Abstract

   Lymphovascular space invasion (LVSI) is the presence of tumor emboli in
   the endothelial-lined space at the tumor body’s invasive edge. LVSI is
   one of three Sedlis criteria components—a prognostic tool for early
   cervical cancer (CC)—essential for indicating poor prognosis, such as
   lymph node metastasis, distant metastasis, or shorter survival rate.
   Despite its clinical significance, an in-depth comprehension of the
   molecular mechanisms or immune dynamics underlying LVSI in CC remains
   elusive. Therefore, this study investigated tumor-immune
   microenvironment (TIME) dynamics of the LVSI-positive group in CC. RNA
   sequencing included formalin-fixed paraffin-embedded (FFPE) slides from
   21 CC patients, and differentially expressed genes (DEGs) were
   analyzed. Functional analysis and immune deconvolution revealed
   aberrantly enriched PI3K/Akt pathway activation and a heterogenic
   immune composition with a low abundance of regulatory T cells (Treg)
   between LVSI-positive and LVSI-absent groups. These findings improve
   the comprehension of LSVI TIME and immune mechanisms, benefiting
   targeted LVSI therapy for CC.

   Keywords: cervical cancer, lymphovascular space invasion, RNA
   sequencing, PI3K/Akt signaling pathway, sustained angiogenesis, immune
   deconvolution, tumor immune microenvironment, TIME, regulatory T cells

1. Introduction

   Lymphovascular space invasion (LVSI) is the presence of tumor emboli
   within dilated endothelium-lined spaces, such as lymphatic vessels or
   small capillaries at the invasive front of a tumor
   [[40]1,[41]2,[42]3,[43]4]. As a crucial initial step of the
   invasion–metastasis cascade, LSVI is an early indication of lymph node
   and other distant metastasis and poor prognosis, such as shorter
   recurrence-free survival (RFS) [[44]4,[45]5,[46]6,[47]7]. With such
   clinical significance, LVSI is a prominent prognostic factor of cancer
   in women, including endometrial, breast, and cervical cancers (CC)
   [[48]5,[49]8,[50]9,[51]10,[52]11,[53]12].

   In particular, LVSI in CC is one of three Sedlis criteria—intermediate
   risk factors concerning early CC prognosis—along with a tumor size
   larger than 4 cm and deep stromal invasion
   [[54]13,[55]14,[56]15,[57]16,[58]17]. As LVSI is an integral route that
   tumor cells disseminate, it guides physicians to inspect the
   possibility of local invasion and distant metastasis as well [[59]18].
   Furthermore, LVSI assists clinicians in surgical methods or patient
   management decisions, such as paraaortic lymphadenectomy, adjuvant
   pelvic radiation, or aggressive chemotherapy
   [[60]16,[61]19,[62]20,[63]21].

   Despite its clinical significance, the specific molecular mechanisms or
   immune microenvironment dynamics underlying LVSI in CC remain poorly
   understood. While some studies have identified genomic alterations
   associated with LVSI, such as cancer cell angiogenesis or the
   enrichment of cell proliferation genes, these findings are
   predominantly observed in LVSI-positive cases of other cancer types,
   such as breast or colorectal cancers [[64]22,[65]23,[66]24]. Limited
   research has explored the gene signatures or immune mechanisms of LVSI
   specifically in CC, such as studies conducted by Huang et al. [[67]25]
   and He et al. [[68]18]. Nonetheless, a comprehensive exploration of
   LVSI in CC is notably scarce within the existent literature,
   particularly concerning tumor-immune microenvironments (TIME).
   Therefore, the present study investigated TIME’s immune composition and
   dynamics within the LVSI-positive group in CC. Moreover, we sought a
   prominent immune marker that highly correlates with LVSI presence in CC
   to facilitate the detection of potential LVSI biomarkers and improve
   LVSI treatment.

2. Materials and Methods

2.1. Patient Selection

   This study selected 21 patients who previously received radical
   hysterectomy for CC. Conventional pelvic lymphadenectomy was performed
   in a standard fashion, involving the removal of lymph nodes from
   external and internal iliac as well as obturator nodal stations. Low
   paraaortic lymphadenectomy added presacral, low paraaortic, and common
   iliac lymph nodes to conventional pelvic lymphadenectomy. Extended
   lymphadenectomy was defined as incorporation of superior and inferior
   gluteal lymph nodes with low paraaortic lymphadenectomy. Adjuvant
   concurrent chemoradiotherapy was administered for positive lymph nodes,
   parametrial invasion, or positive surgical margin, while adjuvant
   radiotherapy and chemotherapy were employed for intermediate risk
   factors.

   Among the 21 patients, 15 had LVSI and were labeled as the
   LVSI-positive group (LVSI+), and the remaining 6 were considered the
   LVSI-absent group (LVSI−). All patients were staged per the revised
   2009 International Federation of Gynecology and Obstetrics (FIGO)
   staging system. The Kyungpook National University Chilgok Hospital’s
   (KNUCH) ethics committee thoroughly reviewed the patient selection
   process following IRB protocol (IRB number approval number:
   2020-10-003). T-tests were conducted for the variables “Age” and
   “Primary tumor size”, and Fisher’s exact tests were applied to the rest
   of the variables.

2.2. Pathological Processing and LVSI Detection

   From surgically resected tissue of 21 CC patients, samples were stained
   with H&E on formalin-fixed paraffin-embedded (FFPE) slides. The
   available H&E slides, ranging from five to ten for each case, were then
   carefully reviewed by a gynecologic pathology specialist from KNUCH.
   LVSI was defined by the presence of tumor cells within
   endothelial-lined vascular/lymphatic spaces at the invasive front.
   Intratumoral LVSI was not considered. The slides were further
   immunostained with vascular endothelial markers CD31 and CD34 and
   lymphatic endothelial marker D2-40 for LVSI detection for cases
   requiring additional confirmation. Due to the spatial heterogeneity on
   the FFPE slides and the controversiality of quantification LVSI in CC,
   any identified focus of LVSI was considered as LVSI positive.

2.3. RNA Sequencing and LVSI DEGs

   RNA was extracted from FFPE slides with the ReliaPrep™ FFPE Total RNA
   Miniprep System (Promega, Madison, WI, USA). Extracted RNA was
   generated into the library using the TruSeq RNA Exome Kit (Illumina,
   San Diego, CA, USA) and further sequenced with the Nova6000 platform
   (Illumina). Raw fastq data were quantified using the STAR [[69]26] and
   RSEM [[70]27], and differentially expressed gene (DEG) analysis was
   accomplished with DESeq2 [[71]28]. DEGs with a threshold of an absolute
   log2 fold change ≥1 with adjusted p-value < 0.05 were considered
   significant. A volcano plot of the DEGs was plotted using the “ggplot”
   in R studio version 4.3.1
   ([72]https://posit.co/download/rstudio-desktop/) accessed on 10 May
   2023.

2.4. Immune Pathways Enrichment Analysis with IMPAGT

   The Immune Pathway and Gene Database (IMPAGT) [[73]29] is an in-house
   immune database referenced from LM22 of CIBERSORT [[74]30]. Unlike
   other immune databases that do not furnish comprehensive lists of genes
   associated with specific immune mechanisms, such as CIBERSORT, IMPAGT
   simultaneously provides immune pathways and their corresponding gene
   sets. Furthermore, IMPAGT offers insights into the involvement of
   individual genes in multiple immune pathways, enabling the
   interpretation of “one gene–many immune pathways” interconnections.
   Contrasting conventional “one gene–one pathway” analyses from other
   immune databases, this unique attribute allows researchers to view and
   interpret TIME from more holistic and integrative perspectives.

   Intersection of LVSI DEGs with a 2534 IMPAGT gene set was performed
   through R studio. Overlapped LVSI DEGs were traced back on IMPAGT, and
   DEGs participating in each pathway were counted. Pathways with more
   than five involved DEGs were considered as significantly enriched TIME
   pathways by IMPAGT. After selecting enriched TIME pathways, the
   pathways were visualized with overlapping LVSI DEGs via KEGG Mapper
   [[75]31]. The overlapped up-regulated DEGs were colored red.

2.5. Functional Analysis of Intersected LVSI DEGs with IMPAGT

   For the functional analysis of LVSI DEGs that intersected with IMPAGT,
   the protein−protein interaction (PPI) network was reviewed via the
   STRING database [[76]32], and gene ontology (GO) terms were analyzed on
   the DAVID database [[77]33]. Among three GO terms (biological process
   [BP], cellular component, and molecular functions), BP terms were
   primarily investigated. Significant BP terms (p-value < 0.05) were
   further investigated through the STRING database.

2.6. Immune Microenvironment Estimation with Immune Deconvolution

   The overall estimation of the 21 CC samples’ immune microenvironment
   was conducted using the R package “Immunedeconv”, which allows seven
   immune deconvolution methods: EPIC, xCell, TIMER, CIBERSORT, MCP
   Counter, and quantiseq [[78]34]. The TPM values of the 21 CC samples
   were utilized as the input. Among the seven deconvolution tools, EPIC
   [[79]35] and xCELL [[80]36] results were further analyzed as they
   provide an “absolute score” of each sample, enabling inter-comparisons
   between the samples. EPIC provided seven immune cell types: B cells,
   cancer-associated fibroblasts, endothelial cells, macrophages, NK
   cells, T cell CD4+, and T cell CD8+. xCell provided 36 immune cell
   types, such as B cells, cancer-associated fibroblasts, endothelial
   cells, eosinophils, macrophages, macrophage M1, mast cells, and
   regulatory T cells (Tregs). K-means clustering was applied to the
   immune deconvolution result matrix of EPIC and xCell. The initial k was
   set as two to investigate whether immune composition grouping
   corresponded to the absence or presence of LVSI (LVSI+/−).

   In order to investigate immune markers highly correlated with LVSI,
   point–biserial correlation—which measures the association between
   continuous numeric and binary categorical variables—was applied to
   xCell immune markers and LVSI. As xCell provided more immune cell types
   than EPIC, xCell results were utilized. Furthermore, all 36 xCell
   immune cell types were evaluated for their association with LVSI+ based
   on a univariate logistic regression model. The t-test was also
   performed to identify differences between LVSI+/− groups. All
   statistical analyses were conducted with the base R function in R
   studio.

3. Results

3.1. Clinicopathological Characteristics of the 21 Samples

   Several clinical parameters for the 21 samples were collected and
   presented relative to LVSI status ([81]Table 1). Compared with the
   LVSI− group, the LVSI+ group was primarily staged in later FIGO stages,
   such as IIA2 or IIB. Most samples with deep stromal invasion,
   parametrial invasion, and lymph node metastasis were from the LVSI+
   group. Also, samples from the LVSI+ group received a broader extent of
   lymphadenectomy as well as adjuvant therapy compared with the LVSI−
   group.

Table 1.

   Clinicopathological characteristics of the 21 samples in this study.
   Variables All
   (n = 21) LVSI−
   (n = 6) LVSI+
   (n = 15) p-Value
   Age (years ± SD) 47.4 ± 11.0 46.8 ± 9.2 47.6 ± 11.3 0.880
   FIGO stage (n, %) 0.586
     IB1 11 (52.4) 4 (66.7) 7 (46.7)
     IB2 2 (9.5) 1 (16.7) 1 (6.7)
     IIA2 4 (19.0) 1 (16.7) 3 (20.0)
     IIB 4 (19.0) 0 4 (26.7)
   Histology 0.862
     Squamous cell carcinoma 15 (71.4) 5 (83.3) 10 (66.7)
     Adenocarcinoma 5 (23.8) 1 (16.7) 4 (26.7)
     Adenosquamous carcinoma 1 (4.8) 0 1 (6.7)
   Primary tumor size (cm ± SD) 3.3 ± 1.6 2.7 ± 1.0 3.5 ± 2.0 0.367
   Deep stromal invasion (n, %) 20 (95.2) 5 (83.3) 15 (100.0) 0.286
   Parametrial invasion (n, %) 9 (42.9) 1 (16.7) 8 (53.3) 0.178
   Positive vaginal margin (n, %) 2 (9.5) 1 (16.7) 1 (6.7) 0.500
   Lymph node metastasis (n, %) 11 (52.4) 3 (50.0) 8 (53.3) 1.000
   Lymphadenectomy 0.084
     Pelvic lymphadenectomy 4 (19.0) 3 (50.0) 1 (6.7)
     Pelvic and low paraaortic lymphadenectomy 7 (33.3) 1 (16.7) 6 (40.0)
     Pelvic and extended lymphadenectomy 10 (47.6) 2 (33.3) 8 (53.3)
   Adjuvant therapy 0.835
     Chemotherapy 5 (23.8) 1 (16.7) 4 (26.7)
     Radiotherapy 2 (9.5) 0 2 (13.3)
     Concurrent chemotherapy 10 (47.6) 2 (33.3) 8 (53.3)
   Recurrence (n, %) 6 (28.6) 4 (66.7) 2 (13.3) 0.031
   [82]Open in a new tab

   For p-value, t-tests were conducted for the variables “Age” and
   “Primary tumor size”, and Fisher’s exact tests were applied to the rest
   of the variables. FIGO, International Federation of Gynecology and
   Obstetrics; (n, %) indicates the number and percentage of samples in
   each parameter; SD, standard deviation; LVSI+/−; lymphovascular space
   invasion positive/absent.

3.2. Representative LVSI Detection

   Representative microscopic features of LVSI are depicted in [83]Figure
   1 through H&E-stained slides.

Figure 1.

   [84]Figure 1
   [85]Open in a new tab

   Representative microscopic features of LVSI (H&E stains; (A), ×100
   magnification; (B), ×200). (A) Malignant cell clusters are identified
   within the lymphovascular spaces at the leading edge of invasive
   cervical cancer. (B) Tumor cell clusters are covered with endothelial
   cells (indicated by black arrows) and intermixed with inflammatory
   cells (indicated by black asterisks).

3.3. Differentially Expressed Genes (DEGs)

   There were 734 DEGs between LVSI− and LVSI+ groups. Among the 734 DEGs,
   708 were up-regulated, and the remaining 26 were down-regulated
   ([86]Supplementary S1). A volcano plot of significant LVSI+ DEGs is
   presented in [87]Figure 2.

Figure 2.

   [88]Figure 2
   [89]Open in a new tab

   A volcano plot of LVSI+ DEGs. The right side of the volcano plot
   indicates up−regulated LVSI DEGs in red, and the left side depicts
   down-regulated LVSI DEGs in blue.

3.4. Intersection of LVSI DEGs with IMPAGT Revealed PI3K/Akt Signaling
Pathway Overactivation

   In order to investigate the TIME dynamics of the LVSI+ group in CC,
   LVSI DEGs that overlapped with the IMPAGT gene set were further
   researched. Overall, 67 LVSI DEGs overlapped with IMPAGT; all
   overlapped with up-regulated LVSI DEGs, and none from down-regulated
   LVSI DEGs ([90]Supplementary S2). Pathways with more than five DEGs
   from IMPAGT were considered significantly enriched and are presented in
   [91]Table 2.

Table 2.

   Significantly enriched pathways from IMPAGT. The listed significantly
   enriched pathways are presented with their corresponding KEGG ID and
   the number of DEGs involved in each.
    KEGG ID              IMPAGT Pathway             Number of DEGs
   Hsa 05200           Pathway in Cancer                  16
   Hsa 04121       PI3K/Akt Signaling Pathway             13
   Hsa 04060 Cytokine-Cytokine Receptor Interaction       9
   Hsa 04514        Cell Adhesion Molecules               7
   Hsa 04640       Hematopoietic Cell Lineage             5
   Hsa 04010         MAPK Signaling Pathway               5
   [92]Open in a new tab

   The most enriched pathway was “Pathway in Cancer” with 16 DEGs,
   followed by “PI3K/Akt Signaling Pathway” with 13 DEGs. The remaining
   enriched pathways were as follows: “Cytokine–Cytokine Receptor
   Interaction”, “Cell Adhesion Molecules”, “Hematopoietic Cell Lineage”,
   and “MAPK Signaling Pathway”.

   Pathway enrichment visualization via KEGG Mapper highlighted where
   overlapped DEGs participated and were activated in each pathway. DEGs
   in “Pathway in Cancer” were primarily related to receptor interactions
   on the cell membrane. Additionally, they were involved in signaling
   pathways, such as phosphoinositide 3-kinase (PI3K)/Akt,
   mitogen-activated protein kinase (MAPK), or vascular endothelial growth
   factor (VEGF) signaling pathways, which eventually directed toward
   “Proliferation”, “Evading Apoptosis”, and “Sustained Angiogenesis”
   ([93]Figure 3A).

Figure 3.

   [94]Figure 3

   [95]Figure 3
   [96]Open in a new tab

   Enriched pathway visualization with KEGG Mapper. Up−regulated LVSI DEGs
   are highlighted in red. (A) “Pathway in Cancer”; all signals lead
   toward ‘Sustained Angiogenesis’, ‘Evading Apoptosis’, and
   ‘Proliferation’, which are highlighted in orange. (B) “PI3K−Akt
   Signaling Pathway”.

   As the “PI3K/Akt Signaling Pathway” was the second most enriched
   pathway and part of the “Pathway in Cancer”, it was also visualized
   with KEGG Mapper ([97]Figure 3B). Highlighted DEGs were mainly involved
   in pathways as growth factors or ECM receptors that activated PI3K/Akt
   signaling via Class 1A PI3K isoforms.

3.4.1. Functional Analysis Revealed 67 LVSI-IMPAGT DEGs Are Related to
Tumoral Angiogenesis in CC

   The 67 LVSI DEGs overlapped with IMPAGT were further investigated on
   the DAVID database for BP terms; significantly identified BP terms are
   presented in [98]Table 3. To examine gene interactions relative to BP
   functions, 67 DEGs were also examined on the STRING database. The
   functional analysis revealed that BP terms were predominantly related
   to tumoral angiogenesis in CC.

Table 3.

   Significant BP terms of 67 up-regulated LVSI DEGs overlapped with
   IMPAGT.
   Group GO: BP ID BP Terms p-Value
   Tumor GO: 0007165 Signal Transduction <0.0001
   GO: 0051092 Positive Regulation of NF-kappa B Transcription Factor
   Activity <0.001
   GO: 0034220 Ion Transmembrane Transport 0.001
   GO: 0001938 Positive Regulation of Endothelial Cell Proliferation 0.002
   GO: 0030334 Regulation of Cell Migration 0.007
   GO: 0007267 Cell–Cell Signaling 0.008
   GO: 0006468 Protein Phosphorylation 0.001
   GO: 0043410 Positive Regulation of MAPK Cascade 0.018
   GO: 0010595 Positive Regulation of Endothelial Cell Migration 0.022
   GO: 0032092 Positive Regulation of Protein Binding 0.028
   GO: 0006816 Calcium Ion Transport 0.037
   Immune GO: 0006955 Immune Response <0.0001
   GO: 0001916 Positive Regulation of T Cell-Mediated Cytotoxicity 0.007
   GO: 0050778 Positive Regulation of Immune Response 0.007
   GO: 0030333 Antigen Processing and Presentation 0.012
   GO: 0002250 Adaptive Immune Response 0.028
   GO: 0030183 B Cell Differentiation 0.038
   [99]Open in a new tab

   The overall PPI network was linked to “Signaling Systems”, such as
   ‘Signal Transduction’ and ‘Cell–Cell Signaling’. Interestingly, the PPI
   network tended to cluster into three representative BP traits: “Immune
   Response”, “Ion Transport”, and “Regulation of Endothelial Cells”
   ([100]Figure 4). First, genes positioned in the right half of the PPI
   network were primarily related to “Ion Transport” BP traits, such as
   “Ion Transmembrane Transport” or “Calcium Ion Transport”. In addition,
   the inner cluster was related to “Regulation of Endothelial Cell”,
   including “Positive Regulation of Endothelial Proliferation” and
   “Positive Regulation of Endothelial Cell Migration”. Lastly, genes
   located in the bottom left of the PPI network were largely related to
   “Immune Response”, as follows: “Positive Regulation of T cell-Mediated
   Cytotoxicity”, “Positive Regulation of Immune Response”, “Antigen
   Processing and Presentation”, “Adaptive immune response”, and “B cell
   differentiation”.

Figure 4.

   [101]Figure 4
   [102]Open in a new tab

   PPI network of the 67 LVSI−IMPAGT DEGs with their significant BP terms.
   While the overall PPI network was related to signal transduction, PPI
   subgroups tended to cluster in three representative BP traits: “Immune
   Response”, “Ion Transport”, and “Regulation of Endothelial Cells”.

3.4.2. Immune Deconvolution Exposed Heterogeneity in the Immune Composition
of the LVSI+ Group with CC

   First, immune deconvolution tools were employed to comprehend the
   overall immune composition of the 21 CC samples. EPIC results presented
   seven major immune cell types, including B, CD4+ T, and CD8+ T cell
   types ([103]Figure 5A). xCell indicated 36 immune cell types with
   specific subsets, such as naïve B, memory B, Th 1 CD4+ T, and
   regulatory T cell types ([104]Figure 5B). The overall immune
   composition results from EPIC and xCELL were clustered to investigate
   whether immune composition differed relative to LVSI, setting the
   initial k as two. The two clusters from both deconvolution results did
   not separate according to the clinical parameter of interest, LVSI+/−.

Figure 5.

   [105]Figure 5
   [106]Open in a new tab

   Immune deconvolution results revealed the overall immune composition of
   21 CC patients. The x-axis indicates the overall 21 CC samples of the
   study, and the y-axis delineates each tool’s absolute immune
   deconvolution scores. (A) EPIC results with seven immune cell types,
   and (B) xCell results with thirty-six immune cell types.

   Although the EPIC results had a distinct separation of two clusters, it
   did not correlate with LVSI+/− ([107]Figure 6A). However, when the
   location of samples from each LVSI+ and LVSI− group were examined,
   samples from the LVSI− group were primarily localized in the middle of
   cluster 2. Regarding xCell result clustering, two clusters overlapped
   ([108]Figure 6B). However, these results revealed that most samples
   from the LVSI− group were a small sub-cluster within cluster 2. The
   adjacent locality of LVSI− samples indicates that the LVSI− group
   shares similar immune compositions.

Figure 6.

   [109]Figure 6
   [110]Open in a new tab

   The k−means clustering plots of (A) EPIC and (B) xCELL immune
   deconvolution results. Black circles indicate LVSI− samples. Although
   the plots did not cluster relative to LVSI, the LVSI− group tended to
   cluster as a sub-group, indicating a similarity in immune compositions.
   On the other hand, the LVSI+ group was dispersed throughout the plot,
   revealing a more “heterogenic” immune composition of the LVSI+ group
   than the LVSI− group.

   On the other hand, samples from the LVSI+ group were dispersed
   throughout the plot, located in clusters 1 and 2 when compared with the
   LVSI− group. This plot dispersion delineates immune composition
   diversity in the LVSI+ group, contrasting the shared similarity of the
   LVSI− group. In short, the LVSI+ group in CC presented more
   “heterogeneity” of TIME composition than the LVSI− group.

3.4.3. Regulatory T Cells as Primary Immune Markers Related to the LVSI+
Group within the Heterogenic Immune Composition of LVSI+ in CC

   After examining the heterogenic immune composition of the LVSI+ group
   in CC, we investigated which immune markers highly correlated with
   LVSI+ parameters within such heterogeneity. As xCell provided more
   immune cell types compared with EPIC, xCell results were taken to
   examine the highly correlated immune markers with LVSI+. First, the
   point–biserial correlation was calculated between LVSI parameters and
   36 immune cell types identified by xCell. A correlation greater than an
   absolute value of 0.35 was considered a significant immune cell marker
   related to the LVSI+ group.

   Three immune cell markers significantly correlated with LVSI+:
   regulatory T (Treg), eosinophil, and T cell CD8+ naïve cells
   ([111]Supplementary S3). Among these immune markers, Tregs presented
   the strongest correlation with LVSI+ with a negative correlation of
   −0.50. The following t-test on Treg abundance score comparison between
   LVSI+/− groups, with a p-value of 0.07, indicated a proclivity of low
   Treg expression in the LVSI+ group ([112]Figure 7). The t-test results
   on the rest of two immune markers were as follows: eosinophil (p =
   0.33) and T cell CD8+ naïve cells (p = 0.03). Furthermore, in order to
   identify whether these immune markers can predict the presence of LVSI,
   the univariate regression analysis was performed on 36 xCell immune
   markers, and it revealed that Treg was the only significant immune
   marker related to LVSI+ (coefficient: −14.09, p-value < 0.02). These
   findings indicate that Treg is the predominant immune cell marker that
   presents less abundance in LVSI+ parameters among the heterogenic
   immune composition of LVSI+ in CC.

Figure 7.

   [113]Figure 7
   [114]Open in a new tab

   A boxplot of the low Treg abundance in the LVSI+ group. The x-axis
   represents LVSI+/− groups, and the y-axis depicts the abundance scores
   from xCell.

4. Discussion

   This study investigated TIME and immune dynamics underlying LVSI in CC.
   TIME is a multifaceted architecture where the crosstalk and interplay
   of various cell types from tumors, stroma, ECM, and immune cells occur
   [[115]37]. As dynamic reciprocal communication between cells in TIME
   switches and favors cancer progression, TIME interpretation and
   comprehension are critical. Therefore, this study elucidates the
   tumor-immune dynamics and mechanisms of LVSI in CC: the aberrantly
   enriched PI3K/Akt signaling pathway and heterogenic immune composition
   of LVSI in CC with low Treg abundance.

   First, this study revealed a highly enriched activation of the PI3K/Akt
   signaling pathway in LVSI+ in CC. From the enriched immune pathway
   analysis, “Pathway in Cancer” was identified as the most enriched
   pathway. Several signaling pathways, such as the PI3K/Akt signaling
   pathway or MAPK signaling pathway, were involved in the mechanism
   responsible for the major cancer hallmarks proposed by Weinberg:
   “Proliferation”, “Avoiding Apoptosis,” and “Sustained Angiogenesis”
   [[116]38]. As the PI3K/Akt signaling pathway was involved in “Pathway
   in Cancer” and was the second most enriched pathway from LVSI+ in CC,
   the PI3K/Akt signaling pathway was investigated further.

   As a primary signaling pathway, the PI3K/Akt signaling pathway
   manifests various cellular processes, such as the cell cycle,
   proliferation, or apoptosis. With such engagement, the PI3K/Akt
   signaling pathway also participates in signaling cascades toward
   prominent cancer hallmarks. As previously mentioned, these markers
   include malignant cell proliferation or anti-apoptosis. Furthermore,
   the PI3K/Akt signaling pathway is also involved in sustained
   angiogenesis. In tumor cells, PI3K/Akt signaling pathway activation
   boosts VEGF secretion by hypoxia-inducible factor (HIF) dependent and
   independent mechanisms. Subsequently, this process increases the
   expression of angiogenic factors, such as VEGF, angiopoietins, or
   nitric oxide [[117]39]. These angiogenic factors then augment vascular
   permeability that contributes to leaky vessels, sluggish blood flow,
   and interstitial pressure of the endothelial vessel, otherwise known as
   sustained angiogenesis [[118]39,[119]40]. Among three PI3K classes, the
   p110a isoform from class 1 PI3K is an essential stimulator of
   angiogenesis by regulating VEGF and dictating overall cancer
   progression [[120]41,[121]42,[122]43,[123]44,[124]45].

   The BP terms analyzed in this study, such as “Ion Transport” and
   “Regulation of Endothelial Cells”, further support and highlight the
   PI3K/Akt signaling pathway’s involvement in “Sustained Angiogenesis.”
   According to the literature, membrane potential (Vm) is depolarized in
   cancer cells compared with normal counterparts [[125]46]. Moreover,
   dysregulated Vm is prevalent throughout all cancer progression stages
   and nearly all cancer hallmarks [[126]47,[127]48,[128]49]. As Ca^2+
   transport was identified as a significant BP term, Ca^2+ entry
   concerning angiogenesis was further investigated.

   There are a few mechanisms that describe calcium entry with
   angiogenesis. Ca^2+ enters the cytosol through Ca^2+ permeable plasma
   membrane channels, such as transient receptor potential (TRP) or
   store-operated Ca^2+ entry (SOCE) [[129]50]. Under hypoxic tumor
   microenvironments (TME), tumor cells secrete growth factors or
   cytokines, including VEGF and epidermal growth factors (EGF). VEGF then
   combines with VEGF receptor 2, which allows Ca^2+ entry via SOCE or TRP
   [[130]51]. In particular, TRP is up-regulated in hypoxic cancer cell
   environments and is considered an “angiogenic switch” that facilitates
   angiogenic mechanisms.

   For example, TRP subtypes—such as TRPC1 and TRPC4—encourage endothelial
   cell proliferation and migration. Similarly, TRPC3 and TRPC6 activation
   increases endothelial cell migration and permeability. Altogether,
   these subtypes enhance pro-angiogenic factor secretion, such as VEGF or
   EGF, that leads to blood capillary sprouts [[131]50]. TRPC4 is related
   to increased intracellular Ca^2+ influx and PI3K/Akt signaling pathway
   activation via VEGF [[132]52,[133]53]. As such, TRPC4 is an essential
   pro-angiogenic determinant that maintains the balance between normal
   and defective vasculatures [[134]54]. Moreover, TRP also forms a
   complex with STIM1 and Orail of SOCE, allowing sustained Ca^2+ influx
   that increases EMT, sustained angiogenesis, and poor prognosis
   [[135]55].

   This study also observed the “heterogeneity” of TIME of LVSI+ in CC.
   Although the clusters did not separate relative to LVSI status, k-means
   plots from the deconvolution results indicated a tendency that samples
   from the LVSI− group position in proximity. This proclivity implies
   shared immune composition characteristics among the LVSI− group. While
   LVSI− presented similar immune compositions, samples from the LVSI+
   group were dispersed throughout the two clusters of the k-means plot.
   Such dispersion suggests a distinctive “heterogenic” TIME composition
   of the LVSI+ group in CC compared with the LVSI− group.

   Regarding TIME heterogeneity, Fujikawa et al. proposed that TIME is
   more heterogeneous in aggressive tumors [[136]8]. In addition, Karar et
   al. discovered that aberrant vasculature and active ECM remodeling
   corresponding to abnormal tumor blood vessels also influenced TIME
   heterogeneity [[137]39]. Furthermore, Li et al. revealed extensive
   cellular heterogeneity in malignant CC epithelial cells via single-cell
   transcriptomics analysis [[138]56]. Our findings corroborate such
   heterogeneity of LVSI+ in CC. Specifically, visualizing the PI3K/Akt
   signaling pathway with LVSI-IMPAGT DEGs highlighted that these DEGs
   primarily act as growth factors or ECM receptors that activate the
   PI3K/Akt signaling pathway, driving cell proliferation or angiogenesis.
   It elucidates atypical enrichment and alterations in PI3K/Akt signaling
   are associated with LVSI+ in CC. Such aberrant shifts within LVSI+ TIME
   may beget atypical immune cell expression, reshaping the TIME
   architecture of LVSI+ into a more heterogeneous landscape than the
   LVSI− group.

   After acknowledging the heterogenic nature of the LVSI+ group’s immune
   composition, we aimed to identify immune markers that were highly
   correlated to the LVSI+ group within heterogeneous TIME. Point–biserial
   correlation and univariate regression revealed that low Treg abundance
   was the strongest correlated immune marker with LVSI+ among its
   heterogenic TIME architecture. A recent study by He et al. also
   supports this observation; they found a suppression of adaptive
   immunity in LVSI+ patients. Their findings presented lower abundance of
   active CD4 T cells, natural killer T cells, effector memory CD8 T cell,
   and Treg in the LVSI+ group in comparison with the LVSI− group
   [[139]18].

   Treg is widely known for its immune suppression in TIME
   [[140]57,[141]58,[142]59]; however, this aspect is controversial
   [[143]60,[144]61,[145]62]. For example, Hori et al. reviewed
   Foxp3+Tregs and debated Treg’s plasticity and heterogenetic role
   [[146]63]. Considering this controversy and outcomes of Treg within
   TIME from various studies, Barua et al. stated the necessity for
   comprehending Treg spatially [[147]64]. Specifically, Luznik et al.
   asserted Treg’s pro- and anti-angiogenic effects [[148]65]. They
   further reported that Treg exerts opposing effects on angiogenic
   function relative to tissue or TIME characteristics in which Treg
   resides, stressing the importance of investigating Treg in a spatial-
   and context-specific manner [[149]65].

   The limitation of this study lies in its relatively small size and
   retrospective nature. The relatively smaller sample size of the LVSI−
   group compared with the LVSI+ group may hold the possibility of
   selection bias. However, it is crucial to highlight that the samples of
   the current study represent invaluable “real-world clinical samples”,
   which are relatively scarce in sample retrieval. Recognizing these
   concerns, ongoing efforts are focused on designing future studies with
   larger samples sizes and a more balanced composition. Integration of
   publicly available datasets related to LVSI+ in CC are actively
   explored to augment the sample numbers. Additionally, public datasets
   of LVSI+ in other cancer types are also being considered to validate
   the specificity of the finding in context of CC. This comprehensive
   approach would bolster robustness of the study while compensating
   current limitations. Furthermore, multi-omics approaches are being
   considered for upcoming studies, such as proteomics, metabolomics,
   spatial transcriptomics, and single-cell transcriptomics. These methods
   would further enable precise immune profiling and assist in the
   investigation and interpretation of specific TIME immune marker roles
   with localization information in LVSI+ in CC.

   Recently, the FIGO made a revision and now requires precise
   documentation on the extent of LVSI when staging II endometrial cancer
   by counting the number of LVSI [[150]66,[151]67]. As such, the 2023
   FIGO revision acknowledges and emphasizes LVSI’s enhanced prognostic
   value, but this revision has yet to be applied to CC. However,
   considering LSVI’s increased significance regarding cancer in women,
   this revision is expected soon. Given the limited research on LVSI+ in
   CC despite its increasing importance, the current findings
   significantly contribute to the medical understanding of LVSI+ in CC
   and further establish a foundational groundwork for comprehending LVSI+
   in CC. Moreover, the findings further contribute to potential LVSI
   targeting therapies. For instance, the identification of an
   over-activated PI3K/Akt signaling pathway suggests a potential avenue
   for therapeutic explorations, such as PI3K inhibitor as EMT modulators
   to mitigate pathway effects. In addition, the “immune-hot” status of
   LVSI+ in CC, characterized by up-regulated immune signals and low Treg
   expression, suggests the feasibility of immunotherapy implementation.
   Modulating the immune response of LVSI+ in CC and the TIME architecture
   could inhibit further LVSI progression, enabling earlier invention
   against lymph node metastasis or distant organ metastasis. Further
   studies building upon these current findings would assist identifying
   more refined and detailed LVSI-targeted therapies.

5. Conclusions

   LVSI, defined as the presence of tumor emboli in the endothelial-lined
   space at the tumor body’s invasive edge, serves as crucial prognostic
   criteria for CC with implications for surgical selection or patient
   management due to its association with poor prognosis. Despite its
   clinical significance, an in-depth comprehension of the molecular
   mechanisms and immune dynamics underlying LVSI in CC still remains
   elusive. Transcriptomic analysis of DEGs in LVSI+/− groups from 21 CC
   samples in the current study revealed the aberrant enrichment PI3K/Akt
   signaling pathway, particularly in relation to tumoral angiogenesis in
   CC. Furthermore, this study discovered the heterogeneity of the LVSI+
   group’s immune composition in CC compared with the LVSI− group. In
   addition, low Treg expression was identified as a major correlated
   immune marker within the LVSI+ group in CC. These findings improve
   comprehension and open avenues for targeted treatments toward LVSI in
   CC.

Acknowledgments