Delayed 18F-FDG PET for the evaluation of suspected recurrent disease in brain metastases treated with radiation

Background and Purpose Differentiating progressive disease (PD) from radiation necrosis (RN) after treatment of brain metastases remains challenging. We evaluated the diagnostic utility of dual-timepoint 18F-FDG PET in distinguishing PD from RN in patients with treated brain metastases showing suspicious MRI findings. Materials and Methods Thirty-four patients with suspicious MRI enhancement following radiation therapy were retrospectively analyzed. Seventeen had PD and 17 had RN, confirmed by a composite reference standard of clinical and radiologic follow-up (all 34 patients) or histopathology (10 patients, 29%). All underwent dual-timepoint 18F-FDG PET imaging (early at ∼77 minutes and delayed at ∼293 minutes post-injection). Tumor-to-background SUVmax ratios (T/B ratios) were calculated for each timepoint. Results T/B SUVmax ratios were significantly higher in PD compared to RN at both early (1.3 ± 0.6 vs. 0.9 ± 0.3, P = .01) and delayed (2.0 ± 1.1 vs. 1.1 ± 0.3, P = .004) imaging, with greater absolute increase from early to delayed imaging in PD (0.7 ± 0.6 vs. 0.2 ± 0.2, P = .007). The delayed T/B ratio achieved an AUC of 0.772 (95% CI: 0.582–0.921). A delayed T/B ratio ≥ 1.5 achieved 94.1% specificity (95% CI: 71.3%–99.9%) with moderate sensitivity (58.8%; 95% CI: 32.9%–81.6%). While no single PET metric outperformed MRI overall, PET correctly reclassified three of four MRI misdiagnoses. Conclusion Dual-timepoint 18F-FDG PET may improve differentiation of PD from RN in equivocal MRI cases, particularly by reducing false-positive diagnoses, with a number needed to image of 8 (95% CI: 3–272). However, all thresholds were data-driven and confidence intervals are wide. Prospective validation with standardized protocols and pre-specified thresholds is warranted.