3.1 Differential and prognostic analysis of ZNF83 in pan-cancer

ZNF83 was found to be upregulated in eight tumor types (particularly KIRC) and downregulated in 22 tumor types (especially LUAD), compared to normal tissues based on pan-cancer datasets (Fig. 1A). Prognostic analysis revealed that higher ZNF83 expression was associated with poorer overall survival (OS) in seven tumor types (LIHC, LUSC, KIPAN, KIRC, LAML, GBMLGG, LGG), but correlated with better OS in two tumor types (BLCA and HNSC) (Fig. 1B). Regarding disease-specific survival (DSS), high ZNF83 expression indicated a worse prognosis in three tumor types (THYM, KIRP, BLCA) and a poorer prognosis in seven others (LUSC, SKCM, GBMLGG, LGG, MESO, THCA, PRAD) (Fig. 1C). Elevated ZNF83 expression was associated with worse disease-free interval (DFI) in PRAD (Fig. 1D). For progression-free interval (PFI), high ZNF83 expression predicted poorer outcomes in six tumor types (LUSC, SKCM, LGG, GBMLGG, ACC, PRAD) and better outcomes in four others (PAAD, KIRP, BLCA, SKCM-M) (Fig. 1E).

Fig. 1

Expression and prognosis analysis of ZNF83 in pan-cancer. (A) ZNF83 was differentially expressed between tumor and normal tissues in pan-cancer. (B) Pan-cancer analysis of ZNF83 for overall survival. (C) Pan-cancer analysis of ZNF83 for disease-specific survival. (D) Pan-cancer analysis of ZNF83 for disease-free interval. (E) Pan-cancer analysis of ZNF83 for progression-free interval. *P < 0.05; **P < 0.01; ****P < 0.0001

3.2 Clinical feature analysis of ZNF83 in pan-cancer

We further examined the relationship between ZNF83 expression and various clinical parameters in cancer cohorts. ZNF83 levels showed a positive correlation with patient age in GBMLGG (R = 0.18), LGG (R = 0.11), THYM (R = 0.24), and TGCT (R = 0.20), but an inverse relationship in SARC (R = -0.25), HNSC (R = -0.12), and LIHC (R = -0.15) (Fig. 2A). Sex-based analysis revealed differential expression of ZNF83 in seven tumor types (LUAD, COADREAD, ESCA, STES, SARC, LIHC, THCA) (Fig. 2B). Tumor grade analysis indicated significant differences in five tumor types (GBMLGG, LGG, UCEC, HNSC, LIHC) (Fig. 2C). ZNF83 expression varied across N-stage in six tumor types (LUAD, KIRP, KIPAN, PRAD, THCA, CHOL) and across M-stage in four (CESC, COAD, COADREAD, SKCM) (Fig. 2D-E). Clinical stage analysis identified differences in six tumor types (LUAD, BRCA, KIRP, THCA, TGCT, ACC), while T-stage analysis revealed differential expression in eight tumor types (LUAD, BRCA, KIRP, KIPAN, STAD, TGCT, BLCA, ACC) (Fig. 2F-G).

Fig. 2

Association of ZNF83 expression with clinical features across human cancers. (A) Correlation between ZNF83 expression and patient age in pan-cancer. (B) Correlation between ZNF83 expression and gender in pan-cancer. (C) Correlation between ZNF83 expression and tumor grade in pan-cancer. (D) Correlation between ZNF83 expression and N stage in pan-cancer. (E) Correlation between ZNF83 expression and M stage in pan-cancer. (F) Correlation between ZNF83 expression and overall clinical stage in pan-cancer. (G) Correlation between ZNF83 expression and T stage in pan-cancer. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

3.3 Correlation between ZNF83 expression and tumor stemness and heterogeneity across pan-cancer types

We analyzed the association of ZNF83 expression with two stemness indices. ZNF83 negatively correlated with DNAss in CESC, LUAD, LAML, BRCA, KIPAN, and LIHC, but positively associated in GBMLGG, LGG, PRAD, HNSC, THYM, TGCT, and PCPG (Fig. 3A). RNAss showed a predominantly inverse association with ZNF83 across 25 cancers, but a positive link in PCPG (Fig. 3B). We next assessed ZNF83’s relationship with two heterogeneity indicators. A positive correlation with TMB was noted in KIPAN, KIRC, and OV (Fig. 3C). ZNF83 expression positively correlated with MSI in LGG, LUAD, PRAD, THCA, READ, and BLCA, and negatively in COAD, COADREAD, KIPAN, and DLBC (Fig. 3D).

Fig. 3

Correlation between ZNF83 expression and tumor stemness and heterogeneity across pan-cancer types. (A) Correlation between ZNF83 expression and tumor stemness based on DNA methylation (DNAss). (B) Correlation between ZNF83 expression and tumor stemness based on gene expression (RNAss). (C) Correlation between ZNF83 expression and tumor mutational burden (TMB). (D) Correlation between ZNF83 expression and microsatellite instability (MSI)

3.4 Correlation between ZNF83 expression with immune microenvironment in pan-cancer

ZNF83 expression levels were positively correlated with most immune checkpoint-related genes in THYM, DLBC, and UVM, but negatively in ESCA, KIPAN, and KIRC (Fig. 4A). It was positively associated with immunomodulatory genes in ACC, BLCA, and SARC, and negatively in GBM, GBMLGG, and LGG (Fig. 4B). Moreover, ZNF83 expression was significantly negatively correlated with stromal, immune, and ESTIMATE scores in several tumor types (ACC, BLCA, GBM, KIRP, LAML, PCPG, SARC, THCA, UCEC), and positively in ESCA and KIRC (Fig. 4C).

Fig. 4

Immunological relevance of ZNF83 expression across pan-cancer types. (A) Correlation between ZNF83 expression and immune regulatory genes. (B) Correlation between ZNF83 expression and immune checkpoint genes. (C) Correlation between ZNF83 expression and three microenvironment scores (immune score, stromal score, and ESTIMATE score) calculated using the ESTIMATE algorithm. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

Using the ssGSEA algorithm, ZNF83 expression showed a positive correlation with T helper and T central memory (Tcm) cells in most tumors, while dendritic cells (DCs), macrophages, Th1, and Th17 cells showed negative correlations (Fig. 5A). The CIBERSORT algorithm revealed positive correlations between ZNF83 expression and CD4 memory resting T cells and resting mast cells across most tumor types (Fig. 5B).

Fig. 5

Correlation between ZNF83 expression and immune cell infiltration across pan-cancer types. (A) Correlation between ZNF83 expression and immune cell populations estimated using the ssGSEA algorithm. (B) Correlation between ZNF83 expression and immune cell populations estimated using the CIBERSORT algorithm. *P < 0.05

3.5 Validation of expression and prognosis in renal cancer cohorts

In the TCGA-KIRC cohort, ZNF83 expression was significantly higher in tumor tissues than in normal tissues (Fig. 6A). ROC curve analysis demonstrated a high diagnostic value for ZNF83 (AUC = 0.910) (Fig. 6B). Higher ZNF83 expression was associated with poorer overall survival (Fig. 6C). Subgroup analyses confirmed this trend in patients with N0 stage, T1–T3 stages, females, and stages I, II, and IV (Fig. 6D-G). Elevated ZNF83 also predicted worse outcomes in White patients, those with M0 and M1 stages, and those aged ≤ 60 (Fig. 6H-J). Prognostic models incorporating ZNF83 and clinical parameters showed strong predictive performance (Fig. 6K–L). In the validation cohort, ZNF83 protein expression was significantly higher in RCC tumor tissues (Fig. 6M–N; Table 1), and patients with high ZNF83 levels had significantly worse outcomes (P < 0.043) (Fig. 6O).

Fig. 6

Expression and prognostic significance of ZNF83 in renal cancer. (A) Differential expression of ZNF83 in the TCGA-KIRC cohort. (B) Diagnostic value of ZNF83 assessed by ROC curve analysis in the TCGA-KIRC cohort. (C) Overall survival analysis based on ZNF83 expression in the TCGA-KIRC cohort. (D) Overall survival analysis of ZNF83 in KIRC patients with N0 stage. (E) Overall survival analysis of ZNF83 in KIRC patients with T1, T2, and T3 stages. (F) Overall survival analysis of ZNF83 in female KIRC patients. (G) Overall survival analysis of ZNF83 in KIRC patients with clinical stages I, II, and IV. (H) Overall survival analysis of ZNF83 in KIRC patients of White race. (I) Overall survival analysis of ZNF83 in KIRC patients with M0 and M1 stages. (J) Overall survival analysis of ZNF83 in KIRC patients aged > 60 and ≤ 60 years. (K) Prognostic model constructed by integrating clinical features and ZNF83 expression. (L) Calibration curve of the predictive model performance. (M) Representative immunohistochemical staining of ZNF83 in renal cancer and adjacent normal tissues. (N) Differential expression analysis of ZNF83 across renal cancer cohorts. (O) Prognostic analysis of ZNF83 in renal cancer cohorts. ***P < 0.001

Table 1 Distribution of clinical features in 26 patients with renal cell carcinoma3.6 Validation of expression and prognosis in lung cancer cohorts

Analysis of the TCGA + GTEx dataset revealed that ZNF83 expression was significantly lower in LUAD tissues compared to normal tissues (Fig. 7A). However, ZNF83 expression did not significantly correlate with OS, DSS, or PFI in LUAD (Fig. 7B–D). In the validation cohort, ZNF83 protein expression was also significantly lower in LUAD tissues (Fig. 7E–F; Table 2). Prognostic analysis showed no significant association between ZNF83 expression and patient outcomes in LUAD (Fig. 7G).

Fig. 7

Expression and prognostic significance of ZNF83 in lung cancer. (A) Differential expression analysis of ZNF83 in the LUAD cohort based on TCGA and GTEx databases. (B–D) Prognostic analysis of ZNF83 expression in the TCGA-LUAD cohort. (E) Representative immunohistochemical staining of ZNF83 in lung adenocarcinoma and adjacent normal tissues (Scale: 200 μm). (F) Differential expression analysis of ZNF83 in lung adenocarcinoma cohorts. (O) Prognostic analysis of ZNF83 in lung adenocarcinoma cohorts. ***P < 0.001; ****P < 0.0001

Table 2 Distribution of clinical features in 92 patients with lung adenocarcinoma