NEW YORK – Newly published validation data and a consensus paper outlining interdisciplinary testing recommendations are the latest positive developments for Castle Biosciences as it seeks to convince the dermatology and oncology fields to use its molecular assay for squamous cell carcinoma (SCC).
Published this month in Dermatology and Therapy, the validation study provided updated and expanded evidence that Castle's DecisionDx-SCC test, a 40-gene expression assay, provides independent, clinically actionable stratification of SCC patients' metastasis risk and improves the predictive accuracy of clinical risk classification systems, potentially guiding treatment.
The consensus recommendation paper, which appeared last week in the Journal of Clinical and Aesthetic Dermatology, represented a group effort between the dermatology and radiation oncology communities to evaluate how Castle's test might best be used in the clinic.
Castle has had some struggles in pushing the SCC test forward, with a negative Medicare coverage determination issued last year, despite continued test volume growth.
Sarah Arron, a dermatologist specializing in Mohs surgery, a microsurgical technique that removes lesions in thin sections aiming for clear margins, and an author on both of the new publications, said that squamous cell carcinoma has been "a little bit understudied" compared to melanoma, which has longstanding stage-based clinical guidelines.
"Up until very recently, it didn't have its own codes and billing," she said. "We had difficulty doing, large epidemiologic studies … so outcomes data were really lacking, and surgeons like myself and oncologists and radiation oncologists that we partner with were really struggling to determine: Were we over treating patients? Were we undertreating them? And how do we catch that small percentage of them that will go on to develop a metastasis?"
Because the mortality from melanoma is so high, a lot of research money has been put into it over the last few decades. By contrast, squamous cell carcinoma is much more common but tends to affect older people as opposed to young, healthy people, Arron said.
"The rate of death from squamous cell is much lower … but the number of people who die every year in the US is roughly the same between these two cancers," she noted. "However, the denominator for squamous cell is so high that people don't think of it as an aggressive cancer or a lethal cancer, and it didn't have the same impetus from drug trials."
Arron has been involved in research on Castle's test from its early development, and she said that immediately after the first publication on the gene-expression classifier, she and her colleagues wanted to validate their findings in a larger cohort.
The Dermatology and Therapy study represents this work, expanding the cohort in which the test has been assessed to almost 900 samples across 58 sites. It also allowed investigators to "dig a little deeper into smaller patient populations," Arron said.
"Now we can look at those combinations between stage and class and start to integrate the Castle calls into our staging systems and improve our predictive values for who might need post-operative adjuvant treatments [and] who might not benefit," she said.
The study evaluated DecisionDx-SCC's accuracy in stratifying risk among patients with one or more high-risk factors across clinically relevant subgroups, including tumors within National Comprehensive Cancer Network high- and very-high-risk groups, lower-stage Brigham and Women's Hospital T1 and T2a tumors, and patients more than 65 years old.
Overall, the investigators found that patients classified as Class 1, Class 2A, or Class 2B by the test had significantly different metastatic risk profiles. Integrating test results with the other clinical risk factors of classification systems demonstrated "significant improvement in accuracy for prediction of metastatic events," the authors wrote.
As that latest study was coming together, Arron said, she and others in the field were debating where this type of test might be most valuable in clinical care. "The biggest question … that I've asked in my practice and that a lot of people were asking is which patients will benefit the most from adjuvant radiation therapy," Arron said, noting that the answer to this question is relevant to both the radiation oncology and dermatology communities.
These specialists may collaborate more closely when they work at large universities with multidisciplinary tumor boards, but a lot of private practice dermatologists and community oncologists don't find themselves in close contact.
"We really wanted to give the opportunity to physicians in both groups to talk to each other. And it was really remarkable in that advisory board setting, how similar our impressions of this data were and how readily these guidelines came together," Arron said.
The opinion-based recommendations published in JCAD are, as described, representative only of the opinions of the group, which in addition to Arron, included radiation oncologists Ramesh Gopal and Michael Marquardt, and dermatologists Gaurav Singh and Stanislav Tolkachjov. They don't represent any official guideline body.
Overall, the multidisciplinary group concluded that there are clinical practice gaps in which Castle's test would have particular utility, including escalation of care for lower-stage patients with high-risk tumors, de-escalation of care for patients for whom the risks of adjuvant radiation may outweigh the benefits, and decision-making regarding elective radiation to the nodal basin. The panel also derived a risk-based clinical workflow for guiding radiation use in patients based on Castle testing coupled with National Comprehensive Cancer Network management guidelines and the most recent staging practices.
"The standard guidelines that we use for considering post-operative adjuvant radiation are extremely broad, and many dermatologists were saying, well, I'm not going to refer every patient that is a high-risk NCCN patient to discuss radiation," Arron added.
Meanwhile, she noted, radiation oncologists were receiving referrals for patients and wondering whether they really needed radiation treatment. "So the addition of a high-risk Castle Class 2 call can kind of help tip us in that direction and help us justify to the patient that this extensive treatment is potentially of value to them," she said.
The second area where Arron sees some consensus on utility is for patients who may meet clinical criteria for radiation, but may want to avoid it for reasons like age or access to care. A Castle Class 1 call could help guide that decision by indicating that these patients are potentially not at a tremendously high risk.
Arron's research on outcomes among DecisionDx-SCC high-risk patients treated with radiation is still unpublished but she said she and her colleagues are working to submit it right now. While previous data has failed to sway payors, this more direct evidence of clinical utility, along with the updated validation data, could provide a tipping point for the company.
"Putting on my epidemiologist hat, utility for a diagnostic is linked to whether or not it changes my patient management," Arron said. "So, to me, the utility of the Castle test is that it helps me when I'm not entirely sure what the optimal treatment for a patient will be."