Skip to main content
Premium Trial:

Request an Annual Quote

Preferred Medicine Building Data for Breast Cancer MicroRNA Screening Test


NEW YORK – Preferred Medicine, a Burlingame, California-based joint venture of two Japanese firms, has begun a multicenter trial to further validate its microRNA-based blood test for the early detection of breast cancer.

The firm — launched by global conglomerate Mitsui and artificial intelligence firm Preferred Networks in 2018 — has begun a new validation study, called DROPLET-BC, to finesse its predictive algorithm to optimally detect breast cancer and to further demonstrate its sensitivity. Investigators are enrolling approximately 1,200 individuals at Northwestern University, MD Anderson Cancer Center, and the Roswell Park Comprehensive Cancer Center — divided equally between women diagnosed with breast cancer and healthy controls.

Similar case-control studies have become the backbone in recent years for emerging early cancer diagnostics, which target a variety of analytes including mutations and epigenetic alterations in cell-free DNA, or, in some cases, RNA, proteins, or other biomarkers.

Grail has become one of the most notable players, with a pan-cancer assay on the market since June, and prospective interventional validation trials ongoing.

Based on its early proof-of-concept data Preferred Medicine believes that its miRNA approach has the potential to outperform DNA-based technologies, at least in breast cancer, where the firm's miRNA classifier performed the best in its pilot study, but which has been more of a challenge for other emerging technologies.

In some of its most recent published data, for example, Grail has reported a sensitivity of only around 30 percent for breast cancer at a specificity of 99 percent. Performance of the DNA methylation-based technology is higher in later-stage breast cancers, but has seemed to struggle, comparatively, in early-stage cases.

Preferred Medicine, in contrast, has reported that it could achieve close to 57 percent sensitivity at 99 percent specificity.

This data has not yet been published in a peer reviewed format, so its takeaways remain limited. The company's sole research presentation, at this year's annual meeting of the American Society of Clinical Oncology, described an analysis of clinical serum samples from cancer patients with breast, colorectal, lung, stomach, and pancreatic cancer, which were split into a machine learning training set and a smaller test set.

The authors reported that they could detect cancers with an average of 88 percent accuracy and that sensitivity remained high regardless of cancer stage, but they did not break down in detail the performance of the classifier in different cancer stages.

Preferred Medicine is now banking on the likelihood that it can replicate and even improve on these limited results as it moves forward with DROPLET-BC.

Raees Ebrahim, the firm's head of commercial strategy and business development, said that the initial proof of concept data was from frozen biobanked samples. With its new multicenter partners, the company will be testing fresh patient samples from a much more diverse cohort. Resulting data will therefore be more likely to reflect true clinical sensitivity.

Ebrahim said the trial could potentially be expanded, but in its current form it should complete by Q1 2022.

"We're hoping to launch our test as an LDT by ASCO next year [and then] we will launch a large clinical utility study," he said.

According to Abhijit Patel, a professor at the Yale School of Medicine and a leader of ctDNA-based early cancer detection efforts, validation in an intended use cohort, not just case-control data, will ultimately be necessary for Preferred's test to prove itself.

However, the early data looks promising, he said in an email.

"Achieving high cancer specificity with circulating microRNA biomarkers has proven to be challenging in many prior studies, but some recent studies incorporating machine learning analysis have been showing good promise," he wrote.

For example, investigators led by Dipanjan Chowdhury at the Dana-Farber Cancer Institute published data in 2017 from a study using neural network analyses to glean a miRNA signature for the detection of epithelial ovarian cancer. The team is now investigating the approach for early detection in high-risk women.

According to Patel, because breast cancers have been significantly "less tractable" with ctDNA-based early detection approaches, they are an especially exciting target for microRNA.

Ebrahim said that the ubiquity of microRNA is part of why it is an attractive target, as long as the appropriate tools are used to glean a predictive signature.

"The power of microRNA is that it's actually ... much more present and in plasma, which is why we think it makes a better marker."

However, Ebrahim argued, miRNAs are difficult to handle from a collection, sample preservation, and stability perspective. "Just finding the right species markers for what you should be looking at" is a challenge.

"Identifying the right MIRs is fundamentally critical ... and the only way you can do that is kind of through a very robust pattern recognition machine learning," he added.

Despite the challenge of identifying predictive miRNA signatures, several other firms are seeing early commercial success in this area. These include Singapore-based MiRxes, founded as a spinoff from Singapore's Agency for Science, Technology, and Research, and US firm MiR Scientific, which markets a test for prostate cancer detection and prognosis.

While not specific to microRNAs, sequencing firm Caris Life Sciences is also poised to launch a comprehensive blood-based assay targeting both DNA and RNA, which it believes it can translate into assays for both minimal residual disease and early cancer detection.

According to Ebrahim, Preferred Medicine considered the possibility of combining microRNA and cell-free DNA analysis but viewed targeting both classes of analytes as too expensive and impractical.

For one, "there are very different workflows in terms of processing and analysis," he said. "NGS is the standard for cell-free DNA, but we are actually trying to develop a non-NGS platform for miRNA that we think could be better for our technology and also more cost sensitive."

The firm hasn't disclosed details of this technology, but Ebrahim said that it is based on flow cytometry with a proprietary reagent methodology.

Although Preferred Medicine is also eyeing other applications, including lung cancer, the initial focus on breast cancer is both a reflection of the firm's higher performance in these tumors in its pilot study, and a corporate strategy, he added.

"From a commercial standpoint, when you look at breast cancer, there's a lot of credence already [there] with the [the US Preventive Services Task Force] in terms of guidelines and so forth … so there's already a lot of positive momentum," Ebrahim said.