Prostate cancer was the most frequently diagnosed type of cancer among men in 112/185 of the countries in the world in 2020 (1). Active surveillance, endocrine therapy, radiotherapy and surgery are all treatments for prostate cancer (2). However, radiotherapy, endocrine therapy and surgery are not the primary treatment options, as they have serious side-effects and can reduce the quality of life (QOL) of patients. Furthermore, patients with low-risk, localized prostate cancer often receive unnecessary radiotherapy, endocrine therapy and surgery. To overcome this problem leading to reduced QOL, recently, active surveillance has been recommended for primary treatment (3). Active surveillance is the monitoring of low-risk localized prostate cancer progression by prostate-specific antigen (PSA) testing and biopsy. Patients treated with active surveillance are monitored until the prostate cancer grows to a point where it can receive active treatments such as endocrine therapy, radiotherapy and surgery. Thus, patients can have a normal life without their QOL being reduced.

The PSA test is an established diagnostic method used for prostate cancer detection and monitoring of cancer recurrence following active treatment (4). However, the PSA test alone is not sufficient for prediction of prognosis, as biopsy-based biomarkers and surgical specimens remain important. Furthermore, as active treatments, such as endocrine and radiation therapy can reduce the patients’ QOL, it is important to determine the appropriate treatment for each patient. Therefore, in order to accurately predict prostate cancer prognosis, the identification of novel biomarkers is required.

Solute carrier family 20 member 1 (SLC20A1) is a gene that encodes a sodium/phosphate symporter (5,6) and has been proposed to be a viable prognostic marker for breast, tongue and pancreatic cancer (7-10). Previous data showed that high SLC20A1 expression also indicates poor clinical outcomes for patients with breast cancer subtypes which are treated with endocrine therapy and radiotherapy (11,12). In addition, high SLC20A1 expression can predict the late recurrence of estrogen receptor (ER) positive-breast cancer treated with endocrine therapy or radiotherapy (11,12). A series of loss-of-function studies have revealed that SLC20A1 is involved in the promotion of the cell cycle progression of HeLa cells, in the viability of HeLa, HepG2, MCF-7, MDA-MB231 and MDA-MB468 cells and in the suppression of tumor necrosis factor-induced apoptosis of HeLa cells (8,13,14). Furthermore, SLC20A1 is involved in the viability and tumor-sphere formation of by aldehyde dehydrogenase 1 (ALDH1)-positive breast cancer stem cells (8). However, the association between SLC20A1 gene expression and the clinical outcomes of endocrine therapy and radiotherapy in prostate cancer remains to be determined.

Analysis of The Cancer Genome Atlas (TCGA) prostate cancer dataset. TCGA prostate cancer dataset (15,16) was downloaded from the University of California, Santa Cruz (UCSC) Xena cancer genomics visualization tools (https://xena.ucsc.edu) (17) on November 7, 2022. This dataset contained mRNA expression data from 52 samples of normal solid tissue and 497 primary prostate tumor samples. The mRNA expression of SLC20A1 was present in both normal and cancer tissues, and was converted to a z-score, and box and bee swarm plots were made.

Analysis of the stand Up to Cancer (SU2C)-Prostate Cancer Foundation (PCF) Dream Team. The SU2C-PCF Dream Team prostate cancer dataset (18) was downloaded from cBioPortal (https://www.cbioportal.org/) (19,20) on October 26, 2022. This dataset contained a total of 208 mRNA expression data from 444 prostate cancer samples. Of these, 71 samples had OS data along with mRNA expression data; the clinicopathological data of these patients are summarized in Table I. The mRNA expression of SLC20A1 was presented using the log₂ median-centered ratio for cancer tissues.

Analysis of TCGA, PanCancer Atlas dataset. The PanCancer Atlas dataset (n=184) (15,16) was downloaded from cBioPortal (https://www.cbioportal.org/) (19,20) on October 26, 2022. The clinicopathological data from these patients are also summarized in Table I. The PanCancer Atlas dataset contained data on both gene alterations (n=494) and mRNA expression levels in primary prostate cancer samples (n=493), and overall survival (OS), disease-specific survival (DSS), disease-free survival (DFS) and progression-free survival (PFS) data according to mRNA expression levels. Patients treated with endocrine therapy were categorized as patients whose treatment type was endocrine therapy and patients without endocrine therapy were categorized as patients whose treatment type was chemotherapy or radiation therapy; patients were categorized as being treated with radiotherapy or not.

Statistical analysis. SLC20A1 mRNA expression was compared between normal solid tissue and primary tumor tissue with box plot and bee swarms using the Mann–Whitney U-test; a two-sided value of p<0.05 was considered to indicate a statistically significant difference. SLC20A1 mRNA expression was compared among tissue sites with box plots using the Kruskal–Wallis test with the Steel-Dwass test; a value of p<0.05 was considered to indicate a statistically significant difference. Cut-off values for dividing the patients into groups with high and low SLC20A1 expression were determined by Youden’s index using receiver operating characteristic curves (Table II). The Kaplan–Meier method was used to analyze the survival curves of OS, DSS, DFS and PFS. For the survival curves, the log-rank (Cochran–Mantel–Haenszel) tests were used to calculate p-values. In order to evaluate the influence of SLC20A1 on prognosis, Cox regression analysis was used to estimate the adjusted hazard ratios (HRs), 95% confidence intervals and p-values, with age at diagnosis in the model as a confounding factor. Cut-off values, Kaplan–Meier curves, and Cox regression analysis were performed using Bell-Curve for Excel version 4.02 (Social Survey Research Information Co., Ltd., Tokyo, Japan). Fisher’s exact and chi-squared tests were used to test for correlations of clinical and pathological factors the high and low SLC20A1 expression; their p-values were calculated by two-sided hypotheses using Excel version 4.02 (Social Survey Research Information Co., Ltd, Tokyo, Japan) and R version 4.1.2 (The R Foundation for Statistical Computing, Vienna, Austria).

SLC20A1 expression is higher in prostate cancer than in normal prostate tissue. To examine SLC20A1 expression in primary prostate cancer and normal prostate tissue from the same patients, TCGA prostate dataset containing data from 549 patients was analyzed. SLC20A1 expression was found to be higher in primary prostate cancer than in normal prostate tissue (Figure 1A). SLC20A1 mutations detected included missense (0.5%; 2/444, SU2C-PCF Dream Team dataset) and deletions (0.2%; 1/494, TCGA PanCancer Atlas dataset). Other gene alterations, including gene amplification and fusion, were not detected. No significant differences in SLC20A1 expression were observed between primary prostate cancer and metastases such as lung, bone, liver and other soft tissue (Figure 1B).

Next, the association between SLC20A1 expression and clinicopathological data, including Gleason score and PSA level, was examined by Pearson’s chi-square and Fisher’s exact tests. SLC20A1 expression was associated with N and T categories (Table III). It was also shown that neither Gleason score nor PSA level were significantly correlated with SLC20A1 expression.

High expression of SLC20A1 indicates poor DSS, DFS and PFS in patients with prostate cancer. To examine the association between SLC20A1 expression and the clinical outcomes of prostate cancer, the SU2C-PCF Dream Team and TCGA PanCancer Atlas datasets were downloaded and OS was analyzed according to SLC20A1 expression status. The results of Kaplan–Meier and Cox regression analyses showed no significant difference in OS between SLC20A1^high and SLC20A1^low groups (Figure 1C and D, and Table IV). Since the SU2C-PCF Dream Team dataset contained only OS data, differences in DSS, DFS and PFS between the SLC20A1^high and SLC20A1^low groups were analyzed using TCGA PanCancer Atlas dataset (Figure 1D and Table IV). It was discovered that SLC20A1^high status was associated with poor DSS, DFS and PFS (p<0.05). Cox regression analysis for DSS, DFS and PFS was also performed and showed high SLC20A1 expression to be significantly associated with poor DFS and PFS; HRs were 2.97 and 2.66, respectively (Table IV). These results suggest that high SLC20A1 expression is associated with poor prognosis, and that it is involved in and contributes to prostate cancer recurrence and progression.

Endocrine therapy is effective for SLC20A1^high prostate cancer. Our previous study showed that endocrine therapy is not effective for SLC20A1^high breast cancer (11). We therefore analyzed whether endocrine therapy was effective for patients with SLC20A1^high prostate cancer (Figure 2A and Table IV). The Kaplan–Meier curves for OS, DSS, DFS and PFS revealed no significant differences between the SLC20A1^high and SLC20A1^low groups in patients with and without endocrine therapy (Figure 2A). Cox regression analysis was also used to study the effects of SLC20A1 expression on OS, DSS, DFS and PFS; it was not possible to analyze OS and DSS in the patients without endocrine therapy (Table IV). Cox regression analysis revealed no significant differences in the effects of SLC20A1 expression on the prognosis of patients with or without endocrine therapy (p>0.05; Table IV). Collectively, these results suggested that endocrine therapy is effective for SLC20A1^high prostate cancer, unlike breast cancer.

Patients with SLC20A1^high prostate cancer treated with radiotherapy tend to have poor clinical outcomes. Our previous study showed that radiotherapy is not effective for SLC20A1^high breast cancer (12). Therefore, the prognosis of patients with and without radiotherapy was next compared by SLC20A1 expression status (Figure 2B and Table IV). The Kaplan–Meier curves for OS and DSS indicate that there were no significant differences according to SLC20A1 expression status in patients treated with and without radiotherapy. The Kaplan–Meier curves for DFS and PFS in the patients treated without radiotherapy indicate that patients of the SLC20A1^high group had a poorer prognosis than those of the SLC20A1^low group (Figure 2B). On the other hand, the curves for DFS and PFS in patients treated with radiotherapy revealed no statistically significant difference between SLC20A1^high and SLC20A1^low groups. However, DFS and PFS of patients with SLC20A1^high prostate cancer treated with radiotherapy showed a tendency towards poorer prognosis (Figure 2B), which suggests that radiotherapy is not sufficient for reducing the recurrence and progression of SLC20A1^high prostate cancer. Cox regression analysis was performed to investigate the effects of SLC20A1 on the prognosis of patients treated with and without radiotherapy (Table IV); it was not possible to analyze DSS for the patients treated without radiotherapy, nor OS and DSS for those treated with radiotherapy. Only the DFS and PFS of the patients without radiotherapy were significant (p<0.05; Table IV). The HRs of patients with prostate cancer treated without radiotherapy were 2.88 for DFS and 2.51 for PFS. Of note, although the Kaplan–Meier curves and Cox regression analysis results revealed no differences between SLC20A1^high and SLC20A1^low groups treated with radiotherapy, the survival rate of patients in the SLC20A1^high group was lower than that of those in the SLC20A1^low group at all time points. Therefore, these results suggest that patients with SLC20A1^high prostate cancer treated with radiotherapy tend to have poor clinical outcomes.

In the present study, SLC20A1 expression in prostate cancer was found to be higher than that of normal prostate tissue (Figure 1A). This is consistent with the result of previous study showing that SLC20A1 is overexpressed in breast cancer (8). Consistent with the findings of the same study on breast cancer, SLC20A1 mutation was also detected at a very low frequency in prostate cancer. Thus, a high SLC20A1 expression in prostate cancer may be associated with transcriptional regulation. Consistent with the findings of previous studies on breast, pancreatic and tongue cancer (8-11), SLC20A1^high prostate cancer was associated with poor DSS. In addition, SLC20A1^high prostate cancer was associated with poor DFS and PFS. Furthermore, SLC20A1 was similarly expressed among prostate, lymph node, bone, liver and other soft tissue in patients with prostate cancer (Figure 1B). Our previous studies showed that patients with SLC20A1^high breast cancer had a high recurrence rate, suggesting metastasis (8,11). SLC20A1 is involved in the promotion of the cell cycle, cell viability, and tumor-sphere formation, and the suppression of apoptosis, in HeLa cells, HepG2 cells and certain breast cancer cell lines (8,13,14). Our findings suggested that high SLC20A1 expression may similarly be involved in cancer progression through prostate cancer cell proliferation, survival and metastasis. It has been reported that high concentrations of circulating phosphorus are associated with an increased prostate cancer risk (21). As SLC20A1 is a phosphorus transporter that brings phosphorus into cells, SLC20A1 might play an important role in the uptake of phosphate required for prostate cancer cell proliferation.

PSA levels were not correlated with SLC20A1 expression (Table III), which indicates that SLC20A1 may play a role in malignancy that is independent of mechanisms regulating the PSA level. SLC20A1 might serve as a biomarker for predicting prostate cancer which cannot be detected by PSA level. However, the specific role of SLC20A1 in prostate cancer cells remain to be elucidated.

Notably, although our previous study showed that endocrine therapy is less effective for SLC20A1^high ER-positive breast cancer, no significant difference in DSS, DFS and PFS was identified between patients treated with endocrine therapy for SLC20A1^high and SLC20A1^low prostate cancer (Figure 2A and Table IV). The difference in sensitivity between prostate and breast cancer for patients with SLC20A1^high disease treated with endocrine therapy, albeit there are differences in androgen receptor-dependent and ER-dependent signaling pathways in these cancer types, should be resolved. On the other hand, as well as breast cancer subtypes such as basal-like, claudin-low, luminal A and luminal B (8,12), in patients treated with radiotherapy, those with SLC20A1^high disease had a poorer prognosis than those with SLC20A1^low disease (Figure 2B and Table IV). ALDH1 is a detoxification enzyme degrading intracellular aldehyde, which is produced by reactive oxygen species (22,23). High ALDH1 expression is correlated with a low level of reactive oxygen species in gastric cancer, lymphoma, breast cancer and melanoma cells (23-26). Cancer stem cells have stemness properties, including resistance to radiotherapy (27-30). SLC20A1 has been shown to be involved in tumor-sphere formation by ALDH1-positive breast cancer stem cells (8). It is also known that ALDH1 is a cancer stem cell marker of prostate cancer (31). However, the specific role of SLC20A1 in ALDH1 positive prostate cancer stem cells remains to be elucidated.

In conclusion, SLC20A1 might serve as a prognostic biomarker for prostate cancer progression. Furthermore, the recommended treatment for patients with SLC20A1^high prostate cancer at the time of diagnosis appears to be endocrine therapy rather than radiotherapy, and SLC20A1^high status might be a biomarker suitable for decisions on selecting patients for endocrine therapy or radiotherapy.

The Authors have no conflicts of interest to declare.

TO, CO, and IM performed the analyses; TO, CO, and KA conceived the study; C M AO, TK, YX, TS, YM, Ke. S, and ST performed validation; TO, CO, and KA drafted the article; TO, CO, IM, C M AO, TK, YX, YH, TS, YN, YM, SM, HI, Ke. S, ST, Ka. S, SO, and KA contributed to the discussion and review of the final article; all Authors approved the final article.

This work was supported by the Grant-in-Aid for Scientific Research (C) of Japan Society for the Promotion of Science (20K07207) (KA), JST Moonshot R&D (JPMJPS2022) (SO), Tokyo University of Science Grant for President’s Research Promotion (KA), Grant-in-Aid for Research Activity Start-up (21K20732) (ST), JST SPRING (JPMJSP2151) (AO), Nagai Memorial Research Scholarship from the Pharmaceutical Society of Japan (AO) and a grant from the Consortium for Training Experts in Statistical Sciences (YN and SM).