Signatures of Adverse Pathological Features, Androgen Insensitivity and Metastatic Potential in Prostate Cancer

Discussion

The main goal of the present study was to investigate the expression of 42 genes which have been shown to be important in prostate cancer (5-9) in a small cohort of patients and in cells lines and to select the most relevant genes for extension studies with archived and prospectively collected material.

We analyzed gene expression in three prostate cancer cell lines and one normal epithelial cell line. These cell lines partly represent the natural course of prostate cancer from healthy epithelial cells to androgen-insensitive tumors with high metastatic potential. This set of cell lines could provide insight into genetic changes taking place in tumors treated with androgens and representing different patterns of invasiveness and metastatic potential. Almost all selected genes, with the exception of CDC6, CDCA5 and BUB1, exhibited differential expression in one or several cell lines (Figure 2), thereby partly confirming previous findings.

Keeping in mind restrictions based on cell line-based studies, we attempted to draw conclusions linking the gene-expression profiles of the tumors to their phenotypes. From a clinical point of view, it is important to be aware of the gene alterations associated with metastatic behaviour. Such knowledge has direct clinical applications (e.g. evaluation of biopsy samples with the aim of individualized treatment). However, many of the detected changes are overcast by the manifestation of androgen insensitivity at the genetic level, which has iatrogenic roots. In the present study, among the genes possibly reflecting the metastatic potential of tumours, the following appeared important: ELOVL fatty acid elongase 7 (ELOVL7), enhancer of zeste 2 polycomb repressive complex 2 subunit (EZH2), gastrulation brain homeobox 2 (GBX2), GOLM1, HOXC6, minichromosome maintenance complex component 6 (MCM6), MKI67, mucin 1, cell surface associated (MUC1), MYCBP2, somatostatin receptor 1 (SSTR1), topoisomerase (DNA) II alpha 170 kDa (TOP2A) and exportin 6 (XPO6). Some of these genes have been previously linked to prostate tumors. ELOVL7 is an important lipid metabolism enzyme, which has been shown to be overexpressed in hormone-naive prostate cancer (18). Our results demonstrated moderate and prominent underexpression of this gene in DU145 (moderate metastatic potential) and PC3 (high metastatic potential) cells, respectively. Thus, this could be related to progressive changes in tumor cells, leading to more aggressive tumoral behaviour. EZH2 is a histone transferase and gene silencer, that is overexpressed in metastatic prostate cancer (19). However, in the present study, it was depressed in the highly metastatic PC3 tumour cell line and unaffected in the DU145 cell line, possibly indicating that the EZH2 gene has an oppositional role in cancer growth (20). The GBX2 homeobox gene, which is necessary for cancer growth (21), was dramatically overexpressed in the PC3 cells in our study. Given that this gene is targetable (22), it might be considered a potential therapeutic target in prostate cancer. GOLM1 could also be a target for silencing due to its overexpression in prostate cancer (23). However, according to our data, it is not overexpressed in metastatic prostate cancer in 100% of cases. HOXC6 has been shown to be associated with adverse cancer features and reduced apoptosis of cancer cells (24, 25). In our study, HOXC6 was uniformly overexpressed in the cancer cell lines. Its expression was significantly lower in cancer cells with higher metastatic potential, indicating the possible prognostic role of this marker. In our study, MKI67 was overexpressed in the DU145 cell line and suppressed in the PC3 cell line. MKI67 is a well-known proliferation marker (26), which warrants further investigation with regard to prostate cancer aggressiveness. Previous research demonstrated that MUC1, an androgen receptor-regulated gene, was related to more aggressive behaviour of prostate tumours (27). In our study, it was underexpressed both in cell lines with moderate and high metastatic potential. However, we found no significant changes in the expression of MUC1 in our clinical cohort, thereby casting doubt on its role in the prognosis of hormone-naive cancer (28). MYCBP2 has not been previously reported to have any significant role in prostate cancer. Our results show that MYCBP2 could be a promising prognostic/therapeutic target and warrants further investigation, especially in view of the fact that it is directly associated with the MYC oncogene and that its overexpression played a discriminatory role in our clinical cohort (Figure 1 and 3B). The SSTR1 gene, encoding the somatostatin type 1 receptor, exhibited interesting bi-directional expression in the androgen-insensitive cell lines. In the DU145 cells, SSTR1 was overexpressed, which is atypical in castration-refractory prostate cancer (29). Nevertheless, if confirmed in human-derived material, this finding might be useful for individualized therapy. The TOP2A gene displayed almost the same pattern of expression, which could be of prognostic relevance. Although recent studies demonstrated overexpression of this gene in aggressive prostate cancer (30, 31), TOP2A was underexpressed in our clinical series and, therefore, potentially linked to advanced local-stage tumors (Figure 3A). Again, further studies might confirm its prognostic importance.

The differential expression of some other genes in the present study may also be clinically relevant and warrant further investigation. For example, previous studies confirmed that the multi-drug resistance protein 4 (ATP-binding cassette, sub-family C (CFTR/MRP), member 4; ABCC4) was important in estimating the efficacy of chemotherapy (32) and in determining the prognosis, with the protein gradually down-regulated in accordance with an increase in the malignant potential of tumour cells (33). We confirmed this finding in our cell line study. From a clinical point of view, our small cohort of patients could be dichotomized according to the expression of ABCC4. The prognostic relevance of this finding remains unclear.

In the clinical arm of our study, the primary aim was to identify which of the selected genes might have prognostic value with regard to adverse pathological features after a radical prostatectomy. Although our cohort was relatively small, we identified six genes which discriminated locally advanced disease and three genes which were differentially expressed in the tumors dependent on the Gleason score (Figure 3). Interestingly, five out of the six genes in the first group and two out of the three genes in the second group seemed to be linked with more aggressive disease in our cell line study. Five out of 28 tumors in our series were positive for gene fusion between transmembrane protease, serine 2 (TMPRSS2) and ERG (TMPRSS2:ERG) positive. TMPRSS2–ERG positivity was not related to any of the clinical or pathological parameters. Four tumors overexpressed serine peptidase inhibitor, Kazal type 1 (SPINK1). Three of these also showed overexpression of ERG, and two of the latter were TMPRSS2–ERG fusion-positive. This confirms that ERG positivity and overexpression of SPINK1 are not mutually exclusive (34). The precise role of overexpression of SPINK1 was not found in our study (too few cases). Previous studies are contradictory, considering the overexpression of SPINK1 as a predictor of aggressive disease and of an increased risk of biochemical recurrence (34, 35).

This study was a pilot and validation study, which aimed to select for genes with previously demonstrated relevance, and only a small group of patients and cells lines were used for the analysis and further validation. The clinical significance of our findings are as follows: some of the genes seem to represent promising candidates in the estimation of tumor aggressiveness. These genes warrant further consideration as tissue biomarkers. Based on an analysis of biopsy material (i.e. after the first minimally-invasive contact with the prostate tumor), these biomarkers could be helpful in the selection of patients for radical treatment or active surveillance. The pool of data on some of the genes (cell line data) related to the androgen-insensitivity of the tumor in the castration-refractory prostate cancer stage of the disease could be valuable for future therapeutic trials.

Our study has certain limitations, which should be considered when interpreting the results. All the cell cultures were cultivated and maintained in vitro for a long time. The passive evolution of tumour cells might result in some changes of the genome, and such changes are hard to predict. Another limitation for the cell lines is the absence of a typical prostate stromal environment. Our clinical cohort was small and not prospectively followed to account for any other hard endpoints, rather than adverse pathology at final investigation after prostatectomy. Nevertheless, the local stage (pT) is considered prognostically relevant (36). Therefore, gene-expression information is important in the prognosis with regard to other hard endpoints. One other limitation is that this study did not consider the intra- and inter-focal heterogeneity of the tumours. An additional limitation of the present study was the absence of a control group of normal tissues, which could be compared to cancerous tissues, obtained from the patients. In the study, we used results derived from a gene-expression analysis of a PreC cell line, which is a pure line of prostate epithelial cells.

Despite the aforementioned limitations, the data obtained from this study could be successfully used for further studies with large cohorts of patients with prostate cancer, both prospectively and with the use of archive material.

In this confirmatory study, a multiple gene-expression analysis was performed of normal prostatic epithelial, prostate cancer cell lines and cancerous tissues of prostatic tumors removed during radical prostatectomy in patients with biopsy-confirmed clinically-localized prostate cancer. The present study identified a set of genes that may have potential relevance for prognosis and tumor characterization. These genes should be included in further large-scale studies of the genetic characterization of prostate tumors that include survival-based end-points.