Introduction
Prostate cancer (PCa) is the primary malignancy among men, responsible for 14.1% of new cases, and ranks 5th in terms of cancer-related deaths (with a mortality rate of 6.8%) worldwide [
1]. In China, prostate cancer accounts for 8.16% of new cancers in men (ranking as the 6th most common malignancy), with a mortality rate of 13.61 (ranks 7th in terms of cancer-related deaths) [
2]. The leading cause of death is metastasis in PCa. The outcome of metastatic PCa is inferior, as only 30% of patients could survive for 5 years [
3]. Gleason score and tumor, node, metastasis (TNM) stage are prognostic factors. Unfortunately, there may be vast differences in clinical outcomes between patients with the same Gleason score, making it essential to identify critical factors influencing prognosis.
It has been found that tumor purity is significantly related to the clinical characteristics and genetic features of patients with tumors. It is possible to develop systematic biases in recurrence risk, tumor genotyping and efficacy prediction by ignoring the influence of tumor purity [
4]. A low-purity tumor sample has a higher mutational burden and more immune cells. Immune cells’ inflammatory response may result in tumor cells mutating more rapidly, which may improve the effectiveness of immunotherapy [
4]. Previous studies have indicated that tumor purity is one way of determining the efficacy of immunotherapy. Gastric and colon cancer prognosis has been demonstrated to positively correlate with tumor purity [
5,
6]. However, few studies have considered the influence of tumor purity in the prognosis of PCa.
A tumor purity calculation was performed using the ESTIMATE algorithm in this study [
7]. The CIBERSORT algorithm was used to verify further whether low- and high-purity tumors had significantly different immune cell infiltration levels. After that, the tumor purity co-expression network was constructed using weighted gene co-expression network analysis (WGCNA) [
8]. The co-expression modules contained genes that were most related to tumor purity. Gene signatures associated with distant metastasis-free survival (DMFS) of prostate cancer were identified using the least absolute shrinkage and selection operator (LASSO)–COX regression analysis. Then, tumor purity score (TPS) was constructed. Kaplan–Meier and receiver operating characteristics curve (ROC) analyses indicated that PCa patients with higher TPS had worse prognoses. A nomogram was created using TPS and clinical parameters. In addition, all hub genes underwent Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Disease Ontology (DO) enrichment analysis. Our study has revealed a relationship between tumor purity and immune cell infiltration in PCa and built a robust predictive model for clinical application.
Discussion
Recently, with the development of precision therapy and immunotherapy for malignant tumors, an important role is played by the immune microenvironment in tumor metastasis, treatment response, and prognosis. In addition, tumor purity can reflect unique characteristics of the tumor microenvironment (TME) [
4]. Meanwhile, the high morbidity and mortality of PCa make it a global public health problem [
1,
2]. Therefore, our study focused on the tumor purity of PCa.
A tumor purity calculation was performed first in this study. By the median value of tumor purity, we divided PCa into low and high groups. We screened out differential genes and obtained key genes with the highest relationship with tumor purity by the WGCNA. The ESTIMATE R package, ssGSEA algorithm and CIBERSORT were performed to uncover TME landscapes of different tumor purity subgroups in PCa. To establish a TPS model relating to DMFS in prostate cancer, LASSO-COX regression was used. Using this model, DMFS of PCa can be predicted independently. An excellent accuracy nomogram that can predict three- and five-year DMFS for PCa patients has been developed and validated.
There was a substantial correlation between tumor purity and immune cell infiltration in prostate cancers, as well as clinical features. With an increase in Gleason score, T stage or M stage, tumor purity was decreased significantly. These results indicate that high tumor purity is related to a favorable outcome of PCa. The tumors with lower purity have a higher degree of malignancy and a worse prognosis. This is consistent with previous findings in gastric cancer [
5], glioma [
23] or colon cancer [
6]. Furthermore, our conclusions are in general agreement with previous results showing that a low Gleason score is a good prognostic factor [
24].
In recent years, computational tools have emerged in an endless stream, and tumor purity estimation methods based on different genetic data types have been proposed. The ESTIMATE algorithm and CIBERSORT algorithm adopted in this study can be used for RNA sequencing analysis [
7,
13]. Our experiments are verified mutually in these two algorithms. The results generated by these two algorithms are in good consistency. In our previous studies, bioinformatics could screen genes and construct features to predict the prognosis of prostate cancer, as well as explore molecular mechanisms of prostate cancer development [
25‐
27]. The radiomics-based survival analysis performed well in predicting the prognosis for PCa patients, with the potential to optimize treatment protocols [
28]. Radiomics combined with bioinformatics can help explore immunotherapy shortly.
The infiltration level of B cells in PCa is relatively higher compared with normal prostate tissue, suggesting that B cells can serve as a therapeutic target [
29]. There are dispersed T-cell populations in both myeloid and blastic prostate cancers [
30]. In metastatic castration-resistant PCa patients, Treg cell aggregation presents in the peripheral blood [
31]. In the process of prostate carcinogenesis, M1 macrophages transform into the M2 phenotype, which promotes an immunosuppressive TME and thus tumor growth and metastasis [
32]. It was proposed the higher the (M1 + M2)/M0 ratio, the worse the prognosis [
33]. Consistently, in the low tumor purity group of our study, M0 cells were significantly decreased and Treg cells were increased considerably, who had a worse prognosis.
Two genes (FCER1G and OLR1) related to TPS were significantly associated with PCa progression and metastasis, as proposed in previous studies. For example, GLRX, SNAP23 and OLR1 are overexpressed, which is related to aggressive metastasis in breast cancer and prostate cancer tissues [
34]. FCER1G is associated with TME in PCa, which may help to predict the prognosis of PCa [
35]. It has been reported that metastasis-associated gene FCER1G was abundantly expressed in circulating tumor cells (CTCs) of a PCa patient who was sensitive to docetaxel, a chemotherapy agent [
36]. There is a significant increase of SLAMF8 in PCa tissues, both at the RNA and protein levels. It is an important metastatic marker worthy of further study.
Significant variations in the immune microenvironment among various tumor purities were observed through the process of enrichment analysis. Low purity tumor exhibits increased infiltration of immune cells and a negative prognosis. Meanwhile, we discovered that hub genes were primarily concentrated in primary immunodeficiency disorder. Accordingly, metastasis of prostate cancer may be linked to immunosuppressive conditions caused by immune microenvironment disorders. Research findings in different types of malignancies strongly support this new perspective, such as non–small cell lung cancer (NSCLC) and melanoma patients with liver metastasis [
37], renal cell carcinoma [
38], lung cancer [
39].
This study has several limitations. Firstly, tumor purity was calculated based on only one set of TCGA transcriptome data. Our finding needs to be validated using more data sets and multiple algorithms. Secondly, this is a retrospective study. A prospective evaluation would enhance the robustness of our findings.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.