Abstract
BACKGROUND: Distinguishing early recurrence from pseudoprogression after chemoradiotherapy remains a major challenge in glioblastoma. Although (18)-F-fluroethyl-L-tyrosine (FET)-PET is increasingly used, reliance on static thresholds limits accuracy in the early post-radiotherapy period. The role of dynamic time-activity-curve (TAC) patterns in this setting is not well defined. We evaluated the diagnostic and prognostic performance of TAC morphology compared with static uptake metrics. METHODS: Sixty-six patients with glioblastoma underwent FET-PET within 6 months of radiotherapy. Final diagnosis was determined by histopathology (n = 15) or predefined clinical endpoints (n = 51). Static metrics (SUV(max), TBR(max)) and dynamic TAC morphology (Types 1-3) were assessed. Diagnostic performance was assessed by receiver-operating-characteristic analysis, logistic regression, and sensitivity analyses. Overall survival (OS) was analyzed with Kaplan-Meier and Cox regression. RESULTS: Thirty-five patients (53%) had recurrence and 31 (47%) pseudoprogression. TAC morphology demonstrated the highest diagnostic accuracy (AUC 0.76; sensitivity 68.8%; specificity 94.7%; PPV 93.6%; NPV 72.9%), outperforming SUV(max) (AUC 0.66) and TBR(max) (AUC 0.57). TAC morphology was the sole independent predictor of recurrence (OR 4.62; P = .010). Median OS was significantly shorter in recurrence compared with pseudoprogression (10.2 vs 20.5 months; P < .0001). TAC type stratification revealed median OS of 11.7 months (Type 1), 9.9 months (Type 2), and 8.8 months (Type 3; log-rank P = .041). Sensitivity analysis excluding patients with diagnostically ambiguous ground truth (n = 14) confirmed that TAC pattern remained a significant predictor of diagnosis (AUC 0.75; P = .012). CONCLUSIONS: Dynamic FET-PET improves diagnostic specificity and provides prognostic insight beyond static uptake metrics. TAC patterns offer a structured framework for improving diagnostic certainty and guiding therapeutic decisions.