Genomic characterization of uterine adenomatoid tumors reveals recurrent alterations in KMT5A and KMT2C.

Genomic DNA was extracted from formalin-fixed paraffin-embedded (FFPE) tumor tissues using the VAHTS ® Universal Pro DNA Library Prep Kit for Illumina (Vazyme, China) in combination with the iGene Tech target capture kit, strictly following the manufacturers’ protocols. DNA purity and concentration were evaluated using a NanoDrop spectrophotometer and a Qubit fluorometer. NGS panel comprising 641 genes (1.7 Mb) was used to identify somatic mutations. The panel included genes involved in chromatin remodeling ( KMT5A , KMT2C ), immune regulation ( HLA-A , HLA-B ), oncogenic signaling, and DNA repair. DNA libraries were prepared using the Agilent SureSelect XT Target Enrichment Kit, and sequencing was performed on the Illumina NovaSeq platform, achieving a mean coverage depth of ≥ 1000× per gene.

Raw sequencing data underwent quality control using FastQC to assess the integrity of the reads. Low-quality reads were trimmed using Trimmomatic [ 23 ]. The processed data were aligned to the human genome (HG19) using the Burrows-Wheeler Aligner (BWA) [ 24 ]. Variants were identified using the Genome Analysis Toolkit (GATK) and annotated using ANNOVAR, leveraging the dbSNP database to identify known mutations [ 25 , 26 ]. Variant pathogenicity was classified according to the guidelines set forth by the American College of Medical Genetics and Genomics (ACMG) [ 27 ].

Key recurrent mutations identified in KMT5A and KMT2C by targeted sequencing were further validated using Sanger sequencing to confirm their authenticity. Primers flanking the mutation sites were designed using standard primer design tools, and PCR amplification was performed using Platinum Taq DNA Polymerase (Thermo Fisher Scientific) according to the manufacturer’s instructions. The PCR products were purified and subjected to bidirectional sequencing on an ABI 3730xl DNA Analyzer (Applied Biosystems). Sequencing chromatograms were inspected to confirm the presence of the corresponding mutations. To place the genomic alterations identified in UATs into a broader molecular context, we analyzed publicly available multi-omics datasets from the cBioPortal for Cancer Genomics platform ( https://www.cbioportal.org ) [ 36 ]. Genomic data were retrieved from several uterine cancer cohorts, including datasets from The Cancer Genome Atlas (TCGA), Memorial Sloan Kettering Cancer Center (MSK), and the Clinical Proteomic Tumor Analysis Consortium (CPTAC). These datasets collectively include multiple uterine tumor types, such as uterine corpus endometrial carcinoma, uterine carcinosarcoma, uterine clear cell carcinoma, uterine endometrioid carcinoma, and uterine sarcoma/mesenchymal tumors, as defined in the original studies. Eight genes recurrently altered in our UAT cohort— KMT5A , KMT2C , HLA-B , ROS1, HLA-A, TRAF7 , SDHA, and KMT2A—were selected for comparative analysis. For each gene, we examined mutation frequency, variant types, and available transcriptomic data across the selected cohorts. Somatic mutations, copy number alterations (CNAs), mRNA expression profiles, and associated clinical annotations were analyzed to evaluate the prevalence and distribution of these alterations in different uterine tumor types. This cross-cohort analysis was performed to determine whether the recurrent alterations observed in UATs are also present in other gynecologic malignancies and to provide a comparative genomic context for the findings in this study.

Clinical characteristics of the 20 patients with uterine adenomatoid tumors are summarized in Table  1 . All cases were pathologically confirmed UATs. The cohort consisted of 20 premenopausal women with a mean age of 39.9 years (range, 26–54 years). All tumors were confined to the myometrium and ranged in size from 2.5 to 11 cm (median diameter, 4 cm). All specimens represented solitary UAT lesions; no cases demonstrated multiple distinct adenomatoid tumors within the same surgical specimen. Table 1 Clinicopathological information of the patients with UATs Sample Age (years) Associated pathologies Size (cm) 1 40 / 7 2 29 / 6 3 46 / 3.5 4 39 Adenomyosis 5.6 5 41 / 5 6 40 Leiomyoma 4 7 44 Leiomyoma 3 8 26 Mesosalpinx-associated paramesonephric duct cyst 5 9 54 / 3 10 39 Adenomyosis 2.5 11 43 Ovarian endometriotic cyst 4 12 40 Ovarian endometriotic cyst 4.5 13 37 Ovarian endometriotic cyst, Mesosalpinx-associated paramesonephric duct cyst 4 14 42 Mesosalpinx-associated paramesonephric duct cyst 4 15 49 Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst 3.5 16 32 Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst 11 17 35 Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst 7 18 37 Leiomyoma 10 19 44 Leiomyoma 9 20 41 Adenomyosis 8 Clinicopathological information of the patients with UATs Ovarian endometriotic cyst, Mesosalpinx-associated paramesonephric duct cyst Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst Leiomyoma, Mesosalpinx-associated paramesonephric duct cyst Histologically, hematoxylin and eosin (H&E)-stained sections demonstrated well-circumscribed nodules composed of cuboidal to flattened mesothelial-like cells. The tumor cells exhibited bland nuclear features, minimal cytologic atypia, and rare mitotic figures. Architecturally, the lesions displayed glandular, trabecular, and slit-like growth patterns within a fibrous to myxoid stroma, consistent with the established morphologic features of adenomatoid tumors. Regarding concurrent gynecological conditions, 25% ( n  = 5) of patients presented with isolated UATs, whereas the remaining 75% ( n  = 15) had coexisting gynecological pathologies. These included uterine leiomyomas (20%, n  = 4), adenomyosis (15%, n  = 3), and ovarian endometriotic cysts (10%, n  = 2). Fallopian tube cysts were observed in 3 cases (15%). Among these, 10% ( n  = 2) had isolated fallopian tube cysts, and 5% ( n  = 1) had concurrent ovarian endometriotic cysts and fallopian tube cysts. The potential biological relationship between UAT and these concomitant lesions remains uncertain and requires further investigation. Targeted panel sequencing revealed recurrent somatic alterations across the 20 UAT samples analyzed. Mutations in KMT5A and KMT2C were detected in all cases (20/20, 100%) (Fig.  1 A). Other frequently altered genes included HLA-B (13/20, 65%), HLA-A (10/20, 50%), ROS1 (9/20, 45%), TRAF7 (9/20, 45%), SDHA (6/20, 30%), KMT2A (6/20, 30%), and HLA-C (6/20, 30%) (Fig.  1 B). The waterfall plot illustrates the mutational distribution across individual tumors. Tumor mutational burden (TMB), defined as the number of somatic mutations per megabase (Mb) of coding sequence covered by the sequencing panel, had a median value corresponding to 17 detected variants per tumor (range: 5–121), with missense mutations representing the predominant alteration type (Fig.  1 B). Fig. 1 Mutation landscape of UATs. A OncoPrint-style mutation heatmap showing the distribution of mutated genes across 20 UAT samples. Red squares represent mutated genes, while white squares indicate wild-type status. Genes with higher mutation frequencies (e.g., KMT5A , KMT2C , HLA-B , ROS1 , HLA-A , TRAF7 ) are labeled. B Waterfall plot displaying tumor mutation burden (TMB, top bar), mutation types (green: missense mutation; black: multi-hit) and frequency distribution of mutated genes (right bar). C Bar plot of the top 10 most frequently mutated genes, with KMT5A and KMT2C mutated in all samples (100%), followed by HLA-B (65%), HLA-A (50%), ROS1 (45%), and TRAF7 (45%). D Variants per sample, with a median of 17 variants. E Overall variant classification, showing predominance of missense mutations. F Box plot summarizing the number of variants for each mutation classification type. G Variant type distribution, with SNPs being the most common Mutation landscape of UATs. A OncoPrint-style mutation heatmap showing the distribution of mutated genes across 20 UAT samples. Red squares represent mutated genes, while white squares indicate wild-type status. Genes with higher mutation frequencies (e.g., KMT5A , KMT2C , HLA-B , ROS1 , HLA-A , TRAF7 ) are labeled. B Waterfall plot displaying tumor mutation burden (TMB, top bar), mutation types (green: missense mutation; black: multi-hit) and frequency distribution of mutated genes (right bar). C Bar plot of the top 10 most frequently mutated genes, with KMT5A and KMT2C mutated in all samples (100%), followed by HLA-B (65%), HLA-A (50%), ROS1 (45%), and TRAF7 (45%). D Variants per sample, with a median of 17 variants. E Overall variant classification, showing predominance of missense mutations. F Box plot summarizing the number of variants for each mutation classification type. G Variant type distribution, with SNPs being the most common Mutation frequency analysis of the top 10 altered genes reaffirmed the universal presence of KMT5A and KMT2C (Fig.  1 C). Distribution analysis showed that most tumors carried fewer than 30 variants, whereas one hypermutated sample exceeded 100 variants (Fig.  1 D). Variant classification confirmed that missense mutations accounted for the majority of alterations, followed by nonsense and nonstop mutations, with missense variants exhibiting the highest counts per tumor (Fig.  1 E, F). In terms of variant type, single nucleotide polymorphisms (SNPs) overwhelmingly predominated across the cohort (Fig.  1 G). Analysis of base substitution patterns revealed that C > T transitions were the predominant substitution type in UATs, followed in frequency by T > C, C > G, C > A, T > G, and T > A substitutions (Fig.  2 A and B). The stacked bar plot demonstrated that this mutational spectrum was largely consistent across samples, with C > T transitions contributing the highest proportion of mutations in nearly all tumors (Fig.  2 A). Quantitatively, C > T transitions were observed 187 times, accounting for nearly half of all detected substitutions, whereas T > C ( n  = 90) and C > G ( n  = 62) occurred less frequently (Fig.  2 B). Boxplot analysis further confirmed that C > T transitions exhibited the highest median proportion among all substitution types (Fig.  2 C). Transition/transversion (Ti/Tv) analysis showed a clear predominance of transitions over transversions, with median proportions of approximately 75% and 25%, respectively (Fig.  2 D). Fig. 2 Mutational spectrum and representative variants in UATs. A Stacked bar plot showing the distribution of base substitution types in each UAT sample, with C > T transitions predominating across the cohort. B Overall distribution of base substitution types in the cohort, showing C > T as the most frequent ( n  = 187), followed by T > C ( n  = 90), C > G ( n  = 62), C > A, T > G, and T > A. C Boxplot illustrating the proportion of each base substitution type per tumor, confirming that C > T transitions have the highest median proportion. D Transition/transversion (Ti/Tv) analysis showing a predominance of transitions (~ 75%) over transversions (~ 25%). E, F Sanger sequencing validation of representative KMT2C mutations, including c.C2656T and c.A2917G. G – I Sanger sequencing validation of representative KMT5A mutations, including c.T995C, c.A719C, and c.G713C Mutational spectrum and representative variants in UATs. A Stacked bar plot showing the distribution of base substitution types in each UAT sample, with C > T transitions predominating across the cohort. B Overall distribution of base substitution types in the cohort, showing C > T as the most frequent ( n  = 187), followed by T > C ( n  = 90), C > G ( n  = 62), C > A, T > G, and T > A. C Boxplot illustrating the proportion of each base substitution type per tumor, confirming that C > T transitions have the highest median proportion. D Transition/transversion (Ti/Tv) analysis showing a predominance of transitions (~ 75%) over transversions (~ 25%). E, F Sanger sequencing validation of representative KMT2C mutations, including c.C2656T and c.A2917G. G – I Sanger sequencing validation of representative KMT5A mutations, including c.T995C, c.A719C, and c.G713C Sanger sequencing was performed to validate selected recurrent variants in KMT5A and KMT2C . In KMT2C , recurrent mutations included c.C2656T and c.A2917G (Fig.  2 E, F), whereas in KMT5A , confirmed variants included c.T995C, c.A719C, and c.G713C (Fig.  2 G–I). Collectively, these results define a mutational spectrum in UATs dominated by C > T transitions, a high Ti/Tv ratio, and recurrent, sequence-confirmed alterations in KMT2C and KMT5A . Lollipop plots were generated to depict the positional distribution of missense mutations within the most frequently altered genes in UATs. KMT5A mutations (mutation rate: 100%) were exclusively missense variants, predominantly clustered within the SET catalytic domain (Fig.  3 A). Likewise, KMT2C mutations (100%) were all missense alterations and were dispersed across multiple functional domains, including zf-HC5HC2H, PHD, HMG-box, FYRN, and SET (Fig.  3 B). Fig. 3 Distribution of missense mutations in frequently altered genes of UATs. A – H Lollipop plots showing the positional distribution of missense mutations within the protein domains of the most frequently mutated genes in UATs. A KMT5A mutations (100%) are predominantly clustered within the SET catalytic domain. B KMT2C mutations (100%) are dispersed across multiple domains, including zf-HC5HC2H, PHD, HMG-box, FYRN, and SET. C HLA-B (65%) and D HLA-A (50%) mutations are concentrated in the MHC_I and IgC_MHC_I_alpha3 domains. E ROS1 mutations (45%) are located in FN3 domain-containing regions. F TRAF7 mutations (45%) are enriched in functional motifs such as zf-TRAF, DD, GKI_COG2319, and WD40. G SDHA mutations (30%) are mapped to the NADB_Rossmann, PTZ00139 , and Succ_DH_flav_C domains. H KMT2A mutations (30%) are distributed across the zf-CXXC, FYRN, FYRC, and SET domains. All mutations are missense variants, many localized within key catalytic or structural regions, suggesting potential impacts on protein function and tumor pathogenesis Distribution of missense mutations in frequently altered genes of UATs. A – H Lollipop plots showing the positional distribution of missense mutations within the protein domains of the most frequently mutated genes in UATs. A KMT5A mutations (100%) are predominantly clustered within the SET catalytic domain. B KMT2C mutations (100%) are dispersed across multiple domains, including zf-HC5HC2H, PHD, HMG-box, FYRN, and SET. C HLA-B (65%) and D HLA-A (50%) mutations are concentrated in the MHC_I and IgC_MHC_I_alpha3 domains. E ROS1 mutations (45%) are located in FN3 domain-containing regions. F TRAF7 mutations (45%) are enriched in functional motifs such as zf-TRAF, DD, GKI_COG2319, and WD40. G SDHA mutations (30%) are mapped to the NADB_Rossmann, PTZ00139 , and Succ_DH_flav_C domains. H KMT2A mutations (30%) are distributed across the zf-CXXC, FYRN, FYRC, and SET domains. All mutations are missense variants, many localized within key catalytic or structural regions, suggesting potential impacts on protein function and tumor pathogenesis In immune-related genes, HLA-B (65%) and HLA-A (50%) mutations were concentrated within the MHC_I and IgC_MHC_I_alpha3 domains, suggesting potential impacts on antigen presentation (Fig.  3 C, D). For ROS1 (45%), missense mutations were distributed across regions containing FN3 domains (Fig.  3 E), while TRAF7 (45%) mutations were enriched in functional motifs such as zf-TRAF, DD, GKI_COG2319, and WD40, which are involved in signal transduction and protein–protein interactions (Fig.  3 F). Among metabolic and chromatin-modifying genes, SDHA (30%) mutations were located within the NADB_Rossmann, PTZ00139 , and Succ_DH_flav_C domains (Fig.  3 G), and KMT2A (30%) mutations were mapped to the zf-CXXC, FYRN, FYRC, and SET domains (Fig.  3 H). All identified alterations were non-random, domain-enriched distributions of missense mutations within frequently altered UAT genes, suggesting targeted disruption of specific functional modules critical for histone modification, antigen presentation, signaling, and metabolism. Histopathological examination of three representative UAT samples (Samples 1–3) revealed characteristic morphological features, including glandular structures embedded within fibrous and smooth muscle stroma, as demonstrated by H&E staining (Fig.  4 A). The clinicopathological data of these patients are summarized in Fig.  4 B, with ages ranging from 29 to 46 years and tumor sizes measuring 3.5–7 cm. Fig. 4 Histopathological features and immunohistochemical characterization of KMT5A -mutated UATs. A Representative H&E staining of three independent UAT samples (Samples 1–3), showing characteristic gland-like structures embedded within fibrous and smooth muscle stroma. Scale bars: 100 μm (left panels) and 50 μm (right panels). B Clinicopathological characteristics of the three representative cases, including patient age, diagnosis, and tumor size. C Representative immunohistochemical staining of KMT5A , TRAF7 , γ- H2AX , CALB2 , and Ki-67 in UAT samples stratified by low versus high KMT5A expression. The images illustrate typical staining patterns observed in the respective groups. Scale bars: 100 μm (left panels) and 50 μm (right panels). D Semi-quantitative analysis of immunohistochemical staining using the H-score method. Compared with the low KMT5A expression group, the high-expression group showed higher staining scores for TRAF7 , γ- H2AX , CALB2 , and Ki-67 . Data are presented as mean ± SD. * P  < 0.05, ** P  < 0.01 Histopathological features and immunohistochemical characterization of KMT5A -mutated UATs. A Representative H&E staining of three independent UAT samples (Samples 1–3), showing characteristic gland-like structures embedded within fibrous and smooth muscle stroma. Scale bars: 100 μm (left panels) and 50 μm (right panels). B Clinicopathological characteristics of the three representative cases, including patient age, diagnosis, and tumor size. C Representative immunohistochemical staining of KMT5A , TRAF7 , γ- H2AX , CALB2 , and Ki-67 in UAT samples stratified by low versus high KMT5A expression. The images illustrate typical staining patterns observed in the respective groups. Scale bars: 100 μm (left panels) and 50 μm (right panels). D Semi-quantitative analysis of immunohistochemical staining using the H-score method. Compared with the low KMT5A expression group, the high-expression group showed higher staining scores for TRAF7 , γ- H2AX , CALB2 , and Ki-67 . Data are presented as mean ± SD. * P  < 0.05, ** P  < 0.01 Although both KMT5A and KMT2C exhibited high mutation frequencies in UATs, only KMT5A was selected for IHC validation and functional exploration in this study, as KMT5A mutations appeared relatively unique to UATs, whereas KMT2C showed a high mutation rate across multiple cancer types. Although both KMT5A and KMT2C exhibited high mutation frequencies in UATs, KMT5A was prioritized for immunohistochemical validation and functional exploration. Subsequent analyses using publicly available datasets suggested that KMT5A alterations were relatively uncommon in other gynecologic tumors compared with the recurrent alterations observed in UATs. Therefore, KMT5A was selected for further investigation to explore its potential biological relevance in UAT pathogenesis. To investigate protein expression patterns associated with KMT5A mutations, IHC staining was performed for KMT5A , TRAF7 , γ-H2AX , CALB2 , and Ki-67 in KMT5A -mutated UATs stratified by low versus high KMT5A expression (Fig.  4 C). Tumors with high KMT5A expression exhibited markedly stronger staining for TRAF7 , γ-H2AX , CALB2 , and Ki-67 compared with those in the low-expression group. Quantitative analysis indicated higher H-scores for all four markers in the high-expression group, with statistically significant differences observed for TRAF7 , γ-H2AX , CALB2 , and Ki-67 (Fig.  4 D). These findings link elevated KMT5A expression in UATs to increased proliferative activity and altered expression of functionally diverse markers, suggesting potential downstream consequences of KMT5A dysregulation. To elucidate the functional role of KMT5A in ISK cells, three independent shRNAs were employed to silence KMT5A expression. Quantitative RT-PCR analysis confirmed significant reductions in KMT5A mRNA levels in the Sh1 and Sh2 groups compared with the NC (Fig.  5 A), and Western blotting further validated a marked decrease in KMT5A protein abundance across all knockdown groups (Fig.  5 B–C). Fig. 5 KMT5A knockdown suppresses proliferation, migration, and colony formation in UAT cells. A Quantitative RT-PCR analysis showing KMT5A mRNA levels in NC and shRNA knockdown groups (Sh1, Sh2). B , C Western blot analysis and quantification of KMT5A protein levels in NC and three knockdown groups (Sh1, Sh2, Sh3), with GAPDH as the loading control. D Representative images of EdU incorporation assays showing proliferating cells (red) and nuclei (DAPI, blue) in NC, Sh1, and Sh2 groups. E CCK-8 cell proliferation curves for NC, Sh1, and Sh2 groups over 96 h. F Quantification of EdU-positive cell proportions in NC, Sh1, and Sh2 groups. G Representative images of Transwell migration assays showing migrated cells in NC, Sh1, and Sh2 groups (crystal violet staining). H Quantification of migrated cell numbers. I Representative images of colony formation assays in NC, Sh1, and Sh2 groups. J Quantification of relative colony counts. Data are presented as mean ± SD; P  < 0.05, * P  < 0.001, ** P  < 0.0001 KMT5A knockdown suppresses proliferation, migration, and colony formation in UAT cells. A Quantitative RT-PCR analysis showing KMT5A mRNA levels in NC and shRNA knockdown groups (Sh1, Sh2). B , C Western blot analysis and quantification of KMT5A protein levels in NC and three knockdown groups (Sh1, Sh2, Sh3), with GAPDH as the loading control. D Representative images of EdU incorporation assays showing proliferating cells (red) and nuclei (DAPI, blue) in NC, Sh1, and Sh2 groups. E CCK-8 cell proliferation curves for NC, Sh1, and Sh2 groups over 96 h. F Quantification of EdU-positive cell proportions in NC, Sh1, and Sh2 groups. G Representative images of Transwell migration assays showing migrated cells in NC, Sh1, and Sh2 groups (crystal violet staining). H Quantification of migrated cell numbers. I Representative images of colony formation assays in NC, Sh1, and Sh2 groups. J Quantification of relative colony counts. Data are presented as mean ± SD; P  < 0.05, * P  < 0.001, ** P  < 0.0001 EdU incorporation assays revealed a pronounced reduction in the proportion of proliferating cells following KMT5A knockdown, as evidenced by diminished red fluorescence in both Sh1 and Sh2 groups relative to NC (Fig.  5 D). Quantitative analysis confirmed significantly lower EdU-positive cell proportions in the knockdown groups (Fig.  5 F). Consistently, CCK-8 assays demonstrated that KMT5A depletion significantly inhibited cell growth over a 96-hour period (Fig.  5 E). Transwell migration assays indicated that silencing KMT5A markedly impaired cell migratory capacity, with significantly fewer migrated cells observed in the Sh1 and Sh2 groups compared with NC (Fig.  5 G–H). Similarly, colony formation assays showed a substantial decrease in both the number and size of colonies in the knockdown groups (Fig.  5 I–J). Thus, genetic silencing of KMT5A robustly inhibits cell proliferation, migratory capacity, and clonogenic survival. To place the genomic alterations identified in UATs into a broader molecular context, we analyzed publicly available datasets of gynecologic malignancies retrieved from the cBioPortal for Cancer Genomics platform and The Cancer Genome Atlas (TCGA). The combined cohort included 4,699 cases across multiple tumor types. Mutation frequency analysis revealed that KMT5A alterations were relatively uncommon in these gynecologic malignancies, occurring in approximately 1.7% of cases (80/4,699). In contrast, alterations in KMT2C (11%, 517/4,699), KMT2A (7%, 329/4,699), and ROS1 (5%, 235/4,699) were observed at higher frequencies (Fig.  6 A). Additional genes showing detectable alterations included SDHA (4%, 188/4,699), HLA-B (1.8%, 85/4,699), TRAF7 (1.8%, 85/4,699), and HLA-A (1.4%, 66/4,699). Most alterations corresponded to missense mutations, with occasional truncating, splice-site, or structural variants. Fig. 6 Mutation profiles of eight frequently altered genes in gynecologic cancers. ( A ) Oncoprint showing mutation types and frequencies of KMT5A , KMT2C , HLA-B , ROS1 , HLA-A , SDHA , TRAF7 , and KMT2A across uterine and endometrial cancer subtypes in cBioPortal cohorts. ( B–E ) Mutation frequencies and types in UCEC, OV, USC, and CESC Mutation profiles of eight frequently altered genes in gynecologic cancers. ( A ) Oncoprint showing mutation types and frequencies of KMT5A , KMT2C , HLA-B , ROS1 , HLA-A , SDHA , TRAF7 , and KMT2A across uterine and endometrial cancer subtypes in cBioPortal cohorts. ( B–E ) Mutation frequencies and types in UCEC, OV, USC, and CESC Further analysis using cBioPortal demonstrated heterogeneous mutation patterns of the eight candidate genes ( KMT5A , KMT2C , KMT2A , SDHA , HLA-A , HLA-B , TRAF7 , and ROS1 ) across several gynecologic malignancies, including uterine sarcoma, uterine corpus endometrial carcinoma (UCEC), and uterine carcinosarcoma (UCS), and other endometrial cancer subtypes (Supplementary Fig.  1 A–H). Subtype-specific analysis of TCGA datasets revealed variable mutation frequencies among different tumor types. In uterine corpus endometrial carcinoma (UCEC, n  = 511), KMT2C mutations were detected in 17% of tumors, followed by KMT2A (14%), ROS1 (13%), SDHA (4%), KMT5A (3%), TRAF7 (3%), HLA-B (2%), and HLA-A (2%) (Fig.  6 B). In ovarian cancer (OV, n  = 407), KMT2C and KMT2A mutations were observed in 4% and 3% of cases, respectively, whereas ROS1 mutations occurred in approximately 1% of tumors and no KMT5A mutations were detected (Fig.  6 C). In uterine serous carcinoma (USC, n  = 57), KMT2C mutations were present in 7% of tumors, followed by ROS1 (4%) and HLA-B (2%) (Fig.  6 D). In cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC, n  = 80), a single ROS1 mutation (1.25%) was identified, while mutations in KMT5A and several other candidate genes were not detected (Fig.  6 E). Collectively, these analyses suggest that KMT5A alterations are relatively uncommon across gynecologic malignancies, whereas KMT2C alterations are observed across multiple tumor types. Pan-cancer transcriptomic analysis suggested that KMT5A and KMT2C expression levels vary across multiple tumor types compared with matched normal tissues. In the analyzed datasets, KMT5A expression appeared lower in UCEC and USC, while KMT2C expression was also reduced in these tumor types (Supplementary Fig.  2 A, B). Kaplan–Meier survival analysis in the UCEC cohort showed that higher KMT5A expression was not significantly associated with overall survival (HR = 0.79, 95% CI: 0.51–1.21, p  = 0.28) (Supplementary Fig.  2 C). In contrast, higher KMT2C expression was significantly associated with longer overall survival in this cohort (HR = 0.57, 95% CI: 0.37–0.90, p  = 0.015) (Supplementary Fig.  2 D). These results suggest that KMT2C expression may have potential prognostic relevance in UCEC, whereas the association between KMT5A expression and patient survival was not statistically significant. We next characterized the mutational landscape of KMT5A and other frequently altered genes in gynecologic malignancies using cBioPortal visualizations. Lollipop plots revealed that KMT5A mutations were predominantly located within the SET domain, with T307I identified as a recurrent variant (Fig.  7 A). KMT2C harbored multiple frameshift alterations, including K2797Rfs26, E2798Gfs11, and K2797Gfs11 (Fig.  7 B). HLA-B and HLA-A mutations were enriched within the MHC_I and C1-set domains, with P209Qfs5 and R205C/H as notable recurrent sites (Fig.  7 C–D). ROS1 alterations were primarily missense mutations, exemplified by the hotspot R1311Q (Fig.  7 E). KMT2A mutations frequently occurred in the PHD and SET domains (Fig.  7 F). TRAF7 mutations clustered within the WD40 repeat and zf-TRAF domains, with A606V detected as a recurrent alteration (Fig.  7 G), whereas SDHA mutations were concentrated in the FAD_binding_2 and Succ_DH_flav_C domains (Fig.  7 H). Fig. 7 Mutation distribution and prognostic significance of frequently altered genes in gynecologic malignancies. A – H Lollipop plots showing the distribution and protein domain locations of mutations in KMT5A A , KMT2C B , HLA-B C , HLA-A D , ROS1 E , KMT2A F , TRAF7 G , and SDHA H across gynecologic malignancies in the cBioPortal database. Recurrent and representative hotspot mutations are annotated. I – N Kaplan–Meier overall survival curves comparing patients with altered versus unaltered gene status for KMT5A I , KMT2C J , TRAF7 K , ROS1 L , HLA-A M , and HLA-B N in TCGA cohorts. Log-rank p -values are indicated. Alterations in KMT2C , ROS1 , HLA-A , and HLA-B were significantly associated with poorer overall survival, whereas KMT5A and TRAF7 alterations showed no significant prognostic effect Mutation distribution and prognostic significance of frequently altered genes in gynecologic malignancies. A – H Lollipop plots showing the distribution and protein domain locations of mutations in KMT5A A , KMT2C B , HLA-B C , HLA-A D , ROS1 E , KMT2A F , TRAF7 G , and SDHA H across gynecologic malignancies in the cBioPortal database. Recurrent and representative hotspot mutations are annotated. I – N Kaplan–Meier overall survival curves comparing patients with altered versus unaltered gene status for KMT5A I , KMT2C J , TRAF7 K , ROS1 L , HLA-A M , and HLA-B N in TCGA cohorts. Log-rank p -values are indicated. Alterations in KMT2C , ROS1 , HLA-A , and HLA-B were significantly associated with poorer overall survival, whereas KMT5A and TRAF7 alterations showed no significant prognostic effect Survival analysis revealed heterogeneous prognostic associations across these genetic alterations. KMT5A alterations were not significantly associated with overall survival (OS) (log-rank p  = 0.282; Fig.  7 I). In contrast, KMT2C alterations showed a significant association with OS (log-rank p  = 1.315 × 10⁻⁴; Fig.  7 J), with the altered group exhibiting a more favorable survival pattern than the unaltered group. Similarly, alterations in ROS1 , HLA-A , and HLA-B were also significantly associated with OS (Figs.  7 L–N), whereas TRAF7 alterations showed no significant correlation with survival ( p  = 0.279; Fig.  7 K). Overall, these findings suggest that selected genomic alterations, particularly KMT2C and several immune-related genes, may have potential prognostic relevance in gynecologic malignancies.

Somatic variant identification was performed using a multi-step filtering strategy to ensure high-confidence mutation calls. Variants were first required to have a minimum variant allele frequency (VAF) greater than 0.4%, consistent with the detection sensitivity of high-depth next-generation sequencing platforms for low-frequency variants [ 28 – 30 ]. This relatively low threshold was permitted due to the high sequencing depth and the application of stringent downstream filtering criteria. Potential germline variants were removed by comparison with population variant databases, including the 1000 Genomes Project and ExAC/gnomAD. Variants with population allele frequencies greater than 5% were excluded to eliminate common polymorphisms [ 31 ]. Additional quality-control filters were applied to minimize technical artifacts and improve variant reliability. Variants located within repetitive genomic regions, including short tandem repeat (STR) loci, were excluded. Variants demonstrating significant strand bias or poor mapping quality were also removed. Synonymous variants and variants lacking functional annotation were discarded. Missense variants predicted to be benign were further filtered using in silico prediction tools, including SIFT, PolyPhen-2, and CADD, which are widely used for functional impact prediction of genetic variants [ 32 – 34 ]. Because DNA derived from FFPE samples is susceptible to cytosine deamination artifacts, particular attention was paid to C > T/G > A transitions, and suspicious variants were manually reviewed using Integrative Genomics Viewer (IGV) [ 35 ]. Variants consistent with typical FFPE-induced artifacts were excluded.

This study included 20 patients diagnosed with UATs at Tongji University Obstetrics and Gynecology Hospital (2023.03-2024.03). The diagnosis of UAT was confirmed by three independent pathologists based on histopathological and immunohistochemical (IHC) criteria. All patients provided written informed consent, and the study was approved by the Institutional Ethics Committee (Approval No. KS2141). Clinical data, including age, tumor size, location, and associated gynecologic pathologies, were collected and analyzed.

In summary, this study provides a comprehensive characterization of the mutational landscape of UATs. Recurrent alterations affecting chromatin-regulatory genes—particularly the frequent co-occurrence of KMT5A and KMT2C mutations—suggest that epigenetic regulatory pathways may contribute to the molecular features of this rare mesothelial neoplasm. Additional alterations in immune-related genes further indicate that multiple biological processes may participate in its development. Although functional experiments suggested a potential role for KMT5A in cellular proliferation and migration, these findings should be interpreted cautiously due to the use of a non–lineage-matched experimental model. Further studies using larger cohorts and lineage-appropriate systems will be necessary to clarify the biological significance of these alterations.

This study provides a comprehensive characterization of the mutational landscape of UATs and explores the potential biological relevance of recurrent alterations involving KMT5A and KMT2C in this rare mesothelial neoplasm. Notably, both KMT5A and KMT2C mutations were detected in all UAT cases analyzed in the present cohort. Such a high prevalence has not been widely reported in other gynecologic tumors and suggests that these alterations may represent a distinctive genomic feature of UATs. In addition, mutations affecting genes such as KMT2A , ROS1 , HLA-B , and HLA-A were also observed, indicating that multiple genomic alterations may coexist within these tumors. A central observation of this study was the recurrent presence of KMT5A and KMT2C mutations. KMT5A encodes a histone methyltransferase responsible for histone H4 lysine 20 (H4K20) methylation, a modification involved in chromatin organization, DNA replication, and regulation of DNA damage repair [ 46 , 47 ]. In contrast, KMT2C is a member of the MLL/COMPASS family that catalyzes histone H3 lysine 4 (H3K4) methylation and regulates enhancer activity and transcriptional programs [ 48 , 49 ]. The frequent occurrence of mutations in these genes suggests that disruption of epigenetic regulatory pathways may contribute to the molecular landscape of UATs. Consistent with this hypothesis, many of the detected mutations localized within functional domains such as the SET domain of KMT5A and the PHD domain of KMT2C , which are known to participate in chromatin-associated regulatory processes. Adenomatoid tumors are benign mesothelial neoplasms that most commonly arise in the male and female genital tracts but have also been reported at less frequent extragenital sites [ 3 , 13 , 17 ]. Previous molecular studies have demonstrated that adenomatoid tumors are frequently characterized by recurrent TRAF7 mutations, supporting the concept that tumors arising at different anatomic sites may share a common molecular pathogenesis [ 17 , 18 ]. Interestingly, well-differentiated papillary mesothelioma, another benign mesothelial neoplasm, has also been reported to harbor recurrent TRAF7 or CDC42 mutations [ 50 ], suggesting partial molecular overlap among benign mesothelial tumors. In contrast, malignant mesotheliomas are typically characterized by alterations involving BAP1 , NF2 , and CDKN2A [ 51 , 52 ], indicating a distinct molecular profile compared with benign mesothelial neoplasms such as adenomatoid tumors. In our cohort, TRAF7 mutations were also detected, although at a lower frequency than KMT5A and KMT2C . The discrepancy between our findings and previous reports may reflect differences in cohort size, sequencing platforms, analytical pipelines, or population-specific genetic backgrounds. Therefore, further validation in larger independent cohorts will be necessary to determine the relative contribution of these genes to the molecular landscape of UATs. Immunohistochemical analyses demonstrated that higher KMT5A expression was associated with increased staining of TRAF7 , γ-H2AX , CALB2 , and Ki-67. γ-H2AX is a well-established marker of DNA damage response [ 41 ], whereas CALB2 is widely used as a diagnostic marker of mesothelial differentiation [ 41 , 53 ]. Although these findings suggest potential associations between KMT5A expression and pathways related to DNA damage signaling and cellular proliferation, the biological significance of these relationships remains to be further clarified. Functional experiments indicated that silencing KMT5A reduced proliferation and migration of ISK cells. However, it should be noted that the ISK cell line represents an endometrial epithelial carcinoma model rather than a mesothelial lineage system. Future studies using lineage-appropriate mesothelial models will be required to validate the biological role of KMT5A alterations in UATs. In addition, although KMT2C mutations were frequently detected in the cohort, the present study did not directly investigate their functional effects, which remain to be explored in future work. Comparative analyses using publicly available cancer genomic datasets, including TCGA and cBioPortal cohorts, showed that KMT5A alterations occur at relatively low frequencies in most gynecologic malignancies, whereas KMT2C alterations are detected across multiple tumor types. Notably, the concurrent presence of KMT5A and KMT2C mutations in all UAT cases analyzed in this study was not observed in the queried malignant datasets. These observations suggest that the co-occurrence of these alterations may represent a distinctive molecular feature of UATs; however, this finding requires confirmation in larger independent datasets. Several limitations of this study should be acknowledged. First, functional experiments were performed using an epithelial endometrial carcinoma cell line rather than a lineage-matched mesothelial model, which limits direct biological extrapolation to UATs. Second, the relatively small sample size, reflecting the rarity of this tumor type, may limit the generalizability of the findings. Future investigations using larger cohorts and lineage-appropriate experimental systems will be necessary to clarify the biological significance of these alterations. Finally, although associations between KMT5A/KMT2C alterations and clinical outcomes were observed in TCGA datasets of gynecologic malignancies, these analyses were performed primarily to provide comparative genomic context. Because UATs are benign mesothelial tumors that rarely undergo malignant transformation, the prognostic implications observed in malignant epithelial tumors should not be directly extrapolated to UAT biology. In summary, this study expands the current understanding of the genomic landscape of uterine adenomatoid tumors and identifies recurrent alterations involving epigenetic regulatory genes. These findings provide a framework for future studies aimed at elucidating the molecular mechanisms underlying the development of this rare mesothelial neoplasm.

Uterine adenomatoid tumors (UATs) are rare, benign mesothelial neoplasms that typically arise within the myometrium of premenopausal women [ 1 – 4 ]. Histologically, they are characterized by tubular or slit-like glandular structures lined by flattened mesothelial-like cells embedded in a fibromyxoid stroma [ 5 – 10 ]. Despite their benign morphology and frequently incidental clinical detection [ 11 ], the molecular mechanisms underlying UAT development remain poorly understood. Adenomatoid tumors are not restricted to the uterus and have been reported in multiple anatomical sites, including the epididymis, fallopian tube, adrenal gland, and other serosal surfaces [ 12 – 15 ]. Across these locations, adenomatoid tumors exhibit consistent mesothelial differentiation and generally indolent clinical behavior [ 16 ]. Recent molecular studies have identified recurrent somatic mutations in tumor necrosis factor receptor–associated factor 7 ( TRAF7 ) in adenomatoid tumors of both the male and female genital tracts, suggesting a shared molecular pathogenesis across anatomical sites [ 17 , 18 ]. However, genomic characterization beyond TRAF7 alterations remains limited, particularly for uterine cases. Advances in high-throughput sequencing technologies have enabled detailed molecular profiling of diverse gynecologic tumors, providing important insights into their genomic architecture [ 19 ]. Nevertheless, UATs have largely been absent from large-scale cancer genome projects, such as The Cancer Genome Atlas (TCGA), resulting in a substantial gap in current knowledge. Given their distinct mesothelial lineage, UATs may harbor molecular alterations that differ from those observed in epithelial gynecologic tumors or smooth muscle–derived uterine neoplasms. Emerging evidence also suggests that dysregulation of epigenetic regulators and immune-modulating pathways contributes to the development of various uterine tumors [ 20 , 21 ]. In particular, genes involved in chromatin modification, histone methylation, and antigen presentation pathways have been implicated in tumor development and immune regulation [ 22 ]. Whether similar molecular processes are involved in the pathogenesis of UATs remains unclear. Given the limited genomic data available for UATs, we sought to systematically characterize their mutational landscape. To address this gap, we performed targeted genomic sequencing of 20 pathologically confirmed UAT specimens. The aims of this study were to describe the spectrum of genomic alterations, evaluate whether chromatin-regulatory and immune-associated genes are recurrently affected, and place these findings into a broader genomic context using publicly available datasets. In addition, immunohistochemical and in vitro assays were conducted to provide preliminary insight into the potential biological relevance of selected alterations.

Total RNA was extracted from cells using TRIzol reagent (Abclonal, China) according to the manufacturer’s instructions, and reverse transcription was performed with a commercial kit to synthesize complementary DNA (cDNA). Quantitative PCR (qPCR) was conducted using SYBR Green Master Mix, with GAPDH serving as the internal control. Relative mRNA expression levels were calculated using the 2^−ΔΔCt method. All assays were performed in triplicate to ensure reproducibility and analytical reliability. The primer sequences were as follows: GAPDH forward 5′-GGAGCGAGATCCCTCCAAAAT-3′ and reverse 5′-GGCTGTTGTCATACTTCTCATGG-3′; KMT5A forward 5′-ACCGACGGGGAGAACGTATT-3′ and reverse 5′-GCATTCCAGAGCATTTGTTCG-3′. Protein lysates were prepared using RIPA buffer containing protease inhibitors, and protein concentrations were quantified using the BCA assay. Equal amounts of protein were resolved via SDS-PAGE (Servicebio, China), transferred onto PVDF membranes, and probed with primary antibodies against KMT5A (CST, USA, 2996) and GAPDH (Servicebio, China, ZB15004-HRP-100). Signal detection was achieved using enhanced chemiluminescence (ECL, EpiZyme, China). Cell proliferation was assessed using the 5-ethynyl-2’-deoxyuridine (EdU) incorporation assay, according to the manufacturer’s protocol (Servicebio, China). After EdU labeling, cells were fixed, permeabilized, and stained with Apollo fluorescent dye. Nuclei were counterstained with DAPI. Fluorescence images were acquired using a confocal microscope, and the percentage of EdU-positive cells was quantified to evaluate proliferative capacity. Cell viability was measured using the Cell Counting Kit-8 (CCK-8, Vazyme, China) per the manufacturer’s instructions. KMT5A -knockdown and control ISK cells were seeded into 96-well plates (3 × 10³ cells/well; Corning, USA). At 0, 24, 48, 72, and 96 h, 10 µL of CCK-8 solution was added to each well and incubated for 2 h at 37 °C. Absorbance was measured at 450 nm using a microplate reader. Experiments were repeated in triplicate to assess time-dependent cell proliferation. Cell migratory ability was evaluated using Transwell chambers with 8 μm pore filters (Corning, USA). Cells were seeded into the upper chamber with serum-free medium, while the lower chamber contained medium supplemented with 10% FBS as a chemoattractant. After incubation, non-migrated cells were removed, and migrated cells were fixed, stained with crystal violet, and quantified under a light microscope. To assess long-term proliferative potential, cells were seeded at low density and cultured for approximately two weeks. Colonies were fixed with paraformaldehyde, stained with crystal violet, and counted. Colonies containing ≥ 50 cells were quantified, and colony formation rate was used as an index of clonogenic capacity. FFPE tissue Sect.  (4 μm thick) were subjected to conventional hematoxylin and eosin (H&E) staining to evaluate histoarchitecture and cellular morphology. Stained sections were examined under a bright-field microscope at ×10 and ×20 magnifications to assess overall tissue architecture and cellular features. Serial FFPE tissue sections were deparaffinized in xylene and rehydrated through graded ethanol. Antigen retrieval was performed using citrate buffer (pH 6.0) in a pressure cooker. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide, followed by nonspecific blocking with 5% normal goat serum. Calretinin is widely used as a diagnostic immunohistochemical marker for mesothelial differentiation, including mesothelioma and other mesothelial tumors[ 38 , 39 ]. Phosphorylation of histone H2AX at Ser139 (γ- H2AX ) represents a well-established marker of DNA double-strand breaks and activation of the DNA damage response [ 40 , 41 ]. Primary antibodies against TRAF7 (Proteintech, China, 11780-1-AP), KMT5A (Proteintech, China, 14063-1-AP), γ- H2AX (phospho-Histone H2AX , Ser139; Proteintech, China, TA311953), CALB2 (Proteintech, China, 82811-1-RR), and Ki-67 (Proteintech, China, 27309-1-AP) were incubated overnight at 4 °C. The γ- H2AX antibody specifically detects the phosphorylated form of histone H2AX at Ser139. On the following day, HRP-conjugated secondary antibodies were applied, and immunoreactive signals were visualized using 3,3′-diaminobenzidine (DAB). Nuclei were counterstained with hematoxylin. Immunohistochemical staining was performed on whole-slide sections, and external positive and negative control tissues were included in each staining run. Immunostaining was semi-quantitatively assessed using the histochemical score (H-score), calculated as: H-score = (1 × % weak staining) + (2 × % moderate staining) + (3 × % strong staining), resulting in a total score ranging from 0 to 300 [ 42 , 43 ]. For comparative analysis, tumors were stratified into low and high KMT5A expression groups based on the median H-score of KMT5A staining across the cohort. The percentage of positive cells was also recorded. Representative images were captured at 10× and 20× magnifications for comparative analysis. All statistical analyses were performed using R software (version 4.1.2) and GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA). Depending on the data type, results were expressed as mean ± standard deviation (SD) or median (interquartile range, IQR). Comparisons between two groups were conducted using Student’s t-test or Mann–Whitney U test, while comparisons among multiple groups were performed using one-way analysis of variance (one-way ANOVA) or Kruskal–Wallis test. Categorical variables were compared using the χ² test or Fisher’s exact test. A two-sided P value < 0.05 was considered statistically significant. Data visualization, including OncoPrint plots, Lollipop plots, and tumor mutational burden (TMB) bar plots, was performed using the ggplot2 and ComplexHeatmap packages [ 44 , 45 ].

In this study, we leveraged the GEPIA2 online platform ( http://gepia2.cancer-pku.cn ) to comprehensively investigate the transcriptomic expression levels of KMT5A and KMT2C across various cancer types and to assess their prognostic relevance. The platform integrates RNA-seq data from both the TCGA and GTEx projects, allowing for robust pan-cancer comparisons [ 37 ]. The Ishikawa (ISK) cell line, a well-characterized human endometrial adenocarcinoma epithelial cell line, was used for in vitro functional experiments as a preliminary gynecologic model. Although this cell line is not derived from mesothelial lineage, it was used to provide initial insight into the potential biological effects of KMT5A alterations. Further studies using lineage-appropriate mesothelial models will be required to validate these findings. ISK cells were cultured in F12 medium (Jiangsu KeyGEN BioTECH, China) supplemented with 10% fetal bovine serum (FBS; Biological Industries, Israel) and 1% penicillin–streptomycin (Servicebio, China) at 37 °C in a humidified incubator with 5% CO₂. To investigate the functional role of KMT5A , three short hairpin RNAs (shRNAs) targeting different regions of the KMT5A transcript were designed: sh KMT5A -1 (CCGAGGAACAGAAGATCAAAG), sh KMT5A -2 (GTTTCCTGAAACTGGGTTAAT), and sh KMT5A -3 (GCCTAGGAAGACTGATCAATC). A non-targeting shRNA was used as the negative control (NC). Knockdown efficiency was evaluated at both the mRNA and protein levels using quantitative real-time PCR (qRT-PCR) and Western blotting.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3