Cancer of unknown primary (CUP) accounts for 2%–5% of new cancer diagnoses and is defined by the presence of metastatic disease without an identifiable primary tumor despite extensive investigation. Due to its complexity and heterogeneity, CUP carries a poor prognosis, contributing to around 8% of all cancer-related deaths, with only 16%–20% of patients surviving beyond one year. Most patients (80%–90%) fall into an unfavorable risk group with markedly reduced median survival.
Medical imaging is crucial in diagnosing CUP, with major advances improving tumor detection. Early 2D methods like X-ray and ultrasound are now obsolete. CT remains the mainstay, tripling or quadrupling detection rates over older tools. MRI offers superior soft-tissue contrast, especially for occult breast (70%–100% detection) and head-neck cancers. ^18F-FDG PET/CT, the most studied modality, combines metabolic and anatomical data, achieving an overall 41% detection rate, rising to 74% in brain metastases.
Future integration of artificial intelligence (AI) and machine learning into image analysis holds great promise for improving diagnostic accuracy, identifying elusive primary sites, and guiding personalized treatment strategies for patients with CUP.