Molecular PET Imaging Versus Conventional Imaging for Detection of Mandibular Metastases in Prostate Cancer: A Paired Retrospective Study

Purpose: Maxillofacial metastases from prostate cancer account for 9–11% of metastatic jaw lesions. The optimal imaging technique for suspected maxillofacial metastases remains undefined. Conventional imaging often lacks sensitivity for early disease, while PET/CT may be limited due to physiologic tracer uptake in adjacent salivary glands. With increasing use of PET/CT in advanced prostate cancer, its diagnostic performance in the mandibular region requires further evaluation. Methods: We conducted a single-institution paired retrospective study using the prospectively maintained Mayo Clinic Advanced Prostate Cancer Registry. Patients with clinically confirmed mandibular metastases who underwent molecular PET/CT ([68Ga]Ga-PSMA-11 or [11C]choline) and conventional imaging (CT or MRI) within 1 month were included. All studies were blindly reviewed by a nuclear medicine physician. Detection rates, anatomic distribution, and quantitative imaging parameters were assessed. Paired detection rates were compared using a one-sided exact McNemar test, with a two-sided exact test performed as a sensitivity analysis. Results: Eighteen patients with clinically confirmed mandibular metastases underwent 36 paired imaging evaluations (18 molecular PET/CT and 18 conventional studies). The median age was 70.5 years (IQR, 60–77), and the median Gleason score was 8 (IQR, 7–9). When mandibular metastasis was detected, 56% of our cohort had hormone-sensitive and 44% had castration-resistant disease. Molecular PET/CT detected mandibular involvement in 18/18 patients (100%; 95% CI, 81–100%) versus 13/18 (72.2%; 95% CI, 47–90%) with conventional imaging. All five discordant cases were detected by molecular PET/CT but not conventional imaging, with no discordant cases favoring conventional imaging (one-sided exact McNemar p = 0.031; two-sided p = 0.063). Mandibular involvement was predominantly unilateral (94.4%), most commonly involving the body (50%), followed by the ramus (27.8%) and condyle (16.6%). The median SUVmax was 6.3 (IQR, 4.4–16.9) for PSMA PET/CT and 4.3 (IQR, 3.8–8.5) for choline PET/CT; the median lesion volume was 5.0 mL (IQR, 1.1–11.3). Conclusions: Molecular PET/CT identified all clinically confirmed mandibular prostate cancer metastases in this paired cohort, including five cases not detected by conventional imaging. Despite physiologic tracer uptake in adjacent salivary structures, mandibular metastatic disease remained detectable on molecular imaging. These findings suggest that molecular PET/CT may provide complementary value when mandibular metastasis is suspected, particularly when conventional imaging is negative or equivocal.