RESUMO
BACKGROUND: In 30% of patients with brain metastasis (BM), neurological symptoms are the first clinical manifestation of systemic malignancy, referred to as BM from cancer of unknown primary site (BM-CUPS). Here, we define the diagnostic value of 18F-fluordesoxyglucose positron emission tomography (FDG-PET/CT) in the workup of BM-CUPS. METHODS: We screened 565 patients operated for BM at the University Hospital Zurich and identified 64 patients with BM-CUPS with data on both FDG-PET/CT and contrast-enhanced chest/abdomen computed tomography (CT) available at BM diagnosis. A cohort of 125 patients with BM-CUPS from Lille and Vienna was used for validation. RESULTS: FDG-PET/CT was not superior to chest/abdomen CT in localising the primary lesion in the discovery cohort, presumably because most primary tumours were lung cancers. However, FDG-PET/CT identified additional lesions suspicious of extracranial metastases in 27 of 64 patients (42%). The inclusion of FDG-PET/CT findings shifted the graded prognostic assessment (GPA) score from 3 with CT alone to 2.5 for PET/CT (p = 3.8 × 10-5, Wilcoxon's test), resulting in a predicted survival of 5.3 versus 3.8 months (p = 6.1 × 10-5; Wilcoxon's test). All observations were confirmed in the validation cohort. CONCLUSIONS: Lung cancers are the most common primary tumour in BM-CUPS; accordingly, CT alone shows similar overall sensitivity for detecting the primary tumour as FDG-PET/CT. Yet, FDG-PET/CT improves the accuracy of staging by detecting more metastases, reflected by decreased GPA scores and decreased predicted survival. Therefore, randomised trials on patients with BM should standardise methods of staging, notably when stratifying for GPA.