p53 immunochemistry is an independent prognostic marker for outcome in conservatively treated prostate cancer


Daniel M. Berney, Barts and the London, Queen Mary’s School of Medicine and Dentistry – Department of Cellular Pathology, The Royal London Hospital, 80 Newark Street, London E1 2ES, UK.
e-mail: danberney@hotmail.com, D.Berney@bartsandthelondon.nhs.uk



To determine whether p53 is an independent biomarker of prostate cancer outcome against currently used biomarkers in a cohort of conservatively treated prostate cancers with long-term follow-up available.


We examined p53 expression by immunohistochemistry in a cohort of 705 patients with clinically localized prostate cancer, who were treated conservatively. Patients were selected through UK Cancer Registries. End-points included prostate cancer death and overall death rates. Standard biological variables, including diagnostic serum PSA, contemporary Gleason scoring, clinical staging and cancer extent were available. p53 expression was measured semi-quantitatively on microscopic examination and compared with current clinical biomarkers.


p53 over expression was a significant predictor of cause-specific survival (hazard ratio [HR] 2.95, 95% CI 2.05–4.25, P < 0.001) and overall survival (HR 2.37, 95% CI 1.84–3.05, P < 0.001). In multivariate analysis including competing biological variables p53 expression was still significantly linked to prostate cancer survival (HR 1.51, 95% CI 1.04–2.19, P = 0.03) and overall survival (HR 1.57, 95% CI 1.21–2.05, P = 0.001).


We conclude that p53 may have a role in the future assessment of newly diagnosed prostate cancer, as it significantly adds to the current prognostic model.


Trans-Atlantic Prostate Group


tissue microarray


hazard ratio


prostatic intraepithelial neoplasia


intensity score.


Prostate cancer has a poorly understood natural history. There are very few biomarkers to differentiate aggressive from non-aggressive tumours after diagnosis. This is critical in early prostate cancer as there is, as yet, little scientific rationale for patient selection for radical vs conservative treatment. It has been calculated that 19 patients need to be treated radically to benefit one man [1]. It is not clear if screening does more good than harm, therefore both under and over treatment remain substantial risks. This was recognized in the recently published National Institute of Clinical Excellence (NICE) guidance on prostate cancer, which acknowledged this was an area where the evidence base was lacking and further research required [2]. The Trans-Atlantic Prostate Group (TAPG) was formed to assemble a large retrospective cohort of prostate cancers with clinical and biomarker data suitable for basic translational science [3,4].

p53 is a tumour suppressor gene, which is mutated in half of human malignancies and has been intensively studied in numerous cancer models [5]. Many of these studies have suggested that p53 may be of prognostic value in prostate cancer after surgery [6–9], radiation therapy [10–12] and hormonal therapy [13,14]. However, to our knowledge, no studies have been performed on a large cohort of tumours that were treated expectantly. As a commonly assessed biomarker in clinical laboratories and a key protein in the regulation of cell apoptosis and proliferation, p53 is of prime interest to assess for biomarker potential in untreated prostate cancer.

We therefore tested the hypothesis that immunochemistry to detect p53 on cases in the TAPG cohort might provide prognostic information in conservatively treated localized prostate cancers additional to that provided by established biomarkers.


The detailed collection methods of the TAPG cohort have been published previously [3]. In short, patient information was collected from six cancer registries in the UK. Men were diagnosed with clinically localized prostate cancer by TURP or needle biopsies in the years 1990–96. The patients were aged <76 years and the initial serum PSA level was >4 ng/mL and <100 ng/mL at diagnosis. Patients who had radical treatment or radiation therapy within 6 months of diagnosis were excluded, as well as patients who showed objective evidence of metastasis or clinical indication of metastatic disease. The clinical stage and Gleason score were reviewed and re-assigned if necessary [15,16]. In all, there were 1656 patients in the study. After 10 years follow-up, a prognostic model and nomogram including Gleason score and PSA level were created [4].

Half of this cohort (810 patients) comprised TURP specimens suitable for conventional tissue microarray (TMA). For TMA analysis, these were placed into 24 wax blocks. Areas of normal prostate away from the tumour were also arrayed as well as TURP cases where no malignancy was identified to act as internal controls. Each tissue core was 0.6 mm in diameter. Multiple samples (usually four) were taken to account for tumour heterogeneity, as well as samples of ‘normal’ tissue away from tumourous areas. In all, there were 3964 cores. These TMAs were cut and immunostained for p53 (DO7, Dako, Ely, Cambridgeshire) using the ABC technique (Vector Laboratories, Orton Southgate, Peterborough) with pressure cooking for antigen retrieval. A dilution of 1:1000 was used. The external positive control was an anaplastic carcinoma of the thyroid.

Each TMA slide was examined by a consultant pathologist (D.M.B). Tissue cores were categorized as tumour, normal, or prostatic intraepithelial neoplasia (PIN), aided by the basal cell marker, CK5/6 on a second section. The immunochemistry was evaluated using a semi-quantitative technique. The percentage staining (P) for each antibody was evaluated for each core semi-quantitatively. The intensity (I) of staining scored: 0, no staining; 1, weak; 2, moderate; and 3, strong. This allowed the calculation of an intensity score (IS), where IS = I × P, where IS could vary between 0 and 300.

The core with maximum staining for the protein was taken as the representative of the patient as it was thought that the maximal intensity for p53 was likely to be the area of tumour most likely to be driving neoplastic progression (Fig. 1) [17].

Figure 1.

Prostatic core from TMA showing strong nuclear staining on immunochemistry for p53.

The association between Gleason score with p53 protein expression was analysed using Pearson’s correlation coefficient. The expression of p53 antibody was evaluated for the two main end-points; prostate cancer survival and overall survival using a Cox proportional hazards model. The other prognostic indicators for the original study, Gleason score and serum PSA, were included, along with p53 antibody expression, in a multivariate model. All P-values were two sided and 95% confidence intervals (CIs) were based on normal distribution.


In all, 1844 cores were diagnosed as cancer, 20 as PIN, and 1246 cores showed normal prostatic epithelium. Information was missing for 831 tissue cores; this was usually because the cores were composed of stroma only with no epithelial element, though common to most TMA studies there was some ‘drop out’ of cores where parts of the TMA section did not transfer to a glass slide.

The analysis included 1844 cancerous cores from 705 eligible patients. The IS for p53 ranged from 0 to 300 with 80% of cores having an IS of 0. Therefore, immunostaining expression was dichotomized using an IS of ≥2 as positive for analyses. The expression of p53 was compared with Gleason score using the Pearson’s chi-squared test. The Gleason score was categorized into three groups including Gleason score <7, 7 and >7. There was a significant relationship between the p53 expression and the Gleason score (Table 1) whatever the cutpoint (P < 0.001). Univariate Cox model analysis (Fig. 2) showed p53 expression was a significant prognostic factor for cause specific survival (hazard ratio [HR] 2.95, 95% CI 2.05–4.25, P < 0.001) and overall survival (HR 2.37, 95% CI 1.84–3.05, P < 0.001). In multivariate analysis including initial serum PSA level, Gleason score, age and extent of disease, p53 expression was still significantly linked to prostate cancer survival (HR 1.51, 95% CI 1.04–2.19, P = 0.03) and overall survival (HR 1.57, 95% CI 1.21–2.05, P = 0.001).

Table 1.  Relationship of p53 IS with other biological variables
Variablep53 immunostainingP
IS ≤1IS >1
  • *

    Clinical stage was assigned by review of the clinical PR examination in patient records.

Mean (sd) age, years 69 (5)70 (5) 0.07
N (%):
 Classes of age, years   0.03
  ≤65119 (91)12 (9) 
  >65–70185 (92)15 (8) 
  >70–73148 (84)29 (16) 
  >73–76169 (86)28 (14) 
 Early hormone management   0.001
  Yes110 (80)28 (20) 
  No511 (90)56 (10) 
 Gleason score  <0.001
  <7337 (98) 7 (2) 
  =7154 (83)32 (17) 
  >7130 (74)45 (26) 
 Clinical stage*  <0.001
  T1183 (94)12 (6) 
  T2134 (84)26 (16) 
  T3 57 (71)23 (29) 
 Baseline PSA, ng/mL  <0.001
  ≤4222 (93)17 (7) 
  >4–10131 (92)12 (8) 
  >10–25131 (89)17 (11) 
  >25–50 85 (78)24 (22) 
  >50–100 52 (79)14 (21) 
 Cancer in biopsy, %  <0.001
  ≤6167 (95) 8 (5) 
  >6–20166 (96) 7 (4) 
  >20–40 90 (88)12 (12) 
  >40–75 79 (76)25 (24) 
  >75–100 112 (78)31 (22) 
Figure 2.

Association of p53 expression with prostate cancer survival and overall survival.

Results were then analysed using sub-cohorts split by Gleason scores. The univariate Cox model analyses (Fig. 3) showed that, for patients with a Gleason score of >7, p53 expression was still a significant prognostic factor for cause-specific survival (HR 1.75, 95% CI 1.12–2.72, P = 0.01) and overall survival (HR 1.87, 95% CI 1.30–2.68, P = 0.001). When age at diagnosis, baseline PSA level and extent of disease are included with p53 expression into a multivariate Cox model, p53 immunostaining was still significantly associated with prostate cancer survival (HR 1.76, 95% CI 1.12–2.76, P = 0.01) and overall survival (HR 1.73, 95% CI 1.19–2.51, P = 0.004). However, significance was not reached in multivariate models for Gleason scores of ≤7.

Figure 3.

Association of p53 expression with prostate cancer survival and overall survival at Gleason score >7.


Wild-type p53 has important roles in the cell cycle, DNA repair and apoptosis. In the presence of DNA damage, p53 is activated and initiates cell cycle arrest. In the absence of wild-type p53, this function is impaired and therefore mutated cells survive continuing their clonal expansion resulting in tumour progression. Mutated p53 has a longer half-life compared with the wild-type and the nuclear accumulation of p53 can be detected via immunohistochemistry. There are numerous possible p53 mutations: point mutations of p53 gene are the most identified in addition to allelic loss, re-arrangements and deletions. These mutations are mostly concentrated on exons 5–8 [18]. It has been shown in breast cancer that the type of mutation affects the risk of tumour progression [19] and therefore in an ideal world, mutation analysis would be possibly a better measure of the likelihood of prostate cancer progression. However, recent studies have shown a high sensitivity of an immunochemical approach when compared with direct sequencing [20]. Although p53 abnormalities are largely a late event, this is not always the case, and may be present in early disease [21]. From a practical standpoint, the assessment of p53 mutations is unlikely to find its way into routine practice in the near future. Small amounts of formalin fixed tissue are not amenable to mutation analysis, and this is normally all that is available in the patient newly diagnosed with prostate cancer. Immunochemistry is a well recognised and understood technique in pathology laboratories, and p53 immunochemistry is already used to assess TCCs in selected situations. Therefore, although immunochemical assessment does not equate with p53 mutation, it may be considered a good surrogate.

The biological role of p53 in prostate cancer is unclear because of the contradictory nature of several studies. Most of the studies on p53 are on radical surgery [22], radiotherapy [12] or hormone therapy patients. Very few studies included patients treated conservatively. One such study on p53 and Ki67 [23] reported that p53 was an independent prognostic marker. Unfortunately, Gleason grading was not used and there was no initial serum PSA levels available, meaning that this series is not based on current powerful prognostic indicators. Studies on radical treatment cases have shown that p53 has the potential to be a useful biomarker of progression, but after radical surgery, options are limited.

We think this is one of the first studies to consider p53 immunochemistry on such a large cohort of conservatively treated patients, showing that p53 may be a useful prognostic marker in addition to Gleason score and serum PSA level. Although significance was maintained at higher Gleason scores, when the cohort was subdivided, p53 analysis failed to achieve significance at lower Gleason scores. This may be a reflection of the smaller number of p53 positive cases at lower Gleason scores: only seven cases had a Gleason score of 6 and had an IS of p53 of >1.

Other biomarkers including the TMPRSS-ERG fusion gene have been assessed in this cohort. The results have shown a strong correlation with outcome. However, measurement by fluorescent in situ hybridization is more difficult on limited samples than immunochemistry, and practical measurement in small biopsy samples may prove challenging [24].

This work needs to be validated in several ways. Firstly, it is planned to use new TMA techniques to examine the needle biopsies in the TAPG series, which comprise 45% of the cohort. This would be a model approximating more to current methods of diagnosis, though it should be noted that comparing the two methods of diagnosis used in the cohort did not reveal significant differences in survival or other parameters. Secondly, the work could be repeated in other conservatively treated prostate cancer series. These have fewer cases than this series, but may have more tissue available due to repeated biopsies on active surveillance [25]. We recognize that current active surveillance techniques are different from older ‘watchful waiting regimens’. In reality, this cohort is likely to have been managed by various monitoring techniques. However, in the absence of other larger series it gives an adequate surrogate for future, more clinically controlled studies.

Prospective studies would obviously be an ideal model for future experiments, but nearly all use needle biopsy as a method of diagnosis and with current histopathological techniques, which thoroughly examine biopsy tissue at multiple levels, there is often little tissue for further analysis. It is likely to be many years before well-planned series such as the Prostate testing for cancer and Treatment (ProtecT) study are mature enough for such assessments [26]. It is therefore crucial that the biomarkers used in these series are well selected as they may provide a ‘one off’ opportunity to provide new tissue biomarkers for possible translation to clinical use. Therefore, for purely practical reasons, further retrospective studies are required, including enlargement of the TAPG series, which is planned. As shown, although the overall number of cases in the TAPG study is high, the number of overall events, especially in the Gleason ≤6 group is relatively low. Therefore testing in a larger model is required to validate this work including further clinical follow-up towards 20 years, similar to older watchful-waiting cohorts [27].

In conclusion, decisions on the treatment of clinically localized prostate cancer are still not evidence based. Only by rigorous application of these techniques to large, well-defined series, will biomarkers be established that can add to the current prognostic model. We think that p53 immunochemistry is a prime candidate for such a biomarker.


This work was supported by Cancer Research UK, The National Institute of Health (SPORE), The Koch Foundation, The NCRI and the Grand Charity of Freemasons. D.B. and S.K. are supported by The Orchid Appeal.


None declared.