Grading and detecting gastroenteropancreatic neuroendocrine neoplasms with dual-tracer ( 18 F-AlF-NOTA-octreotide/ 18...

To evaluate quantitative metabolic indices from dual-tracer 18 F-AlF-NOTA-octreotide ( 18 F-AlF-OC) and 18 F-fluorodeoxyglucose (FDG) PET/computed tomography (CT) for grading gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs), focusing on total lesion glycolysis ratio (TLGR).

This retrospective study included 34 patients with GEP-NENs (198 target lesions) who underwent both 18 F-AlF-OC and 18 F-FDG PET/CT. Parameters included maximum standardized uptake value (SUV max ), total lesion glycolysis (TLG), SUV ratio (SUVR), and TLGR (ratio of 18 F-AlF-OC volumetric somatostatin receptor uptake to 18 F-FDG TLG by conventional TLG framework). Diagnostic performance for differentiating G1/G2 from G3/neuroendocrine carcinoma (NEC) and neuroendocrine tumor (NET) from NEC was assessed using receiver operating characteristic (ROC) analysis. Incremental value was evaluated using Δarea under the curve (AUC) and continuous net reclassification improvement with patient-level cluster bootstrap resampling. Generalized estimating equation (GEE) models accounted for multiple lesions per patient. Internal validation used leave-one-patient-out cross-validation.

18 F-AlF-OC detected more lesions in G1/G2 NETs, while detection was similar in NEC. In ROC analysis, TLGR, SUVR, and TLGR + SUVR yielded AUCs of 0.760, 0.760, and 0.767 for G1/G2 versus G3/NEC, and 0.811, 0.829, and 0.827 for NET versus NEC, respectively. Adding TLGR to SUVR or FDG SUV max provided only modest improvement. TLGR was significant alone in GEE models but attenuated after SUVR adjustment, and cross-validation showed reduced out-of-sample performance.

Dual-tracer PET/CT provides complementary information for GEP-NEN grading. TLGR is an exploratory volumetric dual-tracer parameter, but its incremental value beyond SUVR and FDG SUV max is limited. Further validation is required.