Focal cortical dysplasias and tumors in epilepsy surgery: considerations on experimental and bioinformatic workflows for the detection of somatic mutations in brain and liquid biopsy

Patients with focal epilepsy (FE) often experience drug-resistant (DR) seizures in over one-third of cases, leading to significant morbidity and an increased mortality rate. Drug-resistant focal epilepsy (DR-FE), such as Focal Cortical Dysplasia (FCD) and Long-term Epilepsy-Associated Tumors (LEATs), is frequently associated with somatic mutations localized to the epileptogenic region. These mutations present a detection challenge due to their extremely low variant allele frequency (VAF). To address this issue, a method was designed to detect somatic variants for the genetic characterization of LEAT and FCD mosaicisms. The study utilized surgical specimens from FE patients, comprising primarily matched Fresh Frozen (FF) brain–blood pairs, with a smaller subset of brain-only Formalin-Fixed, Paraffin-Embedded (FFPE) samples. Gene panels targeting 24 genes associated with FCD and LEAT were developed using single molecule Molecular Inversion Probes (smMIPs). The panels were optimized using matched blood-brain samples from two FCD and one LEAT cases to fine-tune the detection of somatic variants at VAFs as low as 1%. The results demonstrated the effectiveness of the approach. A pathogenic variant in the SLC35A2 gene with a VAF of 2.49% was identified in FF lesional tissue of an FCD type I patient, absent in the blood sample. Somatic mutations were detected in the MTOR gene (VAFs ranging from 1.07% to 5.89%) and in the RHEB gene (VAFs from 1.16% to 11.75%). A somatic mutation in the KRAS gene (VAF=6.05%) was found in a FF-FCD sample, while a somatic mutation in the TSC2 gene (VAF=3.89%) was detected in a FFPE-FCD sample. Germline variants in DEPDC5, TSC1, and NPRL3 genes were observed in brain-blood matched samples from FCD type II patients. In patients with LEATs, somatic mutations were identified in BRAF (VAFs ranging from 2% to 35%) and FGFR1 (VAFs between 1.53% and 7.16%) genes, as well as in IDH1 and TP53 genes (25-60% VAFs). Furthermore, our analysis on cerebrospinal fluid (CSF) reported the feasibility of detecting brain somatic mutations in this type of sample, showing that CSF could be an important source of mutated brain DNA, even for low VAF mosaicisms, and may allow us to identify brain somatic variants before epilepsy surgery, representing an early marker. In conclusion, the use of smMIPs proved to be a reliable method for detecting low-VAF somatic variants in FCD and LEAT samples, providing valuable insights into the genetic underpinnings of these conditions.