2017 - Characterizing Molecular Radiosensitivity In Pituitary Neuroendocrine Tumors (PitNETs) Using Next-Generation Sequencing
Presenter(s)
A. Bhandarkar1, L. Yefet2, M. Bettencourt2, A. Lone2, S. Bhandarkar3, P. D. Brown1, D. Donegan4, N. N. Laack II1, E. J. Lehrer1, V. Patil5, and G. Zadeh2; 1Department of Radiation Oncology, Mayo Clinic, Rochester, MN, 2Department of Neurologic Surgery, Mayo Clinic, Rochester, MN, 3Johns Hopkins School of Medicine, Baltimore, MD, 4Department of Endocrinology, Mayo Clinic, Rochester, MN, 5University Health Network and University of Toronto, Toronto, ON, Canada
Purpose/Objective(s): Radiation is an adjuvant therapy for pituitary neuroendocrine tumors (PitNET) with established efficacy in achieving local tumor control after surgery. However, radiation-induced hypopituitarism is a common side effect, the risk of which increases with higher radiation doses. In addition, biochemical control for secreting tumors is suboptimal. In this study, we used RNA-sequencing (RNAseq) to characterize the spectrum of radiosensitivity in resected PitNET samples in order to identify patients who may benefit from dose intensification or reduction to improve biochemical control or reduce risks of endocrinopathies.
Materials/Methods: Publicly available RNAseq data from PitNET samples was aggregated from the NIH Gene Expression Omnibus (GEO). The radiosensitivity index(RSI) was calculated using a clinically validated ranked linear regression model based on the weighted expression of 10 radiosensitivity genes with higher RSI values corresponding to radioresistance. Radiophenotype was determined using a previously validated RSI cutoff of < 0.375 to identify radiosensitive tumors. Clinical annotations for primary cell type and functioning status were manually derived from source studies.
Results: A total of 229 PitNET samples were compiled from nine different source studies of which 91 were corticotrophs (56.0% of which were functioning), 98 were somatotrophs (all of which were functioning),17 were gonadotrophs (none of which were functioning), 4 were lactotrophs (all of which were functioning), and 19 were null cell (none of which were functioning). Corticotrophs had the greatest proportion of predicted radiosensitive tumors (46.2% radiosensitive, median RSI:0.41,IQR:0.28-0.52), followed by somatotrophs (41.8% radiosensitive, median RSI:0.43, IQR:0.30-0.58). None of the included gonadotroph (median RSI:0.55,IQR:0.45-0.62), null cell (median RSI:0.58,IQR:0.50-0.63) or lactotroph samples (median RSI:0.66,IQR:0.63-0.68) were predicted to be radiosensitive. In a multivariable logistic regression model predicting radiosensitivity, functioning status (p = 0.65), gonadotroph status (p = 0.93), lactotroph status (p = 0.997), null cell status (p = 0.99), and somatotroph status (p = 0.99) were not predictive of radiosensitive status.
Conclusion: In this study, we characterized differences in molecular radiosensitivity in a large cohort of PitNETs and found radiosensitive status did not uniquely correspond to one cell type. In the future, these differences in PitNET radiosensitivity identified within RNAseq data from resected tissue may be able to guide studies evaluating dose-adapted treatment to optimize biochemical control and reduce risks of treatment.