Circulating tumor DNA-based genomic profiling and real-world outcomes in cancer of unknown primary (CUP).

4200 Background: CUP accounts for <5% of cancers and carries a dismal prognosis with median overall survival of ~13 months. Studies suggest up to one-third of CUP patients (pts) have a potentially targetable alteration (PTA) that may be eligible for molecularly guided therapy (MGT). Liquid biopsy (LB) is a non-invasive method to identify PTAs and genomic clues regarding primary site via circulating tumor DNA (ctDNA). We characterize the ctDNA landscape of PTAs in CUP and evaluate outcomes for pts receiving MGT. Methods: Real-world data was sourced via GuardantINFORM, a database aggregating insurance claims and de-identified records of pts with clinical LB via Guardant360. PTAs were defined as alterations with FDA approved therapies in non-CUP indications. Pts with CUP and ≥1 treatment claim after LB were included. Outcomes of pts treated with MGT and pts with the same PTA not treated with MGT were assessed via real-world time to treatment discontinuation (rwTTD), real-world time to next treatment (rwTTNT) and real-world overall survival (rwOS) in months. Log-rank test was used to compare time-to-event outcomes. Results: Of 13,324 CUP pts, 50% were male; the median age was 69 years. Majority of pts underwent LB at time of diagnosis (92.1%). In pts with ≥1 genomic alteration identified (91.9%), the most common genomic alterations were TP53 (55%), KRAS (25%), PIK3CA (14%), and EGFR (12%). A PTA was identified in 29.4% of pts, the most frequent being PIK3CA (9.2%), KRAS G12C (4.3%), BRCA1/2 (4%), ERBB2 (3.9%), EGFR (2.8%), IDH1 (2.5%), BRAF V600E (2.4%) and MSI-H (2%). rwTTD was higher for pts with alterations in EGFR, BRAF V600E, and MSI-H receiving MGT; only EGFR reached significance (Table). Similarly, rwTTNT was improved in pts with EGFR, BRAF V600E alterations and MSI-H, but did not reach significance. rwOS was numerically higher for BRAF V600E and ERBB2 mutated pts receiving MGT (Table). Conclusions: This study represents the first large-scale ctDNA genomic profiling of CUP pts with real-world outcomes.LB detected PTAs in 29.4% of CUP pts, similar to tissue-based testing. Our findings support use of LB to identify PTAs in CUP pts; however, prospective trials are needed to assess MGT efficacy. Real-world outcomes (in months): MGT vs no MGT. rwTTD rwTTNT rwOS EGFR (n=46) 5.8 (95CI: 3-10.27) vs 2.8 (95CI: 2.1-4.5), p=0.0043 11.7 (95CI: 6.2-NR) vs 6.5 (95CI: 5.0-NR), p=0.33 12.8 (95CI: 8.9-NR) vs 13.9 (95CI: 7.0-NR), p=0.9 BRAF V600E (n=36) 5.7 (95CI: 4.2-25.1) vs 4.0 (95CI : 3.4-11.4), p=0.48 10.2 (95CI: 6.9-NR) vs 9.1 (95CI: 5.6-NR), p=0.43 35.1 (95CI: 10.2-NR) vs 25.0 (95CI: 7.0-NR), p=0.18 ERBB2 (n=58) 3.9 (95CI: 2.8-5.9) vs 4.2 (95CI: 2.8-6.8), p=0.41 7.0 (95CI: 4.9-NR) vs 6.1 (95CI: 4.2-NR), p=0.96 21.9 (95CI: 11.6-NR) vs 17.0 (95CI: 8.1-NR), p=0.37 MSI-H (n=34) 7.2 (95CI: 4.9-22.5) vs 3.9 (95CI: 2.3-NR), p=0.14 15.4 (95CI: 7.0-NR) vs 10.2 (95CI: 5.5-NR), p=0.77 NR (95CI: 19.0-NR vs 34.5 (95CI: 29.8-NR), p=0.84