NEW YORK – During last week's annual meeting of the Association for Molecular Pathology, diagnostics newcomer Innosign offered a first glimpse at its technology, featuring data from two early adopters who have taken up the company's pathway classification approach to better risk stratify breast and bladder cancer patients.
Results from these efforts, shared in a company-sponsored workshop, suggest that the firm's technology, which it offers as a PCR-based LDT called OncoSignal in the US, could fill in some of the precision medicine gaps left by existing risk classifiers used to guide treatment decisions in early-stage breast cancer. It could also help build similar personalization options for additional tumor types.
Robert Babkowski, Innosign's US medical director, said the company's approach is based on a recognition that gene mutations and protein expression fail to account for all aspects of cancer biology, and that cell signaling pathways can help resolve questions other tools fail to answer.
Originally located in The Netherlands, the Royal Philips spinout now has a second lab in Ohio that became operational in September.
"We've rolled out our first assay, which is for breast cancer, and we've started receiving patient specimens through an early-access program," Babkowski said. Although early, the resulting data has been interesting, he added, with the vast majority of analyses yielding "actionable results."
NYU has been a collaborator in developing this breast cancer application for several years, and Matija Snuderl, a molecular pathologist at NYU Langone Health, shared a glimpse of his and his colleagues' results at the Innosign corporate workshop.
Breast cancer, Snuderl said, is a precision medicine poster child, especially in terms of risk stratification and personalized treatment. Molecular testing, including IHC and FISH-based subtyping, plus gene expression-based recurrence risk tests, have become ubiquitous.
But Snuderl argued that these well-established methods are imperfect. While recurrence risk tests like the Oncotype DX assay can identify a subset of patients at a low risk of recurrence who can forego adjuvant chemotherapy, these women don't have zero risk, and about 40 percent end up seeing their cancer come back despite a promising gene expression signature.
"The question we posed to Innosign a few years ago was, can we further stratify HR-positive, HER2-negative cancers? Can we use RNA expression and pathway analysis to identify patients who have an [Oncotype DX] risk score below 25, but still have an increased risk for recurrence and should get chemotherapy?" Snuderl said.
"There are almost 300,000 women who will be diagnosed with breast cancer in 2023 and … although most of them will have a low risk score below 25, almost 15 percent of those will recur. There's just no way to tell who," he added.
Exploring the potential added value of Innosign's pathway approach, the NYU team collected a retrospective cohort of early breast cancer cases with recurrences and matched controls who did not recur, using RNA seq, and submitting the results for pathway analysis.
The most important finding, Snuderl said, was that the recurrent cases and controls did not vary in some of the expected ways. Pathway activity of MAPK, ER, and AR were all equivalent in cases and controls. HER2 and ESR1 pathway activity also did not differ.
What did distinguish the two groups was the PI3K pathway, which was significantly upregulated in recurring patients.
Snuderl said that while the cohort was small, the discrimination was dramatic enough that NYU is expanding their retrospective study with additional cases and controls and planning a prospective validation.
One of the exciting aspects of the Innosign platform, he added, is the potential to not only improve risk stratification, but potentially to identify treatment targets. But when the NYU researchers dug further into PI3K, they didn't see any difference in gene mutations between cases and controls, which raises some questions about whether the upregulation of what Innosign dubs the PI3K pathway would actually confer responsiveness to a PI3K inhibitor.
According to Babkowski, Innosign has seen some of the same things emerge when running matched specimens internally from the same low-risk patients at their initial diagnosis and from a recurrent metastatic biopsy.
In a separate study of triple negative breast cancers, NYU investigators have seen similar patterns, Snuderl said, highlighting results from an abstract featured in a meeting session on Friday.
The NYU team had previously used methylation analysis to separate a cohort of TNBC patients into three epigenetic clusters, which they then had Innosign analyze for driver pathways. The results indicated that the third cluster — which is enriched for tumors with either BRCA mutations or BRCA promoter methylation — also featured upregulation of the PI3K pathway.
But as in the Oncotype DX cohort, the same was not seen on a mutational level. There was not a single PI3K mutation in that subgroup, Snuderl said.
At NYU, the team hasn't yet dug into the components of Innosign's algorithm to try to figure out what aspect or specific target is actually responsible for the pathway upregulation, but they are working on that now.
"This is really important because it brings up a whole new slew of potential patients who could get these kinds of inhibitors but potentially not the mutation-specific ones," Snuderl said.
Asked about the proprietary aspects of the company's platform, CEO Paul van de Wiel said that its pathway scores reflect the activity of transcription factors, that are algorithmically combined into a pathway readout.
Although breast cancer is the first application where Innosign is promoting its platform in the US, investigators in The Netherlands have also been exploring the technology for risk stratification in bladder cancer.
Erasmus Medical Center PhD student Christiaan de Jong presented some of his results during the company's workshop, highlighting a previously published molecular stratification effort that he and his colleagues are now hoping to adapt to the clinic using Innosign's platform.
Bladder cancer is currently classified based on physical factors, namely whether it has invaded the muscle wall of the bladder or not, and on clinical risk factors. Higher-risk tumors are treated with Bacillus Calmette-Guerin (BCG) immunotherapy, with the highest-risk individual offered a choice between that and a radical cystectomy.
"Of course, you can imagine that patients often opt for BCG because urinary diversion is a procedure with a high morbidity but also a reduced quality of life," de Jong said. Many of these patients end up with that treatment failing them, but clinicians don't have a way of predicting who those will be.
Using RNA-seq, de Jong and colleagues were able to define three molecular subtypes distinguishing BCG-responsive and non-responsive individuals with the BRS1 and BRS2 groups showing better responses than the BRS3 group.
When they looked at the predictive ability of this subtyping to refine risk for the clinically high-risk patients who currently have to choose between trying BCG or opting for a bladder removal, de Jong said the results were striking with clinically high-risk BRS3 patients showing incredibly high failure rates on the immunotherapy, but group 1 or 2 patients having relatively good outcomes on BCG, suggesting they can spare their bladders.
After learning about Innosign's platform, de Jong said he and his colleagues thought it could be good for clinical translation of their approach. After testing this out, with good discrimination in their RNA-seq cohort, investigators then evaluated the firm's digital PCR version of OncoSignal, which would be easier to deploy as a clinical test.
"Ideally, you want to have a bedside test that's usable in the clinic and with RNA-seq, the turnaround time is not ideal," de Jong said.
In a 135-patient cohort, PCR-based pathway analysis offered "highly comparable" results across pathways to the earlier RNA-seq data, equivalently distinguishing the BRCA3 subgroup. The team is currently planning a prospective validation.
In terms of a future applications, Innosign has a large collection of scientific publications and posters from the last several years exploring a variety of applications in other tumor types like ovarian cancer and prostate cancer.
Babkowski said that the company plans to stick with an LDT model in the US going forward unless specific companion diagnostics indications emerge that would require submitting to the US Food and Drug Administration.