NEW YORK (GenomeWeb) – A research team at VU University Medical Center Amsterdam has developed and is validating a new blood-based platform for cancer diagnostics while it explores other application areas such as infectious disease monitoring and prenatal testing.
The approach, dubbed ThromboSeq, enables clinical researchers to identify different cancer types by looking at tumor-educated, platelet-derived RNA using next-generation sequencing.
VU University Medical Center Amsterdam licensed the ThromboSeq IP in 2012 to a spinout called ThromboDx, which Illumina acquired last year for the reported sum of €70 million ($74 million).
An Illumina spokesperson this week confirmed the acquisition but declined to discuss financial details. The company never publicly disclosed the deal. The Dutch financial newspaper, Het Financieele Dagblad, reported on some of the terms of the acquisition at the time.
Illumina's plans for ThromboDx and its technology remain unclear. However, since the acquisition the VUMC team behind the technology has pressed ahead with plans to refine its approach and develop new applications. Myron Best, a PhD student in the neuro-oncology research group at VUMC who has played a lead role in developing the technology, has discussed the development of ThromboSeq publicly, most recently at Cambridge Healthtech Institute's Molecular Medicine Tri-Conference last month in San Francisco.
In his talk, Best detailed the technique's evolution, as well the results of an initial validation study in 283 cancer patients and healthy donors that was published in the journal Cancer Cell in 2015. Best was the lead author on that paper, which demonstrated that by using RNA-seq to profile tumor-educated blood platelets, one could detect early-stage cancer with 96 percent accuracy, and pinpoint the origin of the primary tumor in question with 71 percent accuracy.
"For a long time, platelets were considered to be waste, especially for the detection of biomarkers," Best said in an interview. "We were the first ones to publish in such a large cohort the use of platelets for pan-cancer, multiclass, and molecular pathway diagnostics."
As detailed in the paper, the team relied on a leave-one-out, cross-validation support vector machine (SVM) algorithm to create a pan-cancer classifier test based on 1,072 RNAs. Using the same SVM algorithm, the team was also able to cluster platelet mRNA levels according to six tumor types: non-small cell lung cancer, colorectal cancer, glioblastoma, pancreatic adenocarcinoma, breast cancer, and hepatobiliary carcinomas.
The authors also reported that the same approach could be used to distinguish patients with KRAS mutant tumors from those with KRAS wild-type tumors in relevant cancers, as well as EGFR mutant tumors in NSCLC patients, using algorithms focused on marker-specific gene lists. Also, MET or HER2-positive, and mutant PIK3CA tumors were accurately distinguished using surrogate mRNA profiles. According to Best, the bioinformatics work underlying ThromboSeq was done in house in Amsterdam. The results encouraged the team that tumor-educated platelets are a compelling biosource for oncology testing, "a potential all-in-one platform for blood-based cancer diagnostics, using the equivalent of one drop of blood."
"What we think happens in these platelets is that, as they circulate, they encounter all kinds of cues from the environment from immune cells, epithelial cells, or the tumor," sad Best. "For each disease, a different signature is generated," he said. "For a colorectal tumor, a slightly different splicing program is initiated, resulting in a slightly different RNA profile," he added. "Since these platelets are so reactive and responding to whatever happens in their environment, what happens in the body, they will collect these splicing repertoires and have existing profiles."
To profile blood platelets required innovation, though. Best said that the team at VUMC developed a protocol that allows the sequencing of low quantities of platelet-derived RNA. "Platelets contain only about 2 fentograms of RNA," he said. "The yield of platelet RNA is not that much. So we had to adjust our sequencing protocol by including an amplification step."
Of ThromboSeq's various uses in oncology testing, Best said that the ability to identify tumor origin may in particular complement efforts to test for early-stage cancer using cell-free nucleic acid. "Once you detect cancer, the next question is: Where is it located? Where is the cancer hiding?" said Best. "A platelet-based approach would solve that second question, where the tumor is, and allow one to decide what additional diagnostic test is necessary."
As such, Best advocates pairing the approaches to enable early detection, tumor identification, and therapy selection. "Obviously, blood-based detection of cancer will be the holy grail for curing cancer, by detecting it early when it still can be cured;" said Best. "Here, the platelets will perhaps be a part of the test," he said. "First you can detect cancer at the early stage by detecting the presence of cell-free DNA, and then use a platelet RNA test for determining the tumor type."
He also believes that ThromboSeq could be used to measure response to immunotherapy. "It might be that platelets are involved in immune response and they might have predictive power for immunotherapy response prediction," said Best.
In his view, test developers should move toward a model of combining biosources, rather than focusing solely on circulating tumor cells or cell-free DNA or tumor-educated blood platelets. "We should move to an integrative analysis of multiple biosources together, in this way we could better leverage each individual biosource."
While the 2015 study presented ThromboSeq to the world, Best said that there are many more questions open related to the technology. He noted that peers had questioned the effects of inflammation conditions on the RNA profiles gleaned for sequencing the blood platelets. Also, he said that the sample numbers for the study — less than 300 — were too low to pave its path into regular clinical use.
VUMC recently completed a lung cancer study that analyzed over 700 individuals using ThromboSeq, Best said, and has other studies underway that show, independent of variables, that the approach "still has a high predictive power" for diagnosing cancer with more than 90 percent accuracy. Papers detailing VUMC's work will eventually be published, he said.
In addition to demonstrating ThromboSeq's utility, the VUMC team is studying the biological mechanism behind the profiles. "We are doing experiments to see how these RNA profiles alter both in silico and in vitro to better understand these alterations," said Best. "There are still more things to do here academically."
Thomas Würdinger, who heads Best's group at VUMC, confirmed in an email that the commercial rights of ThromboSeq belong to ThromboDx and Illumina. He said that his group at VUMC will continue its academic work through a wide range of collaborations with a variety of partners.
"It is early days for the platelet work and we are exploring the applicability of this biosource academically across a number of potential applications," Würdinger said.
In 2011, Würdinger and Rolf Jonas Nilsson filed for a broad US patent covering the use of platelets as a source of disease-specific nucleic acids, IP that was subsequently licensed by ThromboDx.
Würdinger, Nilsson, and colleague Rolf Jan Rutten also served as corporate officers for ThromboDx, roles that they relinquished sometime after the Illumina acquisition despite still being listed as such on the ThromboSeq website.
Würdinger and Rutten declined to comment on ThromboDx or its relationship with Illumina.
Illumina's plans for ThromboDx or its technology remain unclear. The potential ability of the ThromboSeq method to pinpoint cancer origin dovetails with the efforts of Grail, the company Illumina helped form last year to develop a pan-cancer screening test based on profiling circulating nucleic acids in blood. Illumina has since reduced its financial stake in Grail significantly.
For instance, the ThromboDx website notes that the technology could be applied not only in cancer diagnostics, but also infectious disease monitoring and prenatal testing as blood platelets can provide a reservoir for nucleic acid from infected cells as well as fetal cells. In 2012, Würdinger and Nilsson also filed for a US patent covering analysis of nucleic acid from blood cells such as thrombocytes for the purpose of detecting a fetal disease or condition.