Choice of control tissue impacts designation of germline variants in a cohort of papillary thyroid carcinoma patients

The term germline is commonly used to refer to any non-tumor control sample analyzed in tumor–normal paired sequencing experiments. Blood is the most commonly utilized control, and variants found in both tumor and blood are considered germline. However, somatic variants accumulate within an organism from embryogenesis throughout life. The resultant mosaicism is extensive and calls into question the assumption that blood, or any somatic tissue, represents the germline. Misclassification of germline and somatic variants has critical consequences for individual patient care and enormous impact on our health care system, given potential screening, counseling, and treatment implications of misidentifying germline variants. Whole-exome sequencing was performed on six separate specimens from each of two patients with papillary thyroid carcinoma, and three specimens each from eight additional patients forming a validation cohort. Tumor variants were compared with each individual non-tumor control and with composite control sets generated as approximations of true germline. For the index patient, parental blood was also sequenced to assess whether patient-only samples could approximate a trio-derived germline. Using different non-tumor control tissues results in altered germline–somatic designation of tumor variants. In patient 1, 82% of variants are labeled germline using blood control, compared with 75.8%, 61.5%, and 49.6% using lymph node, thyroid, and thymus, respectively. In patient 2, the thyroid control resulted in the greatest percentage of germline calls (70.0%), followed by thymus (56.0%), lymph node (50.1%), and blood (44.1%). Composite control sets built from multiple samples can approximate the germline, even in the absence of parental DNA. Misclassification of germline–somatic origin has potential consequences for patient care, informing screening, trial eligibility, prophylactic interventions, and family planning. This study demonstrates the need for caution in interpreting germline–somatic designation if these data are to inform clinical decisions and suggests that improved design of controls can overcome current limitations. •Tumor sequencing is often performed with paired non-tumor samples, allowing distinction of somatic from germline mutations.•Misdesignation of mutation origin has profound impact: screening, prophylactic intervention, clinical trial eligibility.•Even the healthy genome is subject to variation in the course of a lifetime, result