e15166 Background: Predictive biomarker testing for therapy selection is the primary intended use of tissue-based comprehensive genomic profiling (CGP) in advanced and metastatic cancer patients. This study aimed to characterize oncologists’ use of multi-specimen CGP to elucidate diagnostic uncertainty in the course of usual care for advanced and metastatic cancer patients. Methods: To assess the nature and extent to which CGP is potentially used for diagnostic disambiguation, we retrospectively identified patients tested at a reference laboratory with CGP at least twice (Illumina TruSight Oncology 500 assay), and the first two serial CGP tests for each patient were selected for study inclusion. We then explored patient demographics, rare ( < 20 cases per 100k individuals) vs common diagnoses, tested tissue characteristics, time between CGP orders, and differences in genomic variant profiles between test pairs to characterize potential diagnostic use cases. Results: In all, 337/15,717 (2.1%) of patients underwent multi-CGP (June 2021-December 2023). All CGP test pairs were performed using different tissue specimens from 2 separate surgical cases (80.7%) or 2 different blocks for the same surgical case (19.3%). Compared to single-CGP, patients with multi-CGP were younger on average (65 vs 68 years), more often female (54.6% vs 50.6%), and had a greater proportion of lung (45.7% v 39.8%), breast (10.7% vs 6.7%), and unknown primary (7.1% vs 5.2%) tumors. CGP test pairs were grouped for comparison by test order diagnoses submitted by oncologists as: (Group 1) 2 different primary tumors (16%); (Group 2) same primary tumor, different histologic subtypes (19%), or; (Group 3) same primary tumor, same histologic subtype (65%). In Group 1, multi-CGP was performed more frequently for patients with a rare diagnosis for at least 1 test compared to Groups 2 and 3 (81.5% vs 43.8% or 19.6%, p < 0.001), especially for cancers of unknown primary origin (31.5% vs 7.8% and 0.9%, p < 0.001). In Groups 1 and 2, testing for both a primary and a metastatic site was twice as common compared to Group 3 (43% and 41% vs 22%, p = 0.017). In Group 3, 80% of patients were represented by common tumor types, and were most often tested > 30 days apart (64.4%). The mean percent overlap in detected genomic variants was lowest for patients in Groups 1 and 2 where patients had 2 different primary diagnoses or 2 different histologies, and highest in Group 3, where patients had the same primary diagnosis and histology for both tests, (41.5%, 54.5%, 60%, p = 0.008). Conclusions: In practice, most multi-CGP testing is ordered serially for a subset of advanced cancer patients with common tumor types. Use of simultaneous multi-CGP for diagnostic decision-making appears to be limited to a fraction of patients representing more rare tumors with distinct diagnostic conundrums, warranting further exploration of clinical utility.
Read full abstract