Multi-omic analysis in carcinoma of unknown primary (CUP): therapeutic impact of knowing the unknown

Carcinoma of unknown primary (CUP) is a difficult-to-manage malignancy. Multi-omic profiles and treatment outcome vs. degree of precision matching were assessed. Tumours underwent next-generation sequencing (NGS) [tissue and/or blood-derived cell-free DNA (cfDNA)]. Selected patients had transcriptome-based immune profiling and/or programmed cell death 1 ligand 1 (PD-L1) immunohistochemistry analysis. Patients could be reviewed by a Molecular Tumor Board, but physicians chose the therapy. Of 6497 patients in the precision database, 97 had CUP. The median number of pathogenic tissue genomic alterations was 4 (range, 0-25), and for cfDNA, was 2 (range, 0-9). Each patient had a distinct molecular landscape. Food and Drug Administration (FDA)-approved biomarkers included the following: PD-L1(+) >= 1%, 30.9% of CUPs tested; microsatellite instability, 3.6%; tumour mutational burden >= 10 mutations<middle dot>Mb(-1), 23%; and neurotrophic receptor tyrosine kinase (NTRK) fusions, 0%. RNA-based immunograms showed theoretically druggable targets: lymphocyte activation gene 3 protein (LAG-3), macrophage colony-stimulating factor 1 receptor (CSF1R), adenosine receptor A2 (ADORA2) and indoleamine 2,3-dioxygenase 1 (IDO1). Overall, 56% of patients had >= 1 actionable biomarker (OncoKB database). To quantify the degree of matching (tumours to drugs), a Matching Score (MS; roughly equivalent to number of alterations targeted/total number of deleterious alterations) was calculated post hoc. Comparing evaluable treated patients [MS high, > 50% (N = 15) vs. low <= 50% (N = 47)], median progression-free survival was 10.4 vs. 2.8 months (95% CI 0.11-0.64; HR 0.27; P = 0.002); survival, 15.8 vs. 6.9 months (95% CI 0.17-1.16; HR 0.45; P = 0.09); and clinical benefit rate (stable disease >= 6 months/partial/complete response), 71% vs. 24% (P = 0.003). Higher MS was the only factor that predicted improvement in outcome variables after multivariate analysis. In conclusion, CUPs are molecularly complex. Treatments with high degrees of matching to molecular alterations (generally achieved by individualized combinations) correlated with improved outcomes.