The transcription factor E2F3 is overexpressed in prostate cancer and independently predicts clinical outcome

Authors: Sandra Edwards, Colin Cooper

The treatment and management of prostate cancer needs to be improved.   A particular problem that exists is that it is not possible to predict how early prostate cancer still localised to the prostate (for example detected by the PSA test) will behave.   Some cases may remain dormant for many years without progressing while others will progress rapidly to malignancy.   The particular goal of our work is to identify markers that can be used to distinguish dormant from aggressive early prostate cancers.   This is important because such markers can be used to identify aggressive cancers early so that they may be treated and to minimise treatment for dormant tumours.

We have recently demonstrated that the E2F3 transcription factor gene, which has an established role in controlling transition through the cell cycle, is a bladder cancer oncogene that is activated by amplification and overexpression 1 . We have investigated the role of E2F3 in prostate cancer by immunohistochemistry on prostate tissue microarrays.

Tissue microarrays (TMAs) were constructed from TURP and prostatectomy samples that had been taken from a consecutive series of patients (774 cores from 147 patients) diagnosed with prostate cancer who attended the Royal Marsden NHS Trust from 1992. Immunohistochemistry (Figure 1) was carried out using a monoclonal antibody to E2F3 (Upstate, UK).

Figure 1. E2F3 expression detected by immunohistochemistry in formalin fixed tissue in TMAs. (a-c) Primary prostate cancers. Examples of tumours scored as (a, b) positive and (c) negative are shown. (d) Example of a hyperplasia scored as positive for E2F3. (e) Example of morphologically normal epithelia scored positive for E2F3. (f) Example of cytoplasmic staining observed in morphologically normal epithelium. Staining: brown, E2F3; blue, haematoxylin counterstain

Analysis of the TMAs demonstrated intense nuclei staining for E2F3 in 67% (98/147) of prostate cancers (Figure 1a-c), with the maximum proportion of cells containing nuclear staining varying from 5 to 90%. Analysis of non-neoplastic epithelium samples revealed that hyperplastic epithelium (Figure 1d) and morphologically normal epithelium (Figure 1e) exhibited nuclear E2F3 staining in respectively 21/124 (19.8%) and 1/43 (2.33%) of cases. Because of the wide variation in the percentage of nuclei exhibiting staining, the data were stratified into six bands: negative, up to 20, 21-40, 41-60, 61-80 and 81-100%. In some cases, E2F3 was observed predominantly in the cytoplasm (Figure 1f). This pattern was not observed in prostate cancer and was restricted entirely to morphologically normal epithelium (3/43) and epithelia hyperplasia (15/124).

Studies of clinical correlation showed a significant association between the presence of nuclear E2F3 staining and Gleason score (P=0.016), but no association with age (P=0.327), M stage (P=0.574) or AJCC 2 stage (P=0.613)

Figure 2 shows overall survival and cause-specific survival according to E2F3 status. These curves demonstrated significantly increased hazard ratios for patients whose cancers exhibited E2F3 expression for both overall survival (log-rank test, df=1, P=0.0022) and cause specific survival (log-rank test, df=1, P=0.0047) of 1.9290 and 1.8423respectively.

Figure 2. Kaplan-Meier analysis of cause-specific survival in prostate cancer patients that are positive (red) and negative (black) for E2F3 nuclear staining. Vertical lines are error bars.

Figure 3 shows overall and cause-specific survival following stratification of patients according to the maximum percentage of E2F3-positive nuclei observed in their prostate cancer. These analyses demonstrated a significant association between the percentage of nuclear staining for E2F3 and risk of death for both overall (logrank test, df=5, P=0.0014) and cause-specific (logrank test, df=5, P=0.004) survival. Importantly, in multivariate analysis, E2F3 staining was an independent predictor of clinical outcome.

Several lines of evidence now support the view that the pRB-E2F3-EZH2 pathway may represent a key oncogenic axis that has an important role in determining development and aggressiveness of

human prostate cancer (Figure 4).

Figure 4.  Role of the pRB-E2F3-EZH2 cell cycle control axis in determining aggressiveness in human prostate cancer. The INK4A/ARF gene encodes two proteins p16 and p14ARF that are negative regulators of the pRB-E2F3-EZH2 pathway. Genes overexpressed (red) or downregulated (green) in human prostate cancer are indicated. Red stripes indicate that overexpression of p53 measured in studies of clinical correlation was believed to represent underlying mutations in the p53 gene.       signifies that the indicated change in gene expression is associated with adverse clinical outcome

The E2F3 protein has an established role in controlling cell cycle progression and controls the expression of EZH2 which has also recently been implicated in the development of human prostate cancer. Several other alterations of the expression levels of genes in this pathway have also been connected with poor clinical outcome (Figure 4).

In conclusion, we demonstrate aberrant high-level overexpression of E2F3 in 67% of prostate cancer and that overexpression of this protein is an independent predictor of poor clinical outcome in prostate cancer 3 .   Importantly, the higher the level of E2F3 protein, the worse the prognosis.

References

1. Feber A et al. Oncogene. 2004;23(8):1627-30

2. Prostate. In: American Joint Committee on Cancer: AJCC Cancer Staging Manual.

    Philadelphia, Pa: Lippincott-Raven Publishers, 5th ed., 1997, pp 219-224

3. Foster CS et al. Oncogene. 2004;23(35):5871-9