CH Mutations in 16,812 Advanced Cancer Patients–Letter

Magee and colleagues (1) report the finding that 66.5% of somatic BRCA2 and 58.6% of somatic BRCA1 variants detected in their liquid biopsy assay are of clonal hematopoiesis (CH) origin. Furthermore, when considering all BRCA1/2 variants (germline and somatic), they suggest that CH variants account for a large proportion of BRCA2 (39%) and BRCA1 (27%) variants in liquid biopsies. The authors conclude that BRCA1/2 results in liquid biopsies are unreliable, and they imply that about one third of patients with BRCA1/2 mutations detected in liquid biopsies do not, in fact, have BRCA-driven cancers, with critical treatment implications. These results diverge significantly from prior studies, so we investigated their validity.Using the data provided in Supplementary Table S2, we recapitulated the percentages of BRCA1/2 variants reported above to establish that we are examining the same BRCA1/2 variants discussed in the Magee and colleagues study. We found that the vast majority of patients with CH-derived BRCA1/2 alterations also had coexisting germline or tumoral BRCA1/2 alterations in their liquid biopsy (Fig. 1A). At the sample level, only 2.6% (14/531) of BRCA1-altered samples and 1.1% (11/1,033) of BRCA2-altered samples harbored only CH-derived BRCA variants. In other words, 98.4% (1,539/1,564) of patients with BRCA1/2-altered liquid biopsies did have a germline or tumoral BRCA1/2 variant. The authors partially mention this phenomenon in their article:“Interestingly, the co-occurrence rate of germline BRCA and CH BRCA is close to 50%, thus it is important to know if low-VF mutations are from tumor or CH and it is important to report germline alterations when liquid biopsy tests are performed.”However, they do not acknowledge that nearly all of the remaining 50% of samples with CH-derived BRCA1/2 (and without a germline BRCA1/2 alteration) have a tumoral BRCA1/2 alteration (Fig. 1A).This unexpected observation prompted us to delve deeper into the individual BRCA1/2 variants described in this study. Only 2.6% (6/231) of the CH-derived BRCA1 and 2.9% (24/840) of the CH-derived BRCA2 variants would be predicted to be pathogenic or likely pathogenic (Fig. 1B). Of the 231 BRCA1 CH variants identified, 96% (n = 222) were represented by just two intronic variants at the same nucleotide that are not predicted to affect splicing: c.4185+21_4185+22delTG (rs273900723; n = 150) and c.4185+21_4185+22dupTG (n = 72). These intronic variants were reported as a pair in the same liquid biopsies in 53% (82/155) of samples with at least one of these detected. A similar pattern was observed for BRCA2, for which just six recurring intronic variants accounted for 96% (808/840) of the CH-derived BRCA2 variants. In this case, 98% (253/258) and 62% (130/210) of samples with BRCA2 c.1909+22del/dupT or BRCA2 c.68-7del/dupT, respectively, had both variants detected in the same liquid biopsy. In some samples carrying intronic variant pairs, one variant had been assigned as CH-derived and the other as tumor-derived. Based on the distance from the exon–intron junction, these variants are not known or predicted to affect splicing, are listed in ClinVar (2) as likely benign (Fig. 1B), and would not be reported by most testing laboratories. Furthermore, in gnomAD (3), the frequency of BRCA2 c.1909+22delT is 4.0% and that of c.1909+22dupT is 2.8%, but these variants were never assigned as germline in this study (493 CH-derived, 18 tumor-derived, and 0 germline variants), raising the possibility that germline BRCA1/2 variants are being mistaken for CH by this leukocyte sequencing workflow.In our opinion, the study would have benefitted from focusing on pathogenic/likely pathogenic variants and excluding likely benign intronic variants and missense mutations of unknown significance. We confine the focus of this letter to reevaluation of the BRCA1/2 variants, as the authors suggest an impact on therapeutic selection for these variants. However, we note that for most of the genes analyzed in this study, variants designated as benign variants or of unknown significance were included in the analysis. The inclusion of these variants makes it difficult to understand the true extent of CH confounding for clinically actionable pathogenic variants detected on this liquid biopsy platform.We are concerned that Magee and colleagues may have overestimated the extent to which actionable BRCA1/2 variants are CH-derived false-positives on liquid biopsy, with multiple statements and discussion throughout the article and in the accompanying press release (which is intended to reach practicing clinicians beyond the readership of CCR; ref. 4). Inflation of the CH false-positive rate in actionable BRCA1/2 variants poses a risk of causing undue alarm in clinicians who rely on liquid biopsies for BRCA1/2 testing and for clinical decision-making.E.S. Antonarakis reports personal fees and nonfinancial support from Foundation Medicine and nonfinancial support from Caris Life Sciences during the conduct of the study; grants and personal fees from Janssen, Sanofi, Bayer, Bristol Myers Squibb, Curium, MacroGenics, Merck, Pfizer, and AstraZeneca, personal fees from AADI Biosciences, Alkido Pharma, Astellas, Amgen, Blue Earth, Boundless Bio, Corcept Therapeutics, Exact Sciences, Hookipa Pharma, Invitae, Eli Lilly and Company, Menarini Silicon Biosystems, Tango Therapeutics, Tempus, and Z-Alpha, and grants from Novartis, Celgene, and Orion outside the submitted work; and a patent for an AR-V7 biomarker technology issued and licensed to Qiagen. K.A. Reiss reports personal fees from Carisma, AstraZeneca, Bristol Myers Squibb, Guardant Health, Foundation Medicine, Synovation, MOMA Therapeutics, and Merus outside the submitted work. J.E. Berchuck reports grants and personal fees from Guardant Health, other support from Genome Medical, Oncotect, Tracer Biotechnologies, and MusclePharm, and grants, nonfinancial support, and other support from Precede Biosciences outside the submitted work. D.W. Brown reports personal fees from Foundation Medicine and other support from Roche during the conduct of the study. H. Tukachinsky reports personal fees from Foundation Medicine and other support from Roche during the conduct of the study. No disclosures were reported by the other authors.