LiquidCell Dx is developing a cfDNA-based methylation profiling assay, called LiquidTME, to predict responses to immune checkpoint inhibitors by interrogating cellular ecosystems tied to tumor microenvironments. The startup said its underlying methodology was published in Nature, using machine-learning-defined spatial ecotypes from bulk and single-cell spatial transcriptomics. The approach builds on data from 132 primary human carcinoma and melanoma tumors across 10 cancer types, alongside single-cell RNA-sequencing atlases for 144 tumor samples. LiquidCell Dx’s stated plan includes continuing to build regulatory-grade evidence and engaging regulators on commercialization pathways. For oncology diagnostics, the update is notable because tumor microenvironment profiling has historically required invasive sampling and complex workflows. A cfDNA liquid biopsy could potentially broaden access for longitudinal monitoring and for patient selection earlier in care. The company’s next steps—validation of predictive performance and regulator discussions—will likely determine whether LiquidTME can translate the Nature methodology into a clinically actionable product.