Nine Diagnostics co-founders Daniel A. Heller, PhD (CSO) and Mijin Kim, PhD (Scientific Advisor) are co-authors on a study published as the cover article in the February 2026 issue of Nature Nanotechnology: “Machine perception liquid biopsy identifies brain tumours via systemic immune and tumour microenvironment signature” (vol. 21, pp. 277-287). The study represents the second major clinical validation of the core nanosensor technology underlying the Nine Diagnostics platform, and introduces capabilities with broad implications for blood-based cancer diagnostics.
Why Brain Tumors Are the Hardest Case for Liquid Biopsy
Brain tumors pose a uniquely difficult challenge for blood-based diagnostics. The blood-brain barrier (BBB) is a highly selective membrane that tightly regulates the passage of molecules between the central nervous system and the bloodstream. For most extracranial cancers, circulating tumor DNA (ctDNA) and other tumor-derived fragments are reliably detectable in blood, and liquid biopsy has matured into a clinically actionable field. For brain tumors, ctDNA reaches the bloodstream in fewer than half of cases, often at levels too low for clinical use. This makes the brain one of the most diagnostically isolated anatomic sites in oncology and a longstanding open problem for blood-based detection.
An Approach That Reads the Body’s Response
Rather than trying to detect tumor material crossing the BBB, the machine perception liquid biopsy (MPLB) approach detects the tumor indirectly by reading the molecular state of the blood itself. Arrays of quantum-defect-modified single-walled carbon nanotubes stabilized with single-stranded DNA are incubated with plasma. Each nanosensor acquires a protein corona shaped by the proteins circulating in the patient’s blood. Machine learning models trained across these multi-nanosensor fingerprints detect the presence of intracranial tumors and identify tumor type without requiring prior knowledge of which biomarkers to measure.
A Signal From Three Compartments
A subsequent proteomic analysis of the highest-performing nanosensor reveals where the detection signal originates. The study identifies tumor ecosystem-secreted factors from three distinct biological sources: proteins secreted directly by the cancer cells; proteins arising from the tumor microenvironment, including local immune infiltrates, stromal cells, and vasculature physically within the tumor; and proteins reflecting the systemic immune and inflammatory response, which alters the molecular composition of circulating blood even when the tumor is sequestered behind the BBB. Some of these factors were previously described in the literature; others were newly discovered, demonstrating that the platform functions not just as a detector but as a biomarker discovery engine operating from a standard blood draw.
Key Results
Across 739 plasma samples interrogated at Memorial Sloan Kettering Cancer Center, Northwestern University, and NYU, machine learning models achieved 98% accuracy in detecting intracranial tumors and could identify tumor type, including distinguishing glioblastoma from meningioma and detecting indolent slow-growing tumors that shed comparatively few molecules into the bloodstream. Lead author Dana Goerzen of the Heller Lab received the award for oral presentation at the Vincent du Vigneaud Symposium for this work.
What This Means for Nine Diagnostics
Nine Diagnostics is building an AI-enabled multi-omic nanosensor platform grounded in this same core technology. This publication builds directly on the first large-scale clinical validation of the platform, published in Nature Biomedical Engineering in 2022, which demonstrated 87% sensitivity and 98% specificity for high-grade serous ovarian cancer across 269 samples. The brain tumor study extends that foundation in three meaningful ways: it validates the technology in a second cancer type; it achieves the first multi-site clinical validation, demonstrating that performance holds across independent institutions; and it demonstrates for the first time that the platform can surface a blood-based molecular signature for a cancer physically isolated behind the blood-brain barrier, one of the hardest barriers to overcome in oncology diagnostics.
Read the full announcement from Memorial Sloan Kettering Cancer Center and the full paper in Nature Nanotechnology.
About Nine Diagnostics
Nine Diagnostics is an AI-enabled multi-omic nanosensor company advancing precision medicine. The platform simultaneously captures proteomic, metabolomic, and lipidomic signals alongside patient clinical context to generate a multi-omic fingerprint, using machine learning to identify disease-relevant patterns without requiring prior knowledge of which biomarkers matter. This enables pre-treatment patient stratification, on-treatment response monitoring, and post-treatment minimal residual disease detection. Founded by Freddy T. Nguyen, MD, PhD (CEO), Daniel A. Heller, PhD (CSO), and Mijin Kim, PhD (Scientific Advisor), Nine Diagnostics is based in Cambridge, Massachusetts.