Arterial Spin Labeling-Derived Fractional Tumor Burden as a Primary Biomarker for Differentiating Recurrent Tumor from Radiation Necrosis After Stereotactic Radiosurgery.

BACKGROUND AND PURPOSE: Differentiating recurrent tumor from radiation necrosis after stereotactic radiosurgery remains a challenge with conventional MRI. Fractional tumor burden, a voxel-level perfusion biomarker originally developed using DSC-MRI, characterizes intralesional perfusion heterogeneity and has demonstrated good performance and improved interrater agreement in diagnostic interpretations over conventional perfusion metrics. We evaluated arterial spin labeling-derived fractional tumor burden for distinguishing tumor recurrence from radiation necrosis in treated brain metastases. MATERIALS AND METHODS: This retrospective study included 86 patients with 102 brain metastases (68 radiation necrosis, 34 tumor) evaluated with arterial spin labeling-MRI after stereotactic radiosurgery. Ground truth was based on histopathology or clinico-radiologic follow-up. The primary endpoint was fractional tumor burden-high voxel percentage (absolute and normalized). Secondary endpoints included fractional tumor burden-low, mean CBF, maximum CBF, and delta T1. Absolute CBF thresholds of ≤50 and ≥80 mL/100 g/min and normalized thresholds of ≤1.0 and ≥1.4 were defined a priori from prior histopathology-informed data. Diagnostic performance was assessed by receiver operating characteristic analysis and DeLong testing. Between-group comparisons used Mann-Whitney U tests; generalized linear mixed-effects models assessed logistic regression. RESULTS: Normalized fractional tumor burden-high demonstrated good discrimination of tumor from radiation necrosis, with area under the curve 0.81 (95% CI 0.71-0.90), sensitivity 0.79 (95% CI 0.63-0.94), specificity 0.74 (95% CI 0.61-0.86), PPV 0.60 (95% CI 0.43-0.77), and NPV 0.88 (95% CI 0.77-0.96), numerically outperforming absolute fractional tumor burden-high (area under the curve 0.75; 95% CI 0.66-0.84), although the difference was not significant (P = 0.22). Secondary arterial spin labeling-derived metrics demonstrated comparable performance (area under the curves = 0.79-0.83). Normalized fractional tumor burden-high significantly outperformed delta T1 (change in area under the curve = 0.19; 95% CI 0.05-0.31; P = 0.004). Logistic regression confirmed a monotonic association between normalized fractional tumor burden-high and tumor probability (OR = 1.62 per 10% increase; 95% CI 1.33-2.38; P = 0.001). CONCLUSIONS: Arterial spin labeling-derived fractional tumor burden-high provides good lesion-level differentiation between recurrent brain metastases and radiation necrosis after stereotactic radiosurgery, performing comparably to conventional CBF metrics. The voxel-level approach offers a clinically interpretable perfusion biomarker suitable for post-treatment response assessment.