Mark R. Gilbert, MD
The deployment of precision medicine to develop safe and effective therapies to treat malignant brain tumors is an endeavor eagerly embraced by researchers and patients alike, but the effort poses both opportunities and challenges.
“It’s quite a daunting task, but one I think we’re up for,” said Mark Gilbert, MD, in his keynote address at the 20th Annual Scientific Meeting of the Society of Neuro-Oncology (SNO).
Gilbert, a senior investigator and chief of the neuro-oncology branch at the NIH, said that the ultimate goal of these efforts is that “every patient will have an individualized treatment plan based on their unique tumor, genomic landscape, and clinical characteristics.”
However, there are hurdles to overcome. Chief among them is the disease’s heterogeneity. The Cancer Genome Atlas (TCGA) analyses generated four subtypes of glioblastoma multiforme (GBM) and three major groups of signaling pathway abnormalities without much overlap, Gilbert explained.
To address this challenge, Gilbert called for more studies to characterize clinically well-articulated tumor samples on a larger scale, now that TCGA and related efforts have provided proof of principle. Noting the example of HER2 in breast cancer, he said that validation of prognostic characteristics of molecular subgroups is essential. He added that overall, more robust sharing of institutional and clinical trial data, including integration of electronic medical records for population-based analyses, would facilitate the effort.
Citing a recent and “very interesting study,” Gilbert said that it appears although some GBM recurrences are very similar, others share very few molecular similarities, suggesting that these cells may have diverted early from the primary tumor.1
Thus, to better understand the molecular changes that occur with GBM recurrence and progression, Gilbert would like to see more resources devoted to the effort, as well as mandatory acquisition of contemporary tumor tissue for entry onto a hypothesis-based clinical trial.
Another challenge that researchers face in this setting is clinical and genomic patient heterogeneity. He mentioned a number of germline polymorphisms that predict increased risk of treatment toxicity, among them polymorphisms of: the DYPD gene (increases risk for 5-FU toxicity); UGT1A1 (irinotecan toxicity); and CYP39A1 (docetaxel toxicity). Clinical factors that may affect treatment efficacy, pharmacokinetics, and toxicity risk include hepatic enzyme–altering anticonvulsants, renal and/or hepatic dysfunction, bovine serum albumin/lipid burden, and preexisting conditions such as hypertension and coagulopathies.
Gilbert said the patient’s medical condition and medications must be considered upfront in clinical trials. He would like to see comprehensive recording and monitoring of concurrent medications during both early-phase and subsequent efficacy testing; integration of toxicity data in all studies evaluating agents under investigation; and the creation of paradigms to personalize dosing based on a patient’s clinical and his or her tumor’s genomic profile.
Finding the Right Target
Gilbert noted that studies have demonstrated little efficacy with single-agent targeted therapies in neuro-oncology. “We have many agents that have been tested, but response rates have been low, and then we have the additional challenge of limited drug delivery due to the blood–brain barrier … it’s not just getting the drug in, it’s getting it there.”
He suggested drawing lessons from successful treatment targets such as BCR-ABL in chronic myelogenous leukemia, HER2 in breast cancer, and EGFR in lung cancer. “There’s clearly evidence that the targeted approach can work.”
In glioblastoma and lower-grade gliomas, “we have targets,” Gilbert said. These include mutations or translocations involving BRAFV6005, FGFR-TACC
mutations, and EGFRviii
. Investigators can also look to targets identified for other brain tumors, such as medulloblastoma (SHH
), ependymoma (CIMP, RELA
fusion), and meningioma (SMO, Akt
Clinical trial design in neuro-oncology is also an issue: “The typical phase I trial, leading to phase II, then possibly phase III design is inefficient,” Gilbert explained. “Given the number of clinical targets and available treatments, a more rapid screening process is needed,” along with enrichment of the patient population to allow small, efficient trials to discern early efficacy signals.