Getting to Value-Based Cancer Care in the ’Omics Era Requires Data Integration

Wayne Kuznar

December 2015, Vol 6, No 11 - AVBCC 2015 5th Annual Conference


Washington, DC—An evolving knowledge of cancer biology and the availability of comprehensive genetic testing engender a need for tools that help oncologists integrate these data and select therapy, said Gary Palmer, MD, Chief Medical Officer, NantHealth, Los Angeles, CA, in delivering his keynote address at the Fifth Annual Conference of the Association for Value-Based Cancer Care.

He introduced the inspiring concept of an “intelligent” clinical operating system, a portable BlackBerry-like device that holds medical record data and test results, as well as patient-reported data.

Increasingly Complex Data

With the understanding of cancer becoming increasingly deeper and more complex, the data that it generates are stressing the current system, with data collection as one of the primary issues. “That’s an underappreciated problem that we’re going to have in this country if more and more patients get full next-generation sequencing,” explained Dr Palmer. “Where are you going to keep all of these data?”

The information technology solutions will need to integrate myriad data and provide feedback learning based on analyses of the data.

Using breast cancer as an example, Dr Palmer explained that new targets are the goal of next-generation sequencing, yet these targets are transferable across tumor types. In addition, some new technologies, such as those involving natural killer cells, can benefit from genomics testing.

Sequencing the genome in breast cancer, such as delineating estrogen receptor–positive and estrogen receptor–negative cancers and their different subtypes, had treatment implications. Dr Palmer discussed how the Oncotype DX test places women with very early-stage breast cancer into different risk categories, as does MammaPrint. The molecular pathology of breast cancer and the additional risk stratification provided by these tests has led to concern over how the results should be integrated for optimal clinical use and real-time decision-making.

Dr Palmer described a graph that showed the different genes ranked in order of the number of patients or the percentage of genes that have mutations or alterations. He stated that this curve could be replicated for practically every small tumor, displaying a different order of genes for various tumors, but retaining the shape of the curve. Given this amount of information, the debate becomes: When does a comprehensive test of the genomic profile become justified versus doing only a HER2 test, or only ALK and EGFR testing in patients with lung cancer?

Referencing a novel approach to clinical trials, Dr Palmer stated that histology agnostic cancer clinical trials, in which patients are entered across solid tumors and not by anatomic location of the cancer, are becoming more common.

The reports that are generated from DNA sequencing can be lengthy, and oncologists demand salient news on page one. “The responsibility, though, of knowing what’s on page one and incorporating the nuances of some of these finds is really into a trivial matter,” said Dr Palmer.

He noted that not all mutations at the DNA level are transcribed in protein changes or vice versa. “You can have a DNA mutation that doesn’t go anywhere, so to speak, it just stops and doesn’t result in protein changes,” he said. “Vice versa, you can have effect on the protein that is important, for which there are no DNA-equivalent changes. So it makes it more complicated, because now you have to think about doing proteomics to see what the effect is. It’s very difficult to do proteomic testing on every protein.”

The competing treatments for metastatic disease will continue to include chemotherapy, and now both will include target therapy and immunotherapy as well. With the various treatment options, oncologists require an increasing amount of data, according to Dr Palmer. “The more we can eliminate unnecessary treatment and unsuccessful treatment, obviously the better it’s going to be for the patient, but it’s going to be better for the healthcare system, the insurers, and for everybody else.”

Data Integration

Given the huge amount of data from next-generation sequencing, in addition to proteomics and imaging, integration and storage of such data are huge dilemmas. “We’ve invested now in multiple storage centers scattered throughout the country to house genomic data,” noted Dr Palmer. “We’re up now to petabytes of data, which is as far as I knew it went, but we have the projections to exabytes in the not-too-distant future.”

Most healthcare systems do not have the ability to store these data. “Another issue that’s critical is correlating all of this genomic information with clinical outcomes,” Dr Palmer pointed out.

An “intelligent” system that pulls data out of all medical records would organize these data into 1 field or 1 data warehouse for analysis. The recommended screening tests based on patient history and guidelines from medical societies would lead to treatment paradigms. The patient component would electronically signal medication compliance, and the entire medical record and images could be incorporated into a BlackBerry-type device. Financial information would also be incorporated.

“At the backbone of this is the clinical operating system,” said Dr Palmer. “It allows extraction of data from labs, pharmacy systems, electronic medical records in a diagnostic type of fashion, the genomic and proteomic testing, and from imaging systems, as well.” Such an operating system has been developed by Foundation One (Figure).

Table

Dr Palmer said that he believes that the sharing of information would take place on a fiber-optic cable network, via the National LambdaRail system, an abandoned cable system meant to tie together universities to share data. The vision is to extend and collaborate this system internationally.