NEW YORK – A drug screening platform using rat brain tissue may potentially provide more accurate sensitivity scoring than patient-derived cancer models for patients with central nervous system tumors.
In a paper published Monday in Cell Reports Medicine, researchers from the University of North Carolina at Chapel Hill described the development of an organotypic brain slice culture (OBSC)-based platform and algorithm for drug sensitivity testing and reported how these tools performed in predicting drug response in pilot studies using tumor cell lines and patients' uncultured brain tumor samples.
Andrew Satterlee, the paper's lead author and associate director of the Brain Slice Technology Platform at UNC's Eshelman School of Pharmacy, survived a brain tumor he was diagnosed with at age 20. After doctors surgically removed the non-germinomatous germ cell tumor, there wasn't a clear recommended standard treatment for his condition. Satterlee recalled that it ultimately fell to him to decide the course of treatment he wanted because there wasn't consensus among his doctors as to what to do. Satterlee and his doctors eventually settled on carboplatin and atoposide, to which they added iphosphamide only after receiving an email from Jonathan Finlay, a neuro-oncologist at Nationwide Children's Hospital, who wrote in all-caps: "IF YOU DON'T ADD IPHOSPHAMIDE, THE TUMOR WILL GROW BACK."
Sixteen years later, Satterlee is working on taking that kind of guesswork out of treatment decisions for patients. "We have created a tool that we think can eventually be used to help doctors know what treatments they should give their patients," he said.
Drug sensitivity screening using patient-derived models of cancer such as cell lines, organoids, and xenografts have attracted interest in the field of precision medicine because they add a functionality component to conventional classifications of tumor-based histopathology and genomic profiling. However, patient-derived models have some drawbacks, including loss of the genetic and phenotypic heterogeneity of the parent tumor.
When it comes to brain tumors, the most successful models for functional screening have been generated from high-grade tumors such as IDH-wild-type glioblastomas. Satterlee and his collaborators sought to build a platform that is capable of engrafting tissue from all types of brain tumors while conserving key features of the parent tumor, such as genetic heterogeneity, to facilitate functional testing of therapeutics.
"We were looking for different models and different ways to have the benefits of an in vivo system, but the speed and efficiency of an in vitro system," said Shawn Hingtgen, a coauthor on the paper and an associate professor at UNC School of Medicine's division of pharmaco-engineering and molecular pharmaceutics.
Their efforts led them to create OBSCs using slices of brain tissue from 8-day-old rat pups, on which patients' tumor tissue is seeded. "The cells in the brain slice interact with the cells in the human tumor and vice versa," said Satterlee. "You're putting the tumor right out of the brain into something it recognizes, which is brain tissue. In this paper, we show that every tumor we have tried to put down on our brain slice … will persist and is amenable to testing."
Along with the brain slice platform for testing, Satterlee and his collaborators developed an algorithm that factors in 11 different parameters gauged by the assay, including how the tumor is growing on the brain slice, a drug's ability to kill the tumor, and off-target toxicity of the drug affecting the brain slice.
Satterlee and colleagues first tested the platform using immortalized cell tumor lines, showing that the tumor cells were able to establish and grow well on the brain slices, and that the cell lines had responses to drugs approved for central nervous system tumors. Next, they tested their platform with uncultured, cryopreserved brain tumor tissue obtained from patients who underwent surgery at UNC Chapel Hill hospitals.
The 11 patient samples included adult, pediatric, primary, and metastatic brain tumors. Every type of tumor successfully engrafted and persisted on OBSCs at four days after seeding and maintained a "significant genetic resemblance" to the parent tumor, according to the study authors. In comparison, tumor tissue expanded in vitro had a "distinctly different" profile from the parent tumor.
After investigators seeded the patient brain tumor samples onto OBSCs and screened for drug sensitivity, they found associations between drug sensitivity scoring, clinical outcomes, and genomic analysis. Drug sensitivity scores were not used to guide patient treatment, though, according to Satterlee. "Our initial results seem to correlate positively with patient outcomes, even in those cases where genetic testing did not agree with the patient outcomes."
In the case of one pediatric patient, for example, the brain tumor was classified as a diffuse, high-grade, IDH1-mutant glioblastoma with an MGMT promoter methylation profile, TERT promoter wild type, and no 1p/19q codeletion. Immunohistochemistry testing showed "patchy" BRAF V600E positivity even though sequencing did not show a BRAF V600E mutation. Genomic profiling also showed CDK4 amplification and a PIK3CA mutation.
These co-occurring biomarkers could lead doctors to guess at various precision therapy approaches. They could potentially choose an IDH inhibitor, a CDK4/6 inhibitor, a PIK3CA inhibitor, a BRAF inhibitor, or MEK inhibitor. In this case, the physician chose a standard-of-care chemotherapy approach, temozolomide, based on the patient's genomic profiling results showing MGMT methylation, which suggested the chance for a good outcome.
Meanwhile, the drug sensitivity score generated by the OBSC platform suggested that the patient would respond minimally to temozolomide. The patient, in fact, did have a minimal response to the chemo, in line with the OBSC score. According to the OBSC score, a better option would have been Novartis' Mekinist (trametinib), which is approved for BRAF V600E mutated melanoma and killed about one-third of the tumor at the highest dose.
Upon recurrence, this patient again received temozolomide and had minimal response, which the OBSC drug sensitivity screen also predicted. However, the sensitivity score for Mekinist increased by 58 percent in the recurrent tumor compared to the original tumor, suggesting that functional data from OBSC could also be used to update treatment plans after patients fail first-line therapy.
"Because we're actually testing the real tumor cells, that's a bit closer to the answer we're trying to get than the genetic profile," Satterlee said. "The genetic profile may tell you that you have this mutation, so we suggest the tumor may be sensitive to this drug. We say your tumor died from this drug on our slices."
Currently, Satterlee said the drug sensitivity score produced by the OBSC platform does not capture genetic profile information, although that could be incorporated in future versions of the algorithm. Satterlee and Hingtgen said they will further establish the predictive accuracy of the platform and feasibility for clinical use by correlating OBSC-based drug sensitivity scores with patient outcomes and genetic testing scores within a clinical trial. Hingtgen said a tumor board will help determine which drugs to test for each patient's tumor but OBSC scores will still not be used to prospectively guide therapy in the study, which will launch following institutional review board approval. Validation of the platform's ability to guide patient therapy will be established in a future prospective clinical trial.
Satterlee said he and his colleagues have "started some conversations" with potential partners who may test out the capabilities of OBSC, but declined to provide further details. If OBSC reaches the commercialization phase, it will involve using the cryopreservation protocol developed within this research to freeze tumor tissue as soon as it is removed from the patient and shipping it to a central laboratory where the assay would be conducted. The drug screen would cover a customized list of both approved and investigational therapies.
"We envision one central site where the assay takes place, and then any hospital in the world who can cryopreserve the tumor tissue can then send it to us," said Satterlee. "If we get the tumor tissue a day or two after resection, it should take a week to get those results back to the hospital."