Reproducibility and Potential Prognostic Value of Spatial Metabolic PET Metrics in Oral Tongue Squamous Cell Carcinoma

Background/Objectives: Conventional FDG PET parameters quantify metabolic intensity and volume but not the location of the metabolic hotspot, whose peripheral shift may reflect aggressiveness. In surgically treated oral tongue cancer (OTC), we evaluated two spatial PET metrics—the normalised distance from the hotspot to the tumour centroid (NHOCmax) and perimeter (NHOPmax)—in two parallel analyses: prognosis and measurement reproducibility. Methods: We retrospectively analysed 89 patients with preoperative FDG PET/CT. NHOCmax and NHOPmax were derived alongside conventional PET parameters. Associations with disease-free (DFS) and overall survival (OS) were assessed by uni and multivariable Cox models, and reproducibility was assessed across segmentation methods and by intra-reader repeatability. Results: Over a median 83.4/102.1-month follow-up (DFS/OS), 29 DFS and 25 OS events occurred. In the prognostic analysis, higher NHOCmax (OS HR 1.14, p = 0.032) and lower NHOPmax (DFS HR 0.65, p = 0.017; OS HR 0.65, p = 0.031) were associated with poorer survival upon univariable analysis. In the reproducibility analysis, both were robust to segmentation (intra-class correlation coefficient 0.94 for NHOCmax, 0.78 for NHOPmax); intra-reader repeatability was 0.69 and 0.94, respectively. Both metrics were less reproducible in low-volume tumours. Conclusions: NHOCmax and NHOPmax were associated with poorer survival, though not independently of stage and metabolic tumour volume, and were reproducible except in low-volume tumours; they warrant validation in larger cohorts.