3D-tumor organoid, over 2D-cell, is more predictive of in vivo anti-cancer pharmacology in assessing ADC candidates

ADC is a promising anti-cancer modality lately for widened therapeutic window (TW) due to tumor-specific targeting by direct binding to tumor cell-associated antigen (TAA), internalization and payload releasing/cytotoxicity induction. The conventional evaluation of lead ADC candidates usually involves the following steps: 1) ELISA-based binding to recombinant TAA protein; 2) TAA + -2D-cell culture for assessing binding, internalization and cytotoxicity; 3) anti-tumor pharmacology using TAA + xenograft tumor models. This process is largely based on the assumption that 2D cancer cell line culture is predictive of in vivo tumor models. However, 2D-assay is frequently not reflective of tumor pharmaology assay from time to time due to distinct architectures of tumor (cell), rendering the process unproductive. Considering 3D tumor organoid being more reflective of tumor than 2D cell culture, we hypothesized that addition of 3D-tumor organoid culture as a new tier of assessment, right before in vivo tumor pharmacology, could make the process more productive. In this report, we set out to test this hypothesis by using three antibodies (Ab1, 2 and 3) against three different epitopes of PD-L1 and their corresponding ADCs (ADC1, 2, 3) with the same payload/DAR-value for binding/ internalizations/cytotoxicity in several PD-L1 + cancer cell lines. Despite similar high-affinity to PD-L1 recombinant protein or PD-L1 + cells, ADC-1 exhibited the strongest internalization/cytotoxicity in PD-L1 + tumor cell lines, whereas ADC2 and 3 showed significantly lower activity (up to 1000x folds in differences). In another word, only ADC1 looks promising as an ADC candidate for further development. However, when all three ADCs were tested in vitro using 3D-tumor organoids, all ADCs demonstrated potent and comparable activities, sharp contrasting to those seen in 2D-cultures. We then tested them in the same 2D-cell-derived xenograft tumor models (CDXs), the results of which are consistent with those from 3D-tumor organoid cultures, but not those from 2D-cell cultures. These confirmed better transnationality of 3D-organoids over 2D-cells. Together, these findings support a practical workflow in which 2D assays serve as early benchmarks, while 3-D organoids provide a more predictive in vitro readout of ADC pharmacology prior to in vivo confirmation. We believe that the process of 2-D (benchmark) → 3-D organoids (translational) → in vivo (validate) can more efficiently streamline preclinical ADC triage. Furthermore, if the hypothesis that 2D (monolayer) culture is more representative of normal tissues, while 3D-tumor organoid is more representative of tumor architecture, is true, we could further suggest that 3D tumor and 2D cell culture differentiation assay could help to identify more tumor-specific ADC candidates with widen TW, e.g. ADC2 and ADC3 described in this report.