Skip to main content
Premium Trial:

Request an Annual Quote

Study Affirms Platelet RNA's Diagnostic Potential for Early- and Late-Stage Cancer Detection


BALTIMORE – A large, multinational study encompassing thousands of samples has shown that blood platelet RNA can help detect 18 tumor types and determine the tumor site of origin for certain cancers.

Led by researchers from Amsterdam University Medical Center in the Netherlands, the study, published last month in Cancer Cell, demonstrates the potential diagnostic utility of blood platelet RNA as a complementary analyte for pan-cancer liquid biopsy testing. However, work still needs to be done before it can be harnessed for robust and routine diagnostic use.

"It all started with a small observation a couple of years ago that platelets [can] sequester RNAs derived from a tumor," said Myron Best, a neurosurgery resident and post-doc with the Amsterdam UMC who is one of the corresponding authors of the paper.

Highly abundant in the peripheral blood, platelets, albeit lacking a nucleus, contain pre-mRNA transcripts derived from megakaryocytes that can be spliced into mature mRNA upon stimulation.

In 2015, Best and his collaborators published a landmark study that showed tumors can "educate" platelets by altering their RNA profile, and tumor-educated platelet (TEP) RNA signatures can be harnessed as companion diagnostics in both local and metastatic cancer.

"Platelets are highly reactive," Best said. "In the presence of cancer, these platelets are changing from the good guys towards the bad guys." According to him, platelets not only can harvest biomolecules from a tumor but also are found to play a role in inflammation, cancer progression, and metastasis. In addition, the TEP-derived RNA profiles appear to be tumor-specific, offering a unique fingerprint to help identify different cancer types.

Building on the previous findings, in 2017, Best's team published another study that described ThromboSeq, a multiplexed TEP RNA-based workflow that synergizes optimized wet lab protocols with machine learning for cancer detection. The study showed that ThromboSeq can help identify early- and late-stage non-small cell lung cancer based on the patient's platelet RNA signals obtained from next-generation sequencing.

In the meantime, researchers have expanded their research on platelet RNA to more disease types, including multiple sclerosis, pulmonary hypertension, and glioblastoma.

"This study is actually the apotheosis of the [previous] work, in which we put all the resources we have together and find out if we can detect cancer early using the platelets, and in how many tumor types," Best said.

For their latest study, the researchers analyzed blood platelet samples from over 2,400 individuals collected from over a dozen biomedical institutions in Europe and North America. The samples, which covered ages ranging from 18 to 92 years old and both sexes, represented 18 different tumor types, as well as asymptomatic and symptomatic controls. It is unclear from the paper, however, what the racial and ethnic composition of the cohort was.

According to Best, the study was designed to answer at least three main questions. One was to explore whether TEP RNA can be harnessed to develop a highly specific cancer test that is also able to detect cancers at an early stage. Another question was whether it is possible to identify the tumor site of origin using the TEP RNA signals. Finally, the researchers wanted to understand whether metastasis has an added impact on platelet RNA profiles.

Beyond these, Sjors In ’t Veld, a researcher at Amsterdam UMC and the first author of the study, said the paper also aimed to systematically examine whether the previously developed ThromboSeq protocol is both technically and biologically robust for cancer detection.

Overall, the results showed that ThromboSeq achieved 99 percent specificity in asymptomatic controls, and the test correctly identified the presence of cancer in two-thirds of 1,096 blood samples from stage I to IV cancer patients. Specifically, the test correctly detected 46 percent of stage I tumors, 47 percent for stage II, 54 percent for stage III, 72 percent for stage IV, and 61 percent for unknown-stage cancers.

When it came to symptomatic controls, which included participants with inflammatory and cardiovascular diseases and benign tumors, the test's average specificity dropped to 78 percent. The increased false-positive rate indicates that a patient's underlying disease or condition could hamper ThromboSeq's detection accuracy.

Moreover, the study showed that ThromboSeq correctly determined the tumor site of origin for five tumor types in over 80 percent of cancer patients. The researchers also found that, at least for brain cancer, both the primary and metastatic tumor sites influenced the platelet RNA profile.

Interestingly, the researchers also found that the two tumor types that were initially left out of the ThromboSeq machine learning algorithm training process — lymphoma and esophageal cancer — were still "very well recognized" by the algorithm as being cancer, Best said, indicating a generic platelet RNA pan-cancer profile.

"Historically, there was this notion of complete silence within platelets as they circulate throughout the body for nine to 11 days before clearance," Laura Bukavina, a urologic oncologist at Fox Chase Cancer Center who was not involved in the study, wrote in an email. "This has changed over the last several years, as we now know that [platelets] are full of mRNA, ribosomes, template RNA, as many of these have been transferred from the platelet progenitor cell."

Despite platelet RNA's potential diagnostic utility, Bukavina said, compared with circulating tumor DNA (ctDNA), which directly comes from cancer cells, platelet RNA is more of a predictive biomarker of a disease state rather than an indicator of specific mutations in a tumor.

"Clearly, if you think about it, ctDNA is going to be more sensitive for [the] prediction of [a] particular disease as compared to platelet RNA, which is confounded by many disease processes," Bukavina said. "However, ctDNA is rarely used as a screening test, because it is specific for the disease." Platelet RNA, on the other hand, can be potentially used as a screening tool for high-risk populations, she noted.

Still, Bukavina thinks that platelet RNA is "not ready for prime time." While the study showed higher predictive accuracy in blood and lymphatic cancers such as lymphoma or leukemia, she said, detection rates for early-stage and solid cancer types are still "exceedingly low."

Moreover, Bukavina pointed out that although the results showed a relatively high prediction rate for stage IV cancers, many patients at that stage are already symptomatic and a positive test result will likely not alter their treatment and disease course. Therefore, "this test fails to be a screening test at that point," she added.

Bukavina said for platelet RNA to be developed as an effective screening tool, the assay needs to continue improving its sensitivity for early-stage cancer, which "ultimately will depend on increasing [the] number of samples, of all races and sexes."

Best acknowledged the current limitations of ThromboSeq-based cancer detection, including the need to improve specificity and accuracy, especially for patients with other underlying conditions. Additionally, he and In ’t Veld cautioned that the study did not conduct a head-to-head comparison between platelet RNA and other biomarkers such as ctDNA.

Application-wise, Best said one of the potential use cases for ThromboSeq is as a screening test, either for the general population or people who have a predisposition for a certain cancer.

Additionally, he said platelet RNA can potentially aid clinicians in making appropriate treatment plans for certain tumor types. For instance, one challenge with glioblastoma treatment is that when patients receive chemotherapy or radiotherapy of the brain, the tissue might show pseudo-progression, a scenario where the tumor shows progression under imaging, but clinicians are not sure if that is indeed the case.

Because previous studies showed that platelet RNA is likely to reflect the true tumor behavior in glioblastoma, it may complement the current diagnostic tools to help clinicians discern true cancer progression from pseudo-progression, helping make better clinical decisions, Best said.

In terms of ThromboSeq's cost, Best said a rough estimate of the material cost for the test in an academic setting is about €‎100 (roughly $96) per sample. The entire workflow is typically doable within two weeks, he said.

As for commercialization, the authors said the intellectual property pertaining to ThromboSeq was previously transferred to Illumina through its acquisition of ThromboDx, a spinout from Amsterdam UMC that is now defunct. However, Best said the IP has been transferred back to the university, and the technology is not being commercialized by any companies at this point.

Moving forward, the researchers said their hope is to continue validating the diagnostic utility of ThromboSeq in a prospective study with an even larger sample size, including more healthy controls. In the long run, Best thinks platelet RNA will not replace other biomarkers but will rather become an important analyte that can be complementary to other types of liquid biopsies. 

"We were not looking for replacing all the other biomarkers," he said. "With this study, we show that [platelet RNAs] are a serious player in the field and should not be ignored."