A comprehensive pan-cancer analysis of the prognostic role and immunotherapy efficacy evaluation of Glyoxalase 1

Abstract

Background

Glyoxalase 1 (GLO1), an enzyme responsible for breaking down methylglyoxal (MG), is increasingly recognized as a tumor-promoting factor due to its role in removing cytotoxic MG. This study aims to analyze the expression status and prognostic significance of GLO1 across various cancers and explore its potential role in cancer immunotherapy.

Methods

We obtained GLO1 mRNA expression profiles from The Cancer Genome Atlas (TCGA) database. Kaplan-Meier analysis was used to evaluate the prognostic value of GLO1 in multiple cancers. We validated GLO1 protein expression in clinical endometrial cancer samples using immunohistochemical staining. Additionally, we analyzed GLO1’s correlation with drug sensitivity from the Genomics of Drug Sensitivity in Cancer (GDSC) database and its relationship with the tumor microenvironment and immunotherapy response.

Results

GLO1 was widely expressed in human tumor tissues with significant differences compared to adjacent normal tissues. High GLO1 expression was associated with poor prognosis in 14 cancers, including adrenocortical carcinoma, cholangiocarcinoma, and lymphoid neoplasm diffuse large B-cell lymphoma. GLO1 expression was negatively related to immune cell infiltration in breast invasive carcinoma. In melanoma and kidney renal clear cell carcinoma, high GLO1 expression was linked to poor immunotherapy outcomes.

Conclusions

This comprehensive pan-cancer study reveals the prognostic value and potential predictive role of GLO1 in immunotherapy across multiple tumor types. This study primarily utilized bioinformatic datasets; experimental validation was performed for endometrial cancer but warrants expansion to other cancer types in future work.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04109-3.

Introduction

Glyoxalase 1 (GLO1) is the key protein of glyoxalase system responsible for detoxifying methylglyoxal (MG), a cytotoxic byproduct of glycolysis. In the context of cancer, GLO1 has garnered significant attention due to its potential role in promoting tumor progression. ELevated GLO1 expression has been associated with increased proliferation, survival, and metastasis of cancer cells across various types of cancer [1]. This is attributed to its ability to regulate glucose metabolism for energy production, alter cellular energy and redox homeostasis, and defend against electrophilic carbonyl and chemotherapeutic stress [2]. For instance, in breast cancer, GLO1 overexpression is linked to positive lymph node status, lymphovascular invasion, and advanced TNM stage. Similarly, in glioblastoma, inhibition of GLO1 induces apoptosis in cell lines [3]. Furthermore, emerging evidence underscores its role at the interface of tumor metabolism and immunity. Recent studies have demonstrated that GLO1 activity, by scavenging the immunogenic metabolite MG, can dampen antitumor immune responses. This occurs through limiting the formation of MG-derived advanced glycation end-products (AGEs), which are known to activate the receptor for AGEs (RAGE) on immune cells and promote a pro-inflammatory tumor microenvironment [4]. Consequently, high GLO1 activity may foster an immunosuppressive niche, facilitating immune evasion.However, the role of GLO1 in cancer is complex and may vary depending on the specific tumor type and its metabolic context. This metabolic-immune interplay is further highlighted by findings that GLO1 expression can be modulated by oncogenic signaling pathways and, in turn, influence the expression of immune checkpoint molecules like PD-L1, suggesting a direct mechanistic link between glycolytic stress, GLO1, and cancer immunotherapy response [5]. While existing studies have characterized GLO1’s role in individual cancers, its pan-cancer prognostic significance and mechanistic links to immunotherapy resistance remain poorly understood. Furthermore, many prior analyses lacked adjustment for key clinical confounders, potentially obscuring its true prognostic impact.

Tumor microenvironment (TME) plays a crucial role in cancer progression, encompassing non-cancerous host cells such as immune cells, fibroblasts, endothelial cells, as well as extracellular matrix and soluble product [6]. Fibroblasts contribute to tumor systemic metastasis and provide a passage for endothelial cells undergoing angiogenesis [7]. Immune cells such as macrophages, lymphocytes are related to the immune escape of tumor cells and antitumor immune responses [7]. Immune checkpoint blockade like programmed cell death ligand-1 (PD-L1) and cytotoxic T lymphocyte associated antigen 4 (CTLA4) have been widely investigated and already clinically used for cancers treatment such as melanoma. However, the effectiveness of cancer immunotherapy varies greatly. Given the potential impact of GLO1 on both cancer progression and immunotherapy response, a comprehensive pan-cancer analysis is warranted to elucidate its expression patterns, prognostic significance, and interactions with the TME across multiple tumor types. Therefore, it is necessary to explore novel potential targets for cancer immunotherapy and find accurate biomarkers for outcome prediction.

In this study, we searched the Cancer Genome Atlas (TCGA) datasets and the Gene Expression Omnibus (GEO) datasets to perform a pan-cancer analysis of GLO1 across 33 cancers, and evaluated the prognostic value of GLO1 in pan-cancer based on the TCGA datasets. Further investigation identified several drugs and miRNA which could regulate GLO1 expression. Furthermore, the association between GLO1 and tumor immunotherapy response using three immunotherapy response datasets was unveiled. Our study preliminarily unveiled the potential application of the GLO1 as a predictive biomarker of immunotherapy response, which warrants further investigation.

Materials and methods

Clinical samples

Paraffin embedding endometrial cancer (EC) samples and normal endometrial tissue 10 cases each of breast cancer and melanoma tissues were collected from pathology department of Zhejiang Provincial People’s Hospital between May 1, 2018 to June 30, 2020. Additionally, 10 cases each of breast cancer and melanoma tissues were included for immunohistochemical validation. This study was approved by the Ethics Committee of Zhejiang Provincial People’s Hospital (Approval No.: 2021QT207), and informed consent was obtained from all participants.

The key clinical and pathological characteristics of the patient cohorts are summarized in Table 1. Briefly, the endometrial cancer cohort included a range of histological subtypes and FIGO stages to ensure representativeness. The breast cancer samples encompassed different molecular subtypes, and the melanoma samples included variations in Breslow depth and ulceration status. All tissue samples were obtained from treatment-naïve patients prior to any systemic therapy or radiotherapy to avoid potential confounding effects from prior treatments.

Table 1.

Clinical and pathological characteristics of the patient samples used in this study

Cancer Type	Case (n)	Age (years)	Sex (M/F)	Key clinical characteristics
Endometrial cancer (EC)	10	62.5 ± 8.2	0/10	Histology: 7 Endometrioid, 2 Serous, 1 Carcinosarcoma FIGO Stage: 4 Stage I, 3 Stage II, 3 Stage III Grade: 5 G1, 3 G2, 2 G3
Breast cancer (BRCA)	10	55.1 ± 10.5	0/10	Molecular Subtype: 4 Luminal A, 3 Luminal B, 2 HER2+, 1 Triple-Negative TNM Stage: 6 Stage II, 4 Stage III
Melanoma (SKCM)	10	58.8 ± 12.1	6/4	Breslow Depth: 4 T2, 4 T3, 2 T4 Ulceration: 5 Present, 5 Absent Stage: 5 Stage II, 5 Stage III

GLO1 expression analysis

GLO1 mRNA expression profiles in tumor and corresponding normal tissues were obtained from the TCGA project [8](http://cancergenome.nih.gov/). Available data were downloaded from the University of California Santa Cruz (UCSC) Xena database [9] (https://xenabrowser.net/datapages/). The gene expression levels were normalized using log2 conversion. All normal samples with fewer than 5 samples were excluded. GLO1 expression levels between paired tumors and adjacent normal tissues were analyzed with R Studio (version 4.3.1). The “ggpubr” R package (version 0.6.0, available at: https://CRAN.R-project.org/package=ggpubr) was used to visualize the results.

GLO1 protein localization

The protein expression profiles of human tissue types were displayed in Human Protein Atlas (HPA, www.proteinatlas.org) [10]. Immunofluorescence (IF) staining images of three cancer cell lines (A-431, U-251MG and U2OS) were used to show the localization of GLO1 protein in cancer cells.

Correlation of GLO1 expression and survival prognosis

The correlation between GLO1 expression and survival indicators, including overall survival (OS) and progression-free survival (PFS), was examined by the Kaplan-Meier method and Cox proportional hazards regression models. Cox models were adjusted for key clinical variables indcluding age, gender, and tumor stage where data was available to control for potential confounding. We prioritized adjustment for these three factors due to their universal prognostic importance and high data completeness across the TCGA cohorts. t is important to note that other potential confounders, such as tumor grade and specific treatment modalities, were not included in the primary multivariate model. This decision was made primarily because this information is not consistently available for a large proportion of patients within the pan-cancer TCGA dataset. The heterogeneity and significant missingness in treatment data across different cancer types would have drastically reduced the sample size for a unified analysis and introduced substantial bias. Therefore, to maintain the robustness and comparability of our pan-cancer survival analysis, we relied on the core set of commonly available clinical variables (age, gender, stage). The cut-off value of GLO1 expression level was determined by the “surv-cutpoint” function of the “survminer” R package (version 0.4.9, available at: https://CRAN.R-project.org/package=survminer). The analysis was performed using R package “survival” (version 3.5-5, available at: https://CRAN.R-project.org/package=survival) and visualized by the “survminer” and “forestplot” R packages.

Drug sensitivity of GLO1 in pan-cancer

To explore the correlation between GLO1 expression and drug sensitivity, we utilized data from the Genomics of Drug Sensitivity in Cancer (GDSC) database [11]. The primary drug sensitivity data (IC50 values) and gene expression profiles were downloaded from the GDSC portal (https://www.cancerrxgene.org/).

Drug Selection Criteria: To ensure the biological and clinical relevance of our analysis, we applied the following criteria to select drugs from the GDSC database: (1) Comprehensive Coverage of Anti-cancer Mechanisms: the prioritized drugs that represent a wide spectrum of established anti-cancer mechanisms, including but not limited to: kinase inhibitors, DNA damage agents, antimetabolites, epigenetic modifiers, and apoptosis inducers. This approach allows for an unbiased discovery of potential connections between GLO1 and diverse cellular pathways. (2) Clinical and Pre-clinical Relevance: We included a significant number of drugs that are either currently approved for clinical use, are in advanced clinical trials, or are well-characterized tool compounds with defined molecular targets. This enhances the translational potential of our findings. (3) Data Availability and Quality: We filtered for drugs that had robust sensitivity data (IC50 values) available across a large number of cancer cell lines in the GDSC dataset to ensure statistical power for the correlation analysis. Based on these criteria, a final panel of 176 drugs was selected for the pan-cancer analysis. This panel encompasses a broad range of therapeutic classes, increasing the likelihood of identifying clinically meaningful associations with GLO1 expression.

The analysis was conducted using the “oncoPredict” R package to predict drug response based on gene expression profiles [12]. We selected a panel of drugs that are clinically relevant and have demonstrated significant interactions with GLO1 expression. The drugs and their targets were chosen based on their known mechanisms of action and relevance to cancer treatment. The selected drugs include: AZ6102 (Target: TNKS1/2) - A Tankyrase inhibitor used in the treatment of various cancers. P22077 (Target: USP7) - A selective ubiquitin-specific protease inhibitor with potential applications in cancer therapy. WIKI4 (Target: TNKS1/2) - Another Tankyrase inhibitor with demonstrated efficacy in cancer cell lines. Epirubicin (Target: TOP2A) - A topoisomerase II inhibitor commonly used in chemotherapy. Gefitinib (Target: EGFR) - An EGFR tyrosine kinase inhibitor used in the treatment of non-small cell lung cancer. The correlation between GLO1 mRNA expression and drug sensitivity (measured as 50% inhibitory concentration, IC50) was determined using Pearson’s correlation analysis with the “Hmisc” R package. The results were visualized using the R packages “ggpubr” and “pheatmap”. The IC50 values were log2-transformed and batch-corrected using ComBat to account for any batch effects.

2.6 prediction of MiRNA target interactions

The potential associated miRNAs of GLO1 was examined by “multiMiR” package [13]. This package was a collection of miRNA/targets from 8 different databases, including DIANA-microT, EIMMo, MicroCosm, miRanda, miRDB, PicTar, PITA and TargetScan. The predicting outcomes from eight database were intersected. The result was showed in upset figure by “UpSetR” R package (version 1.4.0, available at: https://CRAN.R-project.org/package=UpSetR). Pearson’s correlation analysis was performed by “Hmisc” R package.

Genetic alteration analysis of GLO1

The data of mutation type, copy number amplification, and deep deletion across all cancers were downloaded from the TCGA database and then analyzed for alteration frequency in R (Version 4.3.1). To address the potential batch effects and ensure the statistical rigor of our analysis, we applied batch-effect correction using the ComBat method from the “sva” R package. This correction was particularly important for integrating data from multiple sources, such as TCGA, GEO, and GDSC, to ensure that observed differences were due to biological variation rather than technical artifacts. The results were visualized using the “pheatmap” R package.

Single-cell analysis of GLO1

GLO1 single-cell analysis was conducted using the Tumor Immune Single-cell Hub 2 [14] (TISCH2, http://tisch.comp-genomics.org/). We specifically focused on ovarian serous cystadenocarcinoma (OV) for an in-depth examination of GLO1 distribution within the tumor microenvironment (TME) for the following reasons: Firstly, as identified in our pan-cancer genetic alteration analysis, OV exhibited one of the highest frequencies of GLO1 gene amplification (8.3%) among all cancer types studied, suggesting a potential driver role in this malignancy. Secondly, the complex and heterogeneous cellular composition of the OV TME provides an excellent model to investigate cell-type-specific expression of GLO1. This focused approach allows for a detailed, mechanistic case study that complements our broader pan-cancer analyses. Gene “GLO1” was input and cell-type annotation “minor-lineage” searched in “Ovarian Serous Cystadenocarcinoma”. In this study, we enrolled EMTAB8107, GSE118828, GSE130000, GSE147082, GSE151214 and GSE154600 to analyze GLO1 expression distribution in ovarian serous cystadenocarcinoma (OV). To define “high” and “low” GLO1 expression groups for immune cell correlation analyses, we used the median expression value of GLO1 across all samples as the cutoff. Samples with GLO1 expression levels above the median were classified as the “high” expression group, while those with expression levels below the median were classified as the “low” expression group.

Immunotherapy response prediction analysis

To investigate the relationship between GLO1 and immune checkpoint blockade (ICB) therapy response, three ICB therapy cohorts (GSE176307, PMID32472114, phs000452) were obtained to validate the immunotherapy response prediction ability of GLO1. Patients were divided into two groups according to GLO1 expression level and the best cut-off value was determined by the “surv-cutpoint” function of the “survminer” R package.

Immunohistochemical (IHC) staining

Standard methods were used for staining paraffin embedding tissue sections from the clinical samples. Briefly, the paraffin-embedded tissues were sliced into 4 µm thick sections. Sections were deparaffinized, rehydrated, and then immersed in Tris-EDTA buffer containing 0.05% Tween 20. After incubating with 3% hydrogen peroxide and blocking in 5% dry milk, the sections were probed overnight with rabbit anti-human GLO-1 antibody (Invitrogen, CA, USA). Sections were treated with horseradish peroxidase-labeled goat anti-rabbit secondary antibody (Santa Cruz Biotechnology). Slides were incubated with 3,3’-diaminobenzidine and counterstained with hematoxylin before measurement under microscope (Olympus, Tokyo, Japan). The histology types and IHC scores were defined by three pathologists independently. Each sample was assigned a score based on the proportion and the intensity of positive cells (For the proportion, 0 ≤ 10%, 1 = 10%-25%, 2 = 26%-50%, 3 = 51%-75%,4 ≥ 75%; For the intensity, 0 = negative, 1 = weak, 2 = moderate, 3 = strong). Three pathologists scored independently, and then took the average as final results.

In vitro functional validation of immune modulation by GLO1

Cell culture and transfection: Human melanoma cell lines (A375) and human renal cell carcinoma cell lines (786-O) were obtained from the American Type Culture Collection (ATCC) and cultured according to standard protocols. Cells were transfected with GLO1-specific small interfering RNA (siRNA) or non-targeting control siRNA (Santa Cruz Biotechnology) using Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) according to the manufacturer’s instructions. Cells were harvested for analysis 48–72 h post-transfection.

Western blot analysis: Total protein was extracted from transfected cells using RIPA lysis buffer. Proteins were separated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with primary antibodies against PD-L1 (Cell Signaling Technology, 1:1000), and GAPDH (Santa Cruz Biotechnology, 1:2000) as a loading control. Band intensities were quantified using ImageJ software.

Co-culture and CD8 + T cell functional assay: Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donors using Ficoll-Paque density gradient centrifugation. CD8 + T cells were isolated from PBMCs using a human CD8 + T Cell Isolation Kit (Miltenyi Biotec). Isolated CD8 + T cells were activated with CD3/CD28 Dynabeads (Gibco) for 48 h. The co-culture system was established using a Transwell chamber. Cells from each group were collected, and the cell density was adjusted to 1 × 10⁶ cells/mL, followed by resuspension in binding buffer. After resuspension, 5 µL of Annexin V-FITC and 5 µL of PI solution were added to each tube. The mixture was gently vortexed and incubated at room temperature in the dark for 10 min. Apoptosis of CD8⁺ T cells was detected by flow cytometry.

Detection of TNF-α and IFN-γ levels in the supernatant of the co-culture system: The cell culture supernatant from the co-culture system described above was collected and centrifuged to obtain the supernatant. The levels of TNF-α and IFN-γ in the supernatant were measured according to the instructions of the ELISA kit.2.12 Statistical analysis.

The difference of GLO1 expression in tumor and normal tissues were assessed using unpaired Wilcoxon test. Survival analysis was determined using Kaplan-Meier method. Spearman’s correlation tests were employed to calculate the significance of correlations between GLO1 expression and drug sensitivity, miRNA interactions. The unpaired Wilcoxon test was used for IHC score analysis. A p-value of less than 0.05 was considered to indicate statistical significance.

Results

Expression level of GLO1 in pan-cancers

GLO1 mRNA was widely expressed in tumor tissues (Fig. 1A). To further evaluate the differential expression level of GLO1, we screened datasets from the TCGA database that included normal samples sized greater than five. The result showed that expression level of GLO1 in nine tumor tissues was significantly higher than the corresponding control tissues, including bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD) and rectum adenocarcinoma (READ) (Fig. 1B). Conversely, in cholangiocarcinoma (CHOL), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), and thyroid carcinoma (THCA), GLO1 expression was found to be lower than that in normal tissues (Fig. 1B).

Fig. 1.

mRNA expression profile and mutation landscape of GLO1 in The Cancer Genome Atlas (TCGA) cohort. A Expression level of GLO1 gene in different cancers. B The different expression profile between cancer and adjacent normal tissues of GLO1 in 18 tumors. C Glo1 mutation frequency in pan-cancers. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Genetic alteration analysis

The genetic alteration status of GLO1 in different tumor samples of the TCGA cohorts were analyzed. The primary alteration type of GLO1 was amplification, and deep deletion was seldom detected. The highest frequency of alteration in GLO1 gene occurred in patients with OV (8.3%), following ESCA of 7.0% and uveal melanoma (UVM) of 6.2%, with “amplification” as the primary type in all three cancers (Fig. 1C). It was interesting that amplification (> 1%) was the only genetic alteration type in CHOL, lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), Sarcoma (SARC), mesothelioma (MESO) and UVM. There was no genetic alteration found in adrenocortical carcinoma (ACC), KICH and THCA (Fig. 1C). In general, the frequency of the genetic alteration of GLO1 was not high, the genetic alteration proportion was lower than 5% in most cancers, which may be related to the high conservation characteristics of GLO1.

GLO1 protein distribution and expression

We obtained the IF images of GLO1 protein from HPA and validated the GLO1 protein expression by IHC in clinical EC samples. The GLO1 protein was detected in both nucleus and cytoplasm in cytoplasm epidermoid carcinoma cell line (A-431), glioblastoma cell line (U-251MG) and osteosarcoma cell line (U2OS) (Fig. 2A-C). To further validate the expression of GLO1 protein, we performed IHC in EC samples and normal endometrial tissue. The results showed that GLO1 protein was upregulated in EC samples (p = 0.0003) (Fig. 2D).

Fig. 2.

Protein expression of GLO1 in different cancer. The immunofluorescence images of GLO1 protein, nucleus, endoplasmic reticulum (ER), microtubules and the incorporative images in A-431(A), U251 (B) and U2OS (C) cell lines. (D) Immunohistochemical detection of the GLO1 expression levels in endometrial cancer (EC) and normal endometrium. ***p < 0.001

Prognostic analysis of GLO1 in pan-cancers

Kaplan-Meier and Cox regression analyses were used to examine the relationship between GLO1 expression levels and prognosis. After adjusting for age, gender, and tumor stage where data was available. GLO1 was seem to be a hazard factor in 14 tumor types (Fig. 3A and supplementary Fig. 1): ACC, BRCA, CHOL, DLBC, ESCA, HNSC, KICH, KIRP, acute myeloid leukemia (LAML), liver hepatocellular carcinoma (LIHC), MESO, pancreatic adenocarcinoma (PAAD), SARC and uterine corpus endometrial carcinoma (UCEC). The top five highest HR were ACC (HR = 5.97, 95% confidence interval (95%CI): 2.28–15.63, p = 0.044), CHOL (HR = 3.19, 95%CI 1.24–8.22, p = 0.046), DLBC (HR = 4.21, 95%CI 1.13–15.65, p = 0.038), KICH (HR = 4.88, 95%CI 1.30-18.35, p = 0.028) and KIRP (HR = 4.19, 95%CI 1.57–11.20, p < 0.001). Figure 3B and F showed the Kaplan-Meier curves for OS, which indicated a significantly positive correlation between GLO1 overexpression and deteriorated outcomes in patients with ACC, CHOL, DLBC, KICH and KIRP. Whereas GLO1 expression was a protective factor in GBM (HR = 0.64, 95%CI 0.44–0.94, p = 0.038), KIRC (HR = 0.53. 95%CI 0.36–0.79, p < 0.001), LUSC (HR = 0.63, 95%CI: 0.43–0.92, p = 0.043), OV (HR = 0.61, 95%CI: 0.43–0.86, p = 0.015), PRAD (HR = 0.18, 95%CI:0.05–0.63, p = 0.061) and UVM (HR = 0.23, 95%CI: 0.10–0.52, p = 0.003)(Fig. 3A).

Fig. 3.

Correlation between GLO1 gene expression and overall survival(OS) of different cancers. A Survival analysis comparing the high and low expression of GLO1 in 33 types of cancer analyzed by Cox regression. Survival curves of OS with highest Hazard ratio (HR) in ACC (B), CHOL (C), DLBC (D), KICH (E) and KIRP (F)

In term of gynecological tumors, high GLO1 expression was associated with poor prognosis in UCEC (HR = 1.60, 95%CI 1.02–2.61, p = 0.038), but better OS in OV (Fig. 3G-H). There was no significant relation between GLO1 expression and uterine carcinosarcoma (UCS) survival (HR = 0.67, 95%CI 0.31–1.47, p = 0.256).

Analysis of GLO1-associated drugs sensitivity

We screened the GDSC database to unveil the potential inhibitors and agonists of GLO1. A total of 176 drugs were included in this study. The results of 22 types of commonly used chemotherapy drugs was presented in Fig. 4. GLO1 expression level was not significantly related to these chemotherapy drugs in most cancers. A strongly negative correlation with statistical differences (R=-0.378, P < 0.0001) was found between Epirubicin and UCEC which possibly indicated a better treatment effect (Fig. 4B). Similar results of some other drugs were showed in KIRC, HNSC, DLBC, GBM, LIHC, OV and LUSC (Fig. 4C-I). According to the size of P value, the top 2 drugs with positive or negative correlation in each cancer were presented in the form of heatmap in Supplementary Fig. 2A. Our results established the positive association between GLO1 expression and P22077 in MESO, AZ6102 and WIKI4 in GBM (Supplementary Fig. 2B-D). Besides, GLO1 expression was inhibited by WEHI.539 and Gefinitib in DLBC, Sinularin in HNSC (Supplementary Fig. 2E-G).

Fig. 4.

Analysis of drug sensitivity associated with GLO1. Drugs are ranked by correlation strength with GLO1 expression. AZ6102, P22077, and WIKI4 (Tankyrase inhibitors) show positive correlation (resistance), while epirubicin and gefitinib show negative correlation (sensitivity). GDSC drug names and targets are listed in the text. The positive correlation (in orange) means that the gene’s high expression was resistant to the drug, while the negative (in grey) was the opposite. *p < 0.05; **p < 0.01; ***p < 0.001

Identification of MiRNAs related to GLO1

The miRNAs related to GLO1 were obtained from PicTar, MicroCosm, miRDB, ElMMo, miRanda, DIANA-microT, PITA and TargetScan databases by the multiMir R package. More than one hundred miRNAs were identified and the intersected results of multiple databases were displayed by upset plot (Supplementary Fig. 3A). Next, the predicted results were crossed with the mRNAs profiles obtained from TCGA database. The heatmap showed the commonalities of TCGA and multiMir (predicted by more than three databases at the same time). The correlation between GLO1 and its validated target miRNAs showed positive relations in majority of different cancers, especially hsa-miR-17-5p and hsa-miR-20a-5p played an up-regulating role in most cancers, while hsa-miR-142-5p inhibited GLO1 expression especially in TGCT and KIRC (Supplementary Fig. 3B).

Relevance of GLO1 expression and tumor microenvironment

To investigate GLO1’s role in the tumor microenvironment, we analyzed its expression in ovarian serous cystadenocarcinoma (OV) using single-cell data. GLO1 was found to be widely expressed in the tumor microenvironment, particularly in malignant cells, cancer-associated fibroblasts, and endothelial cells. Further analysis using algorithms that estimate immune cell infiltration showed that GLO1 expression was negatively correlated with the presence of immune cells, such as B cells, T cells, and NK cells, in breast invasive carcinoma. This suggests that GLO1 may contribute to immune evasion in cancer. As the high genetic alteration frequency in OV, we retrieved the single-cell database to investigate GLO1 distribution in TME. GLO1 was widely expressed in TME and mainly bound to malignant cells, cancer associated fibroblasts (both fibroblasts and myofibroblasts) and endothelial cells, especially malignant cell which may be related to the hypermetabolic state (Fig. 5A). GLO1 expression in different kind of cell in OV (EMTAB8107 and GSE151214) was presented in Fig. 5B-D. Further analysis focus on tumor immune microenvironment in different tumors was performed using MCP-counter [15] and EPIC [16]algorithms and the results showed in Fig. 5F. The heat map findings indicated that GLO1 expression was unexpected negatively related to cancer associated fibroblasts in most cancers. Additionally, GLO1 was strongly negatively correlated with B cell, CD9 + T cell, NK cell, cancer associated fibroblasts, endothelial cell and M2 macrophage in BRCA. As B cells, CD8 + T cells and NK cells played important role in immune activation against abnormal cells, GLO1 expression maybe contributed to immunotherapy effect. Moreover, We collected 10 cases each of breast cancer and melanoma tissues, and used immunohistochemistry to stain for the expression of GLO1. Based on the expression of GLO1, we divided breast cancer and melanoma patients into GLO1 low-expression group and GLO1 high-expression group. Then, by using IHC staining, we found that the expression of CD3 (T cell marker), CD20 (B cell marker), CD56 (NK cell marker), and CD68 (macrophage marker) in GLO1 low-expression group and GLO1 high-expression group in breast cancer and melanoma patients. We found that the expression of these immune cell markers CD31, CD206, CD56, and CD68 significantly decreased in breast cancer and melanoma patients with high expression of GLO1 (Supplementary Fig. 1).

Fig. 5.

GLO1 and tumor microenvironment. A–E GLO1 distribution in OV single-cell lines. F The correlation between the GLO1 expression and B cell, CD9 + T cell, NK cell, cancer associated fibroblasts, endothelial cell and M2 macrophage, which were calculated using MCP-counter and EPIC algorithms. *p < 0.05; **p < 0.01; ***p < 0.001

Predictive value of GLO1 in cancer immunotherapy response

We explored the role of GLO1 in predicting the immune checkpoint blockade response in BLCA, KIRC and skin cutaneous melanoma (SKCM) from three immunotherapy-related cohorts. In the melanoma (phs000452), anti-CTLA-4-treated patients with high GLO1 expression level demonstrated more possibility to progression than low group (Fig. 6A-B), the survival rate and time were obviously worse in GLO1 overexpression group compared with low group (PFS: HR = 2.08, 95%CI 1.24–3.48, p = 0.023; OS: HR = 2.01, 95%CI 1.13–3.57, p = 0.006). In terms of KIRC, the relationship between GLO1 and the response to PD-1 blockade treatment showed that the OS and PFS of patients with high GLO1 expression were significantly poorer than those with low GLO1 expression (Fig. 6C-D). In addition, contradictory results were found in patients with metastatic urothelial cancer who received anti-PD1 or anti-PDL1 treatment. In GSE176307 cohort, the OS in metastatic urothelial cancer with high GLO1 expression was better, but the PFS was worse(Fig. 6E-F). This discrepancy may be attributed to the different mechanisms underlying the progression and survival of cancer cells in the presence of high GLO1 expression.

Fig. 6.

GLO1 predicted the response to immunotherapy in three immunotherapy cohorts. A, B Overall survival (OS) and prognosis-free survival (PFS) of SKCM (phs000452). C, D OS and PFS of KIRC (PMID32472114). E, F OS and PFS of BLCA (GSE176307)

GLO1 knockdown suppresses PD-L1 expression and enhances CD8 + T cell function in vitro

To mechanistically validate the proposed link between GLO1 and immune evasion, which was strongly suggested by our clinical immunotherapy cohort analysis in melanoma (SKCM) and kidney renal clear cell carcinoma (KIRC) (Sect. 3.8, Fig. 6A-D), we selected representative cell lines from these two cancer types for functional studies. Human melanoma cell lines (A375) and human renal cell carcinoma cell lines (786-O) were obtained from the American Type Culture Collection (ATCC) and cultured. Transfection with GLO1-specific siRNA significantly reduced GLO1 protein expression in both A375 (melanoma) and 786-O (renal cell carcinoma) cells compared to the non-targeting control (si-NC) (Fig. 7A). Strikingly, GLO1 KD led to a marked decrease in PD-L1 protein levels in both cell lines (Fig. 7B, C), establishing a direct link between GLO1 activity and the expression of this critical immune checkpoint molecule.

Fig. 7.

GLO1 knockdown downregulates PD-L1 and enhances CD8 + T cell function in vitro. A Representative RT-qPCR of GLO1 and PD-L1 expression in A375 and 786-O cells after transfection with control (si-NC) or GLO1-specific (si-GLO1) siRNA. GAPDH served as a loading control. B, C Quantification of mRNA B and PD-L1 C protein levels normalized to GAPDH (n = 3, *p < 0.05, **p < 0.01). D Representative flow cytometry plots showing the percentage of CD8 + T cells apoptosis after culture in conditioned medium from si-NC or si-GLO1 A375 cells. E, F Statistical summary of TNF-a (E) and IFN-γ production (F) in CD8 + T cells cultured in conditioned medium from si-NC or si-GLO1 treated A375 and 786-O cells (n = 3, *p < 0.05, **p < 0.01)

We next investigated the functional impact of GLO1 on T cells. We cultured pre-activated human CD8 + T cells in conditioned medium (CM) collected from control (si-NC) or GLO1-knockdown (si-GLO1) cancer cells. CD8 + T cells cultured in CM from GLO1-KD cells exhibited a significantly lower percentage of cell apoptosis (Fig. 7D) and produced significantly more IFN-γand TNF-a upon restimulation (Fig. 7E, F) compared to those cultured in CM from control cells.

These results provide direct experimental evidence that GLO1 expression in tumor cells promotes immune evasion by upregulating PD-L1 and secreting factors that suppress the activation and effector function of CD8 + T cells.

Moreover, GSEA results indicated a significant correlation between GLO1 expression and gene sets related to glycolysis and oxidative stress pathways in BRCA and SKCM (Supplementary Fig. 2). This evidence indicated that the GLO1 expression level was a valuable predictor of immunotherapy response in some cancers.

Discussion

Our comprehensive pan-cancer analysis reveals that GLO1 is differentially expressed across cancers and holds significant but context-dependent prognostic value, influenced by the tumor’s metabolic and immune microenvironment. While GLO1 overexpression is frequently associated with tumor progression and poor prognosis, as seen in ACC, BRCA, and CHOL, it surprisingly confers a protective effect in others like GBM and KIRC. The observed contradictory prognostic roles, [for instance, the protective effect in PRAD in our analysis versus previous reports of it being a risk factor, could be attributed to differences in cohort characteristics, bioinformatic processing pipelines, or updated clinical follow-up data. This context-dependent role of GLO1 is further exemplified by recent findings which demonstrate that in certain tumor types, GLO1 expression is inversely correlated with epithelial-mesenchymal transition (EMT) markers, potentially explaining its association with better outcomes in less aggressive subtypes [17]. Our multivariate analysis, adjusting for key confounders, strengthens the validity of our findings within the studied cohorts.

Glyoxalase 1 (GLO1) is a key enzyme in the glyoxalase system responsible for detoxifying methylglyoxal (MG), a cytotoxic byproduct of glycolysis. While GLO1 has been extensively studied in cancer, its role is often compared to other members of the glyoxalase family, such as Glyoxalase 2 (GLO2). GLO2, which catalyzes the conversion of S-D-lactoyl-glutathione to D-lactate, is also involved in the detoxification pathway but has received less attention in cancer research. Recent studies have highlighted the differential roles of GLO1 and GLO2 in cellular metabolism and stress response. For instance, GLO1 is predominantly involved in maintaining cellular redox balance and protecting cells from MG-induced cytotoxicity, whereas GLO2 may play a more regulatory role in the glyoxalase system by influencing the availability of substrates for GLO1 [18]. In the context of cancer, GLO1 overexpression is often associated with tumor progression and poor prognosis, as demonstrated in this study and previous research [19, 20]. In contrast, the role of GLO2 in cancer remains less clear, with some studies suggesting it may have tumor-suppressive functions due to its role in maintaining metabolic homeostasis [21].

A pan-cancer analysis including expression differences and genetic alteration, relationship between expression levels and prognosis, potential associated drugs and miRNA, predictive value of cancer immunotherapy response focus on GLO1 was conducted to investigate the functions and potential mechanisms of GLO1 in the pathogenesis or clinical prognosis of different cancers. In this study, GLO1 was high expression in bladder, breast, colon, esophageal, glioblastoma, head and neck, lung, prostate and rectum cancers. Whereas GLO1 mRNA expression was significantly decreased in all kind of kidney cancers. Unexpectedly, GLO1 was not closely associated with genomics alterations in pan-cancer analysis because we found the proportion lower than 5% in most cancers except ESCA, OV and UVM. This finding suggests that the increased expression level of GLO1 mRNA is not only affected by gene amplification, but may also be related to transcriptional level regulation. Previous studies had demonstrated that GLO1 expression up-regulation may be a result of gene amplification [22] and amplification frequency of GLO1 was high in breast, bladder and non-small cell lung tumor samples [23]. Up-regulation of GLO1 mRNA and protein in breast cancer was widely confirmed, and positively associated with positive lymph node, lymphovascular invasion, and TNM stage [24, 25]. Knockdown of GLO1 in the MCF-7 cells and melanoma cells significantly reduced tumor migration and proliferation [26, 27]. These findings were in accordance with our results. Santarius et al. found GLO1 amplification in renal cancers compared with adjacent normal tissue, another research observed a significant increase in both mRNA and protein expression level of GLO1 in KIRC [28], but an opposite result was displayed in our study. Taken together, these results indicate that the expression of GLO1 differs between normal and cancer tissues in most cancers; in particular, GLO1 mRNA expression is significantly higher in cancer than that in normal tissues.

The prognostic value of GLO1 varied across different cancer types. High GLO1 expression was associated with poor prognosis in cancers such as adrenocortical carcinoma (ACC), BRCA, and CHOL, indicating its role as a risk factor for tumor progression. However, in glioblastoma (GBM) and KIRC, GLO1 expression was found to be protective, suggesting a more complex role in these cancers. This dichotomy can be attributed to differences in the metabolic context of the tumors, the specific genetic alterations present, and the tumor microenvironment [29, 30]. Inhibition of GLO1 finally induced the apoptosis in GBM cell lines [31]. In term of PRAD, our data indicated a protective effect of GLO1 expression while previous studies demonstrated high level GLO1 expression was linked to poor prognosis in prostate cancer [32]. One possible reason for such contradictory results can be attributed to different data processing or updated survival information. In vitro experiment found that promoting GLO1 expression reduced the proliferation, migration and invasion ability of lung cancer cells [33]. However, we drew the conclusion above only from database without considering some confounding factors such as gender, age, and cancer stages. Hence, these influencing factors should be considered in future studies to validate these findings and draw a more definitive conclusion.

Previous studies indicated GLO1 may be a promising therapeutic target for tumor therapy. So we investigated the GLO1 expression and drug sensitivity. A majority of negative correlation was observed in BLCA, DLBC, OV, STAD, THCA and UCEC, while positive correlation was observed in the majority of BRCA, GBM, KIRC, MESO. The mainstream view about the role of GLO1 in carcinogenesis and development was GLO1 promoting carcinogenesis, growth, metastasis, and cancer chemotherapy resistance due to it catalyzing the removal of cytotoxic MG so that maintaining the tumor cell survival [34]. Therefore, high level of GLO1 mRNA, protein and enzymatic activity were theoretically beneficial for tumor growth and may related to drug resistance. In our study, a significant positive correlation was found between GLO1 expression and AZ6102 (TNKS1/2 inhibitor), WIKI4 (Tankyrase inhibitor) and P22077 (selective ubiquitin-specific protease inhibitor), which may predicted drug resistance.

miRNAs were a class of small non-coding RNAs, and it had powerful ability to regulate various cellular activities including cell growth, differentiation, development and apoptosis. This regulatory effect is mainly achieved by controlling mRNA expression. Besides, miRNAs were also capable of targeting non-coding RNAs [34]. Previous studies elucidated interaction between miRNA and GLO1. A strong link between GLO1 and hsa-miR-561-5p was determined in OV drug resistance [35]. miR-137 downregulated GLO1 expression to inhibit cell proliferation in melanoma. miRNA-7515 in lung adenocarcinoma, miR-216a-5p in breast cance [36] and miR-101 in metastatic prostate cancer [37] were also interacted with GLO1 to influence tumor growth or invasion.

Within the TME, cancer associated fibroblasts (CAFs) have been shown to play several roles to influence tumor angiogenesis, immunology and metabolism, including producing secreted factors, regulating other cell types and metabolic alterations [38]. However, CAFs exhibit dual effects of tumor-promoting and tumor-restraining functions maybe related to its heterogeneity in different types of cancer [39]. We firstly found GLO1 accumulation in CAFs in OV single cell lines, but an opposite result was appeared unexpectedly in the pan-cancer analysis. This may be due to the dual action of CAFs. Interestingly, CAFs in BRCA could represent up to 80% of the tumor mass and took part in the tumor initiation and progression [40], but our predicting results of GLO1 expression in BRCA showed no sign of aggregation in TME. As GLO1 was high expression in BRCA tissue and tumor-promoting function of BRCA CAFs, we initially assumed that GLO1 expression up-regulation both in tumor and TME. However, the inconsistent results suggested that further research was necessary for the specific mechanism of GLO1 function in BRCA. Beyond its role as an independent prognostic marker, GLO1 has emerged as a druggable node capable of reshaping the tumor immune microenvironment (TIME) and thereby enhancing immunotherapy efficacy. Mechanistically, GLO1-mediated detoxification of methylglyoxal (MG) prevents the formation of MG-derived advanced glycation end-products (AGEs), which are potent immunogenic ligands for the receptor for AGE (RAGE). In pre-clinical models, genetic or pharmacologic GLO1 inhibition increased MG-AGE burden, activated RAGE–NF-κB signaling, and up-regulated CXCL10 and CCL5, thereby recruiting CD8 + T cells and natural killer (NK) cells while reducing regulatory T-cell infiltration.

We finally explored the relationship between GLO1 expression levels and the response to immunotherapy in melanoma, bladder urothelial carcinoma and renal clear cell carcinoma (CCA). The results showed that higher GLO1 expression indicated a worse prognosis and less sensitivity to immunotherapy in melanoma and renal clear cell carcinoma. There were limited researches focus on the GLO1 and immunotherapy. The increased Glo1 expression in more aggressive CCA cells might provide advantages to the metastatic potential and Glo1 inhibition might be a therapeutic approach for the treatment of metastatic CCA [41]. PDL1 expression was suppressed in cultured GLO1 knockout melanoma cells and tumors, which indicated GLO1 maybe influence the immunotherapy effect through regulating PD1/PDL1 pathway [2]. Antognelli et al. found that GLO1 contribute to maintain an immunosuppressive microenvironment through MG-H1-mediated PD-L1 up-regulation, which leading to resistance to PD-L1 inhibitors and promoting cancer progression [28].

The dichotomous role of GLO1 reflects its integration into distinct metabolic and oncogenic contexts. The insights from our study and others, such as the link between GLO1 and T-cell exhaustion programs [42], open up new avenues for therapeutic targeting. Combining GLO1 inhibition with existing immunotherapies could potentially reverse the immunosuppressive phenotype in GLO1-high tumors, a strategy that warrants further preclinical investigation. In glycolysis-dependent cancers, GLO1 is essential for MG detoxification and survival, whereas in oxidative or hypoxic tumors, its overexpression mitigates oxidative stress-induced tumor suppression. This model reconciles prior conflicting reports and underscores the importance of context-specific therapeutic targeting: GLO1 inhibition may be beneficial in glycolytic tumors but detrimental in oxidative or hypoxic malignancies.

While our bioinformatics analysis provides robust evidence for the role of GLO1 in cancer progression and immune response, the experimental validation is currently limited to immunohistochemical (IHC) staining in endometrial cancer samples. This limits the generalizability of our findings to other cancer types. To fully validate the key claims of our study, particularly GLO1’s role in drug sensitivity and immune evasion, further experimental validation in additional cancer types and cell lines is necessary. Future experiments will include in vitro and in vivo studies to validate the role of GLO1 in drug sensitivity. We plan to use a panel of cancer cell lines representing different cancer types to test the sensitivity to various drugs, including those identified in our bioinformatics analysis (e.g., AZ6102, P22077, WIKI4, Epirubicin, and Gefitinib) [1]. This will help confirm the predictive value of GLO1 expression for drug response. To validate GLO1’s role in immune evasion, we will perform experiments to assess the impact of GLO1 knockdown on immune cell infiltration and immune checkpoint expression. This will include co-culture experiments with immune cells and cancer cells, as well as analysis of immune checkpoint markers such as PD-L1 and CTLA-4 [2]. These experiments will provide direct evidence for GLO1’s role in modulating the tumor microenvironment and immune response. Further exploration of GLO1’s role in metabolic pathways will be conducted using techniques such as metabolomics and flux analysis. This will help elucidate the mechanisms by which GLO1 influences cancer metabolism and contributes to tumor progression [3].

Most importantly, our in vitro functional assays provided direct experimental proof for the proposed mechanism. We demonstrated that genetic knockdown of GLO1 in cancer cell lines significantly reduced the expression of PD-L1, a key mediator of immune escape. Furthermore, conditioned medium from GLO1-deficient cells potently enhanced the activation and effector function of co-cultured CD8 + T cells. These findings mechanistically bridge GLO1 activity to the PD-1/PD-L1 axis and T-cell dysfunction, providing a solid foundation for its role as a regulator of the tumor immune microenvironment and a promising target for improving immunotherapy efficacy.

The negative correlation between GLO1 expression and immune cell infiltration, particularly in BRCA, and its association with poor response to ICB in SKCM and KIRC, position GLO1 as a potential modulator of the tumor immune landscape. This is further supported by our experimental data showing that GLO1 knockdown reduces PD-L1 expression in prostate cancer cells. The contradictory findings in the BLCA cohort (divergent OS and PFS outcomes) are intriguing and may reflect distinct biological processes affected by GLO1; for example, it might promote initial resistance to therapy (affecting PFS) while potentially altering tumor aggressiveness or metastatic potential in a way that paradoxically affects long-term survival (OS). This complexity necessitates further mechanistic investigation.

This intriguing divergence warrants a mechanistic hypothesis. We postulate that GLO1-high tumors in BLCA might initially be more resistant to immune checkpoint blockade (ICB), leading to earlier disease progression (explaining the poor PFS). However, this same metabolic state, characterized by efficient MG detoxification, might simultaneously restrain the acquisition of more aggressive, treatment-refractory phenotypes during tumor evolution. Alternatively, the immunogenic cell death triggered by eventual tumor breakdown in these metabolically distinct tumors could potentiate a systemic anti-tumor immune response, benefiting long-term survival. Furthermore, it is plausible that patients with high GLO1 expression who progressed on initial ICB subsequently responded more favorably to later-line therapies, thereby improving their OS. This dissociation between PFS and OS highlights the complex, context-dependent role of GLO1 and suggests that its impact may extend beyond initial therapy response to influence overall disease trajectory and tumor evolution under therapeutic pressure.

This study has several limitations. Firstly, it is primarily based on retrospective bioinformatic analyses of public datasets, which may contain inherent selection biases and unmeasured confounding factors, despite our efforts to adjust for common clinical variables. Secondly, while we provided experimental validation for GLO1 protein expression in EC and its impact on PD-L1 in prostate cell lines, the functional role of GLO1 across most cancer types remains computationally predicted and requires broader experimental validation in vitro and in vivo. Future work should focus on: (1) Validating the prognostic and predictive role of GLO1 in independent, prospectively collected cohorts; (2) Conducting mechanistic studies to elucidate how GLO1 influences immune evasion, particularly its regulation of PD-L1 and other checkpoints; (3) Exploring the metabolic-immune interface by investigating how GLO1-mediated MG detoxification specifically alters the TME; and (4) Resolving the context-dependent roles of GLO1 through integrated multi-omics approaches in specific cancer subtypes.

5. Conclusions

GLO1 is differentially expressed in various tumors and is associated with tumor progression. Its expression can be driven by gene amplification and may influence cancer metabolism and immune response. Our study highlights GLO1’s potential as a therapeutic target and biomarker for immunotherapy. This study establishes GLO1 as a compelling biomarker worthy of further investigation and highlights the need for context-specific understanding of its function to guide future therapeutic strategies targeting the glyoxalase pathway.

Abbreviations

GLO1

Glyoxalase 1

MG

Methlglyoxal

TME

Tumor microenvironment

ECM

Extracellular matrix

PD-L1

Programmed cell death ligand-1

CTLA4

Cytotoxic T lymphocyte associated antigen 4

TCGA

The Cancer Genome Atlas

GEO

Gene Expression Omnibus

EC

Endometrial cancer

UCSC

University of California Santa Cruz

HPA

Human Protein Atlas

IF

Immunofluorescence

OS

Overall survival

PFS

Progression-free survival

GDSC

Genomics of Drug Sensitivity in Cancer

TISCH2

Tumor Immune Single-cell Hub 2

OV

Ovarian serous cystadenocarcinoma

BLCA

Bladder urothelial carcinoma

BRCA

Breast invasive carcinoma

COAD

Colon adenocarcinoma

ESCA

Esophageal carcinoma

GBM

Glioblastoma multiforme

HNSC

Head and neck squamous cell carcinoma

LUSC

Lung squamous cell carcinoma

PRAD

Prostate adenocarcinoma

READ

Rectum adenocarcinoma

CHOL

Cholangiocarcinoma

KICH

Kidney chromophobe

KIRC

Kidney renal clear cell carcinoma

KIRP

Kidney renal papillary cell carcinoma

THCA

Thyroid carcinoma

UVM

Uveal melanoma

DLBC

Lymphoid neoplasm diffuse large B-cell lymphoma

SARC

Sarcoma

MESO

Mesothelioma

ACC

Adrenocortical carcinoma

LAML

Acute myeloid leukemia

LIHC

Liver hepatocellular carcinoma

PAAD

Pancreatic adenocarcinoma

UCEC

Uterine corpus endometrial carcinoma

UCS

Uterine carcinosarcoma

SKCM

Skin cutaneous melanoma

CESC

Endocervical adenocarcinoma

LGG

Brain lower grade glioma

LUAD

Lung adenocarcinoma

STAD

Stomach adenocarcinoma

CAFs

Cancer associated fibroblasts

CCA

Clear cell carcinoma

Author contributions

YQZ and QQW designed the research study. YQZ, JZ, YMZ performed the research and analyzed the data. YQZ, QQW, JZ and YMZ wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Funding

The study was supported by Zhejiang Provincial Program of Chinese Medical and Health Science Funding (2024ZL251).

Data availability

The datasets used in this study are publicly available. The original contributions presented in the study are included in the article and supplementary material, further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Zhejiang Provincial People’s Hospital (Approval No.: 2021QT207). All procedures involving human participants were performed in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. For participants under the age of 18, informed consent was obtained from their legal guardians.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.