The expression profiles, prognostic significance, and immune cell infiltration patterns of LDH gene family members were analyzed using data from the TCGA database ( https://portal.gdc.cancer.gov ) [19] , accessed via the Xiantao Academic platform ( https://www.xiantaozi.com/login ). Transcripts per million (TPM) normalized expression matrices were downloaded and log2-transformed (log2[value + 1]) for downstream analyses. To externally validate the diagnostic and prognostic relevance of LDH genes in CESC, we obtained the GSE63514 and GSE44001 datasets from the GEO database ( https://www.ncbi.nlm.nih.gov/geo ) and plotted receiver operating characteristic (ROC) curves. The area under the ROC curve (AUC) was calculated, and the closer the value was to 1, the higher the diagnostic efficiency. In the analysis, normal controls were pathologically confirmed normal tissues (paired adjacent or healthy donor), with cases being histologically diagnosed CESC samples. The diagnostic value of these genes is significant for P < 0.05.

The Human Protein Atlas (HPA) database ( https://www.proteinatlas.org ) [20] was employed to evaluate LDH family expression across human tissues. Using the “TISSUE” module and GTEx dataset, we queried the official gene symbols of LDH members (e.g., LDHA, LDHB) to generate tissue-specific expression profiles. Using the Xiantao Academic database, we selected three matched pairs of cervical cancer tumor tissues and adjacent normal tissues to compare LDH family gene expression profiles. The GSE63514 dataset was analyzed to confirm differential LDH expression in CESC versus normal cervical tissues. Diagnostic performance was assessed via receiver operating characteristic (ROC) curves using the “pROC” R package.

Immunohistochemical (IHC) staining results for LDH proteins in CESC and normal cervical tissues were retrieved from the HPA database using the “PATHOLOGY” and “TISSUE” modules. Clinical metadata for these samples were systematically curated to ensure relevance.

The Xiantao Academic database analyzed the association between LDH gene expression and three survival endpoints: Overall Survival (OS), Disease-Specific Survival (DSS) and Progress-Free Interval (PFI). Clinical and expression data from GSE44001 were processed via the Sangerbox platform ( http://sangerbox.com/ ) as follows: Dataset and clinical metadata were retrieved using the “GSE44001” identifier. Probe annotation and data normalization were performed in the “Data Tabulation and Interactive Analysis” module. Survival analysis was conducted using the survival R package (v3.3.1), correlating LDH expression with CESC prognosis. Univariate and multivariate Cox regression models were applied to evaluate LDH genes as independent prognostic factors, implemented through the Xiantao Academic database.

To investigate the mutational landscape of LDH family members in CESC, we utilized the cBioPortal database ( https://www.cbioportal.org/ ) [21] . Protein-protein interaction (PPI) networks for LDH family members were constructed using the GeneMANIA database ( http://genemania.org ) [22] and STRING database ( https://cn.string-db.org/ ) [23] . Subsequently, interacting genes were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses via the “ClusterProfiler” R package, with results visualized using ggplot2.

DNA methylation levels of LDH family members were analyzed using the UALCAN online database ( https://ualcan.path.uab.edu/ ) [24] [25] . The GSCALite database ( http://bioinfo.life.hust.edu.cn/web/GSCALite/ ) [26] was employed to assess correlations between mRNA expression and DNA methylation, with results presented as scatter plots.

The Xiantao Academic Database ( https://www.xiantaozi.com/login ) was used to evaluate: Associations between LDH family gene expression and immune cell infiltration in CESC. Correlations between LDH members and immune checkpoint genes, including PDCD1, CD274, PDCD1LG2, CTLA4, LMTK3, LAG3, TIGIT, HAVCR2 and SIGLEC15.2.

Using the Researcher’s Home database ( https://www.home-for-researchers.com/#/ ), we explored LDHC-regulated signaling pathways in cervical cancer progression. CESC patients were stratified into high- and low-LDHC expression groups based on median expression values. Differentially expressed genes (DEGs) were identified using the “DESeq2” R package (|log2FC| ≥ 1.5, *p* < 0.05), followed by GO/KEGG enrichment analyses (“ClusterProfiler”) and visualization.2.

HeLa cells (ATCC) were cultured in DMEM high-glucose medium (Gibco, C11995500BT) supplemented with 10% FBS (Sigma, F8318) and 1% penicillin/streptomycin (Gibco, 15140122) at 37˚C with 5% CO₂. Cells in the logarithmic growth phase were harvested using trypsin (Gibco, 25200056) for experiments, with triplicate wells for all assays.

Cells (3,000/well) were seeded in 96-well plates (Corning, 3599) with 100 µL medium. Five plates were prepared to measure OD450 values at 0 h (post-attachment), 24 h, 48 h, 72 h, and 96 h. Before measurement, CCK-8 reagent (UElandy, C6005M) was diluted (10% medium: CCK-8 = 1:9), and 10 µL was added to each well. After 1-hour incubation, absorbance was measured using a VICTOR Nivo plate reader, and proliferation curves were plotted.

Log-phase cells (3,000/well) were seeded in 6-well plates (Corning, 3516) and cultured for 14 days (medium replaced every 2 - 3 days). When clusters reached 30 - 50 cells/field, cultures were terminated. Cells were: 1) Washed with PBS (Gibco, C10010500). 2) Fixed with 4% paraformaldehyde (biosharp, BL539A) for 30 min. 3) Stained with crystal violet (Solarbio, G1063) for 40 min. 4) Imaged, and colonies (>50 cells) were counted.

Cells were seeded in 6-well plates at a density of 2 × 10⁵ cells/mL and incubated until reaching approximately 90% confluence. A uniform scratch was created using a sterile pipette tip perpendicular to the plate bottom. Initial wound area (0 h) was measured using an inverted microscope (Axio Vert A1, 07060200) by capturing multiple fields. After 48 h incubation, the same fields were re-imaged to measure remaining wound area. The migration rate was calculated as: Scratch healing rate (%) = [(0 h area − 48 h area)/0 h area] × 100ys.

Transwell chambers (JET, TCS003024) were used for both assays. For invasion experiments, 100 μL Matrigel (Corning, 356234) was evenly coated and polymerized in the upper chamber. Cells (2 × 10⁵ cells/mL in serum-free medium) were seeded in the upper chamber, while 10% FBS medium served as chemoattractant in the lower chamber. After 24 h incubation, non-migrated cells were removed with PBS. Migrated/invaded cells were fixed with 4% paraformaldehyde (50 min), stained with 0.1% crystal violet (50 min), and counted from four random microscope fields per chamber.

Total protein was extracted using RIPA lysis buffer (Aase, PC103) supplemented with 1% PMSF (Aase, GRF101). Protein concentration was determined (Aase, ZJ102) and normalized. Samples were separated on 7.5% SDS-PAGE gels (Aase, PG211) at 80 V (40 min) then 120 V (35 min), followed by transfer to methanol-activated PVDF membranes (Millipore, IPVH00010) at 200 mA for 60 min. Membranes were blocked (Yase, PS108P, 40 min), incubated with primary antibodies (Proteintech LDHC: 19989-1-AP, 4˚C overnight) and HRP-conjugated secondary antibodies (Invitrogen 31460/31430, 2 h RT). Protein bands were visualized using ECL reagent (Accelerase, SQ201) and quantified with ImageJ.

Total RNA was extracted using TRIzol (Yingjie Life, 15596026) and reverse transcribed (Tolobio, 22107). qPCR was performed with SYBR Green (Tolobio, 22204-1) under conditions: 94˚C (3 min); 35 cycles of 94˚C (15 s), 64˚C (2 min), 72˚C (2 min); final extension at 72˚C (10 min). Relative expression was calculated via 2^(−ΔΔCt) method. Primer sequences are listed in Table 1 .

The Mann-Whitney U test was used to analyze the expression levels of unpaired samples of LDH family in pan-cancer, and the Wilcoxon singed rank test was used to process the paired samples. The prognostic survival analysis of LDH family in pan-cancer was performed by one-way and multifactorial Cox regression test as the statistical analysis of variance. Immunocorrelation analysis using Spearman correlation coefficient was used to determine the relationship between two variables. Experimental data were taken as mean ± standard deviation after three independent experiments, and paired samples were statistically analyzed using the paired t-test, with p ≤ 0.05 considered statistically significant.

The study design is illustrated in Figure 1 . Analysis of the Human Protein Atlas (HPA) database revealed distinct tissue distribution patterns for LDH family members ( Figure 1 ). LDHA demonstrated predominant expression in retina, liver, and skeletal muscle, while LDHB showed significantly higher expression in midbrain, kidney, and cardiac muscle—tissues with high aerobic metabolic activity. Consistent with previous reports, both LDHAL6B and LDHC exhibited exclusive testicular distribution. LDHD was primarily expressed in liver, cardiac muscle, and skeletal muscle. These findings collectively demonstrate that LDH family members display tissue-specific expression patterns, with particular enrichment in organs exhibiting high oxygen consumption.

Our comprehensive analysis revealed cancer-type specific expression patterns for each LDH family member ( Figure 2 ). LDHA was significantly upregulated in 14 cancer types including BRCA, CESC, and GBM, while being downregulated in KICH ( Figure 2(A) ). LDHAL6A showed elevated expression in CHOL, LIHC and STAD tumors compared to 11 other cancer types where its expression was reduced ( Figure 2(B) ). LDHAL6B expression was increased in GBM, HNSC, KIRC and STAD tumor tissues relative to their normal counterparts ( Figure 2(C) ). LDHB exhibited upregulation in 8 cancer types including CHOL and LIHC, but downregulation in 6 others ( Figure 2(D) ). LDHC was overexpressed in CESC, ESCA and HNSC tumors while being underexpressed in 5 cancer types ( Figure 2(E) ). Notably, LDHD showed tumor-specific expression limited to KICH and LUAD, despite being widely expressed in normal tissues ( Figure 2(F) ).

Comparative analysis between CESC and adjacent normal tissues yielded important findings ( Figure 3 ). TCGA data revealed significant upregulation of LDHA and LDHC in tumor tissues, while LDHAL6B showed higher expression in normal tissue ( Figure 3(A)-(F) ). External validation using GSE63514 dataset confirmed increased expression of LDHAL6A, LDHB and LDHC in tumors, with LDHD being more highly expressed in normal cervical tissue ( Figure 3(G)-(L) ). PCR validation of clinical samples largely corroborated these findings, showing predominant tumor overexpression of most LDH members except for LDHAL6B and LDHD ( Figure 3(A)-(F) ).

ROC curve analysis demonstrated the diagnostic value of LDH family members in CESC. TCGA data revealed exceptional diagnostic performance for LDHC (AUC = 0.950, 95% CI: 0.921 - 0.980) and LDHA (AUC = 0.915, 95% CI: 0.766 - 1.000), with other members showing moderate diagnostic value ( Figure 4(A)-(F) ). The GSE63514 validation cohort confirmed LDHC’s strong diagnostic potential (AUC = 0.884) ( Figure 4(G)-(L) ). These consistent results across independent datasets highlight LDHC as the most promising diagnostic biomarker among LDH family members for cervical cancer.

Immunohistochemical analysis from the HPA database provided protein-level confirmation of our findings. LDHA, LDHB and LDHD all showed significantly higher protein expression in CESC compared to normal cervical tissues ( Figure 5(A)-(F) ). The remaining three LDH family members were not detectable in the database, suggesting potential tissue-specific expression limitations.

Survival analysis revealed important clinical correlations ( Figure 6 ). High expression of LDHA and LDHAL6B was associated with worse overall survival (OS), while elevated LDHB, LDHC and LDHD correlated with better outcomes ( Figure 6(A)-(F) ). Similar trends were observed for disease-specific survival (DSS) and progression-free interval (PFI), with LDHC-high patients showing particularly favorable prognosis ( Figure 6(G)-(L) , Supplementary Figure 4(A)-(F) ). External validation using GSE44001 dataset confirmed the adverse prognostic association of LDHA, LDHAL6A and LDHAL6B high expression (Supplementary Figure 4(G)-(L) ).

Cox regression analysis identified four genes (LDHA, LDHAL6B, LDHC, LDHD) as significant prognostic factors in univariate analysis. Multivariate analysis further established LDHAL6B as an independent prognostic marker for CESC patients.

Comprehensive analysis of mutation patterns in CESC using the cBioPortal database revealed distinct mutation frequencies among LDH family members: LDHA (0.9%), LDHAL6A (0.8%), LDHAL6B (1%), LDHB (2.2%), LDHC (1.1%), and LDHD (1.5%), with gene amplification being the predominant mutation type ( Figure 7(A)-(B) ). Protein-protein interaction (PPI) networks constructed through GeneMANIA and STRING databases demonstrated extensive functional associations between LDH members and other proteins ( Figure 7(C)-(D) ). Subsequent functional enrichment analysis revealed that interacting proteins were primarily involved in critical metabolic processes including dicarboxylic acid metabolism, lactate metabolism, pyruvate metabolism, and energy production ( Figure 7(E) ). KEGG pathway analysis further identified significant enrichment in pyruvate metabolism, cysteine and methionine metabolism, carbon metabolism, glycolysis/gluconeogenesis, and propanoate metabolism ( Figure 7(F) ), highlighting the central role of LDH family members in cellular metabolic regulation.

Methylation analysis using UALCAN database demonstrated significant epigenetic differences between CESC and normal cervical tissues ( Figure 8(A)-(F) ). Notably, LDHB exhibited significantly higher methylation levels in tumor tissues, while LDHC showed marked hypomethylation. Correlation analysis between mRNA expression and methylation status revealed a significant negative association for LDHAL6A, LDHAL6B, LDHB, LDHC, and LDHD ( Figure 8(G)-(L) ), suggesting potential epigenetic regulation of these genes in cervical carcinogenesis. No significant correlation was observed between LDHA expression and methylation status.

Comprehensive immune correlation analysis revealed distinct patterns of association between LDH family expression and immune cell infiltration ( Figure 9 ). LDHA expression showed positive correlation with Th2 cells and neutrophils, while demonstrating negative associations with multiple immune cell types including eosinophils, Tregs, and dendritic cells. LDHAL6A exhibited positive correlations with Thelper cells and Tcm, while LDHAL6B showed associations with Thelper cells and neutrophils. Particularly noteworthy was LDHC’s exclusive positive correlation with NK CD56bright cells and negative association with TFH cells and macrophages. Immune checkpoint analysis demonstrated significant correlations between LDH members and key checkpoint molecules (Supplementary Figure 2(A)-(F) ), with LDHD showing the most extensive interaction profile, including positive correlation with LMTK3 and negative associations with multiple checkpoints (CD274, PDCD1LG2, CTLA4, LAG3, and HAVCR2).

To elucidate LDHC’s functional role in CESC progression, we performed single-gene enrichment analysis after stratifying patients into high-(G1) and low-expression (G2) groups based on median LDHC expression. Differential expression analysis identified 264 significantly altered genes (226 upregulated, 38 downregulated) between the groups ( Figure 10(A)-(B) ). KEGG pathway analysis of upregulated genes revealed enrichment in tyrosine metabolism, transcriptional misregulation in cancer, and sulfur metabolism pathways ( Figure 10(C) ), while GO analysis highlighted involvement in developmental processes including ureteric bud development and sensory organ morphogenesis ( Figure 10(D) ). Downregulated genes were significantly enriched in pathways including α-linolenic acid metabolism and small cell lung cancer ( Figure 10(E) ), with GO terms related to immune regulation such as neutrophil migration and degranulation. These findings suggest LDHC may play multifaceted roles in both metabolic reprogramming and immune modulation during cervical cancer progression.

To validate the expression patterns of LDH family members in cervical cancer, we analyzed clinical CESC specimens and adjacent normal tissues. Our results demonstrated significant upregulation of LDHA, LDHAL6A, LDHB and LDHC in tumor tissues, while LDHAL6B and LDHD showed marked downregulation ( Figure 3 ). Given LDHC’s prominent diagnostic and prognostic value in CESC revealed by our preliminary analyses, coupled with the absence of prior functional studies, we selected this gene for further investigation using the Hela cell line. Successful knockdown of LDHC was confirmed at both transcriptional (qPCR) and translational (Western blot) levels ( Figure 11(A)-(C) ). Functional assays revealed that LDHC depletion significantly enhanced the malignant phenotype of CESC cells. CCK-8 proliferation assays showed accelerated growth kinetics in LDHC-knockdown cells ( Figure 11(D) ). This pro-tumorigenic effect was further supported by colony formation assays demonstrating increased proliferative capacity ( Figure 12(A) ), wound healing assays revealing enhanced migratory potential ( Figure 12(B) ), and Transwell experiments confirming greater invasive capability ( Figure 13(A)-(B) ). Collectively, these findings position LDHC as a potential tumor suppressor in cervical carcinogenesis.