Use of ex vivo/in vitro models to predict in vivo drug responses

Oded Kopper¹, Norman Sachs¹, Chris De Witte², Kadi Lõhmussaar¹, Jose Espejo Valle-Inclan², Wigard Kloosterman² and Hans Clevers¹

¹ Hubrecht Institute, Royal Netherlands Academy of Arts and Sciences, Uppsalalaan 8, 3584 CT, Utrecht, The Netherlands

² Department of Genetics, Center for Molecular Medicine, University Medical Center Utrecht, Utrecht University, Universiteitsweg 100, 3584 CG Utrecht, The Netherlands

Introduction

Ovarian and breast cancer (OC and BC, respectively) are heterogeneous diseases consisting of multiple tumor subtypes. Experimental in vitro models that faithfully capture the hallmarks and tumor heterogeneity of these diseases are limited and hard to establish. As first shown for colorectal cancer, tumor organoid cultures represent robust 3D in vitro systems that faithfully recapitulate the tumor from which they are derived. In our work we have established and characterized organoid cultures for ovarian and breast cancer.

Material and Methods

Tissue and blood were obtained from consenting patients who underwent tumor resection and/or drainage of ascites/pleural effusion. For each case, the available tissue was used for organoid derivation, DNA isolation and histological analysis. Histological and genomic landscape analysis of successfully established organoid lines and the tumor from which they were derived, was performed. Response of different organoid lines to diverse chemotherapies was tested both in vitro and in vivo, following transplantation into immunodeficient mice.

Results and Discussion

OC and BC organoids recapitulate histological and genomic features of the pertinent lesion from which they were derived, illustrating intra- and inter-patient heterogeneity. We show that both platforms can be used for drug screening assays. Moreover, OC organoids capture different tumor subtype responses to the gold standard platinum-based chemotherapy, including acquisition of chemoresistance in recurrent disease. Whereas BC organoid drug screens were consistent with in vivo xenografts and patient response.

Conclusion

OC and BC organoids can be utilized as a valuable tool for research, facilitate drug development and present us with new opportunities for personalized medicine.

Introduction

The limited role of genomic profiling in predicting response to targeted therapies and limitations of pre-clinical models currently used for drug validation represent important obstacles hampering the success of personalized medicine and drug discovery.

In vivo functional genomics has been proposed as a way to overcome some of the hurdles in understanding the molecular complexity of drug response and resistance. Co-clinical trials matching drug response in patients and related pre-clinical models represent a promising strategy to personalize treatment and understand mechanisms of chemo-sensitivity through reverse translation. Most co-clinical trials rely on the use of genetically engineered mouse models or patient-derived xenografts. Although these classes of animal models can sometimes closely mirror clinical scenarios, their use poses several logistic, ethical, and economic issues. Thus, there is an unmet need to develop robust, rapid, and cost-effective pre-clinical models of metastatic cancers.

LGR5+ stem cells can be isolated from a number of organs and propagated as epithelial organoids in vitro. Mouse and human organoids have been used to study the physiology and neoplastic transformation of the liver, pancreas, bowel and prostate among other organs.

Material and Methods

Patient derived organoids (PDOs) were established from ultrasound and/or CT scan guided tissue biopsies from metastatic, chemo-refractory, heavily pre-treated colorectal and gastroesophageal cancer patients recruited in phase I/II clinical trials. Sequential (pre- and post-treatment PDOs were obtained in a number of patients). Genomic and transcriptomic profiling of PDOs were compared to the ones of their parental biopsies and that of the primary cancer where available. High-throughput drug screening using a library of compounds in early clinical trials or clinical practice was used for 3D drug testing.

Results and Discussion

Phenotypic, genotypic and transcriptomic profiling of PDOs was compared to their matching tumour demonstrating high similarity between the two. Drug screening results were matched with the molecular profiling of PDOs in analysing response to targeted agents or drug combinations, supporting the notion that PDOs could complement sequencing approaches in defining cancer vulnerabilities. Ex vivo responses to anticancer agents in PDOs and PDO-based orthotopic xenograft tumour mouse models were compared head-to-head with response observed in their respective patients in the context of clinical trials showing that PDOs have 100% negative and 88% positive predictive power in forecasting treatment responses.

Conclusion

Our data suggest that PDOs are capable of recapitulating responses observed in the clinic, and have the potential to be implemented in precision medicine programs.

Rebecca Marlow¹, Luned Badder¹, Eleanor Knight², Stephen Pettitt², Bram Herpers³, Daniel Larcombe-Young¹, Erika Francesch-Domenech¹, Magdalena Lomzik-Borowik¹, Amelia Rushton², Daniela Novo², Leo Price³, Christopher Lord², Andrew Tutt^1,2

¹ Breast Cancer Now Unit, King’s College London, Division of Cancer Studies, London, SE1 9RT.

² Breast Cancer Now Toby Robins Research Centre, The Institute of Cancer Research, London, SW3 6JB

³ OcellO B.V, Leiden BioPartner Center, 2333CH, Leiden, The Netherlands

It is widely recognised that Patient-derived Xenografts (PDXs) are gold-standard in vivo models of breast cancer, capable of faithfully recapitulating tumour biology. High throughput drug screening methods currently rely on in vitro clonal two-dimensional (2D) cell lines that lack such pathophysiological relevance to parental tumours, limiting their ability to predict patient responses to therapies. There is a need for in vitro models to bridge the gap between such model systems to reduce and refine the number of animal experiments for pre-clinical drug testing, whilst retaining accurate readouts of tumour responses.

Three-dimensional (3D) organoid models have shown promise as disease-relevant models, capable of providing a comparable genetic and phenotypic profile to parental tumours. However, robust protocols for organoid derivation are somewhat limited for breast cancer subtypes, including Triple negative breast cancers (TNBCs), which lack targeted therapies in the clinic. Furthermore, it is yet to be ascertained whether organoid treatment responses correlate with current gold standard PDX models of breast cancer.

In partnership with OcellO (Leiden, Netherlands), we have generated a cohort of patient-derived organoids from both our defined panel of TNBC PDX models, as well as directly from patient tissue. To systematically evaluate whether our models can recapitulate drug responses of companion in vivo PDX models, we have assessed and will present standard of care drug responses of matched organoid and in vivo pairs. This includes characterising BRCA-1 mutant models based on their sensitivities to Poly(ADP-ribose) polymerase (PARP) inhibitors. Thus far, we have found differential sensitivities to PARP inhibition within our BRCA-mutant- derived organoid models, reflective of comparative in vivo studies. We aim to interrogate drug sensitivities of our models further using high throughput phenotypic screening, which will enable us to rapidly test hypothesis-driven drug/target combinations, to establish whether organoid readouts can predict responses in the clinic.