New Patient Tumor Cell Models Rapidly Predict Cancer Therapy Outcomes

Evidence from the clinic and research labs in the last couple of decades shows that there is no single cure for cancer. Within a group of patients, individuals with what look to be identical tumor types react to the same treatment very differently. This often results in a waste of time and quality of life for those that don’t respond, and it has impeded cancer treatment development for years. As a result, precision oncology, the molecular profiling of tumors to identify patient-specific treatments, has become a necessary tool in the battle against cancer.

Recently, researchers developed in vitro patient-derived cell models to guide personalized care in cancer.¹ “The challenge is to make a platform technology that can truly be used in the clinic, and to be very reproducible, fast, and scalable,” said Xiling Shen, a professor in biomedical engineering at Duke University. While training to be an electrical engineer, Shen met Hans Clevers, a former group leader at the Hubrecht Institute and inventor of early patient 3-D cell models. Clevers’ work made Shen realize the potential of manipulating human tissues outside the body and inspired him to shift his focus to study patient tumor cell models. 

In a recent study published in Cell Stem Cell, Shen and his group in collaboration with David Hsu from Duke University and Hans Clevers designed a benchtop device that generates patient-derived micro-organospheres (MOSs) using a high-throughput automated microfluidics droplet platform.² When using this device, researchers add cell suspensions taken from patient tumor biopsies to a 3-D-extracellular matrix before mixing with oil to form microfluidic-based droplet MOSs. Using as few as 15,000 cells, the device generates thousands of MOSs that replicate the patient tumor and are ready for personalized drug screening.

For Shen and his group, generating a device that could be effectively used in a clinical setting came with challenges. After exhaustively trying traditional cell culture protocols and organ-on-chip techniques, the researchers found oil droplets to be the most effective method of deriving the MOSs. The oil, however, eventually suffocated the cells, and traditional removal methods were toxic. To solve this problem, one of the scientists in Shen’s group was inspired by the Deepwater Horizon oil spill cleanup efforts, where clean-up crews used semi-permeable membrane devices to remove oil near sensitive marine habitats. “So, we have these droplets rolling on a membrane where the oil gets sucked out, and that enabled effective [MOS] formation,” Shen said.

To validate the platform’s clinical use, Shen and his group generated MOSs from colorectal cancer biopsies and performed drug screens using FDA-approved therapies. They compared MOS drug sensitivities to real patient responses and found that colorectal cancer patients with MOSs sensitive to the drug Oxaliplatin in vitro responded well to the treatment in the clinic. Patients with MOSs resistant to the drug did not respond to the treatment, showing that the MOSs can reliably test for drug sensitivities and predict patient responses. The entire process took an average of 10 days. With clinical treatment decisions generally made within 14 days of diagnosis, this platform allows clinicians to begin patient-specific therapies much sooner compared to previous methods.

“The question is, can our technology actually recapitulate patient response to immunotherapy? That's the holy grail because it's always the hardest.” 
-Xiling Shen, Duke University

The researchers also wanted to be sure that the MOSs reflected in vivo tumor microenvironments. “The question is, can our technology actually recapitulate patient response to immunotherapy? That's the holy grail because it's always the hardest,” Shen said. By performing single cell RNA sequencing comparing MOSs to the original patient tumor, they found that the tumor spheres maintained their molecular identities and also harbored several major cell types, including immune cells. For Shen, this was a major “aha” moment, and it allowed him to demonstrate that MOSs are vulnerable to T cell therapies and can serve as models for immuno-oncology treatment design.

“I think there's a lot of potential and I think we need more of these trials, where we really try to match a technology with the outcome within the context of a defined intervention,” said Alice Soragni, a professor at the University of California, Los Angeles, who was not involved in the study. “It's going to be very interesting to see and really understand how good this technology is [in the clinic].”

A multi-site clinical trial is already underway for the MOS-generating device and has the potential to affect patient outcomes and allow researchers and companies to rethink how clinical trials are performed. “We can use this technology as a patient avatar to help select the right patients for the drugs and, on both sides, avoid wasting precious patient time and quality of life,” Shen said.