Cancer of unknown primary, favourable subsets

Cancer of unknown primary is a metastatic cancer whose origin cannot be found despite a full work-up. Within it, decades of clinical observation identified subsets that behave like, and respond to treatment for, a specific cancer. The ESMO 2023 guideline lists them: women with adenocarcinoma confined to axillary nodes (treated as breast cancer with axillary dissection, breast radiotherapy or mastectomy and systemic therapy); women with serous papillary carcinoma of the peritoneum (treated as ovarian cancer with cytoreductive surgery and carboplatin-paclitaxel); squamous cell carcinoma in cervical nodes (treated as head and neck cancer, with HPV or EBV testing, neck dissection and chemoradiotherapy) or in inguinal nodes (treated with node dissection and radiotherapy as anogenital cancer); young men with poorly differentiated carcinoma in a midline distribution or raised markers (treated as extragonadal germ cell tumour with cisplatin-based chemotherapy); men with bone metastases and raised PSA (treated as prostate cancer); neuroendocrine carcinoma of unknown primary (treated as extrapulmonary neuroendocrine carcinoma or, if well differentiated, as a neuroendocrine tumour); adenocarcinoma with a colorectal immunoprofile (CK20 and CDX2 positive, CK7 negative; treated as colorectal cancer); a single resectable metastasis (treated with surgery or radiotherapy); and renal-like carcinoma.

Recognising these patterns depends on a disciplined work-up: histology with a directed immunohistochemistry panel, CT of chest, abdomen and pelvis, mammography or breast MRI and gynaecological examination in women, PSA in men, alpha-fetoprotein and hCG in young patients, and PET-CT in cervical node squamous carcinoma and single-site disease. Gene-expression and DNA-methylation tissue-of-origin classifiers can assign a likely primary in most cases, but two randomised trials (GEFCAPI 04 and a Japanese trial) found that classifier-directed site-specific therapy did not beat empirical chemotherapy in unfavourable disease, so the classifiers are used to support rather than replace clinical judgement. Because the subsets are treated as their presumed cancer, their prognosis approaches that of the corresponding metastatic or node-positive disease, which is why every patient with CUP should be reviewed against the list before empirical chemotherapy is started.