Five-layer in-silico validation of the CA9–histone-lactylation–drug-efflux axis in osteosarcoma, with read-level cfDNA fragmentomics as a relapse-monitoring feasibility endpoin
Year: 2026
Type: study-registration
Abstract: Five-layer in-silico validation of the CA9-histone-lactylation-drug-efflux axis in osteosarcoma (OS), plus a read-level cfDNA fragmentomics feasibility endpoint. Orthogonal public-data layers with different confounders are pre-specified to converge on one axis reading: (i) association of CA9 with lactylation/glycolysis module scores across three OS cohorts (TARGET, GSE21257, SGH n=87); (ii) direction via graph-regulatory-network virtual knockdown of CA9; (iii) cellular localisation of ABCB1/ABCG2 within CA9+ malignant cells (17,465 single cells); (iv) confounding control using stroma-free CCLE cell lines (1,019 lines; EBI E-MTAB-2770) against the patient-bulk CA9-ABCG2 negative association; (v) clinical context via pathological necrosis-response strata (TARGET n=21). A LINCS L1000 signature-matching intervention preview ranks candidate axis inducers and reversers. Secondary feasibility endpoint: in perioperative plasma cfDNA (n=11, RRBS/Bismark BAMs), median fragment length is shorter in relapse/metastasis event patients, tested by exact permutation with pre-specified direction. Pre-registered decision rules govern acceptance of the stromal-confounding interpretation and downgrading of the cfDNA endpoint if the signal fails in expanded samples. All data are public or de-identified; analysis code will be released under MIT.
Source: Open Science Framework
Author: Ke Zeng
Language: English
Subfield: Cancer Research
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7225040385