CUPCOMP: A multi-site UK trial in carcinoma of unknown primary: A comparison across tissue and liquid biomarkers.

• DOI: 10.1200/jco.2024.42.16_suppl.3059
• Publication Date: 2024-06-14T17:01:38Z
• Authors: Alicia-Marie Conway; Matthew Robinson; Matthew Concannon; Sally Clive; Tania Tillett; Kai-Keen Shiu; Louise C. Medley; Sonali Dasgupta; Eliyaz Ahmed; Kelly Warrington; Zoulikha Zair; Tess Gillham; Aaron Davis; Julie-Anne Scott; Alison Taylor; Mark Stares; Claire Mitchell; Pedro Oliveria; George Burghel; Natalie Cook

3059 Background: Cancer of Unknown Primary (CUP) is a difficult to treat cancer entity for which the tumour origins remain elusive. There is emerging evidence to support a precision medicine approach to aid treatment decisions, however scarcity of tumour tissue for molecular profiling remains a challenge. CUPCOMP sought to compare the feasibility of molecular profiling from both tissue and blood in patients (pts) diagnosed with CUP. Methods: A non-randomised prospective feasibility clinical trial conducted across 7 UK sites between 06/2021-02/2023. Eligible pts had a histological confirmed diagnosis of favourable/unfavourable CUP based on clinical, radiological, and pathological review at a CUP Multi-Disciplinary Team meeting and ESMO Guidelines for CUP. Pts had mutation profiling performed from circulating tumour DNA (ctDNA) (FoundationOne Liquid CDx) and tumour tissue (FoundationOne CDX or whole genome sequencing (WGS)). All results were discussed at a dedicated Molecular Tumour Board (MTB) to evaluate the activation state of oncogenic pathways. Results: Baseline characteristics of the 117 pts recruited are shown (Table). 114/117 (97%) pts had successful blood (109/117; 93% pts) or tissue (69/117; 59% pts) molecular profiling; both methods were performed in 64/117 (55%) pts. WGS was only successful in 15/21 pts where repeat biopsy was obtained. Where targeted tissue and blood profiling was performed using the same panel (49/117; 42%): 421/550 (77%) Single Nucleotide Variants (SNVs) found in tissue were concordant in ctDNA. ctDNA tumour fraction was ≥10% in 41/109 (35%) pts. In ctDNA, 1556 SNVs, across 271 genes, and 42 rearrangements were identified. 73/114 (64%) of cases carried an actionable alteration in blood or tissue, as determined by MTB. The most frequently reported actionable gene alterations were: KRAS (n=11; 10%); ATM (n=9; 8%); ARID1A (n=9; 8%); PIK3CA (n=10; 9%); NF2 (n=7; 6%); BRAF (n=7; 6%); FGFR2-fusion (n=7; 6%). A tumour mutational burden (TMB) of ≥10 mutations/Mb was detected in 25/117 (22%) of pts. Potential germline variants were found in 8% of pts. 80/114 (70%) pts were alive 6 months after consent to trial. Conclusions: Successful tissue-based molecular profiling is challenging in pts with CUP; achieved in only 59% of pts in this cohort. ctDNA analysis was feasible (93% successful) and reliably concordant with tissue. Potentially actionable alterations were found in 64% of pts. ctDNA molecular profiling should be considered as a reliable alternative to tissue-based testing in pts diagnosed with CUP. Clinical trial information: NCT04750109 . [Table: see text]