NEW YORK – Fellow Health, a company that has established itself in the reproductive health space with a mail-in semen testing service, is now aiming for the molecular diagnostics market, having demonstrated that it can detect DNA fragments in seminal fluid that originate from prostate tissue and could identify prostate cancer.
Company researchers published a proof-of-concept study in PLoS One earlier this month, and will be presenting new data on cancer detection next week at the annual meeting of the American Urological Association.
Kim Clark Langone, R&D leader at Fellow, joined the company two years ago to lead the company's prostate cancer work.
"When I joined back in June of 2023, there wasn't any kind of molecular capability. The team had been working solely in the andrology area, so obviously the first few months of joining … what we were focusing on was really just getting the capabilities up," she said.
Tom Bergstrom, Fellow's chief commercial officer, said that infertility testing was the company's "beachhead" into medical diagnostics, but there had long been discussion of the potential to branch out.
"Traditional labs can't get men to go in and provide a semen sample even when they need it right now clinically, for example after a vasectomy or for infertility … so step one was to unlock the fertility door, which we did with now 2,500 clinics [using our mail-in services] nationwide," Bergstrom said.
"It's always been on our founders' mind to think about what is next beyond fertility, but if you can't get on the fertility beach, you can't get to the prostate cancer beach and you can't get to the other beaches or further inland to be able to discover more," he added.
In the firm's PLoS One study, investigators confirmed previous studies' observation that seminal fluid contains an abundance of cfDNA in both nucleosomal and larger fragments.
Because prostatic fluid makes up 40 percent of seminal fluid, other researchers in the field had hypothesized that it might be a good target for liquid biopsy, but little data existed.
Langone and her team took blood-based liquid biopsy as a model, using methylation signatures to try to deconvolute the cell-free DNA present in semen into different tissues of origin.
They found that high molecular weight (HMW) DNA contained more fragments from somatic cells than from sperm cells. Their expectation had been the opposite.
"Nobody had really done much work in seminal fluid in terms of nucleic acids. There had been a couple of publications where people had identified actually quite high amounts of cell-free DNA in seminal fluid, but nobody really knew where it came from," Langone said.
"We were obviously expecting the large majority of cell-free DNA to come from sperm," said Langone. But the investigators were surprised to find that there was signal from other tissues both in nucleosomal fragments and in HMW fragments.
The proportion of prostate signal in the HMW cell-free DNA was as high as 26.5 percent, investigators reported.
"Initially, when we started seeing the signal, we kept thinking, gosh, there's just all this sperm that's lysing," which prompted at the company to examine vasectomy samples for HMW DNA that was not coming from sperm, said Langone.
When they confirmed its presence, they then developed methodologies for methylation-based tissue deconvolution in both vasectomy and non-vasectomy patients, she added.
The company has since filed patents on a methodology that uses HMW cfDNA to enrich the signal from prostate tissue as well as a proprietary methodology to identify and ignore background noise from sperm DNA based on the way it is packaged in the sperm head.
Bergstrom said that the company has already started its first clinical study assessing the ability of its approach as a reflex test for men recommended to have a biopsy after either a high PSA result or an abnormal digital rectal exam.
"The goal would be to reduce the number of unnecessary biopsies," he said, which is a true unmet need considering the high false-positive rate of PSA as a cancer biomarker.
Looking to the future, Bergstrom said there could be a use case for its platform in monitoring for prostate cancer in individuals under active surveillance, building on the company's mail-in service.
"It might be very attractive to think about trying to do screening with this test, but I think replacing PSA at this point would be very, very challenging," Bergstrom said. "So we want to kind of take baby steps and start with this post-PSA reflex test, then move into surveillance before we try to tackle screening."
At the upcoming AUA meeting, investigators will present two posters describing a study of 276 patients scheduled for a biopsy after either a high PSA result or a physical exam finding.
"We sent them our at-home collection kit. They sent the sample back to us before their biopsy, and then the results from the biopsy were sent back to us, as well. From that sample, we then looked at … methylation, some mutation data, and RNA expression," said Langone.
In abstracts published online ahead of the meeting, authors wrote that the firm's RNA analysis in 148 evaluable samples found that the presence of prostate-cancer associated fusions was not specific to high-grade disease. However, they were able to find a set of RNAs that could distinguish significant prostate cancer from indolent disease or the absence of cancer with high accuracy as represented by an area under the receiver operating curve of about 0.81.
In their methylation study, 168 patients were evaluable, 109 of which had either no cancer or indolent disease with a Gleason grade of 1. Splitting the patients into a training set and a test set, the team settled on a set of methylation markers in 74 genes that could distinguish those non-cancer and indolent cases from prostate cancers. In the 18-patient test set, the AUC was 0.84.
According to the authors, the use of MRI in the study cohort resulted in unnecessary biopsies almost 70 percent of the time.
"Obviously going on from that, we need to enroll additional patients [and] bolster up the number of samples to develop a classifier for differentiating between clinically significant and indolent versus no cancer. And then we need to get this to a robust assay that could be then transferred over to our CLIA lab," said Langone.
Bergstrom reiterated that the company's interest in expanding its clinical impact has been longstanding, and it is dedicated to taking this early work forward.