A method for screening a cancer prognosis marker

Abstract

The application belongs to the technical field of medical treatment and specifically relates to a screening method of a cancer prognosis marker, comprising the following steps: step one) obtaining an imaging mass spectrometry flow cytometry analysis data set; step two) performing single cell segmentation on images in the imaging mass spectrometry flow cytometry analysis data set, and annotating cell types of single cells according to expression intensity of the marker to be screened; step three) identifying an area where epithelial, immune and fibrotic cells intersect, and obtaining marker expression intensity in the area where epithelial, immune and fibrotic cells intersect; step four) obtaining an intercellular interaction analysis result based on cell frequency and marker expression intensity; and step five) performing regression analysis according to cell frequency, marker expression intensity and the intercellular interaction result in different comparison groups, and screening to obtain a prognosis marker.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical technology, in particular to a screening method of a cancer prognosis marker. BACKGROUND

[0002] Gastric cancer (GC) is a global health problem, ranking fifth in incidence and mortality worldwide in 2022. Currently, the main treatment methods for gastric cancer include surgery, chemotherapy and radiotherapy, which have limited effect on advanced patients. Despite continuous advances in clinical and basic research, the five-year survival rate of patients with advanced gastric cancer remains below 40%.

[0003] Given the poor prognosis of patients with advanced gastric cancer, the tumor microenvironment (TME) becomes a key factor affecting treatment effectiveness and the effectiveness of treatment regimens. TME is a complex network composed of cells and non-cellular components, which plays a decisive role in cancer dynamics and treatment effectiveness. Macrophages play a key role in it, significantly affecting tumor behavior and treatment response. Recent studies have revealed the complex relationship between macrophages and gastric cancer, and highlighted new therapeutic targets and mechanisms driving tumor progression.

[0004] The formation and function of tertiary lymphoid structures (TLS) in the tumor microenvironment have also attracted attention. These structures, similar to secondary lymphoid organs, but located outside traditional lymphoid organs, help activate naive T / B cells to enable them to exert tumor immune effects. Studies in various malignancies such as colorectal cancer, lung cancer, breast cancer and malignant melanoma have shown that the presence of TLS is associated with good prognosis.

[0005] The immune microenvironment (especially TLS and specialized immune cell subsets) has a profound impact on the prognosis and treatment outcome of gastric cancer. This understanding lays the foundation for the development of more personalized and effective immunotherapy, which is expected to revolutionize the management strategy of malignant tumors. However, how spatial features (especially the activities of macrophages) affect immune cell function have not been fully studied. To fully analyze the complex relationship between cells and spatial composition in the tumor microenvironment (TME), advanced analytical methods with spatial resolution are needed to effectively reveal spatial heterogeneity. By further combining histology and mass cytometry with imaging mass cytometry (IMC), a single slice can detect 45 metal-labeled antibodies simultaneously, accurately presenting cell interactions and marker co-expression. SUMMARY

[0006] In order to overcome the problem of lack of a method for screening markers based on the cell and spatial composition in the tumor microenvironment (TME) in the prior art, the present application proposes a screening method of a cancer prognosis marker, and the above object is achieved by the implementation of the following technical solutions:

[0007] A screening method of a cancer prognosis marker, comprising the following steps:

[0008] Step one) obtaining an imaging mass cytometry flow cytometry analysis data set, the imaging mass cytometry flow cytometry analysis data set comprising expression intensity of a marker to be screened in a sample and corresponding prognosis of the sample; and dividing different comparison groups according to the sample and the prognosis;

[0009] Step two) performing single cell segmentation on the image in the imaging mass cytometry flow cytometry analysis data set, and annotating cell types of single cells according to the expression intensity of the marker to be screened, to obtain cell frequencies of different cell types;

[0010] Step three) identifying an area where epithelial, immune and fibrosis cells intersect, and obtaining marker expression intensity in the area where epithelial, immune and fibrosis cells intersect;

[0011] Step four) calculating interaction analysis results between different cell types based on the cell frequencies of different cell types and the marker expression intensity in the area where epithelial, immune and fibrosis cells intersect;

[0012] Step five) performing regression analysis according to the cell frequencies, the marker expression intensity and the cell interaction results in different comparison groups, to screen a prognosis marker.

[0013] Optionally, the cancer is gastric cancer.

[0014] Optionally, the step one) comprises the following according to the sample and the prognosis:

[0015] dividing into a tumor group and a tumor surrounding tissue group according to whether the sample belongs to a tumor or a tumor surrounding tissue;

[0016] dividing into an early pathological stage group and a late pathological stage group according to whether the sample belongs to an early pathological stage or a late pathological stage;

[0017] dividing into an early clinical stage group and a late clinical stage group according to whether the sample belongs to an early clinical stage or a late clinical stage;

[0018] dividing into a long survival group and a short survival group according to the prognosis of the patient.

[0019] Optionally, the markers to be screened in step one) include CD45, CD15, CD45RA, CD103, CD68, HLA-DR, CD45RO, B7H4, CD14, CD20, CD57, PD-L1, CD25, CD3, CD4, aSMA, CD8, CD7, FOXP3, KI67, CD163, CD16, CD69, vimentin, CD11b, CD11c, Caspase3, IL-1b, Ki-67, TNF a, Type I collagen, LAG3, GATA3, PD-1, VISTA, CD31, Granzyme B, IL-6, Trypsin, E-Cadherin.

[0020] Optionally, the cell types include:

[0021] Myofibroblast, Type I collagen + Fibroblast, Interstitial cell, Endothelial cell, Epithelial cell, Lymphocyte, Myeloid cell;

[0022] The epithelial cell includes the following types: PD-L1 + Subpopulation, Ki-67 + Proliferative subpopulation, E-cadherin + Subpopulation, E-cadherin - Subpopulation;

[0023] The lymphocyte includes the following types: Regulatory T cell, Effector CD8 + T cell, Memory CD4 + T cell, Memory CD8 + T cell, Effector CD4 + T cell, CD16 - Natural killer cell, CD16 + Natural killer cell, Lag3 + Natural killer cell, B cell;

[0024] The myeloid cell includes the following types: Granulocyte, Monocyte, Ki67 + Proliferative macrophage, CD11c low HLA-DR low Macrophage, CD11c hi Macrophage, HLA-DR hi Macrophage.

[0025] Optionally, the step of annotating the cell types of the single cells in step two) includes the following steps:

[0026] The first round of clustering identifies the main cell populations by the following markers: E-Cadherin, Trypsin, CD3, CD4, CD8, CD20, CD7, CD57, CD45, CD68, CD15, CD14, CD16, CD31, aSMA, Collagen I, and Vimentin;

[0027] The second round of clustering clusters myeloid cells using FOXP3, CD16, CD69, CD4, CD8, Caspase3, B7H4, VISTA, CD7, CD103, LAG3, CD20, Granzyme B, PD-1, KI67, GATA3, CD45RA, CD3, TNFa, TL1b, CD45RO, CD57, CD25;

[0028] Clusters lymphocytes using IL-6, CD14, CD16, Caspase3, CD163, PD-L1, CD11B, CD11C, CD15, Ki-67, HLA-DR, TNFa, and TL1b;

[0029] Clusters epithelial cells by E-Cadherin, PDL1, Ki-67, Trypsin.

[0030] Optionally, the step three) of identifying the region of epithelial, immune, and fibrotic cell intersection comprises the following steps: based on the spatial coordinates, using k- nearest neighbor algorithm to calculate the 20 nearest neighbors of each cell, thereby constructing a local spatial adjacency graph for each cell; for the neighborhood of each cell, aggregating the various marker expression data of adjacent cells, and aggregating according to cell type;

[0031] The local sub-graphs are filtered to ensure that the center node of each sub-graph corresponds to a specific cell type; the sub-graphs are then connected by shared nodes to form a global connection graph, thereby generating a spatial region / patch; the spatial coverage of the constructed patch is expanded by adding n adjacent nodes to the existing patch, thereby expanding the spatial coverage of the constructed patch;

[0032] Based on the spatial connection graph, an immune region, a fibroblast region, and an epithelial region are obtained; the immune region includes lymphocytes and myeloid cells; the fibroblast region includes myofibroblasts and collagen I + fibroblasts; the epithelial region includes epithelial cells;

[0033] The region of epithelial, immune, and fibrotic cell intersection is obtained according to the immune region, the fibroblast region, and the epithelial region, and the region of epithelial, immune, and fibrotic cell intersection includes:

[0034] An immune-fibroblast interface;

[0035] Immuno-epithelial junctions;

[0036] Fibroblast-epithelial junctions;

[0037] The step four) calculates the average number of surrounding cells in each region of interest, and counts the number of interactions between the central cells and the adjacent cells.

[0038] Optionally, the step five) performs regression analysis according to the results of cell frequency, marker expression intensity and intercellular interaction in different comparison groups, specifically including the following steps:

[0039] Variables with a zero value proportion less than 50% are screened out, and LASSO regression is used for cross-validation to identify non-zero coefficients; the screened variables are included in the multivariate Cox regression model, and the proportional hazard assumption is evaluated by the cox.zph function; variables meeting the PH assumption are retained, and the association of the variables with survival is evaluated; the results are visualized by the ggforest function.

[0040] A method for establishing a cancer prognosis diagnosis model, comprising the following steps:

[0041] The cancer prognosis marker is screened by the above method;

[0042] The average expression intensity of each marker in the cancer prognosis marker combination screened from the imaging mass spectrometry flow cytometry analysis data set and the prognosis are analyzed by using the imaging mass spectrometry flow cytometry technology, and a cancer prognosis diagnosis model is established by using a machine learning method.

[0043] A screening system for a cancer prognosis marker, comprising:

[0044] A data acquisition module: used for acquiring an imaging mass spectrometry flow cytometry analysis data set, wherein the imaging mass spectrometry flow cytometry analysis data set contains the expression intensity of a marker to be screened in a sample and the prognosis corresponding to the sample; and different comparison groups are divided according to the sample and the prognosis;

[0045] A cell type annotation module: used for single-cell segmentation of images in the imaging mass spectrometry flow cytometry analysis data set, and annotation of cell types of single cells according to the expression intensity of the marker to be screened, to obtain the frequency of cell subpopulations of different cell types;

[0046] A region expression intensity analysis module: used for identifying regions where epithelial, immune and fibrotic cells intersect, and acquiring the expression intensity of markers in the regions where epithelial, immune and fibrotic cells intersect;

[0047] an interaction analysis module for calculating interaction analysis results between different cell types based on cell frequencies and marker expression intensities of different cell types in the region where the epithelial, immune and fibrotic cells meet;

[0048] a regression analysis module for performing regression analysis on cell frequencies, marker expression intensities and cell-cell interaction results in different comparison groups to screen prognostic markers.

[0049] The present application has the following beneficial effects:

[0050] The present application proposes a method for screening cancer prognostic markers based on cell and spatial composition in tumor microenvironment by imaging mass cytometry flow cytometry technology. The method of the present application can obtain the influence of spatial attributes on immune cells, especially macrophage activity by analyzing gastric cancer tumor samples and prognostic conditions. The screening method of the present application obtains CD11c hi macrophages and type I collagen + The spatial interaction of fibroblasts is the main determinant of poor survival of gastric cancer patients. Based on the above method and conclusion, the prediction model based on the markers screened by the present application shows superior prognostic accuracy compared to the model relying on high-dimensional multi-omics data. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0052] Figure 1 is a cell proportion analysis result diagram of different comparison groups.

[0053] Figure 2 is a box plot showing the distribution of the proportion of selected cell types (percentage of total cells) in the clinical group in different groups.

[0054] Figure 3 is the difference in expression of functional markers in specific cell types in different tissue regions.

[0055] Figure 4 is the average frequency of cell-cell interaction in the fibroblast-immune interface region.

[0056] Figure 5 is the result of the output summary of the multivariate survival model.

[0057] Figure 6FIG. 1 is a result graph of survival analysis in the model validation process.

[0058] Figure 7 FIG. 4 is a comparison graph of the evaluation results of the model in the embodiment and existing prognostic evaluation models. DETAILED DESCRIPTION

[0059] Various exemplary embodiments of the present application will now be described in detail, which should be considered to be illustrative of the application and not restrictive of the application. It should be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0060] In addition, for numerical ranges of the present application, it is to be understood that every numerical range

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present application, preferred methods and materials are described. All publications mentioned herein are incorporated by reference to the extent that the

[0062] As used herein, the terms "comprise", "comprising", "include", "including", "have", "having" or the like are open-ended and do not exclude additional, unrecited elements or method steps.

[0063] The present application discloses a screening method of a cancer prognosis marker, comprising the following steps:

[0064] Step one) obtaining an imaging mass cytometry analysis data set, the imaging mass cytometry analysis data set comprising expression intensity of a marker to be screened in a sample and a prognosis condition corresponding to the sample; and dividing different comparison groups according to the sample condition and the prognosis condition;

[0065] Step two) performing single cell segmentation on the image in the imaging mass cytometry analysis data set, and annotating the cell type of the single cell according to the expression intensity of the marker to be screened, to obtain the cell subpopulation frequency of different cell types;

[0066] Step three) identifying the region where epithelial, immune and fibrosis cells intersect, and obtaining the expression intensity of the marker in the region where epithelial, immune and fibrosis cells intersect;

[0067] Step four) Obtain the results of the intercellular interaction analysis based on the cell frequency and marker expression intensity;

[0068] Step five) Perform regression analysis modeling according to the results of cell frequency, marker expression intensity and intercellular interaction in different comparison groups, and screen for prognostic markers.

[0069] Taking the screening of gastric cancer markers as an example, the specific operation steps are as follows:

[0070] The step one) obtains the imaging mass cytometry analysis data set, which contains the expression intensity of the marker to be screened in the sample and the corresponding prognosis of the sample; and different comparison groups are divided according to the sample situation and prognosis, specifically including:

[0071] To identify regional factors affecting the prognosis and clinical characteristics of gastric cancer (GC) patients, a detailed spatial multi-omics analysis was performed on 205 gastric cancer patients. The imaging mass cytometry (IMC) analysis was performed on 205 tumors and 39 tumor peripheral regions of interest (ROI) of 205 gastric cancer patients. This IMC-based single-cell spatial proteomics technology can identify different regions in the gastric cancer tumor microenvironment (TME). In addition, this technology also helps to reveal other spatial features, including cell-cell interactions, such as tertiary lymphoid structures (TLS). By integrating these spatial features with clinical data (such as prognosis), the most influential spatial determinants can be obtained.

[0072] To explore the complex spatial heterogeneity in the microenvironment of gastric cancer (GC) while preserving the structural integrity of the GC ecosystem, we designed an imaging mass cytometry (IMC) panel comprising 40 markers (Table S2). The panel was tailored to gastric cancer patients and optimized from established protocols. The markers we selected span a broad range of cell types, including epithelial, endothelial, and stromal cells, as well as various immune cells. In addition, key cytokines (e.g., IL-1b, TNF-a, and IL-6), the proliferation marker Ki-67, and the apoptosis indicator cleaved caspase-3 were included. Specifically, CD45, CD15, CD45RA, CD103, CD68, HLA-DR, CD45RO, B7H4, CD14, CD20, CD57, PD-L1, CD25, CD3, CD4, aSMA, CD8, CD7, FOXP3, KI67, CD163, CD16, CD69, vimentin, CD11b, CD11c, Caspase3, IL-1b, Ki-67, TNF a, Collagen I, LAG3, GATA3, PD-1, VISTA, CD31, Granzyme B, IL-6, Trypsin, E-Cadherin.

[0073] Prior to IMC use, each antibody was rigorously validated by immunohistochemistry (IHC) to ensure specificity, followed by coupling to a unique heavy metal. After coupling, each metal-labeled antibody was re-validated for performance by IMC.

[0074] Following staining and IMC scanning, the acquired images underwent a series of preprocessing steps to improve image quality. These steps included compensation, noise reduction, and contrast enhancement. Noise reduction was performed by median filtering to effectively suppress background noise; contrast enhancement was applied using a linear regression model. To accurately segment and analyze individual cells or constituents in different channels of IMC images, we employed a deep learning-based segmentation method. This method enabled precise delineation and quantification of expression levels for each marker, followed by quantification of data into matrix format for further analysis.

[0075] Imaging mass cytometry (IMC) single-cell segmentation:

[0076] To perform single-cell segmentation on IMC images, a pre-trained DeepCell model was employed. The DeepCell model requires two imaging channels for segmentation: the first channel is a nuclear stain channel (typically using DAPI staining) to identify the location of cell nuclei; the second channel is a cell membrane or cytoplasmic region to define the boundaries of cells. In this study, the membrane stain channel was chosen as the second imaging channel for segmentation analysis.

[0077] We established four comparison groups based on key clinical and pathological parameters, including tumor and tumor surrounding tissue (group 1), early and late pathological stage (stage I and II, group 2), early and late clinical stage (stage I and II, group 3), and patients were divided into long survival group and short survival group according to overall survival (OS) (group 4). Patients with overall survival more than five years were classified into long survival group, while patients with overall survival less than one year were classified into short survival group (group 4). The comparison of cell frequency between these groups is shown in Figure 1 .

[0078] The step two) is to perform single-cell segmentation on the images in the imaging mass cytometry flow cytometry analysis dataset, and to annotate the cell types of single cells according to the expression intensity of the marker to be screened, to obtain the cell subpopulation frequency of different cell types; specifically comprising:

[0079] IMC pretreatment and cell type identification:

[0080] Firstly, the marker expression was converted using the hyperbolic arcsine function, and the signal value of each channel was limited to the range of 1% and 99% to set the minimum and maximum values. Then, Min-Max Normalization was applied to each channel respectively to standardize the data. To address possible batch effects, the R package Harmony (version 1.2.3) was used for alignment. Subsequently, cell clustering was completed by FastPG (version 0.0.8) with the number of nearest neighbors set to 100 and two rounds of clustering. In the first round of clustering, the main cell populations were identified by the following lineage-specific markers: E-Cadherin, Trypsin, CD3, CD4, CD8, CD20, CD7, CD57, CD45, CD68, CD15, CD14, CD16, CD31, aSMA, Collagen I, Vimentin. To distinguish between myofibroblasts, Collagen I +fibroblasts, mesenchymal cells, endothelial cells, epithelial cells, lymphocytes, myeloid cells. In the second round of clustering, the myeloid cell subpopulation was re-clustered using additional markers FOXP3, CD16, CD69, CD4, CD8, Caspase3, B7H4, VISTA, CD7, CD103, LAG3, CD20, Granzyme B, PD-1, KI67, GATA3, CD45RA, CD3, TNFa, TL1b, CD45RO, CD57, CD25. Lymphocytes were re-clustered using IL-6, CD14, CD16, Caspase3, CD163, PD-L1, CD11B, CD11C, CD15, Ki-67, HLA-DR, TNFa, and TL1b, etc. Epithelial cells were clustered by E-Cadherin, PDL1, KI67, Trypsin, etc. Finally, the average expression level of each marker in each cluster was visualized by heatmap, and the information was used to annotate the corresponding cell type (CT). The expression here refers to whether there is a relatively higher expression level in all the clustered cell types: it is divided into positive expression (+) and negative expression (-), and the positive expression (+) can be further divided into high expression (hi) and low expression (low). For example, in the heatmap with red-blue corresponding to strong-weak, blue is negative expression (-), red is positive expression (+), high expression (hi) is dark red, and low expression (low) is orange or light red.

[0081] After batch correction, we successfully analyzed 1,562,075 single cells and divided them into seven cell clusters. These cell clusters were mainly annotated by classical cell type markers. This analysis identified cell clusters including E-cadherin + epithelial cells, CD3 + or CD20 + lymphocytes, CD68 + or CD15 + myeloid cells, CD31 + endothelial cells, a-SMA + myofibroblasts, type I collagen + fibroblasts and vimentin + mesenchymal cells. Each cluster was then further subdivided according to specific functional markers, providing a map of cell heterogeneity within the dataset. For example, in our analysis, epithelial cells were classified into PD-L1 + , Ki-67 + proliferation subpopulations and E-cadherin positive or negative subpopulations. Lymphocytes were classified into several functionally specific types: Foxp3+ Regulatory T cells, Granzyme B + Effector CD8 + T cells, CD45RO + Memory CD4 + and CD8 + T cells, Effector CD4 + T cells, and CD16 - , CD16 + , Lag3 + CD57 + Natural Killer cells (NK cells), and CD20 + B cells. Furthermore, myeloid cells were subdivided into CD15 + Granulocytes, CD14 + CD16 + Monocytes, Ki67 + Proliferating Macrophages, CD11c low HLA-DR low Macrophages, CD11c hi Macrophages, and HLA-DR hi Macrophages.

[0082] The distribution of cell subpopulation frequencies under different clinical and pathological conditions is shown in Figure 2 The analysis revealed significant differences between the cell population frequencies in tumor tissue and peritumoral tissue for gastric cancer (GC) ( Figure 1 and 2 ). In the tumor area, a significant increase in the proportion of various macrophage subpopulations as well as CD4 + and CD8 + T cell subpopulations was observed, with the only exception being regulatory T cells (Tregs). Contrary to expectations, Tregs were less abundant at the tumor site, a finding that is atypical of the immunosuppressive characteristics commonly found in the tumor microenvironment (TME), as Tregs are generally considered a key immunosuppressive cellular component ( Figure 1 and Figure 2 ). Furthermore, collagen type I + fibroblasts were more common in the tumor area, which is consistent with the generally enhanced fibrotic characteristics exhibited at tumor sites ( Figure 1 and 2 ). In contrast, natural killer (NK) cell subpopulations were significantly reduced at the tumor site, indicating a weakening of the innate anti-tumor immune response ( Figure 1 and Figure 2 ). Also noteworthy was the greater number of epithelial cell subpopulations in the tumor area, which is consistent with the nature of the majority of tumor cells in these areas, which are primarily derived from the epithelial lineage ( Figure 1 and Figure 2 ).

[0083] A comparative analysis of different clinical stages of gastric cancer (GC) was performed. Notably, it was observed that CD4 + and CD8 + memory T cells were at higher levels in advanced stages of the disease, suggesting an increase in memory T cell phenotypes as gastric cancer progresses. More importantly, in terms of survival outcomes, I found that fibroblasts with type I collagen + were more prevalent in the short survival group, suggesting that it has a negative impact on gastric cancer prognosis. Conversely, stromal cells were less prevalent in short survival patients, suggesting that they can have a protective role in disease progression ( Figure 1 and Figure 2 ).

[0084] A comprehensive study of the frequency distribution of cell subpopulations in the microenvironment of gastric cancer (GC) was performed by imaging mass cytometry (IMC), providing important insights into their dynamic interactions in the tumor environment. We identified multiple key subpopulations, including fibroblasts, various macrophage subpopulations, and T cell subpopulations, which can significantly affect the stage and prognosis of gastric cancer patients.

[0085] After studying the cell frequencies under different clinical conditions, we further analyzed the key spatial features provided by imaging mass cytometry (IMC) at single-cell spatial resolution using spatial proteomics data. Our preliminary study focused on tertiary lymphoid structures (TLSs). TLSs consist of dense aggregates of B and T cells and are believed to have an important impact on treatment outcomes. To verify this hypothesis, we adopted a modified version of the patch algorithm, which is specifically designed to detect "tissue patch" -like clusters characterized by significant B cell concentration.

[0086] No statistically significant differences in the size of TLSs were observed between different clinical or survival groups. Therefore, we further investigated the cell frequencies of different cell subpopulations within TLSs to assess how changes within TLSs affect gastric cancer (GC). Not surprisingly, the majority of cells within TLSs were lymphocytes, mainly B cells and CD4 + T cells, which also verified the accuracy of our analysis algorithm. In addition to lymphocytes, we also observed significant populations of stromal cells, including mesenchymal cells and endothelial cells, in TLSs. Furthermore, we detected HLA-DR hi macrophages in TLSs.

[0087] Subsequently, we performed a comparative analysis of cell frequencies under different conditions. Differences between tumor sites and peritumoral tissues were always significant. In the TLSs of tumor tissues, we noted a significant increase in multiple macrophage subpopulations, including HLA-DR hiand CD11c hi macrophages. This finding highlights the key role of macrophages in the TLS dynamics. Surprisingly, we also detected granulocytes in TLS. This might imply the presence of some degree of immune suppressive function within TLS. Moreover, in tumor tissues, an increased subset of epithelial cells in TLS. This phenomenon might reflect a higher concentration of epithelial lineage cells at the tumor site rather than a specific adaptive feature of the tumor microenvironment. Furthermore, our observation that HLA-DR hi macrophages. This finding highlights the key role of macrophages in the TLS dynamics. Surprisingly, we also detected granulocytes in TLS. This might imply the presence of some degree of immune suppressive function within TLS. Moreover, in tumor tissues, an increased subset of epithelial cells in TLS. This phenomenon might reflect a higher concentration of epithelial lineage cells at the tumor site rather than a specific adaptive feature of the tumor microenvironment. Furthermore, our observation that HLA-DR hi macrophages. This finding highlights the key role of macrophages in the TLS dynamics. Surprisingly, we also detected granulocytes in TLS. This might imply the presence of some degree of immune suppressive function within TLS. Moreover, in tumor tissues, an increased subset of epithelial cells in TLS. This phenomenon might reflect a higher concentration of epithelial lineage cells at the tumor site rather than a specific adaptive feature of the tumor microenvironment. Furthermore, our observation that HLA-DR

[0088] Therefore, we extended our analysis to the expression of functional markers within TLS in different comparison groups. Although no significant differences were observed between short and long survival groups in terms of cell subset frequencies within TLS, we found interesting changes in some functional markers. In the short survival group, CD11c hi macrophages. This finding highlights the key role of macrophages in the TLS dynamics. Surprisingly, we also detected granulocytes in TLS. This might imply the presence of some degree of immune suppressive function within TLS. Moreover, in tumor tissues, an increased subset of epithelial cells in TLS. This phenomenon might reflect a higher concentration of epithelial lineage cells at the tumor site rather than a specific adaptive feature of the tumor microenvironment. Furthermore, our observation that HLA-DR hi macrophages. This finding highlights the key role of macrophages in the TLS dynamics. Surprisingly, we also detected granulocytes in TLS. This might imply the presence of some degree of immune suppressive function within TLS. Moreover, in tumor tissues, an increased subset of epithelial cells in TLS. This phenomenon might reflect a higher concentration of epithelial lineage cells at the tumor site rather than a specific adaptive feature of the tumor microenvironment. Furthermore, our observation that HLA-DR + T cells showed higher levels of resident markers expression, including CD69 and CD103. This suggests that CD4 + The resident status of T cells was associated with better prognosis.

[0089] Upon evaluation of other comparison groups, we observed an increase in the levels of immune checkpoints within tumor tissue TLS, including PD-1, PD-L1, and LAG3. Moreover, there was an increase in the levels of pro-inflammatory cytokines within these TLS, such as TNF-a, IL-6, and IL-1b. This phenomenon was partially reproduced in the advanced pathological stage of gastric cancer, where TNF-a levels were also significantly elevated. Similarly, in the advanced clinical stage, there was an increase in the levels of LAG3 and IL-6. In contrast, in the early clinical stage of gastric cancer, functional markers associated with cytotoxic activity, such as granzyme B and GATA3, were more prominent.

[0090] In further investigations, the focus was on elucidating specific patterns of cell-cell interactions within TLSs. Given that B cells are the predominant cell subset in these structures, it was expected that significant interactions involving B cells with other cell subsets would be observed. Notably, in the short-survival group, interactions between type I collagen + The increased interactions between fibroblasts and E-cadherin- epithelial cells suggest that these non-immune cells play a role in mediating immune suppression within TLSs. Furthermore, in this group, B cells interacted with CD11c low HLA-DR low The high frequency of interactions between macrophages further suggests their contribution to the immunosuppressive environment. In contrast, in the long-survival group, HLA-DR hi Interactions between macrophages and monocytes, as well as between effector CD8 + T cells and effector CD4 + T cells were significantly increased. These interactions suggest that in TLSs, there can be a more robust and coordinated immune response between effector CD4 + and CD8 + T cells.

[0091] In summary, the dynamic changes in the composition of TLSs, particularly the frequency of lymphocytes and their functional markers, as well as cell-cell interactions within TLSs, highlight the complex interactions between tumor microenvironments.

[0092] To understand the broader spatial features beyond TLSs, the tumor microenvironment (TME) was explored, which includes distinct regions such as epithelial, immune, and fibrotic. Previous studies have shown that these regions exhibit different interactions and responses to therapy. Therefore, the analysis was extended to these specific regions, which are characterized by continuous cellular extensions. To improve the accuracy of the analysis, the k- nearest neighbors (KNN) algorithm was used, which aims to identify homogenous clusters of cell types. This algorithm helped to delineate different regions within the tissue samples. Subsequently, overlapping regions were precisely mapped, particularly identifying regions where epithelial, immune, and fibrotic cells intersect, such as epithelial-immune, epithelial-fibrotic, and fibroblast-immune interfaces. This approach allowed for a deeper understanding of cell interactions in these complex microenvironments.

[0093] Step three) identifying regions where epithelial, immune, and fibrotic cells intersect, obtaining the marker expression intensity in the regions where epithelial, immune, and fibrotic cells intersect; specifically comprising:

[0094] Identification of spatial regions:

[0095] First, based on spatial coordinates, the k-Nearest Neighbors (KNN) algorithm was used to calculate the 20 nearest neighbors of each cell, thereby constructing a local spatial adjacency graph for each cell. Next, for each cell's neighborhood, the various marker expression data of neighboring cells were aggregated and aggregated according to cell type. Subsequently, the local subgraphs were filtered to ensure that the center node of each subgraph corresponds to a specific cell type (e.g., B cells). These subgraphs were then connected by sharing nodes to form a global connected graph, thereby generating spatial regions / patches. Finally, the spatial coverage of the constructed patches was expanded by adding n neighboring nodes to the existing patches, thereby expanding the spatial coverage of the constructed patches.

[0096] Downstream spatial analysis:

[0097] Based on the spatial connected graph, three main regions were defined:

[0098] Immune region: including "lymphocytes" and "myeloid cells";

[0099] Fibroblast region: including "myofibroblasts" and "type I collagen + fibroblasts";

[0100] Epithelial region: including "epithelial cells".

[0101] Based on the above region definitions, three overlapping regions were defined: immune-fibroblast interface, immune-epithelial interface, and fibroblast-epithelial interface. By calculating the average number of surrounding cells within each region of interest (ROI), the number of interactions between the center cell and neighboring cells was counted.

[0102] Step four) Based on cell frequency and marker expression intensity, obtain the results of cell-cell interaction analysis; specifically including:

[0103] Differential analysis was performed on the cell frequency, marker expression, and cell-cell interaction within the defined regions. Based on the frequency of cell subsets, patients in the short-term survival group and long-term survival group were compared and ranked. Notably, in the long survival group, the frequency of effector CD4 + T cells, B cells, and NK cell subsets increased significantly in the immune region or immune intersection region. This enhancement of effector cell infiltration was positively correlated with better prognosis. In contrast, the frequency of lag3 + NK cells, Ki-67 + macrophages, regulatory T cells (Tregs), and type I collagen + fibroblasts significantly increased, indicating that cell exhaustion and proliferation of immunosuppressive cells may have a negative impact on patient prognosis.

[0104] Further comparison of the functional marker expression of each cell subpopulation within these regions was performed and the most significantly changed markers were identified (Fig. 2B). Figure 3 At the fibro-epithelial interface, there was a significant change in marker expression, suggesting that the tumor fibrotic environment significantly influenced the dynamic changes in markers. In particular, the expression of markers of exhaustion (e.g., PD-1) and pro-inflammatory markers (e.g., TNF-a) was significantly higher in CD4 + T effector cells, lag3 + NK cells, granulocytes, CD8 + T cells and type I collagen + fibroblasts. This pattern suggests that the presence of markers of exhaustion and inflammation in effector cells, memory cells, myeloid cells, and fibroblasts can be associated with a poorer prognosis.

[0105] In contrast, the expression of markers of activation (e.g., GATA3 and HLA-DR), as well as the proliferation marker Ki-67, was significantly higher in effector CD4 + and CD8 + T cells in the long-term survival group ( Figure 3 ). These findings suggest that the activation and proliferation of effector T cells are associated with a better prognosis. This enhanced effector function and expansion in the tumor microenvironment can play a key role in mediating a more effective anti-tumor immune response. Based on the comparison of cell frequencies and marker expression, an analysis of cell-cell interactions was further performed (Fig. 3). Figure 4 Figure 4 The x-axis represents the initial (central) cell type, and the y-axis represents the adjacent cell type with which the interaction occurs. Statistically significant differences in interaction frequencies (p < 0.05) between the long survival group and the short survival group are highlighted with larger markers. It was observed that in the long-term survival group, HLA-DR hi macrophages interacted more frequently with PD-L1 + epithelial cells, NK cells, B cells, and effector CD4 + T cells. This suggests that HLA-DR hi macrophages play a key role in facilitating interactions between tumor cells and immune effector cells, which can contribute to prolonging patient survival.

[0106] In contrast, in the short-term survival group, there was a significant increase in the interaction between CD11c hi macrophages and type I collagen + fibroblasts ( Figure 4 ). This pattern suggests that CD11c hi macrophages and type I collagen + ​Fibroblast-mediated interactions can promote tumor activity and are associated with poor prognosis. Therefore, the distinct roles of macrophage subsets in tumor-immune interactions appear to be crucial in determining the clinical progression and prognosis of gastric cancer.

[0107] By comparative analysis of regional cell frequencies, marker expression, and cell interactions, different mechanisms associated with patient prognosis can be statistically derived: in the long-term survival group, the frequency and activation status of effector cells are significantly increased, while in the short-term survival group, the main manifestation is the exhausted state of immune effector cells and type I collagen+ fibroblasts. Most notably, HLA-DR hi Macrophages can promote tumor activity by upregulating PD-L1 + Epithelial cells and other immune effector cells interact well. In contrast, in the short-term survival group, CD11c hi Macrophages interact with type I collagen + Fibroblasts promote tumor activity. These findings highlight the key role of specific macrophage subsets in modulating tumor-immune dynamics, which are important factors in determining the clinical prognosis of gastric cancer.

[0108] Step five) regression analysis modeling according to the results of cell frequency, marker expression intensity, and cell interaction in different comparison groups, and screening of prognostic markers includes:

[0109] Multivariate survival analysis:

[0110] After completing the spatial analysis, functional marker expression, cell type frequency, and cell interaction were extracted and integrated for multivariate survival analysis. Features with a zero proportion less than 50% were selected, and LASSO regression was used for cross-validation to identify non-zero coefficients. The selected variables were then included in the multivariate Cox regression model, and the proportional hazards assumption (PH assumption) was evaluated by the cox.zph() function. Variables that meet the PH assumption are retained and evaluated for their association with survival. The results were visualized by the ggforest() function.

[0111] After integrating all spatial features (including cell frequencies, regional subpopulation functional marker expression, and cell-cell interactions), multivariate Cox regression analysis was performed. First, variables were constructed based on marker expression levels of each cell type in different regions, cell interactions, and cell frequencies. Subsequently, Lasso regression was used to select representative variables (non-zero coefficients) by introducing a penalty term, which significantly reduced the model complexity (from 4807 variables to 19) and thus avoided overfitting. Next, the proportional hazard (PH) assumption test was performed to ensure that the hazard ratio in the model remained constant over time. In addition, the multicollinearity assessment was performed to confirm the independence of the selected variables, thereby enhancing the reliability and interpretability of the model. Finally, 16 spatial risk factors were identified, as shown in Figure 5 Figure 1. Among the 16 identified spatial risk factors, HLA-DR hi macrophages mediated spatial features associated with favorable prognosis, HLA-DR hi macrophages were found to promote interactions between PD-L1 + epithelial cells and other immune effector cells. This was consistent with the results of the previous regional analysis. In contrast, CD11c hi macrophages and type I collagen + fibroblasts were found to be the primary spatial high-risk factors associated with poor prognosis ( Figure 5 ), which was consistent with the results of the interaction study described above.

[0112] After single-cell spatial deconvolution, we evaluated the key spatial features identified by IMC analysis in each point, including HLA-DR hi macrophage-mediated positive points, and CD11c hi macrophage and type I collagen + fibroblast interaction points. This approach allowed us to accurately assess the presence and distribution of these key spatial features in tissue sections and provided insights into the mechanisms of this spatial organization in the tumor microenvironment.

[0113] Based on the above analysis, four key cell types associated with gastric cancer prognosis were selected, namely PD-L1 + epithelial cells, HLA-DR hi macrophages, CD11c hi macrophages, and type I collagen + fibroblasts. The markers used for clustering analysis of these cells are the key markers that can be used for gastric cancer prognosis prediction.

[0114] Based on the markers identified by the above imaging mass cytometry (IMC) technology analysis, 7 markers were screened in this embodiment, namely CD68, PD-L1, PAN-CK, HLA-DR, CD45, CD11c, and type I collagen. This is because the above method uses the expression of the above 7 markers for clustering identification of the four key cell types screened out above: epithelial cells, HLA-DR hi macrophages, CD11c hi macrophages and type I collagen + fibroblasts, the interaction between the cells, independent images were extracted from each region of interest (ROI), masks corresponding to these cell types were generated, and a survival prediction model was established.

[0115] The method for establishing the model is as shown in Figure 1 , which includes the following steps:

[0116] From the imaging mass cytometry (IMC), two types of integrated image features are designed: marker intensity features and spatial similarity features.

[0117] The marker intensity feature is the average expression intensity measurement of the 7 markers in the IMC image data, CD68, PD-L1, PAN-CK, HLA-DR, CD45, CD11c, and type I collagen.

[0118] The spatial similarity feature is calculated by the structural similarity index (SSIM), which is used to evaluate the interaction between the epithelial cells, HLA-DR hi macrophages, CD11c hi macrophages and type I collagen + fibroblasts in the IMC image data, reflecting the interaction between the epithelial cells, HLA-DR + macrophages, CD11c + macrophages and type I collagen + fibroblasts. The structural similarity index SSIM is an index for measuring the similarity between two images. Unlike traditional mean square error (MSE) or peak signal-to-noise ratio (PSNR) and other pixel difference-based evaluation methods, SSIM takes into account more image brightness, contrast and structural information, so it can more accurately reflect the perception of the human visual system to image quality. These features and the prognosis of the patients in the dataset are input into a deep neural network, and the model trained can be used for clinical prediction of gastric cancer patients.

[0119] The above markers and models are verified by using tissue samples of 27 patients. The imaging mass cytometry (IMC) image results of the tissue samples of the above 27 patients are obtained, and are respectively input into the traditional survival prediction model based on all IMC markers and the cancer prognosis risk assessment model based on the training of 7 key markers in the application, and the obvious advantages can be obtained by combining the marker expression intensity characteristics and the spatial similarity characteristics. The existing prognosis analysis models are used to analyze the above verification data, including RF random forest (Random Forest), GCN graph convolutional network (Graph Convolutional Network) and DeepSurv deep survival model (Deep Survival Model), and the results show that the model of the application shows significant performance improvement in comparison with the existing research (P < 0.05) Figure 6 , Figure 7 ).

[0120] Based on the above screening method, the application further provides a cancer prognosis marker screening system, comprising:

[0121] A data acquisition module is used to acquire an imaging mass cytometry flow cytometry analysis data set, wherein the imaging mass cytometry flow cytometry analysis data set comprises the expression intensity of a marker to be screened in a sample and the prognosis of the sample corresponding to the sample; and different comparison groups are divided according to the sample and the prognosis;

[0122] A cell type annotation module is used to perform single cell segmentation on the image in the imaging mass cytometry flow cytometry analysis data set, and annotate the cell type of the single cell according to the expression intensity of the marker to be screened, to obtain the cell subpopulation frequency of different cell types;

[0123] A region expression intensity analysis module is used to identify the region where epithelial, immune and fibrosis cells intersect, and acquire the marker expression intensity in the region where epithelial, immune and fibrosis cells intersect;

[0124] An interaction analysis module is used to acquire the cell interaction analysis result based on the cell frequency and the marker expression intensity;

[0125] A regression analysis module is used to perform regression analysis modeling according to the cell frequency, the marker expression intensity and the cell interaction result in different comparison groups, and to screen the prognosis marker.

[0126] The system described above can run on a computer device, which includes one or more processors, memories, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The various components communicate with each other using different buses, and can be mounted on a common main board or otherwise mounted as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of a GUI on an external input / output device, such as a display device coupled to the interface.

[0127] The processor can be a central processor, a network processor, or a combination thereof. The processor can further include a hardware chip. The hardware chip can be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device can be a complex programmable logic device, a field programmable logic gate array, a generic array logic, or any combination thereof.

[0128] The memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method shown in the above embodiments.

[0129] The memory can include a program region and a data region. The program region can store an operating system, application programs required by at least one function; and the data region can store data created according to use of the computer device, etc. In addition, the memory can include a high-speed random access memory, and can further include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some optional embodiments, the memory can include a memory disposed remotely with respect to the processor, and these remote memories can be connected to the computer device through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0130] Obviously, the above embodiments are merely examples for clarity, and are not limiting to the embodiments. Based on the above description, other different forms of changes or variations can be made by those of ordinary skill in the art. All the embodiments do not need to be exhausted, and the obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. A method for screening cancer prognosis markers, characterized in that: The steps include: Step 1) obtaining an imaging mass spectrometry flow cytometry analysis dataset, wherein the imaging mass spectrometry flow cytometry analysis dataset includes the expression intensity of the marker to be screened in the sample and the corresponding prognosis of the sample; and dividing the sample into different comparison groups according to the sample condition and prognosis; Step 2) performing single-cell segmentation on the images in the imaging mass cytometry analysis data set, and annotating the cell types of the single cells according to the expression intensity of the marker to be screened to obtain the cell frequencies of different cell types; Step 3) identifying the region where epithelial, immune, and fibrotic cells converge, and obtaining the marker expression intensity in the region where epithelial, immune, and fibrotic cells converge; Step 4) calculating the interaction analysis results between different cell types based on the cell frequencies and marker expression intensities of different cell types in the region where the epithelial, immune, and fibrotic cells intersect; Step 5) Perform regression analysis based on the cell frequencies, marker expression intensities, and cell-cell interactions in different comparison groups to screen for prognostic markers; The cancer is gastric cancer; The annotation of the cell type of a single cell in step 2) comprises the following steps: In the first round of clustering, the main cell population was identified by the following markers: E-Cadherin, trypsin, CD3, CD4, CD8, CD20, CD7, CD57, CD45, CD68, CD15, CD14, CD16, CD31, αSMA, type I collagen, and vimentin; In the second round of clustering, myeloid cells were clustered using FOXP3, CD16, CD69, CD4, CD8, Caspase3, B7H4, VISTA, CD7, CD103, LAG3, CD20, granzyme B, PD-1, KI67, GATA3, CD45RA, CD3, TNFα, TL1β, CD45RO, CD57, and CD25; Lymphocytes were clustered using IL-6, CD14, CD16, Caspase3, CD163, PD-L1, CD11B, CD11C, CD15, Ki-67, HLA-DR, TNFα, and TL1β; Epithelial cells were clustered by E-Cadherin, PDL1, Ki-67, and trypsin.

2. The method for screening cancer prognosis markers according to claim 1, wherein In step 1), different comparison groups will be divided according to sample conditions and prognosis conditions, specifically including: The samples were divided into tumor group and peritumoral tissue group according to whether they belonged to tumor or peritumoral tissue; The samples were divided into early pathological stage group and late pathological stage group according to whether they belonged to early pathological stage or late pathological stage; The samples were divided into early clinical stage group and late clinical stage group according to whether they belonged to early clinical stage or late clinical stage; According to the prognosis, the patients were divided into long survival group and short survival group.

3. The method for screening cancer prognosis markers according to claim 1, wherein The markers to be screened in step 1) include CD45, CD15, CD45RA, CD103, CD68, HLA-DR, CD45RO, B7H4, CD14, CD20, CD57, PD-L1, CD25, CD3, CD4, αSMA, CD8, CD7, FOXP3, KI67, CD163, CD16, CD69, vimentin, CD11b, CD11c, Caspase3, IL-1b, Ki67, TNFα, type I collagen, LAG3, GATA3, PD-1, VISTA, CD31, granzyme B, IL-6, trypsin and E-Cadherin.

4. The method for screening cancer prognosis markers according to claim 1, wherein The cell types include: myofibroblasts, collagen type I⁺ fibroblasts, stromal cells, endothelial cells, epithelial cells, lymphocytes, and myeloid cells; The epithelial cells include the following types: PD-L1 + Subgroup, Ki-67 + Proliferative subpopulation, E-cadherin + Subpopulations and E-cadherin - subpopulation; The lymphocytes include the following types: regulatory T cells, effector CD8 + T cells, memory CD4 + T cells, memory CD8 + T cells, effector CD4 + T cells, CD16 - Natural killer cells, CD16 + Natural killer cells, Lag3 + Natural killer cells and B cells; The myeloid cells include the following types: granulocytes, monocytes, Ki67 + Proliferating macrophages, CD11c low HLA-DR low Macrophages, CD11c hi Macrophages and HLA-DR hi Macrophages.

5. The method for screening cancer prognosis markers according to claim 1, wherein Identifying the region where epithelial, immune, and fibrotic cells intersect in step 3) includes the following steps: using a k-nearest neighbor algorithm to calculate the 20 nearest neighbors of each cell based on spatial coordinates, thereby constructing a local spatial adjacency graph for each cell; for each cell's neighborhood, summarizing the expression data of various markers of neighboring cells and aggregating them according to cell type; Local subgraphs are screened to ensure that the central node of each subgraph corresponds to a specific cell type; the subgraphs are then connected through shared nodes to form a global connectivity graph, thereby generating spatial regions / patches; the spatial coverage of the constructed patches is expanded by adding n neighboring nodes to the existing patch to extend its spatial extent; Based on the spatial connection map, immune regions, fibroblast regions and epithelial regions are obtained; the immune regions include lymphocytes and myeloid cells; the fibroblast regions include myofibroblasts and type I collagen + fibroblasts; said epithelial region comprises epithelial cells; The area where the epithelial, immune and fibrotic cells intersect can be obtained based on the immune area, fibroblast area and epithelial area. The area where the epithelial, immune and fibrotic cells intersect includes: immune-fibroblast junction; immune-epithelial junction; fibroblast-epithelial junction; In the step 4), the average number of surrounding cells in each region of interest is calculated to count the number of interactions between the central cell and the neighboring cells.

6. The method for screening cancer prognosis markers according to claim 1, wherein The step 5) performs regression analysis based on the cell frequencies, marker expression intensities and cell-cell interactions in different comparison groups, specifically including the following steps: Variables with a zero value ratio of less than 50% were screened and cross-validated using LASSO regression to identify nonzero coefficients. The screened variables were incorporated into a multivariate Cox regression model, and the proportional hazards assumption was assessed using the cox.zph function. Variables that met the PH assumption were retained, and their association with survival was assessed. The results were visualized using the ggforest function.

7. A method for establishing a cancer prognosis diagnosis model, characterized in that: The steps include: Screening and obtaining cancer prognosis markers using the method of claim 1; Imaging mass spectrometry flow cytometry technology was used to analyze the average expression intensity of each marker in the cancer prognostic marker combination screened from the data set, as well as the prognostic situation, and a cancer prognostic diagnostic model was established using machine learning methods.

8. A screening system for cancer prognosis markers, characterized in that: include: Data acquisition module: used to obtain imaging mass spectrometry flow cytometry analysis data sets, which contain the expression intensity of the markers to be screened in the sample and the corresponding prognosis of the sample; and to divide different comparison groups according to the sample conditions and prognosis; Cell type annotation module: used to perform single-cell segmentation on images in the imaging mass spectrometry flow cytometry analysis data set, and annotate the cell type of the single cell according to the expression intensity of the marker to be screened, so as to obtain the cell subpopulation frequency of different cell types; Regional expression intensity analysis module: used to identify the areas where epithelial, immune and fibrotic cells intersect, and obtain the marker expression intensity in the areas where epithelial, immune and fibrotic cells intersect; An interaction analysis module, configured to calculate interaction analysis results between different cell types based on the cell frequencies and marker expression intensities of different cell types in the region where the epithelial, immune and fibrotic cells intersect; The regression analysis module performs regression analysis based on the cell frequency, marker expression intensity and cell-cell interaction results in different comparison groups to screen out prognostic markers.

Images