Cancer of unknown primary (CUP) denotes cancer cases where metastatic spread is histologically confirmed, but no respective primary tumor can be identified. The challenging diagnosis of CUP is further complicated in cases with previously identified malignancies or with dubious clonal relationship between metastatic sites due to ambiguous histology. Our study aims at elucidating clonal relationships by comparing the respective mutational spectra. Targeted next-generation sequencing (NGS) employing formalin-fixed and paraffin-embedded (FFPE) tumor tissue was performed on 174 consecutive CUP patients. Among these, 43/174 (24.7%) patients had a documented prior malignancy. Data on pairwise targeted NGS testing to address clonal relationships between the previous malignancy and the presumed CUP (n = 11) or between different CUP metastatic sites (n = 7) was available in 18 patients. NGS could clarify clonal relationships in 16/18 cases. Among the 11 CUP patients with antecedent malignancies, four cases were clonally independent of the previous malignancy but harbored deleterious germline mutations in BRCA/BAP1/ATM genes. Seven CUP cases were clonally related to the antecedent malignancy, changing the CUP diagnosis to relapse of the prior malignancy. In the seven CUP cases, with doubtfully related metastatic sites, NGS confirmed clonal relationship in five cases and was inconclusive in two. In conclusion, NGS proved an efficient tool to elucidate clonal relationships in clinically challenging CUP cases. Our study cautions against a premature diagnosis of CUP. Relapses of antecedent malignancies should be carefully considered. CUPs clonally independent from the antecedent malignancy should raise a red flag of a potential cancer-predisposing germline mutation.

Cancer of unknown primary (CUP) denotes a malignancy with histologically confirmed metastatic spread while the primary tumor remains elusive. Here, we address prognostic and therapeutic implications of mutations and copy number variations (CNVs) detected in tumor tissue in the context of a comprehensive clinical risk assessment. Targeted panel sequencing was performed in 252 CUP patients. 71.8% of patients had unfavorable CUP according to ESMO guidelines. 74.7% were adeno‐ and 13.7% squamous cell carcinomas. DNA was extracted from microdissected formalin‐fixed, paraffin‐embedded tissues. For library preparation, mostly multiplex PCR‐based Ion Torrent AmpliSeq™ technology with Oncomine comprehensive assays was used. Most frequent genetic alterations were mutations/deletions of TP53 (49.6%), CDKN2A (19.0%) and NOTCH1 (14.1%) as well as oncogenic activation of KRAS (23.4%), FGFR4 (14.9%) and PIK3CA (10.7%). KRAS activation was predominantly found in adenocarcinomas (p = 0.01), PIK3CA activation in squamous cell carcinomas (p = 0.03). Male sex, high ECOG score, unfavorable CUP, higher number of involved organs and RAS activation predicted decreased event‐free and overall survival in multivariate analysis. Deletions of CDKN2A were prognostically adverse regarding overall survival. TP53 mutations did not significantly influence prognosis in the overall cohort, but worsened prognosis in otherwise favorable CUP subtypes. Although not standard in CUP, for 17/198 (8.6%) patients molecularly targeted treatment was recommended and 10 patients (5.1%) were treated accordingly. In conclusion, besides the identification of drug targets, panel sequencing in CUP is prognostically relevant, with RAS activation and CDKN2A deletion emerging as novel independent risk factors in a comprehensive assessment with clinicopathological data.