Researchers recently published details of a diagnostic test and algorithm that enables the identification of biomarkers that could help to diagnose cancer from tumor-educated platelets (TEP).1
The test, thromboSeq, is a multiplexed biomarker signature detection platform that detects lung cancer by examining tumor RNA absorbed by circulating platelets.
“Blood platelets contain endogenous RNAs – which are protected from degradation – that are altered upon the presence of cancer,” explained Myron G. Best, an MD/PhD student at the department of neurosurgery at the VU University Medical Center in the Netherlands, and first author of the study. “Therefore, platelets provide a source of abundant, high-quality RNAs, and platelets can be easily isolated in any clinical laboratory.”
According to Mr Best, the thromboSeq platform offers a unique insight into the RNA composition of blood platelets encountering a tumor. As blood platelets contain RNAs coding for approximately 4500 genes, the thromboSeq platform detects many potential biomarkers at once.
Together with machine learning software, this platform enables analysis of repertoire or signatures of RNAs, Mr Best told Cancer Therapy Advisor.
To develop the inflammatory conditions–controlled non–small cell lung cancer (NSCLC) diagnostics test, Mr Best and colleagues collected more than 700 blood samples from patients with early- or late-stage NSCLC and a group of individuals without cancer matched for age, smoking status, and blood storage time.
The test’s swarm intelligence algorithm scanned the RNA molecules and recorded the small amount indicating a cancerous tumor. The researchers then ran the samples through screenings to diagnose how accurately the TEPs identified cancer.
Analysis showed that the test had an accuracy of 88% in patients with late-stage disease and, in patients with early-stage disease, an accuracy of 81%, independent of potential confounding factors. In a validation control group matched for patient age, smoking status, and blood storage time, the algorithm detected NSCLC with a 91% accuracy, which was only slightly lower compared with a previous non-matched NSCLC algorithm tested in the absence of non-cancer conditions.2
“The algorithms can be tuned towards high sensitivity or high specificity, of which the latter setting is required in a population-based screening setting, at a cost of specificity and sensitivity, respectively,” Dr Best said. “Of note, we now have tested our thromboSeq platform in a cohort that includes all kinds of inflammatory conditions and that is matched for potential confounding factors such as age, blood-storage time, and smoking.”