Researchers looking to cure cancer are constantly searching for disease models that closely resemble human physiology for testing cancer-fighting drugs. Artificial Intelligence (AI) can help researchers sift through large amounts of available data, saving time and costs, to source the most appropriate disease model for oncology drug development.
In early drug development stages, the choice of the disease model used for drug discovery and drug testing can make or break a study. Drugs are only approved for studies in humans based on satisfactory safety and efficacy outcomes from experiments conducted using specific models. This ultimately leads to market approvals for commercializing the drug and helps deliver an optimized product for patients in need.
Using the ‘wrong’ model can be costly for a drug discovery project for multiple reasons. First, disease models themselves can be expensive to source, leaving scientists with less margin for error during selection because study budgets are often limited at this stage.
Second, characteristics of the chosen model, including specific mutations or biological pathways, can heavily influence the efficacy of a drug. Even promising drug candidates can provide conflicting results across different models, which has resulted in some drugs failing human clinical trials despite passing preclinical testing.
Therefore, using a relevant disease model can substantially improve the credibility of the research as well as its findings.
A world of options: from 2D in vivo to 3D in vitro models
Today, there are many different types of preclinical models that researchers can leverage, each coming with its own experimental, monetary, and logistical advantages.
Traditionally, immortalized cell lines have been the most commonly used tool for testing potential drug targets in oncology research. These involve human cells that have been manipulated to multiply indefinitely in a laboratory setting.
Based on the same principle of testing drugs on cultured cells, scientists also use primary patient-derived cell lines. Taken from patients, these cells offer a better 2D representation of real patient physiology.
There are also 3D models, such as spheroids or organoids, which are a step closer to patient physiology than individual cells. These models entail a simplified version of an organ or tumor produced in vitro and aim to simulate the immediate cellular surroundings of the tumor cells in a patient.
An improvement on cultured cell models are in vivo animal models, which capture the tumor cell environment at an organism level. A prominent example are mouse models, which have been an integral part of cancer drug discovery for decades. In fact, syngeneic mice are immunologically compatible and have been selectively bred for drug testing.
“Syngeneic mouse models and patient-derived xenografts (PDX) are popular in vivo models in oncology,” said Javier Pineda, Data Scientist at Scientist.com, an e-commerce platform that connects buyers and sellers of custom research services.
“PDX models involve the implantation of patient cancer tissue into an immunocompromised mouse, such that the tissue is not rejected by the mouse’s immune system.”
PDX models are representative of the shift towards personalized medicine. They enable the selection of the most appropriate therapeutic intervention by truly capturing the 3D environment (albeit without the functioning immune system) within the patient.
More recently, as an improvement on PDX, humanized mouse models with reconstituted human immune systems have also been developed.
The hurdles to choosing the right disease model
With a large number of preclinical disease models available, it can be quite tricky to choose the most appropriate one for a cancer study.
Even knowing how to source a model can be a challenge to begin with, said Pineda. For a single study, researchers regularly end up contacting multiple suppliers to inquire about model availability. This often results in a trial-and-error process that is inevitably time-consuming.
Once a supplier has been identified, getting access to pertinent data that will aid decision-making is another hurdle. Supplier websites and catalogs commonly only provide annotation-level data.
This refers to critical or explanatory notes provided on the disease models and includes information such as tumor type and subtype, type of mouse strain, geographical location of the disease model, and basic patient information.
“It is imperative to study factors like genetic mutations, expression of target genes, or drug sensitivity before sourcing a disease model. Annotation-level data alone is generally insufficient. Even in cases where molecular data is available, due to its proprietary nature, it is often inconsistent across suppliers and hard to compare,” explained Pineda.
This limited data prevents researchers from conducting necessary quality checks as well as making quantitative comparisons between models provided by different suppliers. As a consequence, the choice of model is severely restricted, which can influence the outcome of the experiments conducted.
Lastly, negotiating a project proposal, finalizing a contract, and taking care of regulatory approvals can also be challenging. For some companies, this process takes several weeks to months, causing delays in project timelines.
Disease model selection made easy: AI-driven decision-making
Based on feedback from users on their platform, the Scientist.com team realized the need for a centralized, data-driven approach to disease model sourcing. This triggered the blueprint for the Disease Model Finder, an AI-driven tool that enables researchers to make quantitative model comparisons across suppliers by leveraging machine learning algorithms.
“Our existing relationship with many disease model suppliers has enabled us to aggregate molecular data across providers while ensuring data confidentiality and security,” said Pineda.
“The finder uses a variety of algorithms to process, aggregate, and visualize the data from large, molecular sequencing-based datasets. To efficiently find and source appropriate disease models, the tool enables researchers to filter by mutations, gene expression, cancer types, and more.”
Often, studies require a panel of disease models to test the implications of a target drug across multiple cancer types and subsets. To aid the selection of multiple models, the Disease Model Finder has a comparison function that provides insights into models that may be biologically distinct or similar as per the requirements of the study.
The team behind the Disease Model Finder has aimed to make the computationally driven comparisons accessible to those without a bioinformatics background.
“The goal here is to make the analysis user-friendly so that researchers can improve their search while saving time, without having to rely on external expertise,” explained Pineda.
In the case that a researcher cannot find a model that directly pertains to their study, the Disease Model Finder allows them to create a custom request. This is routed to the platform’s Research Concierge team, who can then identify suppliers that are most likely to have the model in question.
The Disease Model Finder also works in concert with Scientist.com’s COMPLi® functionality, which covers compliance processes during the sourcing of disease models and other regulated services.
“This setup guarantees supplier compliance, allows for automated purchasing, and reduces processing time, enabling researchers to do everything using one platform,” said Pineda.
The evolving role of AI in oncology research
The Disease Model Finder currently hosts an array of PDX models but will be expanded to include more oncology model types in the coming months. Furthermore, the team is also planning to add more bioinformatic tools, such as a recommendation algorithm that uses correlation analysis to suggest disease models of interest.
With the increased use of AI in disease research, including in oncology, Pineda said that the world can expect interesting applications such as AI-based drug-response predictions to become a reality very soon.
“Predicting how a patient will respond to a therapy is a critical research focus in the oncology field. This requires analyzing extensive datasets on drug response and other molecular information,” he continued.
“Incorporating drug response predictions into the Disease Model Finder would significantly improve cancer model sourcing. As a constantly evolving platform, it is a feature that we hope to provide for our researchers in the future,” concluded Pineda.
To learn more about the AI-driven Disease Model Finder and how it can enable your drug discovery research, visit the company’s website. If you would like to partner with Scientist.com and provide scientists worldwide with access to your disease models, get in touch via diseasemodels@scientist.com.
Images courtesy of Scientist.com via metamorworks/iStock