A retrospective analysis of the concordance of diagnosis predicted by cancer type ID test with pathological or clinical diagnosis with incorporation of treatment response data.

Abstract

e18684 Background: Carcinoma of unknown primary site (CUP) is a clinical conundrum, where there is metastatic cancer in multiple organs, but the original site of the cancer cannot be determined either by clinical information, or by pathological specification. In most scenarios, clinical and pathological information may narrow down to 2 or 3 most likely primaries; while in some cases, the pathological diagnosis and the clinical picture may not be in concordance. Cancer type ID test, using a panel of 92 genes detectable by RT-PCR test, and bio-informative algorithm, may provide independent information on the cancer of origin. Incorporating this result, as well as the treatment information, one may be able to get closer to the true site of origin for CUP cases. Methods: Patients who have had a Cancer type ID test (ID test) performed and had received systemic treatment were eligible for this study. We analyzed the concordance of ID result of major probability (> 90% probability) with pathology diagnosis, the concordance of ID result of minor probability (6% probability or “cannot exclude with < 5% probability” with pathology diagnosis, and the concordance of ID result of major probability with final clinical diagnosis (incorporating treatment response data, tumor next gene sequencing data, and genetic test data). We also divided cases into groups of differentiation between 2-3 possible primaries, or true CUPs (as shown in the table below). Results: 26 patients were identified. There were 9 males and 17 females. The median age was 65 years old (ranger 37-92). The final clinical diagnoses were GI (upper and lower intestinal) (n = 6), pancreatic (n = 3), cervical (n = 2), lung adeno (n = 2), lung carcinoid (n = 1), sarcoma (n = 1), GYN (n = 2), Head neck (n = 1) and breast (n = 1). 7 cases remained as CUP. Conclusions: Cancer type ID test, with a combination of 90% probability, the 6% probability as well as the “cannot exclude with a less than 5% probability”, may reach 92% concordance with pathological diagnosis. It has the highest prediction in differentiation of 2-3 primaries, especially when one primary involves pancreas. However, the concordance between the clinical diagnosis and the diagnosis rendered by the ID test major prediction is not optimal, and will need further improvement.[Table: see text]