Molecular classification of cancers with an uncertain diagnosis as candidates for immunotherapy.
e23183 Background: FDA approvals of immunotherapy checkpoint inhibitors (CPIs) are based on organ site and thus cancers of unknown primary or with uncertain tumor type diagnoses have limited access to these revolutionary treatments. The 92-gene assay (CancerTYPE ID) is a validated gene expression classifier of 50 tumor type/subtypes for metastatic patients, six of which (lung adenocarcinoma, lung squamous cell, head & neck squamous cell, urinary bladder, kidney, and melanoma) have FDA-approved indications for CPIs. Study objectives were to analyze the distribution of the 6 tumor types with available CPIs in a large dataset of clinical cases and report patient and disease characteristics for this subset. Methods: A de-identified database was created under an IRB approved protocol that contains pathology and molecular information from consecutive cases submitted for clinical testing with the 92-gene assay. The distribution of molecular diagnoses was analyzed based on biopsy site, age, and gender. Chi-squared tests were used to compare between subgroups. Results: A total of 24,484 patients were available for analysis. Median age was 66y and 46% were male. The assay rendered a molecular diagnosis with an approved CPI in 27% (N = 6536) of cases. The most common biopsy sites were lymph node (23%), lung (15%) and liver (12%). Lung adenocarcinoma (31%) was the most common molecular diagnoses, followed by urinary bladder (20%), and head & neck squamous (16%). The proportion of melanoma molecular diagnoses decreased with age (11% in < 40y vs 4-6% across other age groups, p < 0.001), whereas the proportion of urinary bladder molecular diagnoses increased from 15% in < 40y to 23% in patients > 65y (p < 0.001). Women had a higher proportion of lung adenocarcinoma molecular diagnoses (36% vs 28% in men; p < 0.001), while kidney molecular diagnoses were more common in men (14% vs 8% in women; p < 0.001). Conclusions: These findings highlight the utility of molecular classification to identify metastatic cancers with unknown or uncertain tumor type diagnoses that are eligible for CPI. In addition, there was a significant difference in the distribution of 92-gene molecular diagnoses by age and gender among cancers susceptible to CPI.