Pathological and molecular insights into intravenous leiomyomatosis: an integrative multi-omics study

Abstract

Intravenous leiomyomatosis (IVL) is a histologically well differentiated smooth muscle tumor with aggressive behavior, capable of extending throughout the venous system. Understanding how IVL occurs and develops is really important for diagnosing and treating it. Unfortunately, because IVL is quite rare, there aren’t many comprehensive studies available. In our research, we carried out an extensive multi-omics study, gathering tissue samples from IVL cases, uterine fibroid, and normal myometrium. The single-cell RNA sequencing analysis revealed a notable difference in cell composition between IVL and uterine fibroid. Additionally, H&E staining demonstrated more frequent hydropic changes and hyalinization in IVL tissues, along with a reduced vascular density compared to both normal myometrium and uterine fibroid. In our proteomics analysis of eight paired samples of IVL and normal myometrium fresh frozen tissue, we identified proteins that were differentially expressed, mainly related to focal adhesions and regulation of the actin cytoskeleton. The most frequently involved chromosomes included deletions in 10q22.2, 10q24.32, 13q14, and 13q21-31. Correlation analyses highlighted chromosome 10q as the most frequent cytoband, with corresponding proteins involved in regulating focal adhesions and the cytoskeleton. Integrated analysis between pathological and clinical characteristics indicated that chromosome 10q deletion and vascular morphology in IVL could serve as important markers predicting aggressive behavior. Our study sheds light on the pathological and molecular changes linked to IVL, which could pave the way for new treatment approaches.

Introduction

Intravenous leiomyomatosis (IVL) is an uncommon smooth muscle tumor characterized by histologically well-differentiated smooth muscle disease but exhibiting aggressive behavior [1]. Typically originating in the uterus, IVL extends along the venous system to the iliac vein, inferior vena cava (IVC), right heart, and occasionally the pulmonary artery, resulting in diverse clinical symptoms and, at times, sudden death. Traumatic surgery is imperative for the removal of extra-pelvic disease [2].

Two theories exist regarding the origin of IVL: one predominant theory suggests that IVL originates from uterine fibroid (UF), while the other posits that IVL stems from the smooth muscle cells (SMCs) of the vessel walls [3,4,5]. Although new research methods and animal models for understanding how diseases happen keep coming up all the time [6, 7], limited research has been conducted on IVL tumorigenesis. Ordulu et al. classified IVL into cellular, vascular and usual type based on microscopic characteristics. Their array comparative genomic hybridization studies revealed frequent genetic alterations involving chromosomes 1p, 22q, 2q, 13q, and 14q [8]. Lu et al. utilized Sanger sequencing and short tandem repeat analysis, observed discordance in Mediator complex subunit 12 (MED12) mutation, microsatellite instability, and loss of heterozygosity between UF and IVL [9]. Other studies employing MED12 gene sequencing confirmed IVL as a distinct tumor from UF [10, 11]. At the transcriptional level, Zhang et al. identified differentially expressed genes between IVL and UF, primarily enriched in the extracellular matrix, cell adhesion, and steroid hormone stimulus [12]. Despite these findings, comprehensive multi-omics studies investigating IVL tumorigenesis and development are lacking.

In this investigation, we initiated single-cell RNA sequencing (scRNA-seq) to delineate the cell types in a paired set of UF and IVL fresh tissues. Subsequently, hematoxylin and eosin (H&E) as well as immunohistochemistry (IHC) staining were conducted on 23 paired IVL and normal myometrium paraffin-embedded samples. This aimed to delve deeper into the pathological changes characteristic of IVL, with results validated through an additional cohort. To unveil the molecular alterations in IVL, we employed integrated omics analyses, encompassing tandem mass tag (TMT)-based quantitative proteomics analysis and Whole-Exome Sequencing (WES). This comprehensive approach was applied to eight paired IVL samples (featuring extra-pelvic disease) and normal myometrium fresh frozen tissue. Finally, an integrated analysis was performed, bridging the gap between pathological observations and clinical characteristics. This study not only shed light on the pathological and molecular features associated with IVL tumorigenesis and tumor behavior but also charted new avenues for the treatment of this disease.

Methods

Ethics

Our study adhered to the principles of the Declaration of Helsinki and received approval from the ethical committee of Zhongshan Hospital, Fudan University (Ethics Committee document number: B2021-488R). Written informed consent was obtained from all participating patients.

Study flowchart and tissue sample Collection

The study’s flowchart is illustrated in Fig. 1. Patients pathologically diagnosed with IVL at Zhongshan Hospital, Fudan University, between 2014 and 2020 were included. For scRNA-seq, samples were acquired from freshly removed tissues, including one paired UF and IVL. TMT-based proteomic and WES analyses were performed on eight paired fresh frozen samples, comprising IVL lesions (both extracted from the IVC) and corresponding normal myometrium. These samples were collected during multidisciplinary surgical procedures and stored at -80 °C until use. For H&E and IHC staining, 23 IVL and paired normal myometrium tissues were collected and embedded in paraffin. The validation cohort included 14 paired normal myometrium, uterine fibroid, and IVL tissues. All samples underwent reconfirmation by qualified pathologists, and comprehensive medical records were meticulously reviewed and collected.

Fig. 1

Overview of the experimental design

Single-cell RNA sequencing

In brief, cells were loaded onto the 10X Chromium Single Cell Platform (10X Genomics) at a concentration of 1,000 cells per µL, using the Single Cell 3’ library and Gel Bead Kit v.3, following the manufacturer’s protocol. The process involved the generation of gel beads in emulsion (GEMs), barcoding, GEM-RT clean-up, complementary DNA amplification, and library construction, all in accordance with the manufacturer’s instructions. Library quantification was performed using Qubit before pooling, and the final library pool was sequenced on an Illumina Novaseq 6000 instrument. The targeted mean reads per cell was set to ≥ 25,000, and our samples achieved an average of 31,222 reads per cell.

Data processing of single-cell RNA-seq from chromium system

The cellranger software (version 2.1.0) was utilized for mapping to the GRCh38 human genome, performing quality control, and counting reads of Ensembl genes using default parameters.

Unsupervised clustering and visualization

Unsupervised clustering was conducted using R with the Seurat package (version 2.2). Genes expressed in fewer than two cells were excluded. Cells with more than 200 genes and less than 10% mitochondrial genes were subjected to further processing. Subsequently, the coefficient of variation of genes was calculated using Seurat. Dimensionality reduction of the data was conducted through principal component analysis, focusing on the first 2000 genes with the highest variability. A k-nearest neighbor graph was constructed based on Euclidean distances within the space defined by the first 10 principal components. The Louvain Modularity optimization algorithm was employed to cluster the cells within the graph, and the resulting clusters were visualized using t-distributed Stochastic Neighbor Embedding (tSNE) projection. Cells expressing high levels of genes encoding hemoglobin were removed from the analysis.

Marker gene identification and cell-type annotation

The ‘bimod’ test implemented in the Seurat FindMarkers function was utilized to compute the differential expression of each cluster. Genes exhibiting a log2 average expression difference of 0.585 and a significance level of P < 0.05 were designated as marker genes. Canonical markers of established cell types were employed to annotate cell clusters. Seurat-Bimod statistical test was used to find differentially expressed genes between each group of cells and other groups of cells (FDR ≤ 0.05 and |log2 Fold Change| ≥ 1.5). The TopGO R package was utilized for Gene Ontology (GO) enrichment analysis of these significantly differentially expressed genes, while the Hypergeometric test in R was employed for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. GO terms and KEGG pathways that were significantly enriched were identified based on a threshold FDR (adjusted P-value) ≤ 0.05.

TMT protein labeling and bioinformatics analysis

Following the extraction of total protein from the samples, a portion of the protein was utilized for determining concentration and SDS-PAGE analysis, while another portion underwent trypsin hydrolysis and labeling. Equal amounts of each labeled protein sample were amalgamated for chromatographic separation, facilitating liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis. The qualitative and quantitative data obtained were subjected to analysis. Expression level analysis and functional analysis were conducted subsequent to quality evaluation and preprocessing. The functional annotation encompassed the use of several common data analysis tools, including GO and KEGG pathway analysis. Interaction analysis was then executed for the differentially expressed proteins (DEPs) identified. Furthermore, correlation analysis, expression pattern clustering, and Venn analysis were performed, and heat maps were generated to enhance data interpretation and visualization.

Whole exosome sequencing

Qualified DNA samples underwent random fragmentation into 150 to 220 bp fragments using Covaris. Subsequently, the Agilent SureSelect Human All Exon V6 kit was employed for library construction and capture. The library assembly involved various steps, including DNA end-joining, addition of polyA tails, incorporation of sequencing adapters, purification, magnetic bead capture, PCR amplification, and other processes. A thorough re-evaluation of raw data quality was conducted, considering parameters such as error rate, data volume, comparison rate, and coverage. The resulting high-quality sequences were aligned to the GRCh37.p13 human genome to detect sample variation information, which was subsequently analyzed.

Integrated multi-omics analysis of IVL

To harmonize data across multiple platforms, we compiled a list of all genes exhibiting copy number variation (CNV) and their associated proteins. Correlation (Pearson P-value) was calculated, with significance set at P < 0.05. Enrichment analyses, including GO and KEGG pathway analyses, were performed for positively correlated CNVs/proteins. In exploring the potential molecular mechanisms of IVL tumorigenesis, we enriched the frequency of chromosome positions and analyzed the corresponding proteins within the altered CNVs/proteins.

H&E and IHC staining

Archival paraffin blocks from 23 IVL patients, along with paired normal myometrium samples, were meticulously chosen for this study. Sections of 4 μm thickness were obtained from the paraffin-embedded samples, and subsequent H&E and IHC staining procedures were conducted on these slides. For H&E staining, microscopic characteristics were thoroughly examined in each slide, encompassing the proportion of hydropic change and hyalinization, as well as vascular densities in the tumor parenchyma. Furthermore, all IVL cases were classified based on cellular, vascular, and usual morphology. The extent of these morphological features was scored as minimal (< 5%), focal (5–24%), or diffuse (> 25%) following a prior study ⁶. In the case of IHC staining, the final protein expression score was derived by multiplying the percentage of positive cells (PPC) with the intensity of staining (IS). PPC categories included: 0 = < 10% of positive cells, 1 = 10–25%, 2 = 25–50%, 3 = 50–75%, and 4 = ≥ 75% of positive cells. IS was classified as 0 (negative), 1 (weakly positive), 2 (moderately positive), and 3 (strongly positive). The cumulative score was then categorized as “low” for scores ranging from 0 to 3, “medium” for scores between 4 and 7, and “high” for scores from 8 to 12. Random fields under the 100x microscope were selectively chosen for vessel calculation.

The specific protocol for this study is summarized in the Supplemental Methods.

Results

IVL and UF exhibit profound differences in cellular composition through scRNA-seq

To delve into the cellular heterogeneity between IVL and UF, scRNA-seq profiles were generated from a paired IVL and UF tissue sample. The results, illustrated in Fig. 2A, unveiled the existence of 22 distinct clusters among high-quality cells. These 22 clusters were further categorized into eight subpopulations, encompassing fibroblasts, SMCs, endothelial cells, macrophages, T cells, monocytes, mast cells, and progenitor cells (Fig. 2B). Examining the distribution in Fig. 2C and D, fibroblasts dominated the cellular landscape of the IVL lesion, constituting 84.73% of the identified cells. SMCs and endothelial cells accounted for 4.38% and 2.37%, respectively. In contrast, the UF lesion exhibited a distinct composition, with endothelial cells comprising 56.64% and SMCs making up 37.02%, while fibroblasts constituted only 5.35%. These findings underscore a substantial cellular component disparity between IVL and UF, characterized by reduced levels of SMCs and a markedly increased prevalence of fibroblasts in IVL. Moreover, the observation of limited endothelial cells in IVL suggests a potential hypovascular characteristic. The heat map in Fig. 2E summarizes the top five expressed genes in each cluster. In conclusion, the single-cell analysis underscores the profound differences in cellular components between IVL and UF, emphasizing the necessity for a comprehensive exploration of the pathological and molecular changes associated with IVL.

Fig. 2

Single-cell atlas of IVL and UF. (A) tSNE plot illustrating clusters of high-quality cells from IVL and UF. (B) Identification of cell lineages based on gene expression. (C) Annotation of cell clusters per sample. (D) Distribution of cells per sample. (E) Heatmap of gene expression analyzed by single-cell RNA sequencing, presenting key markers for each cell type. SMC, Smooth Muscle Cell

Pathomorphological alterations in IVL

Subsequently, we scrutinized the pathomorphological distinctions between IVL and normal myometrium in a cohort of 23 IVL patients. The clinical attributes of these patients are succinctly outlined in Supplementary Table 1. Macroscopically, IVL exhibited a “worm-like” tissue appearance, softer in texture compared to both normal myometrium and UF (Fig. 3A). Histologically, IVL was characterized as a benign smooth muscle tumor located within venous vascular spaces (Fig. 3B). Pathomorphologically, numerous IVL tissues displayed hydropic changes and areas of hyalinization in proximity to the tumor parenchyma (Fig. 3C/D). In rare instances, adipose tissue was observed in IVL, accompanied by focal hemorrhage and neutrophil infiltration (Fig. 3E/F). Vascular morphology exhibited variability, with some instances featuring large vessels characterized by thick vascular walls (Fig. 3G).

Fig. 3

Pathological characteristics in IVL. (A) Gross appearance of IVL. (B) Typical microscopic image of IVL. An intravascular tumour plug. (C) Edema change in IVL. The proportions of the edema area between normal myometrium and IVL. (D) Hyaline degeneration in IVL. The proportions of the hyaline degeneration between normal myometrium and IVL. (E) The adipose tissue in IVL. (F) Focal hemorrhage and neutrophils in IVL. (G) Thick vascular wall in IVL. (H) Vascular number between normal myometrium and IVL. (I) The proportions of the edema area, hyaline degeneration and vascular number between N, UF and IVL

Among the 23 IVL cases in our study, the prevailing morphology was vascular (16/23, 69.6%), followed by the cellular (2/23, 8.7%) and usual (2/23, 8.7%) types. Two patients exhibited a combination of cellular and vascular types, while one patient presented with both vascular and usual types.

Utilizing these morphological characteristics, we initially compared the hydropic change, hyalinization, and vascular density in tumor parenchyma between IVL and normal myometrium. In IVL, 8 (34.8%), 10 (43.5%), and 5 (21.7%) patients exhibited hydropic changes as minimal, focal, and diffused, respectively, whereas only 4 patients (3 minimal and 1 focal) showed hydropic changes in normal myometrium. Hyalinization manifested as minimal, focal, and diffused in 7 (30.4%), 11 (47.8%), and 5 (21.7%) patients in IVL, respectively, while only 2 patients (1 minimal and 1 focal) displayed hydropic changes in normal myometrium. The findings imply that the proportions of hydropic change and hyalinization were significantly higher in IVL (Fig. 3C/D). The median blood vessel counts in IVL and normal myometrium tissues were 10 (range 3 to 25) and 20 (range 11 to 30), respectively, suggesting that blood vessel density in IVL was lower than that in normal myometrium (P < 0.001) (Fig. 3H).

To further investigate whether these morphological characteristics differed between IVL and UF, 14 paired normal myometrium, UF, and IVL tissues were selected. The results also indicated that hydropic change and hyalinization were more frequently observed in IVL tissue than in both normal myometrium and uterine fibroids. Conversely, the vascular density was decreased in IVL tissues compared to both normal myometrium and uterine fibroids (Fig. 3I). In summary, when compared with normal myometrium and uterine fibroids, the proportions of hydropic change and hyalinization areas were significantly increased, and the vascular density was decreased in IVL tissues.

Proteomic alterations between IVL and paired normal myometrium

To delve deeper into the molecular changes within IVL, we conducted TMT-based proteomic analysis and WES on eight paired IVL and normal myometrium tissues. The clinical characteristics of these eight patients are summarized in Supplementary Table 2. A total of 263 DEPs were identified, with 133 proteins (50.6%) being upregulated and 130 downregulated (Fig. 4A). The volcano plot and heat map illustrate the distribution of these DEPs (Fig. 4B/C and Supplementary Table 3).

Fig. 4

Overview of the proteomics of IVL patients. (A) Visualization of 263 identified proteins, with orange indicating upregulated proteins and green indicating downregulated proteins. (B) Volcano Plot depicting significantly upregulated (red dots) and downregulated (blue dots) proteins, with gray dots representing proteins with no significant difference. The x-axis represents the fold change (Log2) between two groups, and the y-axis represents the p-value (-Log10) to illustrate the significance of differences. The top 10 upregulated and downregulated DEPs are labeled. (C) Heatmap for hierarchical clustering analysis of DEPs. Each row represents a protein, and each column indicates a sample. Colors indicate expression levels (red for high, purple for low). Violin plots depict the probability distribution of expression values. “+” denotes the median value, and the vertical axis represents protein expression. (D) GO Pathway Analyses of the top 15 enriched terms, classified by biological process (green), cellular component (blue), and molecular function (red). (E) Top 20 enriched KEGG terms, categorized into cellular processes (green), environmental information processing (purple), human disease (orange), and organismal systems (yellow). (F) PPI Network of the top 25 DEPs. DEPs are represented as dots, and edges between proteins indicate potential interactions. Dot size corresponds to combined interaction scores identified by Cytoscape. (G) Vinculin staining in IVL and normal myometrium. (H) Vinculin IHC staining score analysis

Functional annotation analysis using GO was performed to gain a better understanding of the biological functions of the DEPs. As depicted in Fig. 4D, the most downregulated GO terms included cell-matrix adhesion, focal adhesion, and actin cytoskeleton, while the most upregulated GO terms included complement activation and extracellular exosome. Furthermore, KEGG enrichment analysis indicated that the DEPs are primarily involved in the regulation of actin cytoskeleton, extracellular matrix (ECM)-receptor interaction, and complement and coagulation cascades (Fig. 4E and Supplementary Table 4/5). The downregulated proteins in the protein-protein interaction (PPI) network included vinculin (VCL), ACTIN1, and ACTA2 (α-SMA), while the upregulated proteins included C3, C4B, APOA1, APOB, and FN1 (Fig. 4F). Vinculin, a cytoskeletal protein associated with cell-cell and cell-matrix junctions, plays a crucial role in cell morphology and locomotion. To further validate the results of the proteomic analysis, IHC staining of Vinculin was performed on the 23 paired IVL and normal myometrium paraffin-embedded tissues. As shown in Fig. 4G, the expression of Vinculin was significantly lower in IVL than in normal myometrium.

Genetic variants between IVL and paired normal myometrium

WES was conducted to investigate the genetic variants between IVL and the corresponding normal myometrium. As illustrated in Fig. 5A, a total of 381 mutations were identified, with nonsynonymous mutations constituting the most frequent type (61.4%), followed by synonymous mutations (24.4%). Excluding synonymous mutations, a total of 239 potential driver mutations were identified, with a median of 25 mutations in each sample. The prominently mutated genes included PKD1, CAPN 15, ZNF90, SNAPC4, and MUC4 (Fig. 5B), with a mutation frequency of 25.0% (2/8). Details of each mutated gene mentioned above are summarized in Supplementary Table 6.

Fig. 5

Genomic profiles based on WES. (A) Representation of different mutational events in all IVL lesions. (B) Identification of the top 5 mutated genes corresponding to 8 samples. (C) Characteristics of genome-wide Copy Number Variations (CNVs) distribution in 8 samples, with a pooled analysis at the bottom. Red, blue, and green indicate amplification, deletion, and no change, respectively. X-axis represents chromosome cytoband, and Y-axis represents copy ratio (log2). (D) Overall chromosomal distribution of CNVs, where blue represents deletion and red represents amplification

Subsequently, we analyzed CNVs between the two groups. The cytobands of somatic CNVs for each IVL sample are depicted in Fig. 5C. As indicated, the most frequent deletions in chromosomes were concentrated in 10q22.2, 10q24.32, 13q14, and 13q21-31, and observed in 3 of 8 IVLs (37.5%). The top CNV cytobands of gain were observed in 8q24, 11q13, 12q24, 19q13, and 20q13, each of the top CNV cytobands are also observed in 3 of 8 IVLs (Fig. 5D).

Correlation analysis of TMT-based proteomics and WES

To consolidate the findings from both proteomic and WES analyses, we conducted Pearson correlation analysis. The P-value was < 0.05 for 87 CNVs-DEPs, among which 47 exhibited particularly positive correlations (correlation value of > 0) (Fig. 6A). GO analyses were subsequently employed to unravel the potential functions of these 47 CNVs-DEPs. As depicted in Fig. 6B, ECM organization, cell-matrix adhesion, focal adhesion, and integrin binding were the primary enrichments. Subsequently, we delved into the cytoband distributions corresponding to these 47 altered CNVs-DEPs. Of 47 highly correlated CNV-DEPs, 10 (21.3%) were located on chromosome 10q, linking cytoband deletion with lower protein levels (Fig. 6C). Comparing quantity of the 10 candidate proteins in IVL and paired normal myometrium revealed significant difference in 3 out of 10 proteins. The expression levels of GLUD1, PDLIM1, and SLK were notably decreased in IVL (Fig. 6D and E). Significantly, both PDLIM1 and SLK play roles in the regulation of focal adhesion. Intriguingly, in the TMT-based proteomic analysis, we observed that the gene for vinculin, a cytoskeletal protein enriched in focal adhesions, is also located on chromosome 10q. These findings suggest that a chromosome 10q deletion in IVL may lead to the downregulation of several proteins involved in the regulation of the actin cytoskeleton and focal adhesion.

Fig. 6

Correlation analysis of TMT-based Quantitative Proteomics and whole-exosome sequencing. (A) Heatmap illustrating 47 positively related CNVs-DEPs. (B) Identification of the top 30 enriched GO terms related to CNVs-DEPs. (C) Distribution of cytobands associated with CNVs-DEPs. (D) Presentation of 10 CNVs-DEPs with significant protein quantitation. (E) Quantification of the expression of three proteins (GLUD1, PDLIM1, and SLK) between IVL and normal myometrium

Integrated analysis of clinical and pathological characteristics

To explore potential correlations between clinical characteristics and pathological changes, we scrutinized data from 23 IVL patients, stratified by the extent of tumor disease (as illustrated in Table 1). Among the 15 aggressive IVL patients (with extra-pelvic disease), 26.7% exhibited diffuse hyalinization, with an equal percentage showing hydropic changes. These manifestations were observed in 12.5% of patients with intra-pelvic disease. In the subgroup of eight patients with intra-pelvic disease, Vinculin high expression was identified in 37.5% of cases, while the proportion of Vinculin high expression was 13.3% (2/15) in aggressive IVL cases. Despite the small sample size limiting our ability to draw definitive conclusions about the relationship between IVL tumor extent and vinculin IHC staining score, the results suggest that lower vinculin protein expression levels may be positively correlated with a more advanced tumor stage.

Table 1 Clinical characteristics and pathological changes in 23 IVL patients

Among eight non-aggressive IVL patients, cellular, vascular, and usual morphology were observed in 25%, 50%, and 12.5% of patients, respectively. Two patients exhibited both cellular and vascular features. In the group of 15 aggressive IVL patients, a total of 80% were categorized as having vascular morphology, with one patient displaying both vascular and usual features. These findings imply that vascular morphology constitutes a substantial proportion in aggressive IVL cases.

Discussion

IVL exhibits a higher degree of fibrosis and hyalinization compared to normal myometrium and UF

In a recent study involving scRNA-seq analysis of five unpaired UF and normal myometrium samples, the cellular composition showed no significant differences between the two groups [13]. In our study, utilizing UF tissue as a control, we observed a striking difference in cellular composition between one paired IVL and UF tissue through scRNA-seq. Specifically, there was a notable decrease in SMCs and an increase in fibroblasts in IVL. These findings were corroborated by our proteomic analysis, demonstrating a significant decrease in α-SMA (SMCs marker) and an increase in FN1 (fibroblasts marker) in IVL tissue. It’s worth noting that although we initially dissected three paired UF and IVL fresh tissues for scRNA-seq, only one was successfully tested due to the low percentage of tumor cells in IVL tissue. The other two attempts failed because of inadequate cellular numbers, potentially linked to the increased fibrotic and hyalinized nature of IVL tissue.

HE staining revealed a significant increase in the proportions of hydropic change and hyalinization areas in IVL tissues compared to normal myometrium. While hydropic change and hyalinization are features also found in UF, we conducted a comparative analysis to discern differences between IVL and UF in terms of these histological features. The results further indicated a higher prevalence of hydropic change and hyalinization in IVL compared to both normal myometrium and UF. Given that hyalinization involves the transformation of a hard, cellular tumor into softer, acellular material, our findings provide insights into the clinical observation that IVL is noticeably softer than both normal myometrium and UF.

One of the primary functions of fibroblasts is the synthesis of collagen proteins [14, 15]. Collagen, a significant component of the extracellular matrix, contributes to the formation of a three-dimensional meshwork, providing physical support to cells and maintaining structural integrity in tissues [16]. Beyond its role in structural support, collagen plays a crucial role in various cellular processes, including cell migration, adhesion, angiogenesis, tissue development, and repair. Studies by Sahai E have highlighted that during migration, cancer cells utilize collagen fibers as tracks to leave the primary tumor, emphasizing the modulatory role of collagen in cell behavior [17]. Additionally, collagen has been identified as a novel driver of tumor invasion, promoting local invasion and metastasis [18, 19].

In our study, fibroblasts were predominantly identified in IVL. Upon comparing IVL with normal myometrium tissues and UF, we observed higher proportions of hyalinization areas. These areas, characterized by the replacement of smooth muscle cells with collagen, exhibited a similar transformation in blood vessels within regions of hyaline necrosis [20]. Proteomic analysis further revealed an upregulation of proteins associated with collagen and ECM, most probably associated with the increased fraction of fibroblasts in IVL and the tumor’s invasiveness.

In summary, our findings lead us to conclude that IVL exhibits more fibrotic and hyalinization features than normal myometrium and UF. These characteristics may be associated with the tumor development observed in IVL.

IVL exhibits a poorer blood supply compared to normal myometrium and and UF

Our scRNA-seq results indicated significantly lower proportions of endothelial cells in IVL compared to UF. Additionally, HE staining results confirmed a lower blood vessel density in IVL compared to normal myometrium and UF. These findings suggest that IVL experiences inadequate blood supply, potentially explaining the higher frequency of hyalinization in IVL. Insufficient blood supply within the myoma is recognized as a key factor contributing to hyalinization. Interestingly, our results diverge from a previous study suggesting that IVL tumors possess a strong ability to promote angiogenesis. The earlier study proposed that IVL tumors exhibit upregulated expression of proangiogenic factors, including GATA1, LIF, CXCL8, SH2D2A, et al. [21].

To further investigate angiogenesis-associated pathway changes in IVL, we conducted a thorough analysis by identifying enriched angiogenesis-related terms in the GO analysis based on our proteomic data. DEPs associated with GO: 0016525 (negative regulation of angiogenesis) demonstrated significant upregulation in IVL. Notably, APOH (Apolipoprotein H; fold change = 1.72), THBS1 (Thrombospondin 1; fold change = 1.79), HRG (Histidine Rich Glycoprotein; fold change = 1.60), THBS4 (Thrombospondin 4; fold change = 2.19), and THBS2 (Thrombospondin 2; fold change = 1.68) exhibited substantial increases in expression.

Conversely, DEPs associated with GO: 0045766 (positive regulation of angiogenesis) displayed significant downregulation in IVL. Noteworthy examples include HSPB1 (Heat Shock Protein Family B Member 1; fold change = 0.58), RRAS (RAS Related; fold change = 0.59), and AQP1 (Aquaporin 1; fold change = 0.61), all showing marked decreases in expression. Additionally, proteins involved in the general process of angiogenesis (GO: 0001525) underwent significant changes in IVL, with MYH9 (Myosin Heavy Chain 9; fold change = 0.66), FN1 (Fibronectin 1; fold change = 1.93), TGFBI (Transforming Growth Factor Beta Induced; fold change = 1.72), and APOD (Apolipoprotein D; fold change = 1.58) exhibiting altered expression levels.

Collectively, these results align with our earlier findings, supporting the notion that IVL experiences a poor blood supply in comparison to normal myometrium. However, to validate and further elucidate this conclusion, additional studies are warranted.

Dysregulation of focal adhesion and the actin cytoskeleton appears to be closely associated with the development of IVL

Our TMT-based proteomic analysis revealed that the most significantly downregulated signaling pathway was related to the regulation of the actin cytoskeleton and focal adhesion. Notably, key actin cytoskeletal proteins such as vinculin, α-actinin 1, Filamin A, and others exhibited downregulation. Focal adhesions, crucial for ECM interactions, play a fundamental role in maintaining tissue homeostasis and have been implicated in cancer development and pathogenesis [22, 23]. The disassembly of focal adhesions in cells has been linked to cytoskeletal transmission, promotion of epithelial-mesenchymal transition, and induction of cell migration [24,25,26], all of which contribute to cancer formation and progression.

A prior study conducted RNA-seq to explore differentially expressed genes between IVL and normal uterine muscle tissues. Consistent with our proteomic findings, the results from this study also suggested that IVL may disrupt the homeostasis of gene networks involved in ECM and cytoskeleton regulation [12]. These collective observations underscore the potential significance of focal adhesion and actin cytoskeletal dysregulation in the pathogenesis of IVL, emphasizing the need for further investigation into these molecular mechanisms.

Chromosome 10q deletions and vascular morphology could be important markers predicting the aggressive behavior of IVL

WES outcomes revealed an enrichment of chromosome deletions at specific loci, including 10q22.2, 10q24.32, 13q14, and 13q21-31, within IVL tissues. Correlation analyses unveiled that the deletion of chromosome 10q in IVL might lead to the downregulation of proteins crucial in the regulation of the actin cytoskeleton and focal adhesion, which could play a significant role in IVL tumorigenesis. The validity of these findings was substantiated through vinculin IHC staining. Additionally, our investigation unveiled a correlation between lower vinculin protein expression and the distinct vascular morphology with the aggressive behavior observed in IVL.

Chromosome 10q deletions have been implicated in various diseases, including cutaneous T-cell lymphoma [27], low-grade gliomas [28], and prostate cancer [29]. A prior study by Ordulu et al. further supported the association of a 10q deletion in IVL lesions [8, 30]. In Ordulu’s research, hierarchical clustering analysis identified three groups, with all IVL instances in group 3 (n = 5) exhibiting a deletion on 10q. Notably, 80% of IVL cases in group 3 displayed aggressive behavior. In our study, the removal of all eight IVL tissues for TMT-based proteomic and WES from the inferior vena cava already signaled their aggressive nature. Furthermore, Ordulu’s study highlighted that 80% of cases with a vascular morphology among the 5 IVL instances associated with recurrence, aligning with our results where 80% (12 out of 15) of aggressive IVL patients exhibited vascular morphology. Collectively, these findings strongly suggest that chromosome 10q deletions and vascular morphology serve as crucial markers for predicting the aggressive behavior of IVL.

Conclusion

Our study sought to investigate the pathological and molecular alterations in IVL patients through a comprehensive multi-omics approach. However, the study is subject to several limitations that warrant attention in future research. Firstly, the impact of our findings is constrained by the availability of only one paired IVL and UF sample for scRNA-seq. Secondly, to prevent the overinterpretation of IVL origin, further validation of WES results, including genetic variations, is essential with a larger sample size. Lastly, additional basic research is imperative to enhance our understanding of the origin and development of IVL.

In summary, this pioneering study conducted an integrated analysis, incorporating scRNA-seq, H&E, and IHC staining, TMT-based quantitative proteomics analysis, and WES of IVL. Our investigation unveiled both the pathomorphological and molecular characteristics in IVL, paving the way for new avenues in basic research and clinical treatment for IVL.