Clonal origin and genomic diversity in Lynch syndrome‐associated endometrial cancer with multiple synchronous tumors: Identification of the pathogenicity of MLH1 p.L582H

Lynch syndrome‐associated endometrial cancer patients often present multiple synchronous tumors and this assessment can affect treatment strategies. We present a case of a 27‐year‐old woman with tumors in the uterine corpus, cervix, and ovaries who was diagnosed with endometrial cancer and exhibited cervical invasion and ovarian metastasis. Her family history suggested Lynch syndrome, and genetic testing identified a variant of uncertain significance, MLH1 p.L582H. We conducted immunohistochemical staining, microsatellite instability analysis, and Sanger sequencing for Lynch syndrome‐associated cancers in three generations of the family and identified consistent MLH1 loss. Whole‐exome sequencing for the corpus, cervical, and ovarian tumors of the proband identified a copy‐neutral loss of heterozygosity (LOH) occurring at the MLH1 position in all tumors. This indicated that the germline variant and the copy‐neutral LOH led to biallelic loss of MLH1 and was the cause of cancer initiation. All tumors shared a portion of somatic mutations with high mutant allele frequencies, suggesting a common clonal origin. There were no mutations shared only between the cervix and ovary samples. The profiles of mutant allele frequencies shared between the corpus and cervix or ovary indicated that two different subclones originating from the corpus independently metastasized to the cervix or ovary. Additionally, all tumors presented unique mutations in endometrial cancer‐associated genes such as ARID1A and PIK3CA. In conclusion, we demonstrated clonal origin and genomic diversity in a Lynch syndrome‐associated endometrial cancer, suggesting the importance of evaluating multiple sites in Lynch syndrome patients with synchronous tumors.

Lynch syndrome is an autosomal dominant inherited disease that is associated with a germline mutation in DNA mismatch repair (MMR) genes (MLH1, MSH2, MSH6, and PMS2) or EPCAM.The pathogenic variants of these genes lead to microsatellite instability, characterized by a high frequency of somatic mutations, and significantly elevate the risk of developing various cancers.2][3] Endometrial cancer is the most common extraintestinal cancer in Lynch syndrome, and it develops first in approximately 40%-60% of women with Lynch syndrome. 4Therefore, the diagnosis of Lynch syndrome in women with endometrial cancer is crucial to select the optimal treatment and prevent the subsequent development of other associated cancers.However, there are several clinical issues related to the diagnosis of Lynch syndrome.
The first issue is the identification of a variant of uncertain significance (VUS) in genetic testing.The diagnosis of Lynch syndrome is confirmed by the identification of a germline pathogenic variant by genetic testing.However, genetic testing often identifies a VUS in an MMR gene that has not been clearly shown to cause Lynch syndrome.
Even if the variant is determined to be a VUS at the time of testing, it may be reclassified as a pathogenic or benign variant in the future. 5,6 reevaluate the pathogenicity of a VUS, the family history and genetic testing results of family members are useful.Several databases such as the InSiGHT consortium are also informative in investigating relationships between the variant and Lynch syndrome. 7wever, these data are not completely conclusive.Sequencing of tumors of patients with a VUS is useful to search for second events such as somatic mutations and loss of heterozygosity (LOH) in the same gene.Additionally, microsatellite instability and mutational signatures attributable to DNA MMR defects can be precisely estimated by genome-wide sequencing data.Thus, the findings from tumor sequencing provide insights into the pathogenicity of VUS.
The second issue is the staging in Lynch syndrome patients with cancers in two or more organs.For example, endometrial and ovarian cancers are diagnosed simultaneously in approximately 20% of Lynch syndrome cases. 8,9The staging of synchronous endometrial and ovarian cancers depends on whether the two cancers occurred independently or whether they represent a primary cancer and metastatic lesion according to clinicopathological findings. 10,11Generally, postoperative therapy is essential in cases with metastatic lesions.However, postoperative therapy can be considered omitted in cases of synchronous primary cancers if both cancers are at an early stage.Additionally, different cancer types have different available therapeutic agents; in fact, immune checkpoint inhibitors are effective in endometrial cancer but not ovarian cancer. 12,13Genomic analysis can be useful to differentiate primary tumors from metastatic tumors.The presence of shared mutations between tumors indicates a common clonal origin.Furthermore, differences in cancer cell fractions of the shared somatic mutations can reflect the direction of progression from one tumor to another, allowing the prediction of the primary tumor.A recent study conducted massively parallel sequencing for tumors in five cases diagnosed clinically as synchronous endometrial and ovarian cancers in Lynch syndrome. 14ree cases were metastatic tumors and two cases were synchronous primary tumors; all three metastatic cases were identified as primary endometrial tumors.
Here, we present a case of endometrial cancer with multiple tumor sites that was difficult to diagnose as Lynch syndrome.While family history and immunohistochemical analysis suggested Lynch syndrome, genetic testing identified a VUS (MLH1 p.L582H).By integrating the segregation analysis of the VUS in the family, the results from immunohistochemical staining of DNA MMR proteins and microsatellite instability (MSI) assays for tumors of the family members, and mutation profiling from whole-exome sequencing, we examined the pathogenicity of MLH1 p.L582H and the clonal origin and genomic diversity in the Lynch syndrome-associated endometrial cancer with multiple synchronous tumors.

| Patient and samples
A 27-year-old woman with suspected synchronous cancers of the uterine corpus, cervix, and ovaries underwent radical hysterectomy, bilateral salpingo-oophorectomy, and pelvic lymphadenectomy.She was nulliparous, had no significant medical history, and was not immunocompromised.The postoperative diagnosis was stage IIIA endometrial cancer (pT3aN0M0).The histology of the uterine corpus, cervix, and ovaries was all endometrioid carcinoma grade 1 with CD8-positive T cells (Figure 1, Figure S1), and they were pathologically diagnosed as primary endometrial cancer with metastasis in the other sites on the basis of clinicopathological criteria. 10,11The uterine corpus tumor was 7.7 cm in size, with less than one-half of cancer invading the myometrium and continuous extension to the cervix with cervical stromal invasion.The patient underwent six courses of docetaxel and cisplatin combination therapy. 15e patient had a young onset of endometrial cancer.Furthermore, her father had colorectal cancer twice, and her paternal grandfather had gastric and colorectal cancer (Figure 2).These findings were consistent with the Amsterdam II and the revised Bethesda criteria, 16,17 and we clinically diagnosed her with Lynch syndromeassociated endometrial cancer.

| Immunohistochemical staining
Immunohistochemical analysis of MMR protein expression (MLH1, MSH2, MSH6, PMS2) was performed on formalin-fixed paraffinembedded (FFPE) tissue section samples.FFPE tissue sections (3 μm)   were cut with a cryostat and a microtome, and immunohistochemical staining was performed following standard methods.Briefly, after deparaffinization, antigen retrieval was performed.The sections were incubated with primary antibodies and biotinylated secondary antibodies, followed by incubation with a sensitizer.Primary antibodies included antibodies against MLH1 (BD Biosciences, code: 550838) at

| DNA extraction
DNA was extracted from blood with the QIAamp DNA Blood Maxi kit (Qiagen, Hilden, Germany) following the manufacturer's instructions.DNA extraction from tumors was performed with the Tissue Genomic DNA Extraction Mini Kit (Favorgen, Ping Tung, Taiwan), following the manufacturer's instructions.

| PCR-based MSI assay
MSI assay was conducted by Falco Biosystems (Kyoto, Japan).MSI was evaluated in DNA from tumor tissue (FFPE or frozen tissue) by multiplex PCR-fragmentation analysis using five markers with single nucleotide repeats (NR2, BAT-26, BAT-25, NR-24, MONO-27).When two or more markers showed length shifts, the sample was interpreted as MSI-high.

| Sanger sequencing
To validate the mutation status in blood, frozen tumor tissue, and FFPE samples, DNA was extracted from each sample as described above.FFPE serial sections were prepared and tumor epithelium and tumor stroma were isolated by needle microdissection.FFPE DNA was extracted using a QIAamp DNA FFPE Tissue Kit (Qiagen) following the manufacturer's instructions. 18PCR was performed using a KAPA Taq EXtra HotStart ReadyMix PCR Kit (Kapa Biosystems, MA, USA).PCR primers were designed using Primer3 software (http:// bioinfo.ut.ee/primer3-0.4.0/).PCR primers for the target region of MLH1 exon 16 are 5 0 -GCTCCGTTAAAGCTTGCTCC-3 0 (forward) and 5 0 -GGGATTACAGCCATGAGCCA-3 0 (reverse).PCR products were purified and sequenced by GENEWIZ from Azenta Life Sciences (Saitama, Japan).

| Methylation-specific PCR
Tumor DNA extraction was performed with the Tissue Genomic DNA Extraction Mini Kit (Favorgen), following the manufacturer's instructions.Episcope Methylated HCT116 DNA and Unmethylated HCT116 DNA were used as controls (Takara Bio, Shiga, Japan).DNAs were modified with sodium bisulfite using the EpiTect Bisulfite kit (Qiagen) following the manufacturer's instructions.Genomic DNA modified with sodium bisulfite served as template for methylationspecific PCR using primers specific for methylated and unmethylated versions of CpG islands in the MLH1 promoter and the Episcope MSP kit (Takara Bio).Primer sequences were designed by referring to the literature. 19Primer sequences for MLH1 for the unmethylated reaction were 5 0 -TTTTGATGTAGATGTTTTATTAGGGTTGT-3 0 (forward) and 5 0 -ACCACCTCATCATAACTACCCACA-3 0 (reverse), and primers for the methylated reaction were 5 0 -ACGTAGACGTTTTAT-TAGGGTCGC-3 0 (forward) and 5 0 -CCTCATCGTAACTACCCGCG-3 0 (reverse).Amplification was performed at 95 C for 30 s, followed by 30 cycles at 98 C for 5 s, 59 C for 30 s, and 72 C for 60 s.Specific bands were confirmed by electrophoresis.

| Whole-exome sequencing and analysis
1][22] Briefly, DNA samples were fragmented using a Covaris.Target gene enrichment was conducted with SureSelectXT Human All Exon Kit V6 (Agilent Technologies, CA, USA).The libraries were sequenced via an Illumina NovaSeq (Illumina).The Illumina adapter sequences were trimmed using TrimGalore (Version 0.6.3)(https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) as a quality control step.Low-quality sequences were excluded or trimmed with Trimmomatic (version 0.39). 23The filtered sequence reads were aligned to the human reference genome (GRCh38) containing sequence decoys and virus sequences generated by the Genomic Data Commons of the National Cancer Institute using BWA-MEM (Version 0.7.17). 24,25The sequence alignment map files were sorted and converted to the binary alignment map (BAM) file format with SAMtools (version 1.9). 24The BAM files were processed using Picard tools (version 2.20.6)(http://broadinstitute.github.io/picard/) to remove PCR duplicates.Base quality recalibration was conducted using GATK (version 4.1.3.0). 26,27The average depths and the coverages of the target regions were calculated with SAMtools. 24BEDOPS (version 2.4.36) 28and BEDTools (v2.28.0) 29 were used in the handling of FASTA, VCF, and BED files.Alignment of sequence reads was visualized by IGV tools. 30The average depths and the coverages of the target regions in all samples are shown in Table S1.

| Detection of somatic copy number alterations
Somatic copy number alterations (SCNAs) were detected as previously described. 21,22SCNAs were sought using FACETS using the information on the total sequence read count and allelic imbalance in tumor or non-tumor epithelium samples and the matched blood samples. 31Germline polymorphic sites were retrieved from the VCF file generated by the 1000 Genomes Project. 32The absolute value of the log odds ratio of the variant allele read count in the sample and blood pair was used as the degree of allelic imbalance.After excluding the regions affected by SCNAs, the mean of the absolute value of the log odds ratio over the germline heterozygous SNVs in the genome was calculated.The mean was subtracted from each of the absolute values of the log odds ratio to normalize the data to have a mean of zero.

| Detection of MSI with whole-exome sequencing
We applied MSIsensor (version 0.6) for the detection of MSI with whole-exome sequencing, using paired tumor-normal sequence data. 33,34The MSI score was calculated as the percentage of unstable microsatellite loci divided by the total number of microsatellite loci surveyed.The tumors were classified as MSI-high if the score was greater than or equal to 3.5 and microsatellite stable if the score was less than 3.5, which is the suggested cut-off for exomesequenced samples in the MSIsensor publications. 33,3410 | Variant detection and mutation annotation Variant detection and mutation annotation were performed as previously described. 21,22Somatic single nucleotide variants (SNVs) and short insertions/deletions (Indels) in coding exons and splice sites were called using Strelka2 (version 2.9.10). 35We used the information about candidate Indel sites provided by Manta (version 1.6.0)for somatic Indel calling. 36Empirical variant scores provided by Strelka2 of >13.0103 (=À10 Â log10 0.05) were used for subsequent analyses.Additionally, variants with frequencies ≥0.001 in any of the general populations from the 1000 Genomes Project, 32 the National Heart, Lung, and Blood Institute GO Exome Sequencing Project, 37 and the Genome Aggregation Database 38 were excluded to avoid falsepositive variant calls.Functional annotations for protein coding and transcription-related effects of the identified variants were implemented by Ensembl VEP. 39Curated information about cancer-associated genes and their functional roles in cancer development was retrieved from the COSMIC database. 40Tumor mutational burden (TMB) was calculated as the ratio of the number of somatic mutations to the size of the targeted region. 41In the LOH region detected by FACETS, we searched for biallelic mutations within the germline mutations and somatic sites matched there.
To evaluate the extent to which somatic mutations were shared between samples located in spatially separated regions, we compiled mutant allele frequency (MAF) profiles of all the mutation sites for all the samples by counting the sequence reads supporting the reference and mutant alleles with SAMtools mpileup. 24For this analysis, the reads mapped with high confidence (mapping quality >30) were used.
The allele-specific counts were measured using only high-confidence base calls (base quality >20) at the mutation sites.We excluded sites whose MAFs in the matched blood sample exceeded 0.05.We selected informative mutations on the basis of the following criteria: (i) a set of mutations that were shared among a group of samples at MAF of greater than or equal to 0.05; and (ii) mutations with MAF values greater than or equal to 0.10 in at least one sample.
We selected cancer-associated genes that were included in the Cancer Gene Census 42 and the pan-gynecologic cancer-associated genes. 43

| Identification of putative clonal populations
Putative clonal populations were identified as previously described. 20,21The clonal populations present in three samples were detected by clustering somatic mutations (SNVs and Indels) with PyClone. 44The SNVs and Indels were selected by the following criteria: (a) the depth was greater than or equal to 20 in three samples; (b) the MAF was greater than or equal to 0.1 in at least one sample; and (c) the mutations did not overlap with the SCNA detected by FACETS.Clusters with >15 mutations were used for subsequent analysis.The results of the SNVs and Indels clustering together with their MAFs were used as the input to ClonEvol. 45The polyclonal seeding model was implemented.The number of bootstrap samples was set to 10 000.

| Detection of mutation signatures
The mutation signature was detected as previously described. 21,22We used the identified somatic SNVs of clonal populations detected by PyClone described above for mutational signature analysis.The somatic SNVs were classified into 96 mutation classes defined by the six pyrimidine substitutions (C>A, C>G, C>T, T>A, T>C, and T>G) in combination with the flanking 5 0 and 3 0 bases.The 96-mutation catalog was fitted to a predefined list of known signatures for our mutational signature analysis. 46,47We did not select an approach for de novo signature extraction because the number of somatic mutations was not large enough in this study.We used the COSMIC mutational signatures version 3 as a reference set of known mutational signatures. 48We selected 14 SBS signatures (SBS1, SBS2, SBS3, SBS5, SBS6, SBS10a, SBS10b, SBS13, SBS14, SBS15, SBS26, SBS28, SBS40, and SBS44) with activities in uterus adenocarcinoma on the basis of a previous study. 48We implemented a fitting approach using sigfit. 49 ran four Markov chains with a total of 50 000 iterations, including a burn-in of 25 000 samples.We estimated the highest posterior density (HPD) interval for each of the SBS signatures.We considered the SBS signature as a significantly active signature if the 90% lower end of the HPD interval for an SBS signature was above the threshold (0.01, default value).

| Detection of abnormal MLH1 expression, MSI, and MLH1 p.L582H in three generations
A 27-year-old woman with tumors in the uterine corpus, cervix, and bilateral ovaries was pathologically diagnosed with primary endometrial cancer and exhibited cervical invasion and ovarian metastasis (Figure 1).Her family history suggested Lynch syndrome (Figure 2).To

| Sharing of copy number alterations and MSI at multiple sites in the proband
To clarify the pathogenicity of the VUS in MLH1 and clonal relationships among synchronous tumors, we performed sequencing for the uterine corpus, cervical, and left-side ovarian tumor samples of the proband collected from the locations shown in Figure 1.
Because Sanger sequencing suggested the occurrence of an LOH at the MLH1 p.L582H site, we first analyzed for SCNA in the three samples by FACETS (Figure 4A, Table S2).A broad copy-neutral LOH (CN-LOH) in the short arm of chromosome 3 was detected in all three samples (Figure 4B).MLH1 is located within the CN-LOH region.The MAF of MLH1 p.L582H was 0.57 in blood, 0.96 in corpus, 0.93 in cervix, and 0.97 in ovary samples, supporting the presence of the CN-LOH at the mutation site (Figure S4).The cervix sample showed an additional broad CN-LOH in the short arm of chromosome 2 (Figure 4C).We did not find deleterious somatic mutations in cancer-associated genes other than MLH1 in the CN-LOH regions in chromosomes 2 and 3.
Additionally, the result from MSIsensor showed that all three samples had an MSI score of 20 or higher (Table S3), verifying MSIhigh status.These results suggest that MMR abnormalities were caused from the biallelic loss of the wild-type allele of MLH1 by the germline missense mutation (p.L582H) and the somatic CN-LOH.

| Genomic diversity of somatic mutations at multiple sites in the proband
A total of 3888 SNVs and 2388 Indels were detected in the three samples (Table S4).TMB was 24.Another portion of mutations was shared between corpus and cervix or between corpus and ovary, but no mutations were shared between cervix and ovary (Figure 5A, Table S5).The MAFs of the mutations shared between corpus and cervix or between corpus and ovary were lower in corpus (<0.5) and higher in cervix or ovary (close to 0.5), indicating that these mutations were subclonal in corpus and clonal in cervix or ovary (Figure 5B).Additionally, each sample had unique mutations (Figure 5A, Table S5).These results suggest that the three tumors had the same clonal origin.Moreover, this suggests that the tumor of corpus encompassed two major subclones, and each of the subclones metastasized to the cervix and ovary independently.To detect clonal populations and reconstruct clonal evolution trees, the somatic mutations were clustered on the basis of their MAF with PyClone. 44Six clusters were identified (Figure 6A, Table S6).
Cluster 1 contained the mutations shared among all three samples; this cluster seemed to be the founding clone.The mutations in clusters 2 and 3 were shared between corpus and cervix and between corpus and ovary, respectively.Clusters 4, 5, and 6 comprised the private mutations specific to corpus, cervix, and ovary, respectively.
From the mutational clusters predicted by PyClone, clone ordering was conducted with ClonEvol to reconstruct a clonal evolution tree 45 (Figure 6B).We performed mutational signature analysis separately for each cluster of mutations of clonal populations to examine the evolutionary dynamics of mutational processes in the shaping of the genomes of the three tumors (Figure S5 and Tables S7 and S8).SBS6, SBS15, SBS26, or SBS44, which were associated with defects in DNA MMR, were enriched in all clusters, suggesting that somatic mutations from MMR abnormalities accumulated throughout the evolutionary processes of the three tumors.SBS1, which was a clock-like mutational process attributable to the deamination of 5-methylcytosine at CpG dinucleotides, was significantly enriched in the shared mutations (clusters 1, 2, and 3) but not in the private mutations (clusters 4, 5, and 6).
This finding suggests that the metastatic events occurred later in the evolutionary process, and therefore, not so much time had passed to accumulate significant numbers of clock-like mutations after the metastatic colonization.Furthermore, these data suggest that metastatic colonization occurred earlier from the corpus to the cervix and later from the corpus to the ovary on the basis of the number of private mutations in the metastatic sites.

| DISCUSSION
Distinguishing pathogenic variants from VUSs is one of the major issues in clinical genetic testing.1][52] Thus, we hypothesized that the detection of second events hitting the same gene by sequencing for tumors of a patient with a VUS would provide valuable information to evaluate the pathogenicity of the VUS.In this study, unable to explore second hit events because of the limited amount of DNA from the cancer specimens, it is suggested that second hit events may differ among the family members.Just as in our cases, two pedigrees with germline MLH1 p.L582H that met the Amsterdam II criteria are registered on the Universal Mutation Database. 53In the ClinVar, MLH1 p.L582H was recently changed to pathogenic by a single submitter in November 2022 (accession/variation/VCV 001779284.1). 54The findings by us and other groups further strengthened evidence for MLH1 p.L582H being truly pathogenic.
In the case of cooccurrence of endometrial and ovarian cancer, distinguishing between metastatic cancer and double primary cancer is crucial for determining treatment strategies. 10,11,55With the advancement of genomic analysis, a recent study demonstrated that some cases of Lynch syndrome-associated endometrial and ovarian cancer clinically diagnosed as synchronous primary cancer were, in fact, metastatic cancer.Furthermore, the directionality of metastasis was from endometrial cancer to the ovary in all the analyzed cases. 14By focusing on MAF profiles and mutational signatures across multiple sites, we evaluated clonal relationships among corpus, cervical, and ovarian tumors of the patient by whole-exome sequencing.The somatic mutations detected in all tumors exhibited a high MAF and were classified as cancer initiation-associated mutations, providing compelling finding for the same origin of the three tumors.On the other hand, no mutations were shared exclusively between the cervix and ovary.Moreover, the MAFs of mutations shared between the corpus and the cervix or ovary were lower in the corpus, but higher in the cervix or ovary.Additionally, the mutational signatures associated with defects in the DNA MMR system were consistently detected in the three tumors.These signatures are commonly detected in endometrial cancer with an MMR-deficient phenotype but rarely found in primary cervical and ovarian cancers.These findings suggested that the distinct subclones originating from the corpus metastasized to the cervix and ovary.Such information would enable us to proceed with a treatment strategy for endometrial cancer.
Therefore, genomic analysis to clarify clonal relationships among tumors at multiple sites and identify the primary site will play an important role in making clinical diagnosis.
The mutation profiles of representative cancer-associated genes in endometrial cancer differed among each tumor.PIK3CA p.R88Q, ARID1A pQ2070X, PTEN p.K267RfsX9, and PTEN p.G143AfsX4 were shared among the three tumors with high MAF.This indicated that they were the clonal ancestors that emerged at the corpus and played an essential role in cancer initiation.A subclone shared between the corpus and cervix acquired ARID1A p.R2158X and RB1 p.V654CfsX4, while another subclone shared between the corpus and ovary acquired ARID1A p.Q766SfsX67 and RB1 p.R7PfsX24.Because the ancestral clone that gave rise to all three tumors had ARID1A p.Q2070X, the distinct ARID1A mutations that occurred in the two subclones presumably led to biallelic loss of this tumor suppressor gene.As biallelic rather than monoallelic loss of ARID1A plays an important role in cancer development, 56,57 convergent biallelic loss of ARID1A in the two subclones might render molecular traits that facilitated cancer development and metastatic progression in this patient.
PIK3CA showed marked differences across the tumors.While PIK3CA p.R88Q was shared in the three tumors, PIK3CA p.H1047R and p.H1047Y were found only in the corpus, and cervix, respectively.
These findings align with previous studies suggesting that PIK3CA mutations that occur after tumorigenesis contribute to subclonal formation. 58,59The fact that these essential cancer-associated mutations differed by sampling location is highly informative for elucidating clonal evolution.
Recent studies have demonstrated that cancer-associated gene mutations accumulate in normal endometrium. 21,60,61In Lynch syndrome patients, intestinal, endometrial, and gastric epithelial cells in noncancerous tissues harbor cancer-associated gene mutations. 62nomic analysis of normal endometrial cells as well as tumor cells will further our understanding of the evolutionary process from normal endometrium to cancer and the mechanisms of cancer development.We were unable to obtain normal endometrium for analysis because the tumor was diffusely spread in the uterus.
This study has several limitations.First, a detailed analysis of the pathogenicity of MLH1 p.L582H could not be performed on family samples.Because of sample availability limitations, we could not investigate the second hit of the father's and grandfather's cancer as described above.Additionally, examining mutations in cancer-naive individuals, such as the mother and brother of the proband, is important to evaluate co-segregation of MLH1 p.L582H and Lynch syndrome-associated cancer.Unfortunately, we could not obtain cooperation from cancer-naive individuals in the family and therefore could not examine their mutational status.Second, our analysis was limited to a single case.Lastly, functional analysis to confirm pathogenicity could not be performed because of a lack of available resources such as endometrial organoids from the patients.However, our integrated analysis provided clinically valuable results regarding the pathogenicity of the variant classified as VUS and the diagnosis of synchronous tumors.
In conclusion, our study demonstrated the pathogenicity of MLH1 p.L582H and the clonal origin and genomic diversity in Lynch syndrome-associated endometrial cancer.Our results highlight the importance of genomic analysis for cases with a VUS to assess its pathogenicity and conducting sampling from multiple sites to evaluate clonal relationships in cases of Lynch syndrome-associated synchronous cancers.
assistance.We thank Gabrielle White Wolf, PhD, from Edanz (https:// jp.edanz.com/ac)for editing a draft of this manuscript.

F I G U R E 1
Macroscopic and microscopic findings of tissue sampling sites.(A) The macroscopic findings of a surgical specimen from a 27-year-old woman with endometrial cancer with sampling sites.The tumor was continuous from the uterine corpus to the cervix and metastasized to the bilateral ovaries.The sampling sites are indicated: uterine corpus, cervix, and ovary.Each tick mark represents 1 cm.(B) Microscopic findings of each sample are shown.Scale bars represent 100 μm.

F I G U R E 2
The pedigree of the family.The black squares and circles indicate individuals with Lynch syndrome-associated cancer.The index case is the 27-year-old woman who underwent genetic testing, which detected a MLH1 p.L582H variant.
screen the possibility of Lynch syndrome-associated endometrial cancer and identify the genotype, we first performed immunostaining for four MMR proteins (MLH1, MSH2, MSH6, and PMS2).Immunostaining of endometrial cancer tissue showed the loss of MLH1 and PMS2 expression (Figure3A), suggesting MLH1 mutation.PCR-based MSI assay demonstrated that NR-21, BAT-26, BAT-25, NR-24, and MONO-27 showed length shifts, indicating MSI-high status.Genetic testing for MMR genes by Falco Biosystems Ltd identified a missense variant in MLH1 (p.L582H, c.1745T>A) as a VUS.We verified MLH1 p.L582H in blood and uterine corpus tumor epithelium samples by Sanger sequencing.The Sanger sequencing electropherogram showed that the signal for the mutant allele (A) was predominant in the tumor epithelium sample, which suggested a LOH event at the MLH1 site (Figure3B).To rule out the effect of hypermethylation of the MLH1 promoter, we performed methylation-specific PCR.MLH1 promoter hypermethylation was not observed in the tumor sample (Figure3C).Next, we focused on Lynch syndrome-associated cancers in other generations, including the father's colon cancer and the paternal grandfather's gastric cancer.Immunostaining showed the loss of MLH1 and PMS2 expression in both samples (FigureS2).MSI assay showed that all five microsatellite markers showed length shift, indicating MSI-high status.We conducted Sanger sequencing for the cancer tissues of the father and paternal grandfather, and the noncancerous tissue of the grandfather.The noncancerous tissue of the grandfather was used to evaluate the presence of the germline variant.The results showed MLH1 p.L582H in all the analyzed samples, indicating the VUS of MLH1 was inherited from the paternal side of the family.Unlike the results in the proband, LOH was not suggested in these samples (FigureS3).
3 mutations/megabase (Mb) in corpus, 18.2 mutations/Mb in cervix, and 18.2 mutations/Mb in ovary.A portion of the mutations was shared among the three F I G U R E 3 Molecular analysis of MLH1 in the proband's uterine corpus tumor specimen.(A) Immunostaining for MLH1, MSH2, MSH6, and PMS2.Loss of MLH1 and PMS2 expression was detected in tumor cells.Scale bars represent 100 μm.(B) Sanger sequencing chromatograms of MLH1.The results show MLH1 c.1745T>A.Most MLH1 c.1745T>A sites were mutated A bases in the tumor epithelial sample, indicating loss of heterozygosity in the tumor sample.(C) Methylation-specific PCR for the MLH1 promoter.Control unmethylated and methylated DNA samples are shown in the left and center lanes, respectively.U indicates an unmethylated reaction, and M indicates a methylated reaction.The tumor DNA from the proband shows unmethylation.samples, and the MAFs of these mutations were close to 0.5, indicating these mutations reached clonal states in all three tumors.

F I G U R E 4
Landscape of somatic copy number alterations (SCNA) in the three samples.(A) Genome-wide profiles of SNCA for the three samples.The absolute values of the log odd ratios for the variant allele read count at heterozygous single nucleotide variant (SNV) sites in each pair of epithelium and blood samples are plotted in accordance with chromosome coordinates.The log odds ratios were estimated by FACETS.(B) Loss of heterozygosity (LOH) on chromosome 3 was shared in the three samples.The black wavy line indicates MLH1 c.1745T>A mutation detected at the LOH.(C) LOH on chromosome 2 was detected only in the cervix sample.In (B) and (C), regions with SCNA are highlighted in light green; the numbers separated by a colon are the major and minor copy numbers for the regions indicated by the arrows.Representative shared cancer-associated gene mutations are shown in Figure 5C.The mutations shared by the three samples, including ARID1A p.Q2070X, PIK3CA p.R88Q, and PTEN p.K26RfsX9 and p.G143AfsX4, showed high MAF in all samples.The mutations shared between corpus and cervix, including ARID1A p.R2158X and RB1 p.V654CfsX4, and between corpus and ovary, including ARID1A p.Q766SfsX67 and RB1 p.R7PfsX24, had low MAFs in corpus but high MAFs in cervix or ovary.Notably, these distinct mutations in the same genes (ARID1A and RB1) occurred parallelly in the two subclones.The three tumors shared a nonsense mutation in ARID1A (p.Q2070X); therefore, the parallelly-occurred ARID1A mutations (p.R2158X and p.Q766SfsX67) were thought to lead to biallelic loss of ARID1A in each subclone.Furthermore, there were unique cancer-associated gene mutations, including PIK3CA p.H1047R in corpus and PIK3CA p.H1047Y in cervix.
The three samples had a common ancestral clone characterized by the mutations grouped into cluster 1 (PIK3CA [p.R88Q], ARID1A [p.Q2070X], and PTEN [p.K267RfsX9 and p.G143AfsX4]).Driven by the biallelic loss of MLH1 from the germline missense mutation and the CN-LOH at chromosome 3, this ancestral clone was thought to emerge in the corpus and play an important role in cancer development.Two subclones (2 and 3) were estimated to be derived from the ancestral clone 1 in the corpus and metastasized to the cervix and ovary independently.These two clones acquired mutations in the same tumor suppressor genes (ARID1A [p.R2158X] and RB1 [p.V654CfsX4] in clone 2, and ARID1A [p.Q766SfsX67] and RB1 [p.R7PfsX24] in clone 3), suggesting that the convergent evolution of loss of these tumor suppressor genes might be involved in the F I G U R E 5 Sharing of somatic mutation profiles defined by multiregional sequencing.(A) Sharing pattern of somatic single nucleotide variants (SNVs) and short insertions/deletions (Indels) among the three samples.Color density indicates the mutant allele frequency (MAF) of each somatic mutation.(B) Distribution of MAF of the SNVs and Indels as follows: (i) shared in three samples; (ii) shared between corpus and cervix samples; and (iii) shared between corpus and ovary samples.Statistical significance of differences was assessed using the Wilcoxon-Mann-Whitney test.(C) Heatmaps demonstrate the cancer-associated gene mutations.Color and density indicate the type and MAF of each somatic mutation, respectively.acquisition of a metastatic phenotype.Finally, each tumor acquired unique private mutations represented by independent PIK3CA mutations (p.H1047R in the corpus and p.H1047Y in the cervix), which might render a more aggressive phenotype.
we examined the pathogenicity of a VUS in MLH1 (p.L582H) that was inherited in three family members (the proband, her father, and her paternal grandfather) with Lynch syndrome-associated cancers.The cancer samples of the three family members consistently exhibited an MSI-high phenotype and the loss of MLH1 and PMS2 expressions.The result of Sanger sequencing showed that the signal of the mutant allele of the VUS was predominant in the endometrial cancer of the proband, suggesting the presence of an LOH at the mutation site.Whole-exome sequencing for tumors in the uterine corpus, cervix, and ovary of the proband identified a spatially shared CN-LOH in the short arm of chromosome 3 encompassing MLH1, indicating that F I G U R E 6 Clonal relationships among the three samples.(A) Somatic single nucleotide variants (SNVs) and short insertions/deletions (Indels) were classified into six clusters on the basis of their mutant allele frequency (MAF) profiles with PyClone.The prevalence of the six clones in each of the samples was evaluated; clonal ordering was then implemented with ClonEvol to create fish plots showing clonal evolutions within the samples.(B) A branch-based clonal evolution tree was generated with ClonEvol.Mutations in cancer-associated genes were assigned to branches based on the result of mutation clustering by PyClone.MLH1 p.L582H event was manually assigned at the top of the clonal evolution tree because it was a germline mutation.The loss of heterozygosity (LOH) event on chromosome 3 was manually assigned to the initial branch because it was shared at a clonal state among all three samples.The LOH event on chromosome 2 was manually assigned to the cervix branch because it was observed only in the cervix sample.The lengths of the branches are associated with the number of somatic mutations.theCN-LOH was a second hit and led to the complete loss of MLH1 and an MMR-deficient phenotype.Notably, whole-exome sequencing provided additional evidence of the MMR-deficient phenotype in the three tumors by showing that MSI was present in a larger number of microsatellite regions throughout the genome and significant proportions of the somatic SNVs were accounted for by the mutational signatures SBS6, SBS15, SBS26, and SBS44 that were associated with post-replicative MMR deficiency.The electropherogram of Sanger sequencing for tumors of the father and paternal grandfather did not suggest the presence of LOH at MLH1 p.L582H.While we were