Abstract Background: Comparison of a blood or tissue sample with a matched germline control is a key step for accurate detection of somatic mutations. This is particularly important for tracking minimal residual disease (MRD), since alterations erroneously considered to be tumor derived may lead to false detection. Use of blood leukocytes as a germline control is critical not only for appropriate censoring of constitutional allelic variants, but also for addressing those associated with clonal hematopoiesis (CH). However, such matched germline is not always available, or may be suboptimal when contaminated with tumor cells. We hypothesized that accurate tumor genotyping might be feasible without a matched germline, using a dedicated algorithmic framework relying on clonal relationships of alleles of interest (i.e., phylons). Methods: Here, we introduce RePhyNER (Recursive Phylon Nomination, Enumeration, and Recovery), an algorithm for addressing this challenge in serial liquid biopsies. RePhyNER relies on the assumption that when comparing two specimens from the same individual (e.g., before and after therapy), the genomic variants of interests (e.g., tumor somatic mutations) behave differently in their allelic levels when compared to variants needing to be censored (e.g., germline constitutional alleles or CH). Using the change in mean allelic level of all variants in ≥2 samples, RePhyNER relies on a recursive approach to test each candidate somatic variant’s corresponding change gradient against this distribution using count-based statistics. Results: We used simulations to assess the importance of multiple parameters including fraction of contaminating variants or alleles and mean allelic fold-change. These simulations showed substantially superior performance of a recursive approach and revealed that for ≥2-fold changes in the mean VAF in a pair of samples, RePhyNER accurately removed >99% of contaminating alleles while preserving ~100% of true variants. We then applied RePhyNER to 108 plasma cell-free DNA (cfDNA) samples from 47 patients with classic Hodgkin Lymphoma (cHL) profiled by PhasED-Seq and compared MRD-detection performance with or without matched germline. RePhyNER substantially improved specificity by ~30% (68% vs 99%), and Precision by ~50% (46% vs 96%), while only modestly reducing sensitivity (100% vs 91%). We next applied PhasED-Seq to a cohort of cHL patients (n=65) without matched germline information. MRD positivity within the first 2 cycles of therapy was associated with inferior outcome when using RePhyNER (P<0.05 vs ns), correctly reclassifying 26% of detected patients to undetected. Conclusions: RePhyNER enables germline-free and accurate genotyping for MRD detection. Notably, RePhyNER obviates the need for additional germline profiling overcoming key limitations described above and avoids the additional associated costs of sequencing. Citation Format: Mohammad Shahrokh Esfahani, Stefan Alig, Emily Hamilton, Joseph Schroers-Martin, Brian Sworder, Jan Boegeholz, Mari Olsen, Chih Long Liu, David Kurtz, Maximilian Diehn, Ash Alizadeh. RePhyNER: Overcoming limitations of access to matched germline for serial liquid biopsy applications [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1045.
Read full abstract