NEW YORK – Digital pathology diagnostics firm Stratipath continues to establish itself in the clinic with its breast cancer prognostic test with recent data demonstrating strong performance in a cohort of more than 2,000 breast cancer cases.
The Swedish firm, founded in 2019 as a spinout from the Karolinska Institute, received a CE mark for what it calls Stratipath Breast in 2022 and began to amass clinical customers in the spring of 2023. CEO and Cofounder Fredrik Wetterhall said in an interview this week that while its commercial expansion is still in its early stages, with most customers in the Nordic region, it is already looking to develop additional assays for other tumor types.
The company's core technology involves using artificial intelligence to train a discriminator to differentiate higher- versus lower-risk cancer. In the case of breast cancer, the company's goal was to be able to split grade 2 cancers into high- and low-risk groups to offer oncologists the opportunity to better personalize their patients' treatments.
"It's up to 50 percent of breast cancer patients that are receiving inconclusive diagnoses where oncologists don't know how high the risk for relapse is, and then they don't know if they should provide chemotherapy to the patient or not," said Wetterhall.
To do this, the company trained its AI system using pathology images of grade 1 and grade 3 tumors, anticipating that when fed grade 2 slides, it would be able to classify them as closer to one side or the other. In early studies, the strategy proved successful, with initial validation data showing a prognostic strength similar to that achieved by commercial gene expression profiling tests such as Exact Sciences Oncotype DX and Veracyte's Prosigna.
Although these molecular tests are used ubiquitously in the US, their uptake in Europe has been much smaller. Wetterhall estimated that overall penetrance is likely in the single digits on average across the continent. In his opinion, barriers include both cost and turnaround time, both of which Stratipath claims as differentiators.
"The customers we have, they're using our test as an alternative to molecular assays … and I think the main benefit with AI-based precision diagnostics is that they can get results in minutes instead of weeks," he said. "The pricing is also very different. We charge around €500 ($547) per analysis, compared to around €3,000 for molecular or traditional gene expression profiling, which means that you can test a lot more patients."
In the Nordic region, unlike in other parts of Europe, he noted, neither the Stratipath test nor molecular tests for breast cancer are reimbursed, adding "To be able to get into reimbursement, you need to be part of guidelines. And to be part of the guidelines, you need to provide a lot of clinical evidence."
In the company's most recent study, published in August in BMC Breast Cancer Research, investigators from the company and collaborators at the Karolinska Institute gathered pathology slide images, taken from more than 2,700 patients with primary breast cancer treated at two Swedish hospitals.
The researchers analyzed test results against patients' recorded length of progression-free survival (PFS) for ER-positive HER2-negative cases and calculated a hazard ratio of 2.76 after controlling for established clinical risk factors. In other words, patients with a low-risk result had nearly triple the PFS seen in the high-risk group.
Wetterhall said that Stratipath intends to pursue additional validation studies to further shore up the evidence for its test's performance.
Competitors in the molecular space have a long head start in that regard, with Oncotype DX, for example, having been validated prospectively not only as a prognostic but as directly predictive of chemotherapy benefit. Still, the test remains uncovered in many European markets, leaving a hole that Stratipath aims to fill with a budget-friendly alternative.
Wetterhall said that Stratipath believes it is the first company to adapt digital pathology for a precision medicine application in breast cancer, rather than to refine or improve routine pathology, which involves things like determining cell types and classifications, or measuring specifically for the presence or absence of individual biomarkers.
A few other firms have pursued similar predictive and prognostic applications, including Owkin, which also has a CE-marked breast cancer prognostic algorithm. The firm's website currently offers a demonstration of the test, but no test-ordering option.
As it moves forward, Stratipath is hoping to gain direct adoption from hospitals and medical centers but has also made some deals to have its test marketed through other, larger companies, such as digital pathology leader Paige and, more recently, Roche.
Although the company is still early in its clinical foray, Wetterhall said that it believes it can also serve markets where molecular testing has proliferated, including the UK and US, by offering a more rapid option.
As this work progresses, the company is also exploring additional disease targets for its AI digital pathology platform. These include lung, colorectal, and prostate cancer, where the clinical use cases would be similar — differentiating risk to guide treatment decisions, and potentially predicting therapy benefit.