Glioblastoma represents about 15% of brain tumors and resembles cells called astrocytes. Under physiologic conditions, these cells participate in the maintenance of brain homeostasis through different functions such as metabolic support or nervous system repair. In 2016, the World Health Organization defined two major molecular subtypes of glioblastoma: Isocitrate Dehydrogenase (IDH)-wildtype and IDH-mutant. Whereas IDH-wildtype patients have a 15-month median survival, IDH-mutant patients have a better prognosis, with a median survival close to 30 months. Other studies, including one led by Dr. Eric Holland (Human Biology Division), demonstrated that chromosome and gene level copy number variations could even more precisely predict glioma patient survival.
In a recent study published in the Neuro-Oncology journal, Dr. Patrick Cimino and colleagues in the Holland lab, in collaboration with the University of Zurich and Columbia University, “asked whether or not these molecular subtypes are evenly distributed across cohorts [from clinical trials] that select for patients with better outcome”. Clinical trials are inherently biased to select patients with the best performance status to consistently assess the effect of new drugs on patient daily life.
To do so, the team selected two prospective clinical trials. As Dr. Cimino explains, these studies selected “for patients that either do well enough to enter a clinical trial or who are deemed well enough to undergo a surgical re-resection at recurrence.” They compared copy number data from these cohorts to two unbiased global datasets, The Cancer Genome Atlas (TCGA) and the German Glioma Network (GGN) and classified patients according to their previous study. Patients with chromosome 1 gain and CDK4/MDM2 co-amplification belong to group W1 with the worst prognosis, whereas patients with chromosome 19 amplification and CDK4/MDM2 co-amplification have the best prognosis (group W3). Group W2 is composed of patients with chromosome 19 gain but no co-amplification of CDK4/MDM2 and have an intermediate prognosis.
Surprisingly, whereas the overall population of glioblastoma patients predominantly belongs to the W2 group, cohorts from the two selected prospective studies were skewed towards the W3 group (with better survival). In addition, the authors performed a multidimensional scaling analysis (MDS) combining copy number with single nucleotide variations to identify additional subgroups of patients with different DNA alterations. By comparison with TCGA datasets, they demonstrated that half (51%) of the glioblastoma patients with a particular DNA alteration profile were not represented in one of the prospective studies (only one of the two had exome sequencing data). In other words, these clinical trials are missing the patient tumors with the most common DNA alteration profile in the global population of glioblastoma patients, which correspond to the most aggressive subtype within the IDH-WT subgroup.
As Dr. Cimino highlights: “These findings are quite intriguing and have several implications for the design and analysis of glioblastoma clinical trials. Ideally, patients would undergo molecular analysis of their tumors up front in order to inform the clinical trials and make sure that copy number subtypes are evenly distributed across trial arms. It is possible that if molecular subtyping is not addressed initially in glioblastoma clinical trials, then tested therapies that may appear effective overall in early phase II trials, may fail to translate into phase III trial results due to unaddressed shifting molecular distributions not reflective of general population. Additionally, for those patients in the poorest survival molecular group, half do not survive through standard radiation therapy, so perhaps identifying these patients up front can lead to an alternative form of clinical trial or treatment strategy.”
As selection for better performance status is common to most of clinical trials, these findings will probably raise similar questions for many diseases.
This work was supported by the National Institutes of Health.
Fred Hutch/UW Cancer Consortium faculty members Drs Eric Holland, Hamid Bolouri, and Patrick Cimino contributed to this research.
Cimino PJ, McFerrin L, Wirsching H-G, Arora S, Bolouri H, Rabadan R, Weller M, and Holland EC. 2018. Copy number profiling across glioblastoma populations has implications for clinical trial design. Neuro-Oncology. 20(10), 1368-1373. doi:10.1093/neuonc/noy108.
Basic Sciences Division
Human Biology Division
Maggie Burhans, Ph.D.
Public Health Sciences Division
Vaccine and Infectious Disease Division
Clinical Research Division
Julian Simon, Ph.D.
Clinical Research Division
and Human Biology Division
Arnold Digital Library