3D Printed Trimodal Analytical Platform for Simultaneous Molecular Analysis of Protein, RNA, and DNA: A Proof of Concept for Multiclass Biomarker Analysis Applied to Hepatocellular Carcinoma

Abstract Protein, RNA, and DNA are normally quantified by separate methods, each needing its own sample preparation, reagents, and instrument, which limits multiclass biomarker analysis from small, crude samples. Methods: We developed the GPC3-Chip, a monolithic device fabricated by masked stereolithography (MSLA) 3D printing on a desktop printer. A single 20−60 μL sample is routed through three lanes that combine EDC/NHS amine-coupled antibody immunocapture of glypican 3 (GPC3), an on-chip UV photometric channel at 260 and 280 nm for RNA and DNA, and a pillar mesh architecture proposed to route components by size-dependent diffusional residence time. Results: The GPC3 channel gave a concentration-dependent optical response with an empirical blank-based detection limit of 0.012 ng/mL and 97.2−103.4% recovery in spiked, diluted commercial serum; because antibody-free and isotype control surfaces were not tested, this response is not yet shown to be specific to GPC3 recognition. The RNA and DNA lanes agreed with Qubit measurements (r2 = 0.994 and 0.997; LOD 1.8 ng/mL), and fluorescent dextran tracers showed less than 8% overlap between lanes in buffer. In two murine hepatocellular carcinoma models, the GPC3 channel response exceeded the detection threshold about eight days before tumors were palpable, and the combined GPC3, cfDNA, and RNA integrity readout gave an apparent AUC of 0.987 versus 0.922 for GPC3 alone (estimated in the same samples used to build the model, without cross-validation or an independent validation set). Conclusions: The chip is a proof of concept, low-cost route to concentration-dependent protein and nucleic acid readouts in a single device. Specificity of the GPC3 response, independent validation of the diagnostic model, separation of real biomolecules in complex matrices, RNA and DNA selectivity in mixtures, and performance in human samples remain to be validated.