condition," he said.

Building on the prototype described in the JAMIA paper, Warner and his colleagues have begun implementing some of the changes suggested by physicians who responded to their survey. For example, the variant pie chart was activated to direct users to the relevant disease- and variant-specific pages in My Cancer Genome. They have also added icons that direct users to disease- or genotype-specific treatment options available on, a chemotherapy regiment wiki page.

Since its launch, the PCM app was included in the American Society of Clinical Oncology's inaugural interoperability demonstration that aimed to evaluate the ability of currently available products and standards to exchange information in a timely fashion.  Twelve software vendors participated in the demonstration, which focused on sharing data from an imaginary patient with BRAF-mutated colon cancer across multiple systems. Full details of the demonstration and the participating vendors are provided here

"We are in a sort of evolutionary period about how we understand cancer," Warner said. "People are slowly shifting out of the anatomic base that we've used for a long time ... to a more molecular or genomically-informed definition of cancer. Clearly, there's this middle ground where not only anatomy matters ... but the molecular matters as well."

Moving forward, Warner hopes to add several new features but this will depend on available funding and competing priorities. One improvement that the developers are working on is coming up with a mechanism for providing clinicians with drug sensitivity information for somatic mutations within the app. Warner and two colleagues recently published a review article where they proposed an actionability hierarchy that offers fine-grained criteria for categorizing somatic variants and their links to FDA-approved cancer drugs, clinical trial drugs, and drugs approved for off-label use.

In their hierarchy, which is described in an article published last month in Genome Medicine, the list of categories includes variants known to confer sensitivity to an FDA-approved agent for the cancer subtype; variants predicted to confer sensitivity to an FDA-approved agent for the cancer subtype; and variants known to confer sensitivity to an FDA-approved agent for another cancer subtype. An example of the first category is the BRAF p.V600E mutation in melanoma cases, which is known to respond to Vemurafenib. "Those are the kinds of things that we are now trying to integrate into the app," Warner said.

Future versions of the app could also include prognosis information to give patients a sense of what to expect if and when they undergo treatment based on their mutation. The developers also hope to develop additional features that will allow clinicians to view multiple mutations from patients simultaneously — at present, they can only view one mutation at a time. "That's proving more challenging than we thought," Warner said. "We are actually in the process of exploring a variety of potential visualization techniques and actually testing those out with clinicians."

For data security reasons, the PCM app is only available to researchers at Vanderbilt. But the source code for the app and for the SMART on FHIR server used to house EMR data are openly available on Github, so centers interested in setting up their own versions of the PCM app can do so. "Their biggest challenge — and this was our biggest challenge — would be if they had [internal] codes for genes," Warner said. At Vanderbilt, "we had this sort of mishmash of gene codes, and those pretty much had to be translated, many of them by hand, to a standard format."