Early Prediction of Brain Metastases Response to Radiation Therapy by Combination of Changes in Tumor Vascular and Cellularity Properties
R Farjam *, C Tsien, F Feng, D Gomez-Hassan, J Hayman, T Lawrence, Y Cao, The University of Michigan, Ann Arbor, MISU-F-500-2 Sunday 4:00PM - 6:00PM Room: 500 Ballroom
Purpose: To investigate the impact of combining image-based measurements related to alterations of vascular and cellularity properties of brain metastases for early prediction of tumor response to radiation therapy (RT).
Methods: 24 patients undergoing treatment for metastases had perfusion (DCE) and diffusion MRI scans. Tumor subvolumes with high regional cerebral blood volume (rCBV) and high Gd-DTAP transfer constant (Ktrans) were extracted. The apparent diffusion coefficient (ADC) histograms of the tumor volume were categorized into low, normal and high ADC subvolumes. Changes in different tumor subvolumes and their combination from pre-RT to week 2 after start of radiation were evaluated in differentiation of responsive, stable and progressive tumors for patients treated by either whole brain radiation therapy (WBRT, 28 lesions) alone or combined with Bortezomib as a radiation sensitizer (39 lesions). Receiver Operating Characteristic (ROC) analysis compared a combined perfusion/diffusion prediction model with changes in gross tumor volume (GTV) within the same time interval.
Results: For lesions treated with WBRT alone, a decrease in both high cellularity and edema subvolumes were associated with response. A decrease in the tumor subvolumes with high cellularity and vascularity was associated with response in lesions treated with RT combined with Bortezomib. In ROC analysis, areas under curve (AUCs) of 0.96 (WBRT alone) and 0.96 (WBRT + Bortezomib) were seen in prediction of non-responsive lesions when changes in different subvolumes were combined. Also, GTV changes could predict the non-responsive lesions treated with WBRT alone (AUC = 0.91) but failed to predict the response in lesions treated with WBRT combined with Bortezomib (AUC = 0.57).
Conclusion: Combining the changes in tumor cellularity and vascularity could be used for early prediction of brain metastases response to RT and perform better than the GTV changes, suggesting that physiological changes could occur earlier than the morphological changes. (Support: NIH RO1NS064973)
Funding Support, Disclosures, and Conflict of Interest: NIH grant RO1 NS064973