Physiologically relevant tissue-engineered technologies utilized for precision-based drug screening in women’s cancers as an opportunity to advance women’s health

Introduction: Lack of efficacy and a low overall success rate of cancer clinical trials are the most common failures when it comes to advancing cancer treatment. Current preclinical drug sensitivity screenings present several challenges including an inability to encompass the complexity and heterogeneity of tumors, reproducibility, and a lack of translatability. Unfortunately, the most common preclinical cancer models do not deliver full fidelity of the heterogeneous tumors in the context of a tumor-like environment. The recapitulation of cell-cell interactions, the recreation of the extracellular matrix (ECM) architecture, the simulation of drug transport and growth factors, and the amenability to perform high-content screenings are key required features for more suitable preclinical models that would provide an accurate prediction of clinical efficacy. Therefore, there is a critical need for the development and functional characterization of technologies for precision-based drug screenings accounting for physiologically relevant tumor characteristics.

The overall objective of this investigation is to provide a translationally relevant 100% patient-derived 3D culture platform with controlled physiologically relevant physical properties to perform clinically relevant analysis of chemotherapeutic responses in women’s cancers. Specifically, tumoroids (“tumor-like-organoids”) have been shown to retain both histological and genetic features of original tumors and are feasible options for in vitro drug sensitivity assays, recapitulating clinical responses of the matched patients. In the absence of these models, the prediction of the best therapeutic regimen for each patient will likely remain difficult and the improvement of survival rates and reduction of healthcare costs unachievable.

Materials and

Methods: The precision-based tissue-engineered cultures were developed through the crosslinking of patient-derived plasma fibrinogen to fibrin and patient-derived tumoroids from matched gynecological cancers and breast cancer patients were embedded within the 3D matrix (Patient-derived tumoroids, PDT) (Fig. A). This PDT model was fully characterized to show that the model upholds the morphological features of parental tumors (tumor microarrays and immunofluorescence (IF)), and preserves the biochemical composition of the parental tumor while in culture. PDTs were then engineered and characterized to mimic physiologically relevant physical properties including oxygen content verified by oxygen microsensor, ECM composition confirmed by IF, and stiffness shown by atomic force microscopy. Primary biospecimens were categorized as sensitive and resistant by Response Evaluation Criteria in Solid Tumors or RECIST score (defined by pathology, radiology and/or physician decision of follow-up response >6 months to standard of care treatment). PDTs were grown and exposed to standard-of-care carboplatin/paclitaxel treatment at increasing doses for 7 days. To create a predictive score that correlates with the clinical response, several metrics were evaluated including tumor: stroma ratio by QuPath, histopathological screenings for proliferation (Ki67), apoptosis (caspase 3) and CD4/CD8 score, as well as apoptosis screening by flow cytometry. These comparative drug studies utilizing the PDT model could provide novel insights into drug efficacy and drug resistance within patients making it a useful model for investigating advancements in treatment for women’s cancers.

Results, Conclusions, and Discussions: The PDT model recapitulates structural complexity, morphological features, and preserves biochemical composition of the parental tumor. Specifically, cytokeratin (epithelial marker, red) revealed the preservation of the native epithelium (Fig. B), tumor microarrays highlighted conservation of tumor heterogeneity (Fig. C), and cytokine arrays validated no changes to important cytokines/signals in the tumor microenvironment (Fig. D). Importantly, the PDT model was engineered to recapitulate physiological oxygen levels of normal physoxic tissue compared to the hypoxic tissue of tumors and profiling the PDT showed that incubation at 21% O2 captures 6.7kPa oxygen content while 1.5% O2 incubation recapitulates 0.77kPa recapitulating the respective oxygen levels. The ECM production at 1.5% O2 incubation revealed increased deposition of collagen I and laminin compared to the ECM composition at 21% O2 incubation. Additionally, increased stiffness in the matrix correlated with the increased ECM deposition in 1.5% O2 incubation compared to 21% O2 (Fig. E). The functionally characterized PDT model was used for precision-based drug screening and several metrics were evaluated to predict patient treatment response. While a tumor:stroma ratio higher than 0.5 was linked to worse outcomes (Fig. F), histopathological screenings for Ki67, caspase 3 and CD4/CD8 had a moderate prediction (Fig. G). Apoptosis and survival screenings were able to retrospectively match patient response distinguishing sensitive from resistant patients (Fig. H and I). Critically, these results indicate that this model is a unique preclinical model allowing for the recapitulation of physiologically relevant physical properties which could make it a well-suited platform for precision-based prediction of therapeutic efficacy in women’s cancers.

In conclusion, we have optimized a reproducible and clinically translatable preclinical model for assessing effective treatment options by predicting therapeutic efficacy and avoiding treatment for patients with drugs that the tumor could have resistance to. Moreover, our results are expected to have an important positive impact because they will provide a valuable tool for establishing a high-throughput drug screening platform that can be used to proactively predict each individual patients’ response to therapy to assist in therapeutic selection and permit a much more in-depth and clinically relevant analysis of personalized treatment responses than is currently possible.

Acknowledgements (Optional): This project is supported by U54-GM128729 and 5P20GM103548.