A major concern in tissue biopsies with a needle is missing the most lethal clone of a tumor, leading to a false negative result. This concern is well justified, since needle-based biopsies gather tissue information limited to needle size. In this work, we show that molecular harvesting with electroporation, e-biopsy, could increase the sampled tissue volume in comparison to tissue sampling by a needle alone. Suggested by numerical models of electric fields distribution, the increased sampled volume is achieved by electroporation-driven permeabilization of cellular membranes in the tissue around the sampling needle. We show that proteomic profiles, sampled by e-biopsy from the brain tissue, ex vivo, at 0.5mm distance outside the visible margins of mice brain melanoma metastasis, have protein patterns similar to melanoma tumor center and different from the healthy brain tissue. In addition, we show that e-biopsy probed proteome signature differentiates between melanoma tumor center and healthy brain in mice. This study suggests that e-biopsy could provide a novel tool for a minimally invasive sampling of molecules in tissue in larger volumes than achieved with traditional needle biopsies.

Clinical misclassification between cutaneous squamous cell carcinoma (cSCC) and basal cell carcinoma (BCC) affects treatment plans and carries risks of potential for recurrence, metastases morbidity and mortality. We report the development of a novel tissue sampling approach with molecular biopsy using electroporation. The methods, coined e-biopsy, enables non-thermal permeabilization of cells in the skin for efficient vacuum-assistant extraction of informative biomolecules for rapid diagnosis. We used e-biopsy for ex vivo proteome extraction from 3 locations per patient in 21 cSCC and 21 BCC pathologically validated human tissue samples. The total 126 extracted proteomes were profiled using LC/MS/MS. The obtained mass spectra presented significantly different proteome profiles for cSCC and BCC with several hundreds of proteins significantly differentially expressed in each tumor in comparison to the other. Notably, 17 proteins were uniquely expressed in BCC and 7 were uniquely expressed in cSCC patients. Statistical analysis of differentially expressed proteins found 31 cellular processes, 23 cellular functions and 10 cellular components significantly different between cSCC and BCC. Machine Learning classification models constructed on the sampled proteomes enabled the separation of cSCC patients from BCC with average cross-validation accuracy of 81%, cSCC prediction positive predictive value (PPV) of 78.7% and sensitivity of 92.3%, which is comparable to initial diagnostics in a clinical setup. Finally, the protein-protein interaction analysis of the 11 most informative proteins, derived from Machine Learning framework, enabled detection of a novel protein-protein interaction network valuable for further understanding of skin tumors. Our results provide evidence that the e-biopsy approach could potentially be used as a tool to support cutaneous tumors classification with rapid molecular profiling.

No abstract

No abstract

No abstract