br Methods br Consenting patients with a
Consenting patients with a pre-operative diagnosis of EC or complex atypical hyperplasia (CAH) between 5/2005 and 11/2016 were enrolled to a prospectively maintained database at the time of primary surgery at a single institution (Mayo Clinic, Rochester, MN). Information regarding personal and family history of cancer was obtained from a patient-completed questionnaire. Histologic diagnosis was based on the final pathology report of the hysterectomy specimen. Tumors were consid-ered serous if N10% serous histology was reported. Type I EC included tumors with grades 1–2 endometrioid histology, whereas Type II EC in-cluded grade 3 endometrioid and all other non-endometrioid tumors. Histologic information was abstracted from clinical pathology reports.
Germline DNA was extracted from whole blood collected at the time of enrollment. DNA samples were evaluated by sequencing of products from a targeted multiplex amplicon-based QIAseq panel (Qiagen) in-cluding 21 known or suspected cancer predisposition PR 619 (ATM, BARD1, BRCA1, BRCA2, BRIP1, CDKN2A, CHEK2, NBN, NF1, MLH1, MRE11A, MSH2, MSH6, PALB2, PMS2, PTEN, RAD51C, RAD51D, and TP53). Primers covering the coding regions and essential splice sites for each of the genes of interest were included on the panel. Amplicon products were dual barcoded for sample identification. The resulting DNA libraries were sequenced using 150 bp paired-end reads on an Illumina HiSeq4000. Up to 768 samples were sequenced per lane with a median sequence read depth of 200×. Cutadapt v1.10 was used to trim adaptor sequences, and the remaining reads were aligned to the human reference (GRCh37; bwa-mem v0.7.10). Variants were joint-called with haplotype caller and evaluated for depth of coverage using the Genome Atlas Toolkit (GATK) v.3.4–4.6 . Variants were excluded
if fewer than 5 alternate reads or 20 total reads were available. Variants were also called using a single-sample caller (Vardict v1.5.1) using de-fault settings except that variant calls were limited to those with b1% al-ternate allele frequency . The union of variants between the two variant callers was used for downstream analysis. Eight samples (4 pairs) had a high degree of relatedness (Identity By Descent values N 0.5). None were mutation carriers. For each of the 4 pairs, the relatives with the youngest age at diagnosis were retained for the analyses.
Analyses were focused on pathogenic variants in mismatch repair genes associated with LS (MSH2, MSH6, PMS2, and MLH1) as well as other hereditary breast and/or ovarian cancer predisposition genes (ATM, BRCA1, BRCA2, BRIP1, CHEK2, NBN, NF1, PALB2, RAD51C, RAD51D, and TP53) defined as “increased risk” of either breast or ovarian cancer with recommendations for clinical management in NCCN Clinical Prac-tice Guidelines for Genetic/Familial High-Risk Assessment: Breast and Ovarian  in both EC cases and gnomAD non-Finnish European (NFE) public reference controls . The Genome Aggregation Database (gnomAD) is a database of aggregated and harmonized exome and ge-nome sequencing data. With the exception of TCGA, all studies contrib-uting to the database are non-cancer studies. The database includes 55,860 unrelated non-Finnish European (NFE) exomes which were used as reference controls in our study. Variants were classified as path-ogenic/likely pathogenic based on allele frequency (b0.003), effects on protein function, and ClinVar assertions. Rare variants (AF b0.003 in gnomAD) were included in the analyses along with known pathogenic recurrent or founder mutations (e.g. CHEK2 c.1100delC).Stop-gain, frameshift, and essential splice site (+/− 1–2 consensus region) vari-ants were considered pathogenic unless functional evidence or ClinVar assertions from clinical groups (SCRP, Invitae, Ambry Genetics, GeneDx, Emory, InSiGHT) suggested otherwise. If the ClinVar assertions from clinical groups were not in agreement, the more conservative assertion was retained. In addition, nonsense mediated mRNA decay effects were considered for BRIP1, with variants in the last exon or last 55 nucleotides of the penultimate exon excluded. Variants in exons 9 and 11–15 of PMS2 were not included in analyses due to homology with the PMS2L pseudogene in these regions. All variants classified as pathogenic in EC cases underwent visual inspection in IGV 2.4.4 . All variants in EC cases with an alternate allele fraction b0.20 or N0.80 were excluded due to possible mosaicism or clonal hematopoiesis.
Data are summarized using mean and standard deviation (SD) for continuous variables and frequency and percentages for categorical var-iables and mutations. Comparisons between mutation carriers and non-carriers were evaluated using the two-sample t-test for age and BMI, the Fisher's exact test for Caucasian race, and the chi-square test for stage, grade and histology. Mutation frequencies were compared between subtypes and between EC cases and gnomAD reference controls using the Fisher's exact test. All calculated p-values were two-sided. Ninety-five percent confidence interval (95% CI) were constructed using a) an exact method for a binominal parameter for the incidence of each mu-tation and b) an exact method for the odds ratios using an algorithm by Fay . Statistical analysis was performed using the SAS version 9.4 software package or with R package 3.4.2.