Quadratic Phenotypic Optimisation Platform (QPOP) Utilisation to Enhance Selection of Patient Therapy Through Patient Derived Organoids in Breast Cancer

Overview

Based on proof-of-concept study, the investigators hypothesise that the QPOP prediction model can be further extended into use in solid tumors. Using breast cancer as a model, the investigators intend to investigate the feasibility of QPOP as a clinical decision support platform to identify patient-specific drug combinations across a range of breast cancer patients. The investigators propose a pilot phase I clinical study to test the feasibility of using QPOP to guide therapy in patients with advanced breast cancer.

Breast cancer (BC) is highly heterogeneous with more than 20 genetically, morphologically and clinically distinct subtypes. Despite improved diagnosis and a wide repertoire of clinically-approved therapies, metastatic BC patients have a dismal 5-year survival rate of about 20%. Systemic treatment options for breast cancer are empiric based on large clinical studies, and the ability to tailor choices individualized to each patient may aid optimization of patient's responses to treatment. A major obstacle for drug development is the lack of clinically relevant cell model systems. The use of conventional two-dimensional (2D) cell cultures presents limitations in the recapitulation of the primary disease characteristics. In particular, monolayer cultures have been demonstrated for not being able to recapitulate the heterogeneous nature of primary patient tumor tissues. Despite the establishment of the National Cancer Institute 60 Panel (NCI-60) in an attempt to capture the heterogeneous nature of nine different cancer types, it is still unable to comprehensively represent the diversity which cancer patients exhibit, and is currently being phased out. Studies comparing the molecular profiles of cell lines with primary tumors have reported that conventional cell lines are unable to represent all cancer subtypes. Recently, patient-derived organoids (PDOs) has been demonstrated to faithfully preserve and maintain heterogeneity of several primary cancer types including BC and are being used to evaluate drug sensitivity and their associated genomic variations. PDOs have also been shown to retain the molecular profiles of the parental tumors and can be effectively used to investigate drug responses and mechanisms ex vivo. Thus, human PDOs can provide more clinically relevant models that can recapitulate disease heterogeneity for more accurate studies of drug responses as compared to conventional cell line-based models. Identifying the most suitable patient-specific drug combinations remains a challenge due to the complex molecular networks that contribute to feedback mechanisms of drug resistance and compensatory oncogenic drivers that limit efficacy of targeted inhibitors. Pharmacogenomics is highly useful in predicting drug sensitivity and clinical outcome and define appropriate subpopulations for specific drugs based on genetic biomarkers. These approaches are largely limited to monotherapy and cannot identify patient-specific drug combinations nor do they consider other genomic alterations that are unaccounted for by the specific biomarkers used. In order to address this critical deficit in combination therapy, the investigators have developed an experimental-analytical hybrid platform, Quadratic Phenotypic Optimisation Platform (QPOP), that can rank potential drug combination response in biological model systems. The investigators initially applied QPOP towards identifying novel drug combinations against bortezomib-resistant multiple myeloma. As the investigators have improved the efficiency of the QPOP platform and applied QPOP towards primary patient samples and patient-derived organoids, it has become clear that QPOP may be useful as a clinical decision support platform that may be able to suggest clinically actionable and suitable patient-specific drug combinations derived from drug sensitivity tests using patient-derived materials. The investigators hypothesise that the QPOP prediction model can be further extended into use in solid tumors. Using breast cancer as a model, we intend to investigate the feasibility of QPOP as a clinical decision support platform to identity patient-specific drug combinations across a range of breast cancer patients. The Investigators propose a pilot phase I clinical study to test the feasibility of using QPOP to guide therapy in patients with advanced breast cancer. Eligible patients will undergo a fresh biopsy of tumour lesion to obtain cells that will be used to generate patient-derived tumour organoids. After successful organoid generation, organoids will be subjected to a 12-drug panel screening including 10 fixed drugs and 2 drugs that may be selected by physician based on treatment history of individual patients. The drug panel is curated taking into account standard drugs commonly used in breast cancer treatment as well as drugs/combinations being tested at ongoing therapeutic trials at the National University Cancer Institute Singapore (NCIS) which enrolled patients can gain access to. Fixed drug panel will include chemotherapy (cisplatin, 5-fluorouracil, paclitaxel), endocrine therapy (tamoxifen, fulvestrant), HER2-directed therapy (lapatinib) and small molecule tyrosine kinase inhibitors (abemaciclib, olaparib, alpelisib, lenvatinib). QPOP analysis results will be reported to treating physician once available to aid selection of therapy. To prevent delay in patient treatment while awaiting QPOP results, patients are allowed to receive next line of therapy at physician's discretion (termed "empirical therapy"), and QPOP analyses results can be used to guide further lines of treatment upon progression on "empirical therapy".