Redmond, Kristin Janson

We introduce tumor connectomics, a novel MRI-based complex graph theory framework that describes the intricate network of relationships within the tumor and surrounding tissue, and combine this with multiparametric radiomics (mpRad) in a machine-learning approach to distinguish radiation necrosis (RN) from true progression (TP). Pathologically confirmed cases of RN vs. TP in brain metastases treated with SRS were included from a single institution. The region of interest was manually segmented as the single largest diameter of the T1 post-contrast (T1C) lesion plus the corresponding area of T2 FLAIR hyperintensity. There were 40 mpRad features and 6 connectomics features extracted, as well as 5 clinical and treatment factors. We developed an Integrated Radiomics Informatics System (IRIS) based on an Isomap support vector machine (IsoSVM) model to distinguish TP from RN using leave-one-out cross-validation. Class imbalance was resolved with differential misclassification weighting during model training using the IRIS. In total, 135 lesions in 110 patients were analyzed, including 43 cases (31.9%) of pathologically proven RN and 92 cases (68.1%) of TP. The top-performing connectomics features were three centrality measures of degree, betweenness, and eigenvector centralities. Combining these with the 10 top-performing mpRad features, an optimized IsoSVM model was able to produce a sensitivity of 0.87, specificity of 0.84, AUC-ROC of 0.89 (95% CI: 0.82–0.94), and AUC-PR of 0.94 (95% CI: 0.87–0.97).

Abstract BACKGROUND In treating glioblastoma, irradiation of the neural progenitor cell (NPC) niches is controversial. Lower hippocampal doses may limit neurocognitive toxicity, but higher doses to the subventricular zones (SVZ) may improve survival. OBJECTIVE To prospectively evaluate the impact of limiting radiation dose to the NPC niches on tumor progression, survival, and cognition in patients with glioblastoma. METHODS Patients with glioblastoma received resection followed by standard chemoradiation. Radiation dose to the NPC niches, including the bilateral hippocampi and SVZ, was minimized without compromising tumor coverage. The primary outcome was tumor progression in the spared NPC niches. Follow-up magnetic resonance imaging was obtained bimonthly. Neurocognitive testing was performed before treatment and at 6- and 12-mo follow-up. Cox regression evaluated predictors of overall and progression-free survival. Linear regression evaluated predictors of neurocognitive decline. RESULTS A total of 30 patients enrolled prospectively. The median age was 58 yr. Median mean doses to the hippocampi and SVZ were 49.1 and 41.8 gray (Gy) ipsilaterally, and 16.5 and 19.9 Gy contralaterally. Median times to death and tumor progression were 16.0 and 7.6 mo, and were not significantly different compared to a matched historical control. No patients experienced tumor progression in the spared NPC-containing regions. Overall survival was associated with neurocognitive function ( P ≤ .03) but not dose to the NPC niches. Higher doses to the hippocampi and SVZ predicted greater decline in verbal memory ( P ≤ .01). CONCLUSION In treating glioblastoma, limiting dose to the NPC niches may reduce cognitive toxicity while maintaining clinical outcomes. Further studies are needed to confirm these results.