Clonal Relatedness Index: Practical and Reliable Tool for Establishing Clonal Relatedness in Gastrointestinal,...

Although accepted as gold standard, conventional morphology and immunohistochemistry-based methods of determining clonal relatedness between two neoplasms are inconclusive in a subset of patients. Next-generation sequencing data can be used for this purpose; however, this approach has not been widely clinically validated for gastrointestinal (GI)/hepatopancreaticobiliary (HPB)/peritoneal carcinomas. This study evaluated tumor-specific alterations in 120 paired GI/HPB/peritoneal tumors from 60 patients, previously established as clonal (n = 52) or non-clonal (n = 8) based on their clinicopathologic characteristics. A clonal relatedness index (CRI) was calculated as proportion of shared over the average of total alterations and was compared versus phenotype-based clonality prediction. Based on 1508 assessed alterations, the median CRI was 0.85 (interquartile range, 0.69 to 0.95) for clonal cohorts and 0.24 (interquartile range, 0.06 to 0.43) for non-clonal cohorts. CRI receiver-operating characteristic analysis yielded area under the curve of 0.986 (95% CI, 0.962-1.0; P < 0.01), with optimal sensitivity (Sn) and specificity (Sp) at CRI = 0.53. In the overall cohort, phenotype predicted clonality with Sn = 71.2%, Sp = 62.5%, and positive predictive value (PPV) of 92.5% (P = 1), versus CRI with 96.2%, 100%, and 100% (P < 0.001), respectively. In tumors associated with hypermutability, Sn of phenotype dropped to 42.9%, Sp = 33.3%, and PPV = 60.0% (P = 1), whereas CRI maintained Sn = 85.7%, Sp = 100%, and PPV = 100% (P = 0.03). CRI performance remained robust in a validation cohort and with the addition of alternative genomic alterations. CRI is a practical, clinically accessible, and highly relevant tool that outperforms the phenotype-based method and is essential in reliably establishing clonal relatedness in GI/HPB/peritoneal carcinomas.