Downregulation of PPA2 expression correlates with poor prognosis of kidney renal clear cell carcinoma

Tissue specimens

A KIRC tissue microarray (HKidE180Su02) comprising 150 KIRC tissues and 30 normal renal specimens was purchased from Shanghai Outdo Biotech Co., Ltd (http://www.superchip.com.cn/), which was used to measure the protein expression level of PPA2 via immunohistochemistry (IHC). Another cohort of 10 paraffin-embedded KIRC tissues and paired normal renal tissues from patients at Meizhou People’s Hospital was used for external validation. Ethical approval was granted by the Medical Ethics Committee of Meizhou People’s Hospital (2020-CY-06) and written informed consent was obtained from all patients.

Immunohistochemical staining and scoring

All specimens were confirmed by two independent pathologists after hematoxylin and eosin staining. The samples were first deparaffinized with dimethylbenzene and rehydrated in an ethanol-water gradient. Antigen retrieval was performed by microwave. The samples were then treated with 3% hydrogen peroxide to inhibit the activity of endogenous peroxidases, and subsequently blocked with 5% bovine serum albumin for 1 h. The samples were incubated with anti-PPA2 primary antibody (diluted 1:200, ab180859, Abcam) overnight at 4 °C. Finally, the samples were stained with 3,3′-diaminobenzidine solution for visualization after incubation with horseradish peroxidase-conjugated secondary antibody (diluted 1:10000, ab205718, Abcam) (Sun et al., 2017).

The level of PPA2 expression was evaluated based on the degree of staining and the percentage of positively stained cells. The staining degree was scored as 0 (no staining), 1 (light yellow), 2 (dark yellow), or 3 (brown). The percentage of positively stained cells was scored as 0 (<1%), 1 (1%–25%), 2 (26%–50%), 3 (51%–75%), and 4 (76%–100%) (Niu et al., 2018). We multiplied the two above scores to obtain a final IHC score ranging from 0 to 12.

Construction and evaluation of a PPA2-based predictive nomogram

We used the identified independent prognostic factors to construct a predictive nomogram. The concordance index (C-index) and calibration plots were used to evaluate the constructive nomogram, as previously reported (Liu et al., 2019).

Gene Set Enrichment Analysis (GSEA)

GSEA was conducted to explore the potential molecular mechanisms linking PPA2 expression to KIRC prognosis (Subramanian et al., 2005). We first divided the patients into two groups according to the median PPA2 expression level. We then identified the differentially expressed genes between the low and high expression groups and performed GSEA. The gene set permutations were conducted 1000 times for each analysis. A false discovery rate of < 0.25 and P value < 0.05 were considered to indicate statistical significance (Liu et al., 2020).

Calculation of stromal and immune scores

We calculated the stromal scores and immune scores by applying the “Estimation of STromal and Immune cells in MAlignant Tumors using Expression data (ESTIMATE)” algorithm (https://bioinformatics.mdanderson.org/estimate/) to infer the fraction of stromal and immune cells and predict tumor purity for each KIRC sample in The Cancer Genome Atlas (TCGA) (Yoshihara et al., 2013). Then we downloaded the gene expression profile of TCGA-KIRC dataset from UCSC Xena platform (https://xena.ucsc.edu/) (Goldman et al., 2020). We matched the samples’ gene expression data and ESTIMATE algorithm results. Next, we divided the KIRC samples into two groups according to the median value of stromal score, immune score, and ESTIMATE score, respectively. Finally, we explored the association between PPA2 expression and algorithm results.

Statistical analysis

All data processing and statistical analyses were performed using R software V3.6.1 (R Core Team, 2019) and SPSS V25 (SPSS Inc.) (Chan, 2018; Liang, Fu & Wang, 2019). Differential expression analysis was performed by the “limma” R package and the survival analysis was performed using the “survival” and “survminer” R packages (Ritchie et al., 2015; Wang et al., 2020). Student’s t-test or analysis of variance was used to evaluate the differences in PPA2 expression among the different subgroups. The paired t-test was used to explore the difference in PPA2 expression between the ten paraffin-embedded KIRC tissues and paired normal renal tissues. The chi-square test or Fisher’s exact test was used to investigate the relationship between PPA2 protein expression and the clinicopathological features of KIRC. The Kaplan–Meier method and log-rank test were used for the survival analyses. The Cox regression model was used for univariate and multivariate survival analyses. For all analyses, a P-value < 0.05 was regarded as statistically significant.

Clinicopathological characteristics

A total of 150 KIRC patients whose specimens were present in the tissue microarray were enrolled in this retrospective study. All patients underwent nephrectomy or partial nephrectomy. The median age was 57 years (range, 24–83 years). Sixty-three patients (42%) were aged above 60 years. Specimens from 103 patients (68.7%) were histologically graded as relatively well-differentiated (G1/2), whereas those from the remaining 47 patients (31.3%) were relatively poorly differentiated (G3/4). Most of the patients (138 cases, 92%) had localized tumors of American Joint Committee on Cancer (AJCC) stage I/II, whereas the other 12 patients (8%) had AJCC stage III/IV tumors. The detailed clinicopathological information of the enrolled patients is summarized in Table 1.

PPA2 expression profiles and their relationship with KIRC clinicopathological parameters

Immunohistochemical staining on the tissue microarray showed that PPA2 was primarily expressed in the cytoplasm of KIRC cells, and that the protein expression level of PPA2 in tumor tissues was significantly lower than that in normal renal tissues (Fig. 1A). To corroborate our results, we first validated the differences in PPA2 expression at the gene level using data from TCGA on the Gene Expression Profiling Interactive Analysis (GEPIA2) platform (Tang et al., 2019) (http://gepia2.cancer-pku.cn). This analysis showed that PPA2 was expressed at lower levels not only in KIRC, but also in two other common kidney cancer types (Fig. 1B). We then used another cohort containing ten paraffin-embedded KIRC tissues and paired normal renal tissues to validate these results at the protein level. This analysis confirmed that PPA2 expression was lower in the KIRC tissues (Fig. 1A).

The 150 patients enrolled in our study were divided into high and low PPA2 expression groups based on their median expression values. Details of the association between PPA2 expression levels and clinicopathological features of KIRC are shown in Table 1. PPA2 expression was lower in tumors with histologic grade G3/4 than in those with grade G1/2 (p < 0.014). However, there was no significant correlation between PPA2 expression and patient AJCC stage, possibly due to the small sample size of patients with stage III/IV KIRC. Considering the importance of the AJCC stage for outcome prediction, we further explored the association between PPA2 expression and AJCC stage using data from the TCGA KIRC cohort (including 523 patients with KIRC) on the GEPIA2 platform. We found that PPA2 expression was also significantly different among patients with different AJCC stages (p < 0.001, Fig. 1C).

Prognostic potential of PPA2 in KIRC

Kaplan–Meier survival analysis revealed that patients with low PPA2 expression exhibited worse overall survival (OS) than those with high PPA2 expression (p = 0.004, Fig. 2A). To confirm these results, we further explored the prognostic potential of PPA2 in a pan-cancer dataset (including 33 cancer types and 9,486 patients in total) from TCGA using the GEPIA2 platform. Interestingly, PPA2 also displayed prognostic value in several other cancers, including breast invasive carcinoma, colon adenocarcinoma, kidney renal papillary cell carcinoma, brain lower grade glioma, and uveal melanoma (Fig. 2B, Table 2).

Independent prognostic value of PPA2 in KIRC

Cox survival analysis revealed a significant survival difference between the high and low PPA2 expression groups in both univariate (hazard ratio [HR], 0.287; 95% confidence interval CI [0.116–0.708]; p = 0.007) and multivariate (HR, 0.126; 95% CI [0.039–0.403]; p < 0.001) analyses (Figs. 3A, 3B). PPA2 expression, histologic grade, and AJCC stage showed independent prognostic values in KIRC. Moreover, we performed further stratified analysis based on independent prognostic factors, histologic grade and AJCC stage. This analysis showed that patients with low PPA2 protein expression had poorer OS than those with high PPA2 protein expression when bearing tumors of histologic grade 1/2 (p = 0.007), grade 3/4 (p = 0.001), or AJCC stage I/II (p = 0.027), but not stage III/IV (p = 0.273) (Figs. 3C, 3D).

Construction and evaluation of a predictive nomogram

Nomograms are widely used to predict the outcomes of cancer patients (Iasonos et al., 2008). The C-index and calibration plots are often used to investigate the discrimination ability and calibration of nomograms (Fenlon et al., 2018; Longato, Vettoretti & Di Camillo, 2020). In this study, patient age, histologic grade, AJCC stage, and PPA2 expression were used to construct a nomogram to predict the 1-year, 3-year and 5-year OS probability of KIRC patients, which could provide helpful information for individualized clinical evaluation and therapeutic strategies (Fig. 4A). The C-index of our nomogram was 0.795, which was significantly higher than that of any of the other independent prognostic factors (age, 0.582; histologic grade, 0.695; AJCC stage, 0.623; and PPA2 expression, 0.569). The calibration plot indicated that the 1-year, 3-year, and 5-year OS probabilities predicted by our nomogram were consistent with the actual outcomes (Fig. 4B). Furthermore, the area under the curve of our nomogram for the 1-year, 3-year, and 5-year OS of KIRC patients was 0.910, 0.847, and 0.819, respectively, displaying greater values than those found for any prognostic factor, including age, histologic grade, AJCC stage, and PPA2 expression (Fig. 4C). Taken together, these results suggest that our nomogram could accurately predict the OS of KIRC patients with KIRC.

GSEA

To understand the potential molecular mechanism underlying the prognostic value of PPA2 in KIRC, we conducted GSEA using transcriptome data from the TCGA KIRC cohort, downloaded from the Xena platform of the University of California, Santa Cruz (Goldman et al., 2020) (https://xena.ucsc.edu/). The results of GSEA suggested that the epithelial-mesenchymal transition (EMT) and allograft rejection pathways were possibly activated in patients with low PPA2 expression, whereas the pathways of oxidative phosphorylation, protein secretion, bile acid metabolism and fatty acid metabolism were possibly suppressed (Fig. 5).

Differential PPA2 expression between different stromal/immune score groups

The ESTIMATE algorithm is helpful for us to understand the landscape of stromal and immune cells in the tumor microenvironment. The higher stromal/immune scores mean more stromal/immune cells in tumor microenvironment, whereas higher ESTIMATE scores mean lower tumor purity (less tumor cells). Our study showed that PPA2 was significantly higher expressed in samples with lower stromal scores or ESTIMATE scores (Fig. 6). It suggested that PPA2 might be expressed highest in tumor cells.