To develop a machine-learning approach for predicting immunotherapy response in melanoma using blood-derived DNA.
Key Findings:
Identified nine conserved ecotypes linked to tumor biology and survival outcomes.
High levels of inflammatory ecotypes SE7 and SE8 correlated with durable benefit and longer survival post-therapy.
Ecotype SE4, associated with wound-healing, linked to resistance and poorer survival.
cfDNA-based approach outperformed established biomarkers like tumor mutational burden and PD-L1 expression.
Interpretation:
The study suggests that spatial ecotypes provide broader biological insights into tumor microenvironments, enhancing the predictive power for immunotherapy responses.
Limitations:
Findings require validation in larger prospective cohorts.
Potential challenges in standardizing cfDNA analysis across different laboratories.
Conclusion:
This research indicates a promising future for liquid biopsy in oncology, enabling noninvasive monitoring of tumor microenvironments and improving patient stratification and treatment monitoring.
