Non-invasive profiling of the tumour microenvironment with spatial ecotypes
A blood test capturing tumor structure through DNA patterns may help predict which melanoma patients benefit from immunotherapy without invasive biopsies.
This Nature paper presents a first-of-its-kind computational framework that distills the spatial organization of tumours into 9 ecotypes recoverable from a simple blood draw, using cfDNA methylation and deep learning. In ~100 melanoma patients, these blood-based ecotype signals predicted immunotherapy response, establishing a new paradigm for non-invasive tumour microenvironment profiling.
What the study was
- Study design
- Multi-cohort computational + retrospective clinical validation
- Population
- Diverse human carcinomas and melanomas; ~100 melanoma patients with cfDNA for immunotherapy response validation
- Sample size
- 10000000
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- Nature
Why it surfaced
Landmark Nature paper establishing cfDNA-based non-invasive spatial TME profiling; 9 reproducible spatial ecotypes with immunotherapy response associations in melanoma; enables precision oncology from liquid biopsy without tissue biopsy
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.