Method Profile
Integrate expression, exosomal miRNA and drug-response profiles across matched primary and metastatic models.
Generate hypotheses about molecular features and pharmacology associated with colorectal liver metastases.
Five patient-matched primary colorectal cancer/liver-metastasis sets, represented by 11 organoid models. Four sets were generated from early-passage cell lines and one directly from tissue.
Whole-exome sequencing, RNA sequencing, exosomal miRNA microarrays and 21-compound dose-response profiles, followed by DIABLO, a supervised multi-omics integration method in the mixOmics R package, and exploratory classification.
Drug screen
Six serial dose points at a 1:3 dilution ratio, with drug added 72 hours after seeding. At least triplicate screens produced dose-response AUC profiles.
Computational inputs
DIABLO used 319 mRNAs, 296 miRNAs and 21 drug-response features. Automatic modeling used 102 predictors, a 70% training split and fivefold settings.
Use primary/metastatic labels for supervised integration. Identify cross-layer correlations and within-cohort discriminating features, without treating these as prospectively validated treatment predictions.
Patient-paired, multi-layer model comparison and exploratory computational association.
Five patients and 11 cultures, with at least triplicate drug screens. Parental-tumor genomic concordance could not be tested because the original tissue was not preserved.
Matched primary-site versus liver-metastasis organoids; solvent controls; a separate public expression dataset for selected genes.
Models within a patient shared many mutations and expression features. DIABLO linked miRNA, mRNA and drug profiles; regorafenib, KRAS-related scores, TGF-β and apitolisib were discriminating features within the cohort.
Five patient-matched model sets combined sequencing, exosomal miRNA and 21-drug screening. The associations nominated metastasis-related features, but were not validated as predictors of patient treatment response.
