Abstract Background Nasopharyngeal carcinoma (NPC) is a malignant tumor with strong tendency of metastasis and recurrence. Therefore, prognostic indicators for NPC are urgently needed. DGKA is overexpressed in many cancers. However, the role of DGKA in NPC remains unclear. The aim of this study is to investigate the expression patterns and prognostic value of DGKA in NPC. Methods The mRNA expression of DGKA was analyzed from the dataset of The Cancer Genome Atlas (TCGA). Immunohistochemistry was performed to detect the expression of DGKA in 124 paraffin-embedded samples. Survival analysis was conducted according to the Kaplan–Meier method. Univariate and multivariate Cox regression analyses were used to identify the independent prognostic factors. Results We found the mRNA expression of DGKA was significantly higher in tumor tissues than in adjacent tissues in TCGA datasets. Besides, the comparison of the different clinicopathological features of NPC between high and low DGKA expression groups revealed that high DGKA expression was related to the survival status (p = 0.008). Statistical analysis indicates that high expression of DGKA is associated with poor prognosis in NPC. A monogram model was successfully established and demonstrated favorable predictive performance. Bioinformatics analysis showed that DGKA was associated with the activation of oncogenic signaling pathways and the reduction of anti-tumor CD8 + T cell infiltration. Conclusion High expression of DGKA is associated with poor overall survival of NPC patients. DGKA is expected to be a potential biomarker and therapeutic target for NPC. Supplementary Information The online version contains supplementary material available at 10.1007/s12672-025-03110-0. Keywords: DGKA, Nasopharyngeal carcinoma, Clinical prognosis, Bioinformatic analysis Introduction Nasopharyngeal carcinoma (NPC) is a unique malignant tumor that originates from the nasopharyngeal epithelium. It exhibits a significant regional prevalence, with a high incidence in China and Southeast Asian countries, followed by North Africa [[34]1, [35]2]. According to the GLOBOCAN data released by the International Agency for Research on Cancer in 2020, there were approximately 133,354 new cases of nasopharyngeal carcinoma worldwide. More than 70% of these cases were concentrated in high-risk regions [[36]3]. The occurrence of nasopharyngeal carcinoma is related to many factors, including genetic susceptibility, Epstein-Barr virus infection, and chemical carcinogens. Most NPC cases are undifferentiated or poorly carcinoma [[37]4]; therefore, it is highly sensitive to radiotherapy and chemotherapy [[38]5, [39]6]. Although great progress has been made in tumor treatment over the past few years, approximately 20–30% of NPC patients still develop distant metastasis within 5 years after the initial treatment [[40]7]. The prognosis for patients with recurrent or metastatic NPC is poor, with a median overall survival (OS) of approximately 20 months [[41]8]. Therefore, there is an urgent need to identify new biomarkers for early diagnosis and improved prognosis of nasopharyngeal carcinoma. Diacylglycerol kinases (DGKs) are enzymes involved in lipid metabolism that catalyze the transfer of γ-phosphate from adenosine triphosphate (ATP) to the hydroxyl group of diacylglycerol (DAG), resulting in the generation of phosphatidic acid (PA) and participating in the phosphoinositide (PI) cycle [[42]9]. Diacylglycerol kinases (DGKs) are a family of ten PI metabolic enzymes (α, β, γ, δ, ε, ζ, η, θ, ι, and κ) [[43]10, [44]11]. DGKA, the first member of the DGK family identified, is a key regulator of polarized secretion of the exosomal, which has specialized fat and protein content [[45]9, [46]12]. Previous studies have demonstrated that DGKA plays a significant role in T-cell receptor signaling, anergy, and T-cell proliferation [[47]13, [48]14]. Furthermore, DGKA has been found to play important roles in various other biological processes. These include the activation of Rac and membrane folding induced by hepatocyte growth factor, cell migration, regulation of DAG-sensitive isoforms of protein kinase C, and angiogenesis induced by vascular endothelial growth factor [[49]13, [50]14]. In terms of tumors, a series of studies have reported that abnormal expression of DGKA in some types of tumors is related to survival prognosis, such as hepatocellular carcinoma, non-small cell lung cancer, esophageal squamous cell carcinoma and gastric cancer [[51]15–[52]21]. However, the role of DGKA expression in nasopharyngeal carcinoma is still unclear. Therefore, in this study, we firstly comparatively analyzed the expression levels of DGKA in head and neck cancer tissues and normal tissues. Next, the expression of DGKA in the tissue sections of NPC patients was detected by immunohistochemistry, and the relationship between its expression level and the clinicopathological features, laboratory examination, and prognosis of NPC patients was studied to clarify the clinical significance of DGKA in NPC patients. Lastly, we performed an analysis of DGKA’s role in signaling pathways and immune functions through public databases. Materials and methods Data collection and processing Due to the nasopharyngeal carcinoma dataset in the TCGA database was included in head and neck cancer cohort, we downloaded the TCGA-HNSC dataset to analyze the nasopharyngeal carcinoma as previously reported, which could be seen in serval literatures [[53]22–[54]25]. The RNAseq profile and clinical data of HNSC patients of The Cancer Genome Atlas Esophageal Cancer cohort (TCGA-HNSC) and the DGKA expression data in normal and tumor tissue of Genotype-Tissue Expression (GTEx) were obtained from the University of California Santa Cruz (UCSC) Xena database ([55]https://xenabrowser.net/datapages/, accessed on 1 February 2024) in the fragments per kilobase million (FPKM) format. Then, the data were transformed into transcripts per million (TPM) format and normalized into a log2 base before further analysis [[56]26]. To validify our findings in TCGA-HNSC, we perform most analysis in the GEO dataset again. Eight transcriptome sequencing datasets ([57]GSE134886, [58]GSE13597, [59]GSE40290, [60]GSE53819, [61]GSE64634, [62]GSE68799, [63]GSE12452, [64]GSE103249) containing nasopharyngeal carcinoma and normal nasopharyngeal tissues were obtained from the GEO database ([65]https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE134886, [66]GSE13597, [67]GSE40290, [68]GSE53819, [69]GSE64634, [70]GSE68799, [71]GSE12452, [72]GSE103249). Identification of DEGs between high and low DGKA expression group To identify the DEGs between the high DGKA expression and low DGKA expression group divided above in the TCGA-HNSC cohort, the limma package of R was applied. Genes with the expression levels of mRNAs with|log2(FoldChange)| >1 and p-value < 0.05 were regarded as the DEGs, where log2(FoldChange) represents the value of the fold change of the mean between groups transformed into the base logarithm by 2. The DEGs were visualized with the volcano plot. The Kaplan-Meier survival analysis of the TCGA-HNSC To evaluate the prognosis difference between high and low DGKA expression groups, with the “survival” and “survminer” packages of R software, the high and low DGKA expression groups were divided according to the auto select best cutoff-value of DGKA, then the HR with 95% CI and p-values were calculated and generated the Kaplan-Meier plot finally. Pathway enrichment analysis The functional annotation of biological process (BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis for DEGs between high DGKA expression and low DGKA expression group were conducted with the “clusterProfiler” package of R based on Org.Hs.eg.db comment pack. To further explore the pathway, Gene Set Enrichment Analyses (GSEA) was carried out based on the gene set “c2.cp.kegg.v7.5.1.entrez.gmt” and “c2.cp.reactome.v7.5.1.entrez.gmt” downloaded from the Molecular Signatures Databases (MSigDB), with the log2 (FoldChange) data from the DEGs analysis between high DGKA expression and low DGKA expression group. The normalized enrichment scores (NES) > 1 and nominal p-values < 0.05 were used to determine the enrichment levels and statistical significance. R packages “GSEABase” and “clusterProfiler” (version 3.10) were used to perform GSEA and create the GSEA plot. CIBERSORT CIBERSORT is an algorithm based on the principle of linear support vector regression to deconvolute the expression matrix of immune cell subtypes, and use RNA-seq data to estimate the immune cell infiltration of tumor tissue and normal tissue [[73]27]. To explore the relationship between DGKA expression and 20 immune cell subtypes, the CIBERSORT algorithm was utilized to estimate the proportions of 22 immune cell subtypes and set the perm at 1000 and the threshold at p < 0.05. Then, the t-test was used to determine differences between immune cell subtypes in high and low DGKA expression groups. Spearman correlation was used to measure the linear correlation between DGKA expression and immune cell subtypes. Single-sample gene set enrichment analysis Single-sample gene set enrichment analysis (ssGSEA) is a common immune infiltration evaluation method, which is designed for the inability of a single sample to do GSEA. As an extension of GSEA, it allows to definition of an enrichment score, which represents the absolute enrichment degree of each sample gene set in a given data set. The ssGSEA algorithm of the R package “Gene Set Variation Analysis (GSVA)” was used to explore the infiltration level of different immune cell types. Then, Pearson’s correlation coefficient was used to calculate the correlation between DGKA and immune cell types. Tissue microarray and data collection A tissue microarray containing 124 human NPC sample tissues that were collected between May 2010 and May 2014 at Sun Yat-sen University Cancer Center in China was obtained. All patients included were pathologically reassessed according to the 2019 World Health Organization (WHO) criteria by an experienced pathologist, and TNM staging was reperformed according to the eighth edition AJCC guidelines. None of the patients had received treatment. All the clinicopathological data were obtained from the electronic medical record system of Sun Yat-sen University Cancer Center and experiment data was collected from the laboratory information system of Sun Yat-sen University Cancer Center. We collected demographics (including age, gender, patient status, BMI, alcohol, smoking, hypertension, diabetes, TNM stage), histology, comorbidities (including lung metastasis, liver metastasis, bone metastasis), Blood Routine examination data (including WBC, ANC, ALC, MO, NLR, LMR, PLR, PLT), Biochemical Routine examination data (including AST, ALT, LDH, GGT, TBA, ALP). All participants provided written informed consent for study participation. The study was approved by the Ethics Committee of The Sun Yat-sen University Cancer Center(B2023-398-01). Immunohistochemical staining Immunohistochemistry (IHC) was carried out as follows. First, Paraffin-embedded tissue sections were dewaxed in dimethyl benzene and rehydrated in gradient ethanol solutions. Thereafter, endogenous peroxidase activity was then blocked using 3% hydrogen peroxide solution for 10 min, and rinsed in phosphate-buffered saline (PBS) 3 times for 5 min each. Then the tissue sections were heated at 100 °C for 5 min in citrate (10 mmol/L, pH 8.0) solution to retrieve the antigens. After cooling to room temperature, the tissue sections were blocked and incubated with 10% fetal bovine serum for 30 min. The primary antibody DGKA (ab197249; Abcam, Cambridge, MA, USA, 1:200 dilution) was added and incubated at 4 °C overnight, and the tissue sections were subsequently incubated with HRP-labeled universal anti-mouse/rabbit secondary antibody in the dark. Immunostaining was visualized with diaminobenzidine (DAB), and the tissue sections were counterstained with hematoxylin, then dehydrated and mounted. Immunohistochemical scoring Based on the proportions of positive cells and staining intensity, the level of DGKA expression was assessed by two independent pathologists who were blinded to the clinicopathological information. The percentage of positive cells was evaluated quantitatively and scored as 0 (0–5%), 1 (6–25%), 2 (26–50%), 3 (51–75%), and 4 (76–100%), whereas staining intensity was rated as 1 (weak), 2 (moderate), and 3 (strong). A total “staining score” of 0–12 was calculated by multiplying the staining intensity and percentage scores. The final DGKA expression was defined as follows: negative (−), 0 points; low expression (+), 1–8 points; and high expression (+ + +), 9–12 points. All NPC patients were divided into two groups of low and high DGKA expression according to the X-tile determined cutoff value of 8 for the DGKA IHC score. The X-tile software adopts the enumeration method, and replaces each value into the calculation. The value corresponding to the smallest p-value is the best cut-off value. Follow-up All patients were followed up every three to six months in the outpatient clinic or by telephone, with the last telephone follow-up in February 2023. The time of death was recorded for patients who died, and the last follow-up date and status of patients who could not be reached were obtained from the hospital system. The outcome was overall survival (OS), defined as the time from the date of diagnosis to death or the last follow-up. Statistical analysis All RNAseq bioinformatic work was done in R (version 4.2.0) environment. GraphPad Prism9 (GraphPad Software, Inc.) and SPSS (IBM, USA) were used for data analysis. Continuous variables with normal or nonnormal distribution are expressed as median and IQR. Categorical variables are shown in frequency and percentages Continuous variables with normal or abnormal distribution are compared by the Students’ test or Mann–Whitney U-test. χ2 test or Fisher’s test were used to compare Categorical variables respectively. The cut-off value of the expression of the marker was based on X-tile (Version 3.6.1, New Haven, USA). Cox proportional hazards regression was performed to identify prognostic factors. Variables with significant differences in the univariate analysis were further analyzed with a multivariate analysis to determine the independent prognostic factors. Based on the prognostic factors, the nomogram was formulated using the “survival”and “rms” package in R. Each predictor included in the nomogram was represented on one row, and a corresponding number of points was assigned to different magnitudes of the predictor. The cumulative point axis was represented at the end of the nomogram, and higher total points indicated a worse survival outcome. The discriminative ability of the model was evaluated with Harrell’s concordance index (C-index) and receiver operating characteristic (ROC) curve analysis. A larger C-index value and a greater area under the curve (AUC) in the ROC curve indicated better discrimination ability. Calibration curves (1000 bootstrap resamples) were generated to evaluate the calibration ability of the nomogram. Kaplan–Meier method with long-rank was used to estimate the survival. P-value < 0.05 was considered statistically significant. Results DGKA is highly expressed in pan-cancer tissues Using the RNA-seq data of TCGA, we first analyzed the gene expression levels of DGKA in different human tumor tissues and normal tissues. The results showed that the mRNA expression of DGKA was significantly higher in multiple types of tumor tissues than corresponding normal tissues, including HNSC (Supplementary-Figure 1A-B). In [74]GSE134886, the expression of DGKA in NPC tumor tissues was significantly higher than in normal nasopharyngeal tissues (p < 0.05) (Supplementary-Figure 1C). However, in [75]GSE12452, [76]GSE64634 and [77]GSE68799, there was no difference in the expression of DGKA between normal tissues and tumor tissues. In [78]GSE13597, [79]GSE40290 and [80]GSE53819, the expression of DGKA in normal tissues was higher than that in tumor tissues (Supplementary-Figure 2A-B). In addition, the prognostic value of DGKA expression was reached using the Kaplan–Meier survival analysis based on the survival information of the TCGA-HNSC cohort. The analysis showed such a trend that the survival time of patients with high DGKA expression was shorter than that of patients with low DGKA expression, though there was no statistical significance (p = 0.096; Fig. [81]1D). However, prognostic analysis of [82]GSE102349 revealed no survival difference between high and low DGKA expression groups (p = 0.24) (Supplementary-Figure 2C). We also investigated the power of DGKA in distinguishing the tumor and normal tissue. The ROC analysis revealed that DGKA exhibited a limited sensitivity and specificity in telling patients apart from normal people with AUC: 0.650(95% CI: 0.455–0.869) (Supplementary-Figure 1E). Based on the fact that DGKA was highly expressed and related to poor prognosis in HNSC, we also explored the genetic alterations of DGKA. We performed OncoPrint analysis using cBioPortal, and the OncoPrint landscape showed that patients with higher mutation count had higher rates of metastasis and poor survival (Supplementary-Figure 1F). Those results demonstrated the high level of DGKA expression and its poor prognosis value in HNSC. Fig. 1. [83]Fig. 1 [84]Open in a new tab (A) HIC staining for low and high expression of DGKA (200X and 400X). (B) Differences in scores between DGKA high and low expression groups Clinicopathological characteristics of patients A total of 124 patients with NPC were included in this study. Table [85]1 summarizes the clinicopathological characteristics of these patients. Most NPC patients (70.2%) were infected with the Epstein-Barr virus (EBV), and more than half (70.2%) were male. The tumor histopathological type of all NPC patients was undifferentiated non-keratinizing carcinoma (UNKC). There were 90 cases (72.6%) with advanced T stage (T3, T4) and 106 cases (85.5%) with advanced clinical stage (III, IV). Until the last follow-up, 22 patients (17.7%) had bone metastasis, 13 patients (10.3%) had lung metastasis, and 9 patients (7.3%) had liver metastasis. A total of 55 patients (44.4%) died. Table 1. Clinical data of 124 cases of nasopharyngeal carcinoma Characteristics Total (n,%) Age, year ≤ 46 67(54.0%) >46 57(46.0%) Gender Male 87(70.2%) Female 37(29.8%) BMI ≤ 22.3 70(56.5%) > 22.3 54(43.5%) EBV No 37(29.8%) Yes 87(70.2%) Alcohol No 96(77.4%) Yes 28(22.6%) Smoking No 77(62.1%) Yes 47(37.9%) Hypertension No 111(89.5%) Yes 13(10.5%) Diabetes No 117(94.4%) Yes 7(5.6%) T category T1-T2 34(27.4%) T3-T4 90(72.6%) LN metastasis N0-N1 52(41.9%) N2-N3 72(58.1%) Distant metastasis Bone metastasis Absent 102(82.3%) Present 22(17.7%) Lung metastasis Absent 111(89.5%) Present 13(10.5%) Liver metastasis Absent 115(92.7%) Present 9(7.3%) Clinical stage I-II 18(14.5%) III-IV 106(85.5%) Death No 69(55.6%) Yes 55(44.4%) [86]Open in a new tab Correlation between DGKA expression and clinicopathological characteristics of NPC patients As shown in Fig. [87]1A, immunohistochemical staining demonstrated positive cytoplasmic staining of DGKA in tumor cells. DGKA was expressed in all 124 cases of nasopharyngeal carcinoma tissues. However, DGKA expression was undetectable in adjacent normal tissues. The cut-off value of the DGKA immunohistochemical score determined by x-tile was 8. The score greater than 8 was considered highly expressed. Among them, 94 cases (75.8%) of NPC tissues were DGKA highly expressed. 30 cases (24.2%) of NPC tissues were DGKA lowly expressed. There was a significant difference in the scores between the DGKA high and low expression groups (Fig. [88]1B). Next, we evaluated the association of DGKA expression in NPC with the following clinicopathological features and laboratory findings: Gender, age at diagnosis, smoking, drinking, BMI, hypertension, diabetes, T stage, N stage, bone metastasis after diagnosis, lung metastasis and liver metastasis after diagnosis, clinical stage, survival status, EBV, WBC, ANC, ALC, MO, PLT, NLR, LMR, PLR, AST, ALT, LDH, GGT, TBA, ALP. We found that high expression of DGKA was associated with patient prognosis (p = 0.008) (Table [89]2). Table 2. Correlation between the expression of DGKA and clinicopathologic characteristics in nasopharyngeal carcinoma Characteristics Total DGKA expression p value Low (n = 30,%) High (n = 94, %) Age, (median), n(%) ≤ 46 67(54.0%) 18(60.0%) 49(52.1%) 0.451 >46 57(46.0%) 12(40.0%) 45(47.9%) Gender, n(%) Male 87(70.2%) 22(73.3%) 65(69.1%) 0.663 Female 37(29.8%) 8(26.7%) 29(30.9%) BMI, (mean), n(%) ≤ 22.3 70(56.5%) 15(50.0%) 55(58.5%) 0.413 > 22.3 54(43.5%) 15(50.0%) 39(41.5%) EBV, n(%) No 37(29.8%) 12(40.0%) 25(26.6%) 0.162 Yes 87(70.2%) 18(60.0%) 69(73.4%) Alcohol, n(%) No 96(77.4%) 23(76/7%) 73(77.7%) 0.91 Yes 28(22.6%) 7(23.3%) 21(22.3%) Smoking, n(%) No 77(62.1%) 21(70.0%) 56(59.6%) 0.305 Yes 47(37.9%) 9(30.0%) 38(40.4%) Hypertension, n(%) No 111(89.5%) 26(86.7%) 85(90.4%) 0.808 Yes 13(10.5%) 4(13.3%) 9(9.6%) Diabetes, n(%) No 117(94.4%) 27(90.0%) 90(95.7%) 0.464 Yes 7(5.6%) 3(10.0%) 4(4.3%) T category, n(%) T1-T2 34(27.4%) 11(36.7%) 23(24.5%) 0.192 T3-T4 90(72.6%) 19(63.3%) 71(75.5%) LN metastasis, n(%) N0-N1 52(41.9%) 15(50.0%) 37(39.4%) 0.304 N2-N3 72(58.1%) 15(50.0%) 57(60.6%) Distant metastasis Bone metastasis, n(%) Absent 102(82.3%) 22(73.3%) 80(85.1%) 0.142 Present 22(17.7%) 8(26.7%) 14(14.9%) Lung metastasis, n(%) Absent 111(89.5%) 26(86.7%) 85(90.4%) 0.808 Present 13(10.5%) 4(13.3%) 9(9.6%) Liver metastasis, n(%) Absent 115(92.7%) 27(90.0%) 88(93.6%) 0.794 Present 9(7.3%) 3(10.0%) 6(6.3%) Clinical stage, n(%) I-II 18(14.5%) 6(20.0%) 12(12.8%) 0.495 III-IV 106(85.5%) 24(80.0%) 82(87.2%) Death, n(%) No 69(55.6%) 23(76.7%) 46(48.9%) 0.008 Yes 55(44.4%) 7(23.3%) 48(51.1%) WBC 6.85(6.00-8.30) 7.25(6.27–8.36) 6.64(5.90–8.30) 0.228 ANC 4.35(3.63–5.60) 4.75(3.78–5.63) 4.30(3.58–5.65) 0.538 ALC 1.78(1.43–2.20) 1.95(1.48–2.60) 1.70(1.40–2.10) 0.287 MO 0.40(0.30–0.50) 0.45(0.30–0.53) 0.40(0.30–0.50) 0.459 PLT 244.92(± 64.70) 245.53(± 59.62) 244.73(± 66.55) 0.953 NLR 2.54(1.92–3.57) 2.44(1.67–3.95) 2.54(1.95–3.50) 0.711 LMR 4.25(3.25–5.31) 4.78(2.77-6.00) 4.23(3.31-5.00) 0.23 PLR 134.59(99.80-179.11) 129.89(93.74–172.00) 135.84(100.89–184.80) 0.407 AST 21.00(17.90–25.90) 20.85(18.38–26.78) 21.10(17.83–25.900 0.951 ALT 21.30(14.75–36.10) 23.90(14.95–38.83) 21.00(14.70-34.68) 0.543 LDH 176.50(158.53-216.38) 171.25(155.73-204.93) 180.15(160.48-217.43) 0.282 GGT 24.50(16.30–38.60) 27.10(18.28–39.53) 23.95(16.08–38.40) 0.519 TBA 2.70(1.43–4.58) 2.35(1.38–3.73) 2.80(1.50–4.90) 0.337 ALP 74.80(60.40-91.13) 72.80(56.35–87.13) 74.80(60.80-93.05) 0.349 [90]Open in a new tab The highlighted text indicates that the P value is less than 0.05. The results show statistically significant differences. Association between DGKA expression and survival outcome in NPC To determine whether DGKA expression, clinical pathological characteristics, and laboratory findings can serve as independent risk factors in NPC, we performed univariate and multivariate analyses based on Cox regression analysis. In univariate analysis, we found that age (p = 0.028), bone metastasis after diagnosis (p = 0.031), lung metastasis after diagnosis (p = 0.027), liver metastasis after diagnosis (p = 0.029), LDH (p = 0.001), ALP (p = 0.004), and high expression of DGKA (p = 0.023) were significantly associated with OS. Multivariate COX model analysis showed that liver metastasis after diagnosis (p = 0.010, HR: 3.196; CI: 1.315–7.770), LDH (p = 0.039, HR: 1.003; CI: 1.000-1.005), and high expression of DGKA (p = 0.027, HR: 2.520; CI: 1.112–5.711) were independent predictors of OS (Table [91]3). Table 3. Univariate and multivariate Cox regression analysis of overall survival in patients with nasopharyngeal carcinoma Variable Univariable analysis Multivariate analysis HR(95% CI) P-value HR(95% CI) P-value DGKA expression 2.521 (1.138–5.582) 0.023 2.520 (1.112–5.711) 0.027 Sex 0.872 (0.491–1.549) 0.641 Age 1.030 (1.003–1.057) 0.028 1.018 (0.993–1.044) 0.155 BMI 0.928 (0.835–1.031) 0.163 Alcohol 0.827 (0.426–1.604) 0.574 Smoking 1.125 (0.653–1.936) 0.672 Hypertension 0.923 (0.367–2.320) 0.865 Diabetes 0.259 (0.036–1.875) 0.181 T category 1.882 (0.966–3.666) 0.063 LN metastasis 1.391 (0.799–2.421) 0.244 Distant metastasis Bone metastasis 1.964 (1.065–3.620) 0.031 1.601 (0.825–3.108) 0.164 Lung metastasis 2.255 (1.099–4.628) 0.027 2.054 (0.963–4.380) 0.063 Liver metastasis 2.596 (1.104–6.102) 0.029 3.196 (1.315–7.770) 0.010 Clinical stage 1.801 (0.765–4.240) 0.178 EBV 1.202 (0.687–2.102) 0.519 WBC 1.036 (0.925–1.160) 0.543 ANC 1.008 (0.873–1.163) 0.918 ALC 1.354 (0.895–2.047) 0.151 MO 1.790 (0.554–5.776) 0.330 PLT 1.001 (0.997–1.005) 0.662 NLR 0.857 (0.705–1.041) 0.120 LMR 1.033 (0.911–1.172) 0.612 PLR 0.998 (0.994–1.002) 0.253 AST 0.992 (0.965–1.019) 0.539 ALT 0.996 (0.983–1.009) 0.558 LDH 1.004 (1.002–1.006) 0.001 1.003 (1.000-1.005) 0.039 GGT 0.999 (0.993–1.005) 0.716 TBA 0.993 (0.951–1.037) 0.760 ALP 1.015 (1.004–1.025) 0.004 1.012 (0.999–1.024) 0.063 [92]Open in a new tab Further, we evaluated the prognostic performance of DGKA in predicting OS outcomes in all NPC patients. Kaplan-Meier survival analysis showed that patients in the low DGKA expression group had better survival results in terms of OS(p = 0.018) (Fig. [93]2A). Next, we analyzed the OS outcomes between the DGKA low and high expression groups in patients with different phenotypes. In NPC patients without liver metastasis, the OS of patients with high DGKA expression was significantly lower than that of patients with low DGKA expression p = 0.001 (Fig. [94]2B), but there was no difference in patients with liver metastasis p = 0.944 (Fig. [95]2C). Fig. 2. [96]Fig. 2 [97]Open in a new tab Kaplan–Meier survival curves grouped by high and low DGKA expression in NPC patients. (A) Overall survival (OS) in all patients. (B) Kaplan–Meier survival for OS in subgroups stratified by no liver metastasis, (C) liver metastasis Construction and efficacy evaluation of nomogram model Based on the prognostic factors related to OS screened by multivariate COX regression analysis, a Nomogram model for predicting 3-year overall survival rate and 5-year overall survival rate was constructed using DGKA expression, liver metastasis after diagnosis, and LDH (Fig. [98]3). Each variable in the Nomogram model has a different range of values, and the corresponding score is related to the HR value of COX regression analysis. Each patient was scored by the Nomogram model according to their own risk factors, and the total score was projected to the column of probability, so that the probability of predicting 3-year overall survival and 5-year overall survival could be obtained. Figure [99]4 shows the ROC curve and calibration curve of the 5-year overall survival probability and the 3-year overall survival probability model of the model. The AUC of the ROC curve of the 3-year overall survival probability model was 0.690, the AUC of the ROC curve of the 5-year overall survival probability model was 0.673, and the c index was 0.633, indicating that the model had a good prediction effect. Fig. 3. [100]Fig. 3 [101]Open in a new tab Nomogram model for predicting 3-year and 5-year overall survival Fig. 4. [102]Fig. 4 [103]Open in a new tab ROC curve and calibration curve of Nomogram model (A) ROC curve of 3-year OS probability; (B) ROC curve of 5-year OS probability; (C) Calibration curve for predicting 3-year OS probability; (D) Calibration curve for predicting 5-year OS probability The differentially expressed genes and pathway enrichment analysis between high and low DGKA expressed patients in HNSC and NPC To further explore the pathway changes caused by the high expression of DGKA, we firstly obtained the differentially expressed genes (DEGs) between the high and low DGKA expression groups and showed the DEGs with a volcano plot (Supplementary-Figure 3A). Then, we performed GO analysis for the DEGs, including the biological process (BP), molecular function (MF) and cellular component (CC). The BP analysis showed that DEGs were mainly enriched in cell-cell adhesion via plasma-membrane adhesion molecules and homophilic cell adhesion via plasma membrane adhesion molecules (Supplementary-Figure 3B). The MF analysis showed that DEGs were mainly enriched in antigen binding and immunoglobulin receptor binding (Supplementary-Figure 3C). The CC analysis showed that immunoglobulin complex and circulating immunoglobulin complex were the most abundant terms for DEGs (Supplementary-Figure 3D). Besides, the REACTOME pathway analysis of all DGEs showed that most of the DEGs involved in the degradation of the extracellular matrix, extracellular matrix organization, collagen formation, anchoring fibril formation, laminin interactions, the Nonintegrin membrane the ECM interaction, the assembly of collagen fibrils and other multimeric structure (Supplementary-Figure 3E). To better estimate the status of pathways, we performed GSEA enrichment analysis based on the log2(Fold change) of the DEGs. The enrichment of KEGG pathways identified 7 important signaling pathways activated in the high DGKA expressed group, such as ERBB signaling pathway, metabolism of xenobiotics by cytochrome P450, pathways in cancer and so on. It was found 10 pathways suppressed in the high DGKA expressed group, such as ribosome, allograft rejection, type I diabetes mellitus, graft versus host disease, systemic lupus erythematosus, oxidative phosphorylation and so on (Supplementary-Figure 3F). In the field of REACTOME pathways, the analysis showed that 10 processes were activated in the high DGKA expressed group, such as type I hemidesmosome assembly, apoptotic cleavage of cell adhesion proteins, RND3 GTPase cycle, GTPase cycle, RHO GTPase cycle, formation of the cornified envelope, keratinization, etc. It was found to be suppressed in 10 aspects, such as CD22 mediated BCR regulation, FCGR activation, scavenging of heme from plasma, creation of c2 and c4 activators and so on (Supplementary-Figure 3G). For the exploration of pathways at different angles, the GSEA analysis was also performed with the terms of MF and BP. The GSEA analysis of MF showed that helicase activity, ATP dependent activity acting on DNA, cadherin binding, protein serine kinase activity, and ATP hydrolysis activity were activated, while structural constituent of ribosome, oxidoreduction driven active transmembrane transporter activity, NADPH dehydrogenase quinone activity, and rRNA binding were suppressed (Supplementary-Figure 3H). In terms of BP analysis, the results showed that 10 biological processes were activated, including the regulation of water loss via skin, golgi organization, ERBB signaling pathway, keratinocyte differentiation, epidermis development, epidermal cell differentiation and so on; As for the suppressed processes, the enrichment showed that phagocytosis recognition, complement activation, B cell receptor signaling pathway and humoral immune response mediated by circulating immunoglobulin were suppressed (Supplementary-Figure 3I). The same method was used to conduct pathway enrichment analysis on the DEGs of the between the high and low DGKA expression groups in [104]GSE134886. A volcano plot showed the DEGs (Supplementary-Figure 4A). The BP analysis showed that DEGs were mainly enriched in cilium organization, cilium assembly, microtubule-based transport and so on (Supplementary-Figure 4B). The CC analysis showed that DEGs were mainly enriched in ciliary tip and intraciliary particle (Supplementary -Figure 4C). Besides, the KEGG pathway analysis of all DGEs showed that most of the DEGs involved in chemical carcinogenesis-DNA adducts, metabolism of xenobiotics by cytochrome P450, drug metabolism- cytochrome P450 and drug metabolism-other enzymes (Supplementary-Figure 4D). To better estimate the status of pathways, we performed GSEA enrichment analysis based on the log2(Fold change) of the DEGs. The GSEA analysis of BP pathways showed that cilium organization, microtubule-based movement and so on were activated, while positive regulation of interferon gamma production, interferon gamma production, phagocytosis recognition and humoral immune response mediated by circulating immunoglobulin were suppressed (Supplementary-Figure 4E). The enrichment of REACTOME pathways identified 5 important signaling pathways activated in the high DGKA expressed group, such as organelle biogenesis and maintenance, cilium assembly, biological oxidations and so on. It was found 10 pathways suppressed in the high DGKA expressed group, such as CD22 mediated BCR regulation, FCGR activation, scavenging of heme from plasma, creation of c2 and c4 activators and so on (Supplementary-Figure 4F). In the field of KEGG pathways, the analysis showed that 4 pathways were activated in in the high DGKA expressed group, such as metabolism of xenobiotics by cytochrome P450, drug metabolism- cytochrome P450, retinol metabolism and steroid hormone biosynthesis. It was found to be suppressed in 7 aspects, such as antigen processing and presentation, leishmania infection, toll like receptor signaling pathway and so on (Supplementary-Figure 4G The relationship between the expression level of DGKA and the immune infiltration level in NPC and HNSC Apart from the activation of oncogenic signaling pathways, the immune microenvironment also plays an important role in the progress of tumor. We further investigated whether certain links exist between genes and immune cells. Firstly, we analyzed the DGKA expression of immune cells in tumor and normal tissue by GEPIA website tool. The boxplot showed that the tumor infiltrated-Treg cells expressed higher level of DGKA than normal tissue-infiltrated, but the tumor infiltrated-M1 cells expressed lower level of DGKA than normal tissue-infiltrated, while the DGKA expression in CD8 + T cells, activated NK cells and M2 cells had no significant changes (Supplementary-Figure 5 A). To further explore whether there was a difference in the infiltration of the immune cells, two immune infiltration estimation algorithms, CIBERSORT and ssGSEA, were applied, and the patients of TCGA-HNSC were divided into high and low DGKA expression groups based on the survival analysis with the best cutoff value. In the results of the CIBERSORT, the proportions of various immune cell infiltrations in tumor tissues with high and low expression of DGKA were evaluated. It was found that the DGKA low expression group had a significantly higher proportion of CD8 + T cells compared to the DGKA high expression group. Conversely, the DGKA high expression group showed a higher proportion of activated DC cells and Tregs T cells compared to the DGKA low expression group (Supplementary-Figure 5B). For further argument, another estimation algorithm ssGSEA was performed, the dot plot showed that the infiltration of a variety of immune cells in DGKA low expression group was significantly higher than that in the DGKA high expression group, such as activated B cells, activated CD8 + T cells, M1, M2 cells, etc. (Supplementary-Figure 5 C). The same method was used to evaluate the infiltration of immune cells in [105]GSE134886. The results of [106]GSE134886 CIBERSORT2 showed that infiltrated-M1 cells and T cells follicular helper infiltrated more than in the DGKA low-expression group than in the DGKA high-expression group (Supplementary-Figure 5D). The dot plot of [107]GSE134886 ssGSEA showed that the infiltration of a variety of immune cells in DGKA low expression group was significantly higher than that in the DGKA high expression group, such as activated CD4 T cells, activated CD8 + T cells, nature killer T cells, M1 cells, etc. (Supplementary-Figure 5E). Based on the above results, it can be seen that in TCGA-HNSC and [108]GSE134886, the infiltration of CD8 + T cells, M1 cells and NK T cells in the low-expression group of DGKA was significantly more than that in the high-expression group of DGKA. Finally, the correlations between DGKA expression and immune infiltration score were explored. In terms of infiltration of CD8 + T cells, DC cells and Tregs cells, the CIBERSORT infiltration score of CD8 + T cells (R = −0.12, p = 0.0064) and Tregs cells (R = −0.098, p = 0.03) was negatively correlated with the expression of DGKA, expect the DC cells (R = 0.16, p = 0.00042) (Supplementary-Figure 5 F). The same results were obtained from the ssGSEA, the ssGSEA infiltration scores of CD8 + T cells (R = −0.14, p = 0.0018), DC cells (R = −0.15, p = 0.0012), and Tregs cells (R = −0.15, p = 0.00079) were negatively correlated with DGKA expression. (Supplementary-Figure 5G). Discussion Due to the hidden location, mild early symptoms, and strong invasiveness of NPC, most patients are diagnosed at an advanced stage [[109]28]. Despite considerable progress in surgical resection, radiotherapy, chemotherapy, and immunotherapy for NPC, the clinical efficacy of these therapeutic options for advanced NPC remains disappointing [[110]7, [111]8]. Therefore, this study aims to explore new biomarkers of NPC to improve early diagnosis and prognosis of patients. We first analyzed the mRNA expression levels of DGKA in the pan-cancer samples obtained from the TCGA dataset. We found that the expression level of DGKA was significantly increased in many tumor tissues, compared with normal tissues. A number of previous studies have used TCGA HNSC dataset to study nasopharyngeal carcinoma-related markers [[112]26, [113]29–[114]31]. Given the absence of dedicated nasopharyngeal carcinoma datasets in the TCGA database, we utilized HNSCC data as a surrogate for NPC in the analytical workflow. In addition, we obtained eight transcriptome sequencing datasets containing nasopharyngeal carcinoma and normal nasopharyngeal tissues from the GEO database. Next, we analyzed TCGA HNSC cohort and found that DGKA expression was significantly higher in head and neck tumor tissues than in normal tissues. While the analytical results from these transcriptome sequencing datasets show inconsistencies with the initial results, this precisely further emphasizes the importance and novelty of clarifying the role of DGKA in the expression and prognosis of nasopharyngeal carcinoma. The experimental validation provided in this study carries stronger evidentiary weight. Some studies have reported that high expression of DGKA in hepatocellular carcinoma, non-small cell lung cancer and esophageal squamous cell carcinoma is associated with poor prognosis [[115]15–[116]17]. However, high DGKA expression was protective in gastric cancer, and low DGKA expression was associated with poor prognosis [[117]32]. In this study, COX regression analysis showed that DGKA as an independent prognostic marker for OS in NPC. Kaplan-Meier survival analysis showed that DKGA high expression was closely related to poor OS. In conclusion, validation using clinical NPC samples confirmed that DGKA is highly expressed in tumor tissues and associated with poor prognosis, marking the first such verification in real-world specimens. Nomogram model, a recognized evidence-based prognostic prediction method, has been widely used in a variety of malignant tumor related prognostic studies [[118]33]. Therefore, based on COX regression analysis, this study successfully constructed a Nomogram model by incorporating the independent risk factors affecting the OS of patients with NPC. In previous studies on the Nomogram model of NPC, the Nomogram model of Tang Linquan’s and Yang lin’s studies did not include systemic inflammatory parameter indicators [[119]34, [120]35]. Jian peili’s study included systemic inflammatory parameters and other indicators, including age, TNM stage, EBV-DNA, LDH, hs-CRP, HDL-C, HGB, and LMR to construct a Nomogram model. The c index of the modeling cohort model was 0.800, and the c index of the validation cohort model was 0.831. The Nomogram prediction model has high accuracy and can be used to predict the prognosis of NPC [[121]36]. Liww et al. first reported the association between pretreatment with serum LDH and nasopharyngeal carcinoma [[122]37]. At present, many studies have shown that LDH level at diagnosis is of great prognostic significance for NPC patients [[123]37–[124]39]. This study collected the indexes of sex, age, BMI, smoking, drinking, hypertension, diabetes, TNM stage, survival status, EBV, WBC, RBC, ANC, ALC, MO, PLT, NLR, LMR, PLR, AST, ALT, LDH, GGT, TBA, ALP. The model included DGKA expression, liver metastasis after diagnosis and LDH. Although the number of indicators included in the model was less than that in previous research models, DGKA expression was included in the model for the first time. The mechanism by which high DGKA expression is associated with poor prognosis in NPC remains unclear. We sought to explore the potential mechanisms by analyzing public data. A large number of studies have confirmed that DGKA participates in signaling pathways to promote the occurrence and development of tumors. For example, Takeishi et al. reported that DGKA promotes the progression of hepatocellular carcinoma by activating the Ras-Raf-MEK-ERK pathway [[125]15]. Lingyi Fu et al. reported that DGKA interacts with SRC/FAK to promote metastasis of non-small cell lung cancer [[126]16]. Jie Chen et al. reported that DGKA stimulates the Akt/NF-κB pathway under upregulation of inflammatory factors, thereby promoting the progression and metastasis of esophageal squamous cell carcinoma [[127]17]. Kenji et al. reported that DGKA inhibits tumor necrosis factor-α-induced apoptosis in human melanoma cells by activating NF-κB [[128]18]. Charli L et al. identified DGKA as a potential therapeutic target for glioblastoma [[129]19]. Jie Li et al. reported that DGKA was involved in platinum resistance in ovarian cancer by activating the c-JUN-WEE1 signaling pathway [[130]20]. In 3D colon and breast cancer models, DGKA has been reported to promote cell survival by regulating SRC signaling [[131]21]. In this study, the differentially expressed genes between DGKA high expression group and low expression group were obtained, and then pathway enrichment analysis showed that ERBB-related signaling pathway was highly active in DGKA high expression group. The ERBB family of transmembrane receptor tyrosine kinases consist of epidermal growth factor receptors EGFR(ERBB1), HER2(ERBB2), HER3(ERBB3), and HER4(ERBB4) [[132]40]. Activation of HER2 and EGFR stimulates intracellular pathways such as RAS/RAF/MEK/ERK, PI3K/AKT/TOR, SRC kinases, and STAT transcription factors [[133]41]. Some scholars have confirmed that the occurrence and development of nasopharyngeal carcinoma is related to PI3K/AKT, MAPK and other signaling pathways [[134]42]. In conclusion, DGKA may be involved in ERBB-related signaling pathways to promote the occurrence and development of NPC. Studies have shown that DGKA is highly expressed in T cells in addition to its expression in tumor cells [[135]43, [136]44]. DGKα promotes an immune anergy (nonproliferation) state known as T-cell clonal anergy [[137]45–[138]47]. T cell anergy induction is a major mechanism through which advanced tumors avoid immune responses [[139]48]. DGKα limits anti-tumor immune responses by impeding tumor-infiltrating CD8 + T cells [[140]49]. Studies have shown that during tumor immunotherapy, DGKα mediated T-cell dysfunction by exacerbating the exhaustion of reinvigorated tumor-specific T cells [[141]50]. Indeed, inhibiting the activity of DGKα is considered to enhance T cell activity, which regulates cancer immunity [[142]12, [143]51–[144]53]. In the field of tumor immune infiltration, we found that the infiltration of CD8 + T cells, M1 cells and NK T cells in the low-expression group of DGKA was significantly more than that in the high-expression group of DGKA and CD8 + T cell was negatively related to the expression of DGKA, while DC cells and Treg T cells showed an inconsistent correlation within different analyses. This suggests that DGKA may promote T cell anergy in the immune environment of NPC as one of the underlying mechanisms. In this study, we reveal a negative effect of DGKA in the NPC, which might contribute to the progress of the tumor. However, further investigation of the biological functions of DGKA in promoting NPC development and its potential regulatory mechanisms are warranted. This study has several limitations that warrant careful consideration. First, this study is subject to inherent methodological constraints due to the analytical substitution of HNSCC data from TCGA-HNSC project for NPC research, necessitated by the scarcity of publicly available NPC-specific multiomics databases. Second, this study lacks basic research to support the mechanism of DGKA promoting the occurrence and development of nasopharyngeal carcinoma. Last, the investigation is constrained by a limited sample size and the single-center origin of the clinical data. The retrospective design of this study inherently introduces selection and information biases that require rigorous quantification. In conclusion, our study showed that DGKA is highly expressed in NPC tissues and high expression is closely related to poor OS of NPC. Therefore, DGKA is expected to be a potential biomarker and therapeutic target for nasopharyngeal carcinoma. Electronic supplementary material [145]Supplementary Material 1^ (325KB, jpg) [146]Supplementary Material 2^ (128KB, jpg) [147]Supplementary Material 3^ (308.1KB, png) [148]Supplementary Material 4^ (266.5KB, jpg) [149]Supplementary Material 5^ (422.7KB, jpg) Acknowledgements