By Allison Proffitt
May 4, 2017 | Caris Life Sciences was born out of loss. David Halbert, Caris’ Chairman and CEO, lost his mother to cancer, David Spetzler, President and CSO of Caris Life Sciences, told Diagnostics World. The molecular profiling business he started in 2008 has now tested over 110,000 cancer patients.
“This is a pursuit of passion, and has been for a very long time,” he said. “At the forefront of everything we do is the patient. It’s not really about anything other than that patient sitting in front of the doctor right now.”
The company’s Caris Molecular Intelligence test suite provides relevant, clinically actionable, and individualized treatment information to personalize cancer care for all solid tumors. The company takes a “holistic approach” Spetzler says, that’s not limited by technology. They include next-generation sequencing of DNA and RNA, broad gene panels, and specific tests like PDL1 tests for lung cancer, HER2 and AR in breast cancer, and MMR markers for colorectal cancer. Tests are marketed to the oncologist. The company has insurance contracts covering about 100m lives, as well as patient assistance programs if insurance won’t cover a test. “We are not in the habit of rejecting patients,” Spetzler said.
“We always try to stay ahead of things,” Spetzler said, citing PDL1 testing as an example. The company has tested over 45,000 patients with PDL1 over the past four year. “Even before PDL1 was approved as the companion for Keytruda, we were doing a lot of PDL1 testing.”
The company’s headquarters is Irving, Texas, and the lab is in Phoenix, Arizona. Most of the company’s 300 employees work to process about 950 tests per day, seven days a week, in two shifts. Last year, the company averaged about 10 tests per patient, conducting 230,000 tests. Turnaround time is eight days.
“We do a fair amount of next-generation sequencing, not just of DNA, but of RNA to add in the protein components. We’re really trying to push the boundaries of that to expand it even further to integrate more of a systems biology perspective to understand the underlying perturbations within the molecular system that are driving that person’s cancer. With a greater understanding of what’s gone wrong at a molecular level, we hope to improve patient outcomes by directing patients to the most appropriate treatment.”
Physician Perspective
A bioinformatician recommending the Caris Molecular Intelligence test recently called the test “well-designed” and praised the simple report. It was feedback Spetzler was happy to hear.
“The report, in many ways, is our product. We continually evolve it and try to make it better. At the end of the day we tried to make things very simple and easy for that busy oncologist so they can glance at the report and know what to do, and maybe more importantly, what not to do.
Reports are color coded—green and red—categorizing treatment options at a glance. Gray sections mark treatment options for which data are less clear, for “when things get messy.” The first page of the report only includes drugs approved for an indication, but other options are included within. All biomarker results are included on the front page. Right now, Spetzler says, the company’s decision-making is based about half on proprietary data and about half on published data.
“We’ve tried design that front page where it can be a tool by which physicians can engage with patients to help them understand more about their particular disease.”
And do physicians listen? Years ago, Spetzler said, doctors followed treatment recommendations resulting from molecular testing about half the time. Now, he said, they adjust course about 80% of the time.
“It becomes difficult when the molecular profile argues against the standard of care,” he said. “That puts everyone in a tough spot.”
For the past seven years, Caris has been tracking patients in observational studies. Participants are enrolled by their oncologist and consented into the study. The oncologist sends updates to Caris every month. Their findings among epithelial ovarian cancer (EOC) cases were published in March 2016 in Oncotarget (doi: 10.18632/oncotarget.7835).
“When patients are treated in a manner consistent with their [molecular] profile, they live on average about nine months longer… and at the same time receive one fewer lines of therapy,” Spetzler said. “That’s very important from the patient’s perspective. Being on a chemotherapy that’s ineffective is the very worst thing for quality of life.”
Parts vs Whole
The company’s tests have historically been on solid tumors, but Caris is developing its ADAPT Biotargeting platform—Adaptive Dynamic Artificial Poly-ligand Targeting—to support several liquid assays to diagnose complex conditions, monitor disease progression, and assess therapeutic response.
The platform uses very small synthetic pieces of DNA, essentially aptamers, as “fingerprint dust” to understand what is happening to that entire molecular system. The system is complex: genes to RNA transcripts to protein isomers. “Those proteins work together in complexes and pathways, and we really don’t have a good understanding of how many different combinations there actually are,” Spetzler said.
Looking at just DNA or RNA or proteins only reveals parts of the story, he explained. “If we were to do this in a Ford dealership, we’d find a lot of carburetors and transistors and maybe hubcaps and steering wheels. It doesn’t tell us if we’re looking at a full car. Is it turned on? Is it running? Do the brakes work?”
Precision medicine, Spetzler believes, won’t be fully realized until we can manage this level of complexity.
The ADAPT platform contains a library of 1013 unique binding elements, single-stranded oligodeoxynucleotides (ssODNs). That’s enough, he says, to encompass the complexity of the system. “We try to use those numbers to our advantage to get a sense of what’s happening at the systemic level; that’s really where the power is.”
Caris is putting the platform to work on blood-based assays for Herceptin response and breast cancer diagnosis. Early data on the breast cancer test were published in Nature’s Scientific Reports in February 2017 (doi:10.1038/srep42741). Caris collected blood samples from both positive and negative breast biopsies as well as samples from women with no history or indication of cancer. “We were able to achieve an AUC of just over .7,” Spetzler said. “That’s encouraging—it tells us we’re on the right track—but it’s not quite good enough for us to feel comfortable launching that as an assay.”
Spetzler wants a test much better than that. “There’s a societal responsibility associated with screening, and we really need something much better. Our effort is on improving that performance,” he said.
So Caris is training the platform. “One of the neat things about this technology is we can take the group of patients that we missed, and continue to train and evolve our libraries,” he said. “There’s a process similar to machine level that’s actually happening at the physical level. We’re able to bring in new molecules that recognize the aberrations in those sub-populations of patients that we missed.”