Organoids Fuel Development of Gene Expression Approach for Tailoring Pancreatic Cancer Treatment

NEW YORK – Gene expression signatures uncovered using tumor-derived organoids could help guide pancreatic cancer patients to therapies to which they are more likely to respond.

Standard-of-care therapies for pancreatic cancer include gemcitabine/nab-paclitaxel or combination therapy Folfirinox, but what treatment patients receive often depends on their overall health and whether they can endure aggressive therapies. Pancreatic ductal adenocarcinoma has a five-year survival rate of less than 8 percent, underscoring the need to connect patients to treatments that might be best suited for their own disease.

Organoids derived from patients' tumors might be able to pinpoint those treatment approaches that patients are more likely to respond to as they may better reflect the complexities of individual patients' tumors.

"Nobody has really proven which patients should receive which chemotherapy," said Dennis Plenker, a postdoc in David Tuveson's lab at Cold Spring Harbor Laboratory. "If there is a way to have a phenotypic assay or a functional assay — which organoids would be — this would help stratify patients and give them the right chemotherapy upfront."

He and his colleagues have been studying organoids derived from pancreatic cancer patients, and through that analysis they homed in on gene expression signatures that indicate whether patients' tumors are sensitive to particular treatments. These signatures, he said, could be implemented as an array or PCR-based test, bypassing the need to generate organoids for each and every patient, a prospect that may be difficult to do in a clinically relevant timeframe.

"If we can just use a genetic test, for example, on patients' biopsies to stratify them [to therapies] that would help a lot," Plenker added.

Organoids can be generated from tumors that have been surgically resected, but also from biopsies, including from fine-needle aspirates. These biopsies can then be grown so they contain cellular features and architecture of the original tissue samples. In this way, they may better mimic what occurs in the body than two-dimensional cultures. 

Tumor-derived organoids also harbor the genomic alterations present in the original tumors and enable more-in depth sequencing analysis, Plenker said, since researchers can isolate a purer population of tumor cells from them than they can from biopsies. Since only around 20 percent of biopsy tissue is pure from the tumor, he added that organoids can serve as a sort of living biobank.

Researchers like Plenker and his colleagues have been using organoids to gauge which treatment approaches might be the most suitable for particular patients for several years. As they reported in Cancer Discovery in 2018, Plenker and his colleagues generated 114 organoids from 101 patients with pancreatic cancer. They tested how these organoids responded to a number of various standard-of-care treatments and compared that to how the patients in the cohort fared. 

They conducted pharmacotyping for five common pancreatic cancer therapies — gemcitabine, paclitaxel, irinotecan, 5-fluorouracil, and oxaliplatin — on 66 patient-derived organoids. The patient-derived organoids varied in their response to these treatments but corresponded with how the patients responded to treatment in the clinic.

"The organoid seems to be really predictive in that space," Plenker said.

As Plenker and his colleagues also sequenced the transcriptomes of these organoids, they began to uncover genes whose expression corresponded with how responsive the organoids were to particular treatments. 

"As you get higher resolution, you can treat these organoids and put them in categories of resistant, sensitive, or [intermediate]," he said. "And then we can use them to predict sensitivity in patients, as well as a tool for biomarkers."

They applied these gene expression signatures to real-world data generated by a Canadian team they collaborated with to find that these organoid-derived transcriptomic signatures also reflected patients' treatment response. He and his colleagues are now further investigating how these gene signatures could be used to guide treatment.

Going forward, rather than generating organoids for each patient and testing various treatment approaches on them, Plenker said patients' biopsies could instead be directly tested for these treatment response-linked gene signatures using, for instance, a microarray- or qPCR-based assay. This is something Plenker said the Tuveson lab plans to pursue following additional validation.

"Obviously the goal is to have a test that is as fast as possible. If we had a three-day test that would be great, but that would have to be something that doesn't involve growing organoids from a patient," he added. Growing organoids is tricky, he noted, estimating that only about half of researchers who try to establish organoids are successful.

He added that this approach could be adapted for other targeted therapies beyond chemotherapy. Ninety-six percent of the pancreatic cancer patient-derived organoids they generated harbored KRAS mutations, and other patient-derived organoids had mutations in other known cancer driver genes, such as HER2, which can be targeted by a number of drugs. In this initial case, Plenker noted that the organoids with those HER2 alterations were indeed most sensitive toward a HER2 inhibitor. 

"You could give a patient obviously something like chemotherapy," he said. "But if you know this is the tailored drug for that specific tumor for the patient, there is a big advantage for these patients to give them the right drug."