Genomic Concordance Patterns Distinguish Multiple Primary Tumors from Metastatic Recurrences

Background Multiple primary cancers (MPC), wherein two or more synchronous or metachronous malignant tumors in the same or different organs are present within a single patient (pt), are observed in ~10-25% of pts with cancer. Differentiating MPCs from metastatic disease has implications for staging, treatment selection and prognosis. Here, we evaluated whole exome sequencing (WES) data of tumor tissues from pts known to have MPCs and metastatic recurrence (MR). Methods Pairs of distinct primary tumors from 76 pts with MPCs and primary/metastatic site tumor pairs from 134 pts with MR were compared. Tumor tissue WES was performed as part of either commercial circulating tumor DNA testing (Signatera TM , Natera, Inc) or Altera TM Comprehensive Genomic Profiling (Natera, Inc). To enable intertumor comparison, normalization involved restricting variants to shared genomic regions (defined by a BED file) and normalizing variant allele frequencies in each sample by their respective median. Concordance analyses included shared variants (Jaccard index), VAF correlation coefficients, TMB comparisons (mut/Mb) and single-base substitution (SBS) mutational profile. Outlier pairs were identified by >5 mut/Mb TMB difference or Pearson VAF correlation <0.3. Results MPC and MR cohorts had 44.7% and 48.9% males, respectively; median pt ages were 69 (40-90) years and 72 (39-90) years, respectively. No primary tumor pairs were from the same organ. MPC tumor pairs exhibited minimal somatic mutation overlap [median Jaccard index = 0.00 (0.00-0.43)], with only 5% of patients having >1 shared driver mutations. Nearly 77% (59/76) MPC tumor pairs had >2-fold differences in TMB. Additionally, the cosine similarity distribution of SBS mutational profiles across MPC tumor pairs was broad and relatively low [median: 0.49 (0-0.95)], indicating distinct mutagenic processes. In contrast, MR tumor pairs demonstrated markedly higher genomic similarity, with high proportion of shared mutations [median Jaccard index = 0.352 (0- 0.965)], high VAF correlations (mean Pearson r = 0.62) in shared mutations, and >1 shared driver mutations in 61.9% of tumor pairs. Further, MR presented a strongly right-skewed distribution of SBS mutational profiles cosine similarity [median: 0.77 (0-0.98)]. TMB was largely concordant between MR paired tumors, though 15% of pairs exhibited divergence >5 mut/Mb. Conclusions MPC and MR tumor pairs showed markedly distinct patterns of genomic concordance. MPC tumors have low mutation sharing and minimal similarity in mutational profiles, suggesting independent mutagenic processes. Conversely, MR tumors display high clonal continuity across all genomic metrics, indicating a shared origin. VAF correlation and mutational signature profiling provides a robust framework to molecularly distinguish MPCs from MR, addressing a critical unmet need in tumor classification and precision oncology.