Organoids for Colorectal Cancer Drug Development using a KRASmut Screen

Drug development is notoriously inefficient, slow, and expensive. In oncology, only 5% of new compounds at the drug discovery phase successfully make it to clinical trials. This high attrition is largely due to over-reliance on 2D cell cultures or animal models which fail to recapitulate the patient tumor biology and physiology. Adoption of more advanced, clinically relevant preclinical models at earlier stages of drug development can drive the identification and prioritization of agents that have better chances of success in clinical trials.

Our patient-derived organoids faithfully recapitulate the genetic and phenotypic features of original tumors. Due to their predictive power and amenability to large-scale drug screens, in this study they were successfully used to develop a clinical candidate that inhibits tumor growth and metastasis in colorectal cancer.

Looking to use HUB Organoids^® for your drug development program? Speak to an expert now!

Section Overview

• Colorectal Cancer Drug Development Challenges
• What are Patient-Derived Organoids?
• Organoid Drug Screening Methods
• Results
• Conclusion

Colorectal Cancer Drug Development Challenges

Colorectal cancer (CRC) is the third most common cancer in the world, with an estimate of 1.9 million new cases annually and 850,000 deaths. Of new colorectal cancer diagnoses, 20% of patients have metastatic disease at presentation and another 25% who present with localized disease develop metastases later.

Metastatic CRC (mCRC) remains largely incurable and constitutes a major unmet need in cancer therapy. The primary treatment for unresectable mCRC is systemic cytotoxic chemotherapy, but clinical trials have demonstrated that tailoring treatment to the molecular and pathologic features of the tumor improves overall survival. About half of mCRC patients with unresectable tumors present activation in the EGFR signaling pathways and can be treated with cetuximab, however, only a subset of patients derives meaningful therapeutic benefits. For the remaining 35% to 40% of patients with KRAS or NRAS sequence variations, effective targeted therapies are not yet available.

WNT signaling is known to play a key role in stem cell homeostasis in the normal intestine. The dysregulation of WNT signaling in combination with aberrant activation of receptor tyrosine kinases such as EGFR is believed to drive cancer stem cell expansion and lead to tumor growth.

Advanced preclinical models, directly derived from CRC patients that preserve patient tumor genetic make-up, are sorely needed to advance novel therapies targeting EGFR and WNT in CRC. Immortalized cell lines conventionally used in preclinical testing lack physiological relevance and genetic congruence to the original patient tumor and fail to recapitulate the complex mitogenic signals occurring in vivo.

On the other hand, more physiologically relevant patient-derived models, such as primary cells or patient-derived xenografts (PDX) lack scalability and are therefore not amenable to high-throughput drug screening. The tox profile of these models has not always correlated with humans in clinical trials, they can be resource intensive to establish and maintain, and often display poor establishment rates from patient material.

What are Patient-Derived Organoids?

Patient-derived organoids (PDOs) are mini-organs derived from patient tissue resections or biopsies. These models are more clinically relevant than standard cell lines, with the ability to faithfully recapitulate the phenotypic and genotypic features of the patient tumor, including key driver mutations. PDOs are faster to develop than other in vivo models, can be more easily expanded than other patient-derived models with a low batch-to-batch variation even after long-term culturing and cryopreservation, and can be developed from both healthy and diseased tissue, facilitating the identification and prioritization of agents with a low toxicity and high therapeutic index.

Here we demonstrate the feasibility of replacing traditional cell models with organoids for target and lead validation, reducing animal usage in translational studies that can instead be performed on a predictive, clinically relevant, and easily expandable 3D in vitro system. This case study also demonstrates a significantly shortened timeline to clinical trials resulting in cost-saving and patient benefit.

Organoid Drug Screening Methods

The SupressSTEM consortium¹ – which includes HUB Organoids^® – was formed to develop therapeutics that address important resistance mechanisms exploited by colorectal cancer cells to overcome conventional cancer treatments. To increase the likelihood of translating the preclinical findings to the clinic, drug candidate screening and selection was performed directly on organoids using novel imaging readouts.

A library of bispecific antibodies (bAbs) from Merus (now part of Genmab) that selectively block EGFR in cancer stem cells (CSCs) was developed and screened on a biobank of normal and CRC organoids to identify the best performing candidate. This lead agent was further validated, optimized, and evaluated for efficacy and off-target toxicities using paired normal colon organoids.

The screening clearly identified a clinical candidate which was evaluated on patients in clinical trials and was shown to have positive results in five head and neck cancer patients². This candidate has received a breakthrough therapy designation by the US FDA, is approved for head and neck cancer, and is now being evaluated in colorectal cancer patients.

Results

The screening clearly identified a clinical candidate which was evaluated on patients in clinical trials and was shown to have positive results in five head and neck cancer patients². This candidate has received a breakthrough therapy designation by the US FDA, is approved for head and neck cancer, and is now being evaluated in colorectal cancer patients.

Tumor Heterogeneity and Organoids

More traditional preclinical tumor models, including 2D cell lines or animal models, can fail to capture the diversity of human malignancies, leading to poor prognosis of drug response and treatment resistance. For this application, the lead compound would not have been identified using traditional 2D cell lines as the cell lines do not preserve the stem cell markers. Accounting for tumor heterogeneity early in the compound discovery stage can help to develop therapeutics that have better chances of performing in the clinic.

Patient-derived HUB Organoids^® display a spectrum of genetic alterations that largely overlap with the ones reported for CRC by The Cancer Genome Atlas (TCGA)³. These genetic alterations also include the different components of the WNT, RTK/RAS, TP53, and transforming growth factor-β pathways, which are known to be the four main drivers of colorectal cancer (Figure 1).

This comparative display of clinically relevant genetic alterations in the CRC HUB Organoids^® biobank reinforces them as excellent models to investigate therapeutic efficacy before patient application.

Large Drug Candidate Panel Testing

In this study, our CRC HUB Organoids^® biobank was leveraged to screen a large library (>500) of bispecific antibodies targeting a CSCs surface marker (LGR4, LGR5, ZNRF3 RNF43) and a growth factor receptor (EGFR or HER3) involved in growth and proliferation. From the first screening, performed in 3 CRC models, 28 bispecific antibodies were selected based on their ability to inhibit tumor organoid growth using high content-based imaging.

The efficacy of these 28 lead candidates was then re-evaluated in a CRC biobank composed of 25 models that captured most of the known genetic alterations and subtypes of human CRC. From this secondary screening, the most potent and most broadly reactive bispecific antibody with activity on a subset of KRAS mutant organoids was selected for further optimization and development (Figure 2).

Our portfolio also includes a new pre-set screen for metastatic colorectal cancer drug screening with patient-derived organoids to help researchers make mCRC drug decisions and evaluate novel compound efficacy.

Organoids as a Platform for Lead Optimization

Our CRC HUB Organoids^® biobank was used to compare lead candidate efficacy in inhibiting CRC organoid growth against current standard of care therapeutic, cetuximab. The biobank includes metastatic and non-metastatic models as well as models with activating KRAS mutation.

The experimental candidate outperformed cetuximab across the CRC organoid panel, showing efficacy on a subset of models displaying KRAS activating mutations where cetuximab treatment has failed to show any benefit (Figure 3).

HUB Organoids^® are amenable to genetic manipulations conventionally applied to standard cell lines. Knockdown of LGR5 (shRNA) in PDOs resulted in insensitivity to the bispecific antibody (Figure 4A). Further clarifying the mechanism of action, direct targeting of LGR5 or EGFR independently by a different antibody achieved little to no response in CRC organoids. Similarly, targeting both pathways at the same time by two different antibodies resulted in moderate tumor growth inhibition compared to the experimental candidate (Figure 4B), thus indicating simultaneous engagement within the same cell of LGR5 and EGFR is required to effectively induce tumor growth inhibition. High-content imaging of normal vs colorectal cancer organoids post-treatment allows the visualization of the antibody mechanism of action and shows delocalization of EGFR receptor from the surface of cancer cells but not normal colorectal organoids. This suggests successful receptor internalization by the antibody treatment which correlates with tumor shrinkage (Figure 4C).

Organoids as a Model for Drug Toxicity Testing

HUB Organoid Technology enables the development of matched tumor/normal organoids, thus allowing the assessment of off-target toxicities in the context of a patient-relevant platform. When tested on a set of matched normal/CRC organoids, the test candidate outperformed the standard of care by showing better efficacy on tumor organoids and reduced toxicity on normal organoids (Figure 5).

Conclusion

The case study demonstrates the application of HUB Organoids^® at multiple stages of the preclinical drug development pipeline, from target validation to lead optimization and mechanism of action investigation. The ability to better predict patient drug responses helps reveal a revolutionary approach to drug development, resulting in the potential for lower drug attrition in the clinic. By virtue of its genetic stability and ability to preserve key patient features like the expression of cancer biomarkers, HUB Organoids^® can be easily expanded for large-scale drug screenings and catalyze the discovery of drugs in a short timeframe.

Additionally, thanks to better recapitulating the heterogeneity, organization, and vulnerabilities of the tumor of origin, PDOs enable discoveries that cell lines selected to expand in standard culture conditions may fail to reveal. Here we present a path for therapeutic intervention for a subset of CRCs patients carrying KRAS-activating mutations who still display EGF dependency. This feature has not been shown before using more traditional models and marks a critical step forward in oncology and precision medicine.

Acknowledgements

The work described was the result of work by an international consortium, SuppresSTEM, coordinated by Merus NV. (The Netherlands) and including as participants the Institute for Research in Biomedicine (Spain), the Welcome Trust Sanger Institute (United Kingdom), the Hubrecht Institute (The Netherlands), HUB Organoids (The Netherlands) and OcellO B.V. (now Crown Bioscience).

HUB and consortium partners led the design of the experiments that were executed using the high content screening platform from Crown Bioscience on HUB Organoids. These studies enabled the discrimination of antibodies that were active against a broad range of mutational profiles and tumor subtypes, with hit validation and mechanism of action studies that enabled the selection of optimally performing antibodies.