Because of the different molecular mechanisms and the low proportion of PRCC in RCC, patients with PRCC have been excluded from large clinical trials of targeted drugs, such as sorafenib and sunitinib, and research on PRCC is always less studied than clear cell RCC and progresses slowly [21]. Although some patients with PRCC can be diagnosed by ultrasonography and receive surgery at an early stage, a significant number of advanced patients die due to postoperative recurrence, metastasis, and resistance to chemotherapeutic drugs, underscoring the importance of exploring the molecular mechanisms and prognostic factors of PRCC [22].
Recently, studies have been reported involving gene signatures for prognostic prediction in human cancers [23–25]. Immune cells were found in human solid tumors, and the immune pattern of the tumor microenvironment is a major predictor of patient survival in most primary tumors [26]. Previous studies have shown a major prognostic value of the immune pattern (CD8+/DC-LAMP + cell densities) in colorectal carcinoma and RCC, reproducible from primary to metastatic tumors [27]. The immune checkpoint molecules on expanded T cells in patients with advanced RCC were higher than those on unexpanded T cells before surgery [28]. Considering the importance of the immune environment in the progression of cancer, it is essential to identify immune-related biomarkers to evaluate the prognosis of patients with PRCC, which may also play a significant role in immunotherapy. The objective of our study was to recognize the immune-related genes (IRGs) associated with prognosis and to construct a dependable model to predict the overall survival (OS) of patients with PRCC.
First, we obtained 371 DEIRGs, including 232 upregulated and 139 downregulated genes, based on 289 PRCC tissues and 32 normal kidney tissues that were downloaded from the TCGA database. We then performed Cox and Lasso regression analyses to assess the relationship of these DEIRGs with the prognosis of patients with PRCC, and 4 PDEIRGs of interest (IL13RA2, CCL19, BIRC5, and INHBE) were ultimately determined. All 4 PDEIRGs have been reported to be involved in the immune microenvironment and inflammatory response [29–32]. IL13RA2 could mediate resistance to sunitinib in certain populations of ccRCC by avoiding sunitinib-induced apoptosis [33]. BIRC5 was overexpressed in patients with breast cancer and was responsible for shorter relapse-free survival, worse overall survival, reduced distant metastasis-free survival, and increased risk of metastatic relapse events [34]. In addition, INHBE emerged as a candidate hepatokine associated unexpectedly with whole-body energy metabolism under obese insulin-resistant conditions, which could decrease fat utilization and increase fat mass [35]. Moreover, CCL19 has been regarded as one of the immune-related risk genes that can predict PRCC patient survival [18]. From this, a prognostic prediction model was constructed, and the risk score of patients was calculated.
Next, we examined the value of the risk score model for the prognostic prediction of patients by survival analysis. The results showed that patients in the high-risk group had significantly poorer OS outcomes than those in the low‐risk group, suggesting that the model was associated with the prognosis of patients with PRCC. We then further analyzed the reliability and stability of the model and validated it. Our results indicated that the model could accurately discriminate patients with different survival outcomes. Combining univariate and multivariate Cox regression analyses, the model was demonstrated to independently predict the prognosis of patients with PRCC. Thus, our model can be used to identify patients with PRCC at high risk for death and to carry out early interventions to improve the prognosis of patients in clinical work.
Previous studies have demonstrated that immune infiltration is an important determinant of the therapeutic response and prognosis of cancer [36, 37]. Li et al. found that higher enrichment of multiple immune/inflammatory cells, such as Th2 cells and macrophages, was associated with poor prognosis in breast cancer [38]. G. Drake et al. reported that the high infiltration of CD8 + T cells in RCC is related to worse outcome [39]. Therefore, we also analyzed the relationship between the risk score and immune cell infiltration and found that the risk score correlated positively with the infiltration of B cells and CD4 + T cells. The tumor mutation burden (TMB) might predict clinical response and be associated with survival in patients taking immune checkpoint inhibitors (ICIs) across a wide variety of cancer types [40, 41]. Thus, we speculated whether our model reflected TMB and found that the mutation burden was higher in the high-risk group than in the low-risk group. These results suggested that the model can be used to distinguish patients with different sensitivities to immunotherapy and to develop individualized treatment strategies.
Recently, the risk models according to PDEIRGs have attracted wide attention and revealed the tremendous potential in prognosis prediction of patients with cancer. Wang et al. constructed a prognostic risk model screening 15 PDEIRGs in PRCC and verified that the model could independently distinguish patients with different risks of death [18]. Wan et al. applied Cox and Lasso regression to identify 7 PDEIRGs for establishing a risk model for the prognostic stratification of patients with ccRCC and found that the model could predict immune cell infiltration, the mutation burden and the progression of ccRCC [17]. Zhang et al. established a prognostic prediction risk score model based on the expression profiles of 14 IRGs in lung adenocarcinoma that showed high prediction accuracy and stability in identifying immune features [42]. Our research differed from previous studies in several ways. First, there were fewer IRGs in our model than in the previous models, and we focused on only 4 IRG expression patterns in PRCC. Second, the IRGs in our model did not overlap with those in the previous models. Third, we used multiple algorithms (including univariate Cox, multivariate Cox and Lasso regression) to identify PDEIRGs for the model. Therefore, our study was more accurate and reliable than the others.
There are still some limitations in our study. First, all of the investigative data were completely acquired from public databases. Second, we evaluated the performance of the risk model by the full TCGA cohort lacking further validation, owing to limited patient numbers in the validation datasets. Third, the biological functions of 4 PDEIRGs in PRCC require further examination by a series of experiments.