NCAPG as a novel prognostic biomarker in numerous cancers: a meta-analysis and bioinformatics analysis

Literature search

Two authors independently searched four databases, namely, Pubmed, Embase, Web of Science, and the Cochrane Library for studies published before April 30, 2022. The following search terms were used: (“Neoplasms” OR “Carcinoma” OR “Prognosis” OR “Diagnosis” OR “Survival”) AND (“non-SMC condensin I complex subunit G” OR “NCAPG”).

Inclusion criteria

The following inclusion criteria were considered when screening the databases: (1) the original literature was in English; (2) cancers with abnormal NCAPG expression were investigated; (3) high and low NCAPG expression was delineated; and (4) HR and 95% CIs of the OS can be obtained or calculated from the survival curve.

Exclusion criteria

The exclusion criteria were as follows: (1) reviews, publication letters, retracted literature, and case reports; (2) insufficient data; (3) bioinformatics analysis; and (4) studies not relevant to NCAPG.

Data extraction

Data extraction was performed by two investigators for all included studies and submitted to a third researcher to resolve disagreements. The following data were extracted according to the inclusion criteria: first author, publication date, country of origin, cancer type, number of cases, follow-up time, measurement method of NCAPG expression, outcome measures, HR and 95% CIs for OS.

Quality assessment

The quality of the literature was evaluated using the Newcastle–Ottawa Scale. The evaluation was conducted independently by two investigators, and when disagreements arose, a third investigator participated in the discussion. The total score was 9 points, and a score of ≥6 points indicated high-quality research [15].

Validation of the bioinformatics database

The GEPIA2.0 database (http://gepia.cancer-pku.cn/index.html) is a platform for sequencing and expression data, that includes most tumors and normal tissue samples [16]. Moreover, this study used the “Expression DIY” module to explore the differences between NCAPG transcripts from cancer tissue samples and normal tissue samples. In addition, we downloaded tumor transcription samples from the TCGA database (https://portal.gdc.cancer.gov/) [17] and used R (survival and timeROC packages) to perform a cox regression and ROC analysis of the survival rate. Next, we used the Kaplan-Meier Plotter database (https://kmplot.com/analysis/) to analyze the effect of the NCAPG gene on the survival rate in different cancers for additional data supplementation [18]. To validate the prognostic tumor status of NCAPG, we utilized the PrognoScan database to verify the survival information of this gene in multiple cancer datasets. We used the UALCAN database (http://ualcan.path.uab.edu/index.html) [19] to validate clinical information on NCAPG expression in tumors and explore the methylation and phosphorylation of NCAPG in these tumors.

Molecular role and functional enrichment analysis of NCAPG

GeneMANIA (http://www.genemania.org) is an online tool for protein-protein interactions that helped in identifying genes with similar functions relative to that of the NCAPG gene in this study [20, 21]. We used the STRING database (https://string-db.org/) for GO and KEGG pathway analyses of NCAPG [22]. Finally, we used the multiMiR package (version 4.12) to identify competitive endogenous RNA of NCAPG and constructed a NCAPG network of competitive endogenous RNA interactions of target genes through Cytoscape software (version 3.8.2, https://cytoscape.org/) [23].

Data processing and statistical analysis

The K-M curves of the included studies were processed by Enguage Digitizer 11.3 software to obtain HR values and 95% CIs. Further data analysis was performed using Review Manager 5.3 software. Survival outcomes were calculated by logarithmic HR values and their standard errors. In addition, the correlation between NCAPG upregulation and clinicopathological parameters of cancers (age, gender, degree of differentiation, TNM stage, metastasis, vascular invasion) was assessed by calculating the ORs and 95% CIs. Cochran’s Q test and I² test assessed the heterogeneity to determine the effect model. A fixed model was used if the included studies had no significant heterogeneity (I² < 50%, P > 0.1), while a random model was used otherwise. A sensitivity analysis of the included studies was performed using STATA 12.0 software to assess the stability of the results. Publication bias was evaluated using Begg’s rank correlation and Egger’s linear regression. At P < 0.05, publication bias was observed. If publication bias was present, then the trim-and-fill method was used to further assess the stability of the pooled results.

Data availability statement

All relevant data is contained within the article: The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding authors.

Literature selection

A flow chart depicting the literature screening process is shown in Figure 1. A total of 340 articles were screened, and 180 duplicate articles were excluded by searching PubMed, EMBASE, the Cochrane Library, and Web of Science. After scrutinizing the titles and abstracts, 133 more studies were excluded. As eight studies did not record the hazard ratios (HRs) and/or Kaplan-Meier (K-M) curves for overall survival (OS), 11 articles were excluded from bioinformatics analysis. Finally, we included eight studies in the meta-analysis [5, 6, 8, 9, 12, 24–26], all of which were cohort studies, and all outcome measures were OS.

Figure 1. Flow chart of literature screening for this meta-analysis.

Study characteristics and quality assessment

The characteristics of the included studies are presented in Table 1. All the studies were conducted in China, and 1096 patients were recruited. The types of cancer included non-small cell lung cancer (NSCLC), glioma, breast cancer, gastric cancer, and hepatocellular carcinoma. All included studies assessed the association between OS and NCAPG expression, with follow-ups ranging from 80 to 140 months. All but one study reported clinicopathological parameters. All studies measured NCAPG expression via immunohistochemical staining, except for one study that performed RNA sequencing instead. Newcastle–Ottawa Scale scores were ≥6, indicating that the included studies were of moderate to high quality.

Correlation of NCAPG expression with OS and cancer type

Eight studies reported that NCAPG upregulation was associated with tumors. Therefore, a pooled analysis of the eight studies was performed. As shown in Figure 2, a random model was used because of the absence of obvious heterogeneity (I² = 56%, P = 0.02). The pooled results suggested that high NCAPG expression rates were correlated with worse OS in patients with different cancers (HR = 2.90, 95% confidence interval (CI) = 2.06–4.10, P < 0.00001).

Figure 2. Forest plot of the pooled OS for subgroup analysis.

Due to the existence of heterogeneity, a subgroup analysis was performed to further explore the impact of high NCAPG expression on survival in different types of cancer (Figure 2). Four studies [5, 9, 25, 26] investigated cancers of the respiratory system, two studies [8, 24] investigated cancers of the digestive system, and the remaining two studies investigated other types of cancers, namely, glioma [12] and breast cancer [6]. According to the forest plot, upregulation of NCAPG expression in cancer tissues was related to worse OS regardless of the group (respiratory cancer, HR = 2.27, 95% CI = 1.73–2.98, P < 0.00001; digestive cancer, HR = 2.12, 95% CI = 1.39–3.22, P = 0.0005; breast cancer, HR = 7.57, 95% CI = 3.13–18.30, P < 0.00001; glioma, HR = 9.34, 95% CI = 3.75–23.26, P < 0.00001).

Correlation of NCAPG expression with clinicopathological parameters

As all studies included in this meta-analysis reported clinicopathological parameters, we analyzed the high expression of NCAPG and these parameters. As shown in Figure 3, the upregulation of NCAPG was not significantly correlated with gender (male vs. female, odds ratio (OR) = 1.20, 95% CI = 0.92–1.56, P = 0.19, fixed model; Figure 3A), age (young vs. old, OR = 0.68, 95% CI = 0.44–1.06, P = 0.09, random model; Figure 3B), vascular invasion (yes vs. no, OR = 2.23, 95% CI = 0.81–6.11, P = 0.12, random model; Figure 3D), differentiation (well differentiated vs. poorly differentiated, OR = 1.20, 95% CI = 0.50–2.90, P = 0.68, fixed model; Figure 3E), TNM stage (III–IV vs. I–II, OR = 1.64, 95% CI = 0.92–2.92, P = 0.09, random model; Figure 3G), or T classification (T3+T4 vs. T1+T2, OR = 1.68, 95% CI = 0.83–3.40, P = 0.15, random model; Figure 3J), although it was correlated with distant metastasis (yes vs. no, OR = 5.65, 95% CI 2.60–12.26, P < 0.0001, fixed model; Figure 3C), lymph node metastasis (yes vs. no, OR = 2.08, 95% CI = 1.54–2.80, P < 0.00001, fixed model; Figure 3F), relapse (yes vs. no, OR = 2.62, 95% CI = 1.02–6.70, P = 0.04, random model; Figure 3H), and clinical stage (III–IV vs. I–II, OR = 2.23, 95% CI = 1.44–3.44, P = 0.0003, fixed model; Figure 3I). The results of database validation of the clinicopathological characteristics and NCAPG expression data (Supplementary Figure 1) are consistent with most of our previous meta-analysis results.

Figure 3. Forest plot of the relationship between high NCAPG expression and clinicopathological parameters. (A) gender, (B) age, (C) distant metastasis, (D) vascular invasion, (E) differentiation, (F) lymph node metastasis, (G) TNM stage, (H) relapse, (I) clinical stage, (J) T classification.

Sensitivity analysis and publication bias

To verify the robustness of the results, we performed a sensitivity analysis (Figure 4). Removal of one or more articles did not significantly affect the results, indicating that the results were relatively stable. Begg’s and Egger’s tests indicated publication bias for OS and distant metastasis (Table 2, Supplementary Figure 2). We analyzed the stability of the study further using the trim-and-fill method. For OS, the results indicated that the estimated number of missing studies was 0 and the adjusted HR was 2.90 (95% CI = 2.06–4.10, P < 0.001), which indicated that upregulation of NCAPG was associated with poorer OS, suggesting that our result is reliable. In addition, for distant metastases, two studies were estimated to be missing and the adjusted OR was 4.85 (95% CI = 2.48–9.58, P < 0.001), which indicated a reliable result.

Figure 4. Sensitivity analysis. (A) OS, (B) gender, (C) age, (D) distant metastasis, (E) vascular invasion, (F) differentiation, (G) lymph node metastasis, (H) TNM stage, (I) relapse, (J) clinical stage, (K) T classification.

Validation of NCAPG expression against public databases

We evaluated the NCAPG expression levels and performed a survival analysis in various cancers using the GEPIA2.0 database to validate our results. The findings showed that, compared with normal tissues, the expression of NCAPG was significantly upregulated in tumors, including BRCA, GBM, LIHC, LUAD, LUSC, and STAD (Figure 5A). We next performed a univariate Cox survival analysis on the above tumors, and it showed that the OS of LIHC and LUAD was related to NCAPG (Figure 5B). The progression-free survival (PFS) of BRCA, LIHC, LUAD, and STAD was related to NCAPG (Figure 5C). The relapse-free survival (RFS) of LIHC, LUSC, and STAD was related to NCAPG (Figure 5D), and the disease-specific survival (DSS) of LIHC and LUAD was related to NCAPG (Figure 5E). We then performed receiver operating characteristic (ROC) survival analysis of the data on the above tumors (Supplementary Figure 3), and the area under the curve of GBM, LIHC, and STAD was greater than 0.7. Therefore, we concluded that NCAPG may be a good prognostic indicator of various cancers. In addition, we analyzed the survival prognosis of the above tumors using the Kaplan-Meier plotter database based on the median cutoff value of NCAPG expression (including probes 218663_at and 218663_s_at) in cancer. The results showed that NCAPG expression can be used to predict OS in breast cancer (P < 0.01; Figure 6A), RFS (P < 0.01; Figure 6B), PPS (P < 0.01; Figure 6C), DMFS (P < 0.01; Figure 6D), liver cancer, PFS, PFS, and DSS (P < 0.01; Figure 6E), lung cancer (P < 0.01; Figure 6F), FP (P < 0.01; Figure 6G), and PPS (P < 0.01; Figure 6H), gastric cancer (P < 0.01; Figure 6I), FP (P < 0.01; Figure 6J,), and PPS (P < 0.01; Figure 6K). Moreover, the findings were validated against multiple cancer datasets using the PrognoScan database. We collected 29 datasets on breast and lung cancers. As shown in Table 3, NCAPG expression in these tumors significantly affected prognosis-related indicators, such as OS, DSS, RFS, and PFS (P < 0.05).

Figure 5. (A) Expression levels of NCAPG in cancer tissues and normal tissues in GEPIA2. From left to right are gastric cancer (STAD), lung cancer (LUAD), liver cancer (LIHC), glioma (GBM) and breast cancer (BRCA). The red box represents the expression level of NCAPG in cancer tissues; the gray box represents the expression level of NCAPG in normal tissues, the screening criteria were log2FC|>1 and P < 0.01, (B) OS of BRCA, GBM, LIHC, LUAD, LUSC and STAD, (C) PFS of BRCA, GBM, LIHC, LUAD, LUSC and STAD, (D) DFS of BRCA, LIHC, LUAD, LUSC and STAD, (E) BRCA, GBM, LIHC, LUAD, LUSC and STAD's DSS.

Figure 6. (A) OS of NCAPG (218663-s-at) in BRCA (n = 1879), OS of NCAPG (218663-at) in BRCA (n = 1879), (B) RFS of NCAPG (218663-s-at) in BRCA (n = 4929), RFS of NCAPG (218663-at) in BRCA (n = 4929), (C) PPS of NCAPG (218663-s-at) in BRCA (n = 458), PPS of NCAPG (218663-at) in BRCA (n = 458)(D) DMPS of NCAPG (218663-s-at) in BRCA (n = 2765), DMPS of NCAPG (218663-at) in BRCA (n = 2765), (E) OS (n = 364), PFS (n = 316), PFS (n = 370)and DSS (n = 362)of NCAPG in LIHC, (F) OS of NCAPG (218663-s-at) in lung cancer (n = 1925), OS of NCAPG (218663-at) in lung cancer (n = 1925), (G) FP of NCAPG (218663-s-at) in lung cancer (n = 982), FP of NCAPG (218663-at) in lung cancer (n = 982), (H) PPS of NCAPG (218663-s-at) in lung cancer (n = 344), PPS of NCAPG (218663-at) in lung cancer (n = 344), (I) OS of NCAPG (218663-s-at) in STAD (n = 592), OS of NCAPG (218663-at) in STAD (n = 592), (J) FP of NCAPG (218663-s-at) in STAD (n = 358), FP of NCAPG (218663-at) in STAD (n = 358), (K) PPS of NCAPG (218663-s-at) in STAD (n = 221), PPS of NCAPG (218663-at) in STAD (n = 221).

Molecular role and functional enrichment analysis results of NCAPG

We used the GeneMANIA database for the protein-molecular interaction analysis of NCAPG and its related molecules, such as NCAPG2, NCAPH, and SMC4 (Figure 7A). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional and pathway enrichment analyses of NCAPG were performed using the STRING database. The most abundant GO terms were nuclear division, cell division, and cell cycle process (Table 4). In addition, the KEGG pathway analysis confirmed that these co-expressed genes were significantly involved in the p53 signaling pathway, cell cycle, and cellular senescence (Table 4). These results indicate that NCAPG is involved in the biological pathways of cancer. Additionally, we used the multiMiR package to identify NCAPG-related miRNAs and lncRNAs (the screening criterion for miRNAs was a predicted cutoff of 500,000, and the screening criteria for lncRNAs were lnc_mi$pancancerNum>10 and lnc_mi$clipExpNum>4) that may interact with NCAPG (Figure 7B). We identified 20 miRNAs and 13 lncRNAs, which could provide direction for future experimental designs. DNA methylation directly affects the occurrence and progression of cancers. We used the UALCAN database to investigate the DNA methylation of NCAPG. Our results showed that NCAPG methylation levels were significantly reduced in BRCA, GBM, LIHC, LUAD, and LUSC tissues compared to normal tissues (Figure 8A–8F), which may explain the difference in NCAPG expression between BRCA, GBM, LIHC, LUAD, and LUSC tissues and normal tissues. Post-translational modification is a key molecular mechanism underlying NCAPG activation. Therefore, we examined the changes in NCAPG phosphorylation levels between tumor tissues and normal tissues. The Clinical Proteomic Tumor Analysis Consortium database includes four cancers: BRCA, GBM, LIHC, and LUAD. Compared with the normal samples, the phosphorylation levels at 674/973/975/984 of NCAPG were higher in BRCA, GBM, LIHC, and LUAD, respectively (Figure 8G). The specific results are shown in Figure 8H–8M. Since the p53 signaling pathway is highly enriched, we also explored the phosphorylation of NCAPG in this pathway, and the results are shown in Figure 8N–8R.

Figure 7. Network analysis between NCAPG and target genes (A) PPI network for KIF23 was constructed in Gene MANIA, Different colors of the network edge indicate the bioinformatics methods applied: physical interaction, co-expression, predicted, co-localization, pathway, genetic interaction and shared protein domains. Abbreviation: PPI: protein–protein interaction. (B) The relationship between NCAPG and non-coding RNA, the red square represents the target gene NCAPG, the blue oval represents miRNA, and the yellow triangle represents lncRNA.

Figure 8. DNA methylation features of NCAPG in BRCA (A), GBM (B), LIHC (C), LUAD (D), LUSC (E) and STAD (F). Phosphorylation of NCAPG in several selected cancers according to the CPTAC database. (G) The schematic diagram and phosphorylation sites of the NCAPG protein are shown. The phosphorylation of NCAPG at S674, S973, S975 and 984 in BRCA (H–J), S674 in GBM (K), S674 in LIHC (L), S674 in LUAD (M). The P53 pathway phosphorylation of NCAPG at S674, S973, S975 and 984 in BRCA (N–P), S674 in GBM (Q), S674 in LUAD (R), from the UALCAN database. ^*p < 0:05, ^**p < 0:01, and ^***p < 0:001, Abbreviation: ns: No statistical significance.