Performance of a PLK1- based immune risk model for prognosis and treatment response prediction in breast cancer

Objective: Polo- like kinase 1 (PLK1), a serine/threonine- protein kinase, functions as a potent oncogene in the initiation and progression of tumor. The aim of this study is to assess potential correlations between PLK1 expression and immune infiltration in breast cancer (BRCA) and construct a PLK1- based immune risk model applicable for prognosis and treatment


| INTRODUCTION
Breast cancer (BRCA) is one of the most prevalent types of cancer, with the second-highest morbidity and the fourth highest mortality across all human cancers. 1 Currently, surgery is the preferred mode of treatment for majority of BRCA cases, followed by a combination of adjuvant or neoadjuvant therapy and radiotherapy. 2 However, patients with distinct subtypes and pathological characteristics require a more flexible treatment regime. For example, patients are often treated with endocrine therapy or combined chemotherapy when the tumor is estrogen receptor/progesterone receptor (ER/PR)-positive or treated with human epidermal growth factor receptor 2-(HER2) targeted drugs (trastuzumab or pertuzumab) or concurrent chemotherapy when the tumor is HER2 positive. If the tumor is both HER2-and ER/PR-positive, endocrine therapy is given separately. Chemotherapy is preferred for patients with triple-negative breast cancer (TNBC), which is a more aggressive subtype of BRCA, exhibiting early onset, rapid clinical progression, and a poor prognosis. 3 The median survival time of metastatic TNBC patients is only 10-13 months. 2 Multimodal combination systemic therapy has shown remarkable efficacy in BRCA. Among these, immunotherapy has achieved the most extraordinary response and produced revolutionary outcomes in the treatment for various solid tumors, particularly improving the overall survival (OS) of patients. [4][5][6] Studies have demonstrated that the OS of TNBC patients with positive programmed cell death ligand 1 (PD-L1) could be prolonged following a single use of programmed cell death protein 1 (PD-1)/PD-L1 inhibitors, such as atezolizumab or pembrolizumab, with a higher efficacy achieved by a combination of chemotherapy and targeted drugs. 7,8 However, immunotherapy is not recommended for all BRCA patients due to different tumor immune microenvironment (TIME) conditions. Therefore, there is an urgent need for robust indicators of BRCA for prognosis and response prediction for efficient treatment of BRCA patients.
Polo-like kinase 1 (PLK1), a serine/threonine-protein kinase, serves as a regulator of mitotic G2/M progression, centrosome maturation, spindle assembly, cell cycle, chromosome segregation, DNA replication, cytokinesis, and meiosis. 9 In addition, PLK1 expression, which is observed in various malignant tumors, often suggests poor prognosis. 10,11 The latest evidence implies an oncogenic role of PLK overexpression in cancer, resulting in defects in mitosis, cytokinesis, centrosome, and cell cycle checkpoints or inducing a persistent increase in chromosome instability, culminating in aneuploidy and tumor formation. 12 Meanwhile, it has been shown that PLK1 stimulates tumor cell proliferation, inhibits apoptosis, and promotes the epithelial-mesenchymal transition (EMT). [13][14][15] Moreover, PLK1 overexpression has been associated with chemoresistance, which can reduce the efficacy of chemotherapeutic agents. 16,17 Therefore, PLK1 is a critical target for cancer treatment. 18 PLK1 also controls the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) and interferon regulatory factor 3 signaling pathways, indicating that it may play a role in cellular immunity and inflammatory signaling, thereby influencing immune infiltration in the tumor microenvironment (TME). 13 However, the association between PLK1 and TME is poorly known and requires further investigation. In conclusion, PLK1 may not only regulate tumor development, growth, and metastasis, but may also have substantial immunological implications in tumors.
Associations between PLK1 and TIME in all cancer types have not been thoroughly studied. In this study, we assessed PLK1 expression and its association with immune cell infiltration and immune regulatory factors in BRCA and developed a risk model based on our screening of PLK1-related immunomodulator hub genes for BRCA progression. Additionally, we assessed the predictive performance of this model by assessing its correlation with clinicopathological factors and the gene mutation spectrum of patients and subsequently constructed a nomogram. Furthermore, we evaluated the efficacy of the model in immunotherapy or chemotherapy response prediction, and our results revealed that this model has a high predictive value for determining treatment regimes and can be implemented for personalized prescription for BRCA patients.

| Data acquisition and differential and survival analysis of PLK1 expression in BRCA
RNA-seq, clinical, and gene mutation data associated with BRCA were obtained from The Cancer Genome Atlas (TCGA) (https://portal.Gdc.Cancer.Gov/). The RNA-seq data consisted of PLK1 gene expression data from 1109 breast cancer (BRCA), immune infiltration, polo-like kinase 1 (PLK1), prognosis, risk model BRCA and 113 normal tissue samples. Clinical data, including survival time, survival status, age, gender, clinical stage, and TMN stage, were collected from 1097 BRCA patients. Gene mutation data included single-nucleotide mutation of genes in 980 patient samples.
BRCA and control breast tissues were compared for differences in PLK1 gene expression. Kaplan-Meier (KM) survival curves and receiver operating characteristic (ROC) curves were plotted in conjunction with the survival data to explore the prognostic significance. Simultaneously, the results were reconfirmed by online differential and survival analysis of PLK1 expression in BRCA using Gene Expression Profiling Interactive Analysis 2 (GEPIA2; http://gepia2.cance r-pku.cn/#index). Additionally, survival analysis of PLK1 expression was utilized to confirm the aforementioned findings in the GSE1456 and GSE7390 datasets.

| Identification of co-expressed genes associated with PLK1
Data on genome sequencing, chromosome copy number, and large-scale parallel sequencing were collected from 1019 human cancer cell lines from the Cancer Cell Line Encyclopedia (CCLE) (https://porta ls.Broad Insti tute.Org/ccle/data). Using CCLE-BRCA RNA-seq data, we searched for genes that were substantially coexpressed with PLK1 (correlation coefficient [Cor] >0.5 and p < 0.001). Enrichment analysis of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) was performed using clusterProfiler and pathview software package in R to elucidate the molecular mechanisms of the highly co-expressed genes associated with BRCA development or progression (p < 0.05). Gene Set Enrichment Analysis (GSEA) was utilized to uncover signaling pathways associated with abnormal PLK1 expression in BRCA.

| Identification of tumor-infiltrating lymphocytes in BRCA
The cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT) algorithm was used to describe immune cell infiltration by calculating the proportion of each subpopulation of the 22 immune cells in BRCA compared to the normal breast tissue samples, 19 with the filtering standard set at p < 0.05. For each immune cell type, the RNA-seq data was transformed into a score using the CIBERSORT method. A p-value of ≥0.05 was considered statistically insignificant. Lastly, for the purpose of identifying tumor-infiltrating lymphocytes (TILs) in BRCA, 727 BRCA and 70 normal breast tissue samples were analyzed. Additionally, we compared the amount of immune cell infiltration in various PLK1 expression groups.
From the 28 TILs, those associated with PLK1 expression in BRCA were confirmed by correlation analysis using TISIDB (http://cis.hku.hk/TISID B/index.php), a portal website for interactions between tumor and immune system. Additionally, the data on clinical stage, immunophenotype, and molecular typing associated with PLK1 expression were acquired. TIMER2.0 (http://timer. comp-genom ics.org), a resource portal that systematically analyzes immune infiltration in pan-carcinomas, was used to assess the associations of PLK1 gene copy number variations (i.e., normal diploid, arm deletion, arm amplification, and high amplification) with the selected TILs (e.g., CD4 + , CD8 + T cells, and B cells) in BRCA.

| Correlations between PLK1 and immunomodulators
We initially selected 42 immunostimulators and 21 immunoinhibitors from the TISIDB and assessed their associations with PLK1 expression in BRCA. GO and KEGG enrichment analysis was performed for PLK1-associated immunomodulators to determine their roles in oncogenetic mechanisms associated with PLK1 in BRCA using WebGestalt (http://www.webge stalt.org/; an online portal for enrichment analysis), with the FDR <0.05. Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; https://strin g-db.org/) was used to resolve interactions between the selected immunomodulators and PLK1.

| Establishment of a PLK1-based immune risk model
Univariate Cox regression analysis was performed on PLK1related immunomodulators (p < 0.05) to screen prognostic genes. Variables with statistical significance were subjected to multivariate Cox regression analysis to determine whether the PLK1-based immune model or the risk score were more powerful in prognosis prediction. BRCA patients were categorized into a high-or low-risk group according to the median risk score of each patient, and KM survival curves and risk curves were drawn. The PLK1-based immune risk model was confirmed using the GSE58812 dataset. We also conducted an infiltrating duct carcinoma and a lobular carcinoma KM survival analysis for distinct pathological types. In addition, correlation analysis was carried out to describe correlations between the model and clinical features, and the independence of this model as a robust predictor for BRCA patients was assessed using the univariate and multivariate CoX regression analysis. ROC curves were drawn to determine the prediction accuracy of the model. Finally, gene mutations in BRCA and tumor mutation burden (TMB) in patients were systematically evaluated. Additionally, the survival difference between the high-and low-TMB groups was analyzed.

| Construction of a nomogram of prognosis prediction
A nomogram for BRCA prediction was constructed based on regression analysis of the model and clinical features using the rms package in R, and the corresponding correction curves were plotted. Each predictive index was scored according to its contribution degree to OS. The OS rate of patients was calculated based on the sum of each index score (or the total score). The correction curve depicting a difference between the actual and predicted risk was plotted to determine the prediction accuracy of the nomogram.

| Response assessment of immunotherapy and chemotherapy for BRCA using the PLK1-based immune risk model
We first identified immune checkpoints that were significantly associated with the risk model from the eight common immune checkpoints in BRCA, PD-1, PD-L1, PD-L2, cytotoxic T-lymphocyte associated protein 4 (CTLA4), lymphocyte activating 3 gene (LAG3), TIM-3, T cell immunoreceptor with Ig and ITIM domains (TIGIT), and lemur tyrosine kinase 3 (LMTK3), to determine the patients' immune status. To determine the patients' response to immunotherapy, the relative immune genome analysis results of next-generation sequencing analysis and immunophenoscores of BRCA patients were available in The Cancer Immunome Atlas database (https://tcia.at/home). The potential difference in response to immunotherapy was predicted between patient groups stratified using the risk model. Lastly, the pRRophetic package in R was employed to predict the drug sensitivity of BRCA patients to four first-line chemotherapy drugs, 20 paclitaxel, cisplatin, doxorubicin, and gemcitabine, based on the 50% concentration inhibition (IC 50 ) values.

| Statistical analysis
Commercially available R (version 4.0.2) and GSEA (version 4.1.0) software were used for all statistical analyses.
Survival differences between high-and low-risk BRCA groups were assessed using the Log-rank test. Continuous variables were evaluated using the Wilcoxon test. The chi-square test was utilized for discontinuous variables. Spearman's rank correlation test was employed for all the correlation analyses. All tests were two-tailed tests and a p-value of <0.05 was regarded as statistically significant.

| PLK1 gene expression in BRCA tissues and several BRCA subtypes
Using TCGA data, we initially discovered that PLK1 expression differed between the BRCA and normal breast tissue samples, with the PLK1 expression much lower in the normal tissues compared to the BRCA tissues (p < 0.001). KM survival analysis revealed a shorter survival time in the high PLK1 expression group versus the low PLK1 expression group (p = 0.021, n = 1097). Furthermore, the ROC curves suggested a moderate accuracy of PLK1 gene expression in predicting BRCA prognosis (AUC = 0.649; Figure 1A). GEPIA2 analysis supported PLK1 upregulation in BRCA and poor survival in PLK1 overexpression cases ( Figure 1B). These findings were further confirmed in the GSE1456 and GSE7390 datasets, indicating consistent survival (p < 0.001) and strong predictive effectiveness (AUC = 0.791, 0.700; Figure 1C,D). Furthermore, there was a significant difference in PLK1 expression between different clinical stages and BRCA subtypes ( Figure 1E-G). PLK1 expression levels were significantly higher in stage II, III, and IV patients than in stage I patients, and the most significant increase was observed in stage IV patients (p = 0.00826). PLK1 expression was exceptionally high in HER2-positive and luminal B-like, particularly basal-like, BRCA samples compared to the normal control samples. Usually, basal-like BRCA is TNBC because they are highly overlapping and have a poor prognosis. 21 In addition, PLK1 expression was significantly correlated with immune subtypes, with the lowest expression in the C3 subtype and the highest expression in the C2 subtype.

| Co-expression genes of PLK1 in BRCA
We analyzed potential correlations between the RNA-seq data of 51 BRCA cell lines, obtained from the CCLE database, with PLK1 expression, and initially identified 348 genes that were highly co-expressed with PLK1. GO enrichment analysis of these genes revealed that under biological functions the genes were associated with organelle fission, nuclear fission, and chromosome segregation; under cell components the genes were enriched in chromosomal regions and spindle apparatus; and under molecular functions the genes were associated with tubulin binding and ATPase activity (Figure 2A,B). The KEGG enrichment analysis of the co-expressed genes revealed that they are may be associated with cell cycle, cellular senescence, and P53 signaling pathways in BRCA ( Figure 2C,D). The results of enrichment analysis supported the regulatory role of PLK1 as a serine/threonine-protein kinase at the M phase of cell cycle, indicating that the co-expressed genes were actively involved in oncogenetic role of PLK1 in BRCA. In addition, the co-expressed genes regulating cell cycles were highly positively correlated with Cyclin B (CycB), cell division cycle 25 B/C (Cdc25B/C), MCM, and origin recognition complex subunit 1 expressions levels, and negatively correlated with Mpsl, stromal antigen 1 (Stag1), and Stag2 ( Figure 2E). Additionally, the coexpressed genes were positively correlated with CycB, G2, and S-phase expressed 1 (also named B99), P53R2, and protein kinase Chk1 (CHK1) that are involved in the P53 signaling pathway ( Figure 2F).

| Association of PLK1 expression and immune cell infiltration in BRCA
GSEA enrichment revealed that PLK1 not only disturbed cell cycle (NES = 2.23, p < 0.001), but affected various immune-related signaling pathways, including the T-(NES = 1.84, p < 0.001) and B-cell receptor signaling pathways (NES = 1.75, p < 0.001). Among others, the chemokine signaling pathway (NES = 1.60, p < 0.001) was particularly essential ( Figure 3A) as it implied the role of T cells ( Figure 3C). Moreover, the degree of CD8 + T cell, CD4 + T cell, neutrophil, macrophage, and myeloid dendritic cell infiltration was significantly associated with the copy number variations of PLK1 in BRCA ( Figure 3D).
In addition, the correlation analysis highlighted 14 tumor-infiltrating immune cells associated with PLK1 expression in BRCA ( Figure 4A The results from the correlation analysis between PLK1 expression and 28 TILs using the TISIDB database were nearly consistent with the above findings ( Figure 4B), demonstrating that PLK1 expression was involved in regulating immune cell infiltration during the occurrence and development of BRCA.

| PLK1 regulates the immunomodulator network in BRCA
Forty-six immunomodulators associated with PLK1 expression in BRCA were obtained from the TISIDB database, including 30 immunostimulators, such as CD80 and CD8, and 16 immunoinhibitors, such as CD274, CTLA4, and IL10 (Figures 5 and 6). Protein-protein interaction (PPI) networks were plotted for the 46 immunomodulators ( Figure 7A). The GO and KEGG enrichment analysis indicated that PLK1 might participate in response to stimulus, protein binding, and cytokine-cytokine receptor interactions ( Figure 7B,C) via regulating these immunomodulators.

| Establishment of a PLK1-based immune risk model
Considering the direct or indirect susceptibility of a single gene to environmental factors, a multiple-gene signature was considered for a stable prediction. The univariate Cox regression analysis revealed 12 immunomodulators most related to BRCA prognosis, and their relevancy was shown using the forest map ( Figure 8A). Thereafter, seven genes (PLK1, IDO1, KLRC1, LTA, RAE1E, TNFRSF13C, and TNFRSF14) were confirmed by the multivariate Cox regression analysis ( Figure 8B). The risk score of each patient was calculated based on the expressions of these genes as follows: risk score = PLK1 × 0.1916 + IDO1 × (−0.1807) + K LRC1 × (−0.6780) + LTA × 0.6634 + RAET1E × 0.3713 + TN FRSF13C × (−0.4095) + TNFRSF14 × (−0.3574). All BRCA patients from TCGA were assigned to the high-(n = 545) or low-risk (n = 545) group based on the risk score, and the prediction performance of the model was assessed. The KM survival analysis revealed a shorter OS of the highrisk patients compared to the low-risk patients (p < 0.001) ( Figure 8C). Additionally, univariate and multivariate Cox regression analysis were used to compare the risk model to clinical characteristics, and the findings suggested that this model is independent of clinical features in predicting prognosis ( Figure 8D,E). Furthermore, as evidenced by ROC curves, this model demonstrated acceptable accuracy, which was considerably greater than PLK1 overexpression (AUC values: 0.690 vs. 0.649; Figure 8F). AUC of 0.798 was observed when the risk model was merged with clinical characteristics such as age, clinical stage, and TNM stage, improving the performance of the combined model in prognosis prediction. Additionally, to achieve more effective predictive performance, we further defined the PLK1-associated model as having a survival significance consistent with the above model in both lobular and infiltrating duct carcinoma ( Figure 8G,H), with an AUC value of 0.798 in lobular carcinoma ( Figure 8I) indicating higher sensitivity. The risk curve showed higher mortality in high-risk patients compared to the low-risk patients ( Figure 9A). Intriguingly, our model also had associations with the clinical stage (p = 0.013), particularly the N stage (p = 0.012), of the high-risk BRCA patients ( Figure 9B). Our results ascertained a good performance of the PLK1based immune risk model in predicting BRCA progression. Lastly, the PLK1-based immune risk model was evaluated using the TNBC dataset GSE58812 and demonstrated consistency (p = 0.002, AUC = 0.725; Figure 9C).
On the basis of the total score, a predictive nomogram integrating the risk score, age, and clinical stage was built to estimate the 1-, 3-, and 5-year survival probability of BRCA patients ( Figure 10A). The calibration curve revealed a minor discrepancy between the nomogram's predicted 5-year survival rate (gray line) and the actual 5-year survival rate (red line) of the patients ( Figure 10B). As a result, the prognostic nomogram demonstrated a high degree of predictive potential for BRCA survival.

BRCA patients
The gene mutation profiles of BRCA patients were also evaluated using the risk model. It was shown that the PI3KCA, TP53, and TTN genes have common missense mutations in the patients. Among these, TP53 and TTN mutation rates were observed in the high-risk group, while PI3KCA and CDH1 mutation rates were observed in the low-risk group ( Figure 10C,D). In contrast, TMB demonstrated no correlation with patient prognosis (p = 0.195; Figure 10E). However, high-risk BRCA individuals had a significantly higher TMB value (p < 0.001; Figure 10F). The raw data of TMB for the selected BRCA cases are shown in Table S1.

| The PLK1-based immune risk model predicts immunotherapy and chemotherapy response of BRCA patients
We carried out the assessment for immunotherapy response prediction using the risk model and eight immune checkpoints (PD-1, PD-L1, PD-L2, CTLA4, LAG3, TIM-3, TIGIT, and LMTK3), which were significantly upregulated in low-risk patients ( Figure 11A). The immunotherapy analysis revealed an insufficient response to anti-CTLA4 and/or anti-PD1 drugs in high-risk BRCA patients ( Figure 11B). For chemotherapy response, the drug sensitivity analysis based on IC 50 values showed a good response to paclitaxel, cisplatin, and gemcitabine in low-risk patients compared to the high-risk patients; however, there was no significant difference in patient response to doxorubicin ( Figure 11C). These findings suggest that our model can serve as a useful tool for guiding therapeutic decision-making for BRCA patients.

| DISCUSSION
The immune system has a critical role in the onset and progression of BRCA. Immune cell infiltration, particularly of CD4 + or CD8 + T cell, B cell, monocyte, macrophage, dendritic cell, and NK cell infiltration, develops in lockstep with progressive cancerization of breast tissues. [22][23][24] Several studies have demonstrated that TIL levels have tight associations with prognosis and improvements in adjuvant or neoadjuvant treatment response of BRCA patients, with higher TIL levels representing a lower recurrence and mortality risk. [25][26][27] However, different subtypes of TILs have different functions. CD8 + T and T-bet + cells are indicators of a better prognosis or disease-free survival, 28,29 while FOXP3-positive TILs like Tregs often indicate poor OS and a higher risk of recurrence. 30 Different immunomodulatory factors can stimulate an anti-tumor immune response or restrain tumor immune escape during carcinogenesis or tumor progression. Immune checkpoints have been extensively studied in clinical and basic research, particularly PD-L1, whose expression is associated with poor differentiation and prognosis in BRCA. 31 However, when paired with chemotherapy, the PD-L1 inhibitor, atezolizumab, has shown significant effectiveness in advanced TNBC. 8,32 Patients with advanced BRCA have poor prognosis and limited treatment options. The use of immunotherapy, especially immune checkpoint inhibitors (ICIs) combined with chemotherapy or targeted drugs can be an effective treatment for these patients; however, the overall response to ICI is insufficient. 33,34 Therefore, the first priority is to identify patients who may benefit from ICI therapy by analyzing their TIME conditions before the detection of any signs of tumor development, as this may aid in appropriate therapeutic decision-making. However, the absence of reliable biomarkers for predicting BRCA prognosis, TIL distribution, and immunotherapy or chemotherapy response impede the resolution of this issue. Our study aimed to develop an immunological risk model based on PLK1 expression that could account for the aforementioned factors. PLK1 is involved in a variety of tumorogenic processes, including carcinogenesis, proliferation, EMT, and chemoresistance. [13][14][15][16] PLK1 is also involved in the regulation of NF-κB signaling pathway, which is crucial for the immunological response or inflammation in the body. 13,35 Therefore, PLK1 is a promising biomarker for TIME assessment in BRCA patients. First, we confirmed that PLK1 overexpression is associated with shorter OS in BRCA patients. Although the survival curves intersected, we noted that in the 20-year follow-up of the BRCA cohort, which included 1090 people, the number of persons with 8 years of follow-up (102) was <10%, and the number of persons with 10 years of follow-up (40) was <5%. As a result, we concluded that a cohort with such a minimal number of observations already has a significant bias. The results indicated that during the ninth year, the survival rate of the otherwise smooth low-PLK1 expressing group decreased abruptly (with three deaths recorded in the low-expression group and one death recorded in the high-expression group) and intersected for the first time with the curve of the high-expression groups, which was interpreted as a chance result due to the small sample size. Additionally, the findings were confirmed in a  wider queue. Increased PLK1 expression was also associated with clinical stage, particularly stage IV BRCA. These data suggest an oncogenic role of PLK1 overexpression in BRCA, which promotes tumor initiation and progression and predicts a poor prognosis. Our results corroborate those obtained in previous investigations of PLK1 expression in malignant tumors. 12,17 Additionally, we observed a differential expression of PLK1 across BRCA subtypes, with an exceptionally high level of expression in basal-like BRCA, suggesting that PLK1 may be a therapeutic target in these individuals.
Comparison of the BRCA tissues with the normal breast tissues revealed substantial shifts in the levels of various TILs, including CD8 + T cells, NK cells, Tregs, and dendritic cells, across time. There is a possibility that these TILs are directly involved in the emergence and evolution of BRCA. However, studies have revealed no link between PLK1 expression and immune cell infiltration. This study has identified 14 types of immune cells in BRCA, including Tregs, 36 which have been shown to assist in the creation of an inhibitory microenvironment and demonstrated to be positively associated with PLK1 expression in BRCA. NK cells are responsible for tumor monitoring and destruction, whereas M2 macrophages often promote tumor development and dissemination. 37 According to the TISIDB study, PLK1 expression is positively correlated with the infiltration of activated CD8 + T cells, which have a substantial anti-tumor role. However, the associations between PLK1 and immune cell infiltration indicated above are similar to those found in the TISIDB analysis. Therefore, there are two lines of evidence supporting the role of PLK1 in both immunosuppression in TIME and the anti-tumor immune response. A high enrichment of the PLK1 gene in T cell receptors, B cell receptors, or molecules of the chemokine signaling pathway was found using GSEA enrichment analysis, demonstrating that PLK1 is involved in controlling the immunity against BRCA via the regulation of immune cell infiltration. Further investigations are required to confirm the regulatory mechanisms of PLK1 in immune cell infiltration.
This was the first study to demonstrate that PLK1 was associated with a wide range of immunostimulators (e.g., CD80 and CD86) and immunoinhibitors (e.g., PD-1, PD-L1, and CTLA4). However, PPI networks revealed a very complex protein network between these components, making it impossible to discern the regulatory functions of PLK1 in tumor immunostimulation and immunosuppression. However, enrichment analysis revealed that immunomodulatory variables associated with PLK1 were mostly engaged in cellular response to stimuli, protein binding, and cytokine-cytokine receptor interactions. This finding suggested that PLK1 was extensively involved in TIME regulation in BRCA. We identified seven genes of PLK1-associated immunomodulators (PLK1, IDO1, KLRC1, LTA, RAE1E, TNFRSF13C, and TNFRSF14) for construction of a Cox regression model, which was expressed by a hazard function. Subsequently, survival analysis and ROC curves exhibited an excellent performance in predicting BRCA prognosis, suggesting that they might be used as independent prognostic factors for BRCA survival prediction. Although clinical stage was more predictive than signature in the first 3 years, the AUC value of signature surpassed that of clinical stage after 3 years, showing that it has greater long-term predictive importance. More importantly, the risk model demonstrated outstanding predictive performance for lobular carcinoma, which is predicted to result in more effective treatment implications for lobular carcinoma patients. Furthermore, the model with integrated clinical characteristics had even greater prediction accuracy. In comparison to low-risk individuals, high-risk individuals may have more objective signs, such as clinical stage and lymph node involvement. This data implies a more rapid clinical development in high-risk BRCA individuals, emphasizing the vital need of early intervention for treatment effectiveness. Additionally, we developed a prognostic nomogram for estimating 1-, 3-, and 5-year survival rates, making the risk model prediction accessible.
Missense mutations are the most common type of mutations in BRCA, followed by frameshift and nonsense mutations, 38 and prior investigations have regularly revealed PIK3CA, TP53, GATA3, and CDH1 mutations in BRCA. 39,40 To understand the immunological characteristics of our risk model at the gene level, we compared the difference in gene mutations between the high and low-risk groups. Our results revealed that PIK3CA, TP53, TTN, CDH1, and GATA3 mutations were most frequent (>10%) in BRCA, in which missense mutations are common. Among them, the probability of TP53 mutation was exceptionally high in high-risk patients. TP53 is a key tumor suppressor gene, and TP53 expression in noncancerous tissues is closely associated with cell-cycle arrest, DNA repair, and apoptosis, all of which contribute to the acceleration of tumor incidence and progression. There is evidence that the TP53 mutation confers a poor prognosis for BRCA 41 patients, which is consistent with our survival data.
Additionally, we tested the predictive ability of the model for immunotherapy response. Anti-PD-1 and/or anti-CTLA4 drugs had a substantial advantage on the low-risk BRCA patients identified by our algorithm. Immune checkpoint expression, especially PD-L1, is a useful biomarker for monitoring ICI therapy, 42 and there is evidence that PD-L1 positive BRCA individuals respond well to ICI treatment. 8,32 In this study, eight immune checkpoints, including PD-1, PD-L1/PD-L2, and CTLA4, were observed to be significantly upregulated in the low-risk patients, indicating a potential association between PD-L1 expression and ICI treatment benefits in BRCA, as reported previously. TMB is a critical biomarker for identifying cancer patients who may benefit from ICI treatment because it increases tumor immunogenicity primarily by increasing the amount of mutant proteins that generate novel antigenic peptides. 43,44 In this study, we evaluated the overall prognostic significance of TMB before establishing its predictive value for ICI treatment. We found that TMB had no significant effect on the OS, excluding the interference from TMB, which can be used as an indicator for ICI treatment efficacy. A cohort study involving 8207 patients with various tumor types, including bladder cancer, non-small cell lung cancer, melanoma, etc., showed that high TMB was associated with prolonged OS after anti-PD-1/PD-L1 and/or anti-CTLA4 treatment, and the treatment benefits increased with the increase in TMB value. 45 Moreover, a study on TMB in BRCA patients, in conjunction with other latest evidence, shows that high TMB is also related to longer progression-free survival following anti-PD-1/PD-L1 treatment. 46 However, it was shown that following immunotherapy, there was no discernible difference in the survival of BRCA patients with varying TMB levels. 45 We observed a greater TMB level in patients with a high-risk score, but the usage and research of TMB in BRCA are presently inconclusive for a variety of reasons. For instance, BRCA is a less immunogenic tumor, and TMB levels are often lower in BC, 47 indicating an inadequate specificity for predicting immunotherapy benefits. Therefore, further study is required to establish the importance of TMB in BC.
For BRCA treatment, majority of doctors recommend chemotherapy, particularly in advanced patients. To assess the model's capacity to predict the effectiveness of chemotherapy, we compared the IC 50 values of four commonly used BRCA chemotherapeutic agents in high-and low-risk individuals. Paclitaxel, cisplatin, and gemcitabine had considerably lower IC 50 values in the low-risk group, suggesting that this group had a greater sensitivity to these agents. In general, the risk model successfully identified low-risk populations that responded well to either common immunotherapy agents or chemotherapy drugs, indicating that it is a promising biomarker and a cost-effective method of providing early prognosis prediction and response guidance for individual patients.
In summary, PLK1 upregulation is essential in TIME regulation in BRCA and facilitates in the determination of TIL distribution and immune regulating factors expression in BRCA patients. The immunological risk model we developed on the basis of PLK1 is capable of effectively predicting patient prognosis, tumor growth, the extent of immunotherapy and chemotherapy response, and identifying gene alterations. Its predictive ability may be enhanced by including clinical characteristics. The related nomogram graphically depicts the survival prediction data, making it practical and useful for BRCA therapy. In clinical practice, refractory BRCA and TNBC pose many challenges in the determination of treatment strategies, and our research provides a novel strategy for BRCA treatment, suggesting the use PLK1 inhibitors in combination with ICIs. We expect that this novel treatment strategy can significantly improve the prognosis of BRCA patients. AUTHOR CONTRIBUTIONS Yan Chen: Resources (equal); validation (equal). Yiqing You: Software (equal); writing -original draft (lead). Qiaoling Wu: Data curation (equal); software (equal). Jing Wu: Formal analysis (equal); validation (equal). Shujing Lin: Validation (equal). Yang Sun: Conceptualization (equal); supervision (equal); writing -review and editing (equal). Zhaolei Cui: Conceptualization (equal); funding acquisition (equal); writing -review and editing (equal).