Abstract
Aim: Biomarkers with prognostic and predictive value can help stratify patients with colorectal cancer (CRC) into appropriate treatment groups. We sought to evaluate the clinical utility of P53 protein expression as a biomarker in VICTOR, a large phase III trial of rofecoxib in stage II and III CRC. Patients and Methods: Tissue micro arrays were constructed from 884 tumors and the expression of P53 was examined by immunohistochemistry. Tumors were dichotomised as either P53-positive (nuclear expression in >10% of cells or the ‘absent’ pattern, both representing TP53 mutation) or P53-negative (nuclear expression in <10% of cells). Results: Aberrant P53 expression was found in 65% (482/740) of patients. It was associated with distal location (p<0.001) and stage III disease (p<0.001). No effect was observed on disease-free or overall survival, and there was no interaction with chemotherapy or radiotherapy. Conclusion: Analysis of P53 expression in the patients recruited to the VICTOR trial confirmed that P53 expression is associated with site and stage of CRC. However, independently, this biomarker has neither prognostic nor predictive utility in this cohort of patients.
Colorectal cancer (CRC) is the second most common cause of cancer-related death in the Western countries (1, 2). Management of patients with CRC is dependent on the staging of the resected tumor by a pathologist. Patients with early-stage CRC (i.e. stage I) have a good prognosis and will not receive adjuvant therapy (3). In cases with more advanced-stage disease (i.e. stage II and III), the prognosis is more variable and consequently the management is more complex. Treating all cases of stage II disease with adjuvant chemotherapy gives a net benefit of around 4% improvement in survival (4). This is most probably due to the fact that around 70% of patient with stage II CRC will be disease-free five years later and thus will not benefit from adjuvant therapy. It would be expedient to target only the 30% of stage II cases who are at a high risk of recurrent disease. However, discriminating between the two groups is not easy, although certain pathological features (such as extramural vascular invasion) are associated with poorer outcome, pathological evaluation is operator-dependent and can be a highly variable method of gathering prognostic information (5). Furthermore, pathological staging does not provide any information on tumor biology.
Almost all patients with stage III disease are given adjuvant chemotherapy, as the overall prognosis is quite poor (6). However, not all tumors are the same and it is not known which tumors will respond to which specific therapy. There is, thus, a need for robust predictive and prognostic biomarkers which can add to current standard pathological analysis (7).
TP53 is a tumor-suppressor gene and it is mutated in 60-70% of CRCs (8). It can be inactivated by truncating mutations or by missense mutations and these lead to aberrant P53 protein expression. In the former case, there is complete loss of expression of P53 protein, whilst in the latter case, there is post-translational stabilisation leading to gross overexpression of the protein (9). These changes can be detected by immunohistochemistry (IHC), thereby allowing tumors containing mutant TP53 to be distinguished from those which are wild-type for TP53 (10).
We aimed to test the clinical utility of aberrant P53 expression as a prognostic and predictive marker in CRC. Although there exist previous studies investigating this biomarker, published data are not completely conclusive. Studies have been confounded by a variety of factors such as pooling data from multiple trials and technical variation in the laboratory methods. This study was performed on tissue obtained from patients recruited to the VICTOR trial (a single large randomised phase III trial run in the UK) (11) and all the IHC was performed in a single, fully accredited, diagnostic laboratory.
Patients and Methods
Patients. The VICTOR trial was a phase III randomised, placebo-controlled double-blind trial of rofecoxib (VIOXX®) in patients with stage II or III CRC who had undergone potentially curative surgery and completion of adjuvant therapy (if it was given). The trial was terminated prematurely in 2004 but up to that point, it had recruited 2434 patients from 151 hospitals in the UK. The rofecoxib-treated population comprised 1167 patients and there were 1160 placebo control patients. The trial showed no survival benefit of rofecoxib in the overall population and a lack of prognostic or predictive significance of Cycolo-oxygenase 2 expression (12).
Tissue microarrays (TMAs) and IHC. Formalin-fixed paraffin-embedded (FFPE) tumor samples were collected (with full consent for research use) from 1,006 patients. Of these, 884 were suitable for use in TMAs and the characteristics of this population were typical of the whole VICTOR population. Three cores of tumor (plus, if available, one core from adjacent normal tissue) were taken from each block and a total of 29 TMAs were assembled at the Astra-Zeneca Oncology Molecular Pathology Laboratory (Alderley Park, Cheshire, UK) and Oxford Radcliffe Bio-bank Oxford University using standard techniques. All IHC was performed using compact polymer technology on an automated Bond-Max (Leica Microsystems, Milton Keynes, Buckinghamshire, UK) using consumables provided by the manufacturer (Leica Bond Refine Detection kit; DS9800).
Sections (3 μm-thick) were de-waxed using Leica Dewax solution (AR9222) for 30 s at 72°C followed by the antigen retrieval step by incubating for 30 min at 90°C with Epitope Retrieval solution 1 (AR9961). After washing in Leica Bond wash solution (AR9590) sections were immersed in peroxide block solution (Kit DS9800) to block endogenous peroxidase for 5 min at room temperature (RT). The sections were incubated with a 1:100 dilution of primary antibody (Clone NCL-L-P53-D07; Leica Microsystems) for 15 min at RT. The post primary and polymer detection system were incubated each for 8 min at RT, before addition of 3,3’-Diamino-benzidine (DAB) for 10 min at RT followed by the DAB enhancer (AR9432) for 5 min at RT. The sections were counterstained with haematoxylin for 5 min at RT. Finally, sections were removed from the Bond staining machine, dehydrated in three baths of 100% Industrial methylated spirits (Genta Medical, York, Yorkshire, UK), cleared in Xylene (Genta Medical) and permanently mounted under glass coverslips using Pertex (Histolab, Hemel Hempstead, London UK). The stained TMA sections were scanned (at ×40 magnification) using a NanoZoomer Digital slide scanner (Hamamatsu, Hamamatsu City, Shizuoka Pref., Japan) and the digital images were uploaded into the SlidePath digital pathology system (SlidePath, Dublin, Ireland) and scored using Distiller software (SlidePath). Blinded scoring for P53 was performed by two independent observers (W.F and M.M).
Criteria for the scoring of the IHC. Scoring for P53 expression was based on the intensity, pattern and percentage of stained tumor nuclei and was scored dichotomously as either P53+ or P53−. Non-neoplastic colonic mucosa, inflammatory and stromal cells adjacent to neoplastic cells exhibited weak staining and served as positive internal controls. Using these as a reference, two patterns of aberrant P53 expression were seen. The more common pattern was strong nuclear expression in >10% of the tumor cells and this has been shown to be indicative of missense mutations of TP53 (10). In addition, we and others have described a pattern characterised by a complete loss of P53 expression in the tumor cells, which we term the ‘absent’ pattern and this has been correlated with truncating TP53 mutations (13, 14). Either of the aberrant patterns was classed as P53+, whilst other patterns were classed as P53− (Figure 1). There were tumors where neither the tumor cells nor the stromal cells exhibited any P53 expression and these cases were considered as technical failures and deemed unclassifiable.
Statistical analysis methods. Cases exhibiting the unclassifiable pattern were excluded from further analysis. The P53absent and P53+ cases were grouped together (as they all represent mutant P53) and called P53+ and these were compared against P53− cases. Overall survival (OS) was the primary endpoint and disease-free survival (DFS) was the secondary endpoint. OS and DFS are displayed using Kaplan-Meier plots by expression category together with the associated log-rank p-value. Univariate and multivariate analyses were undertaken using the Cox proportional hazards model adjusting for treatment and for clinical factors including gender, tumor site (proximal, defined as proximal to the sigmoid colon; distal, defined as within the sigmoid and rectum), tumor stage, age and prior chemotherapy/radiotherapy. Hazard ratios and 95% confidence intervals (CI) are presented and analysis was carried out using STATA version 11.0 (StataCorp, College Station, Texas, USA).
Results
Clinicopathological features and outcomes associated with P53 expression. Scoring for P53 expression was dichotomised as either P53+ (including both patterns associated with TP53 mutation) or P53− (associated with wild-type TP53, Figure 1). Overall, 65% (482/740) of the tumors were P53+ (Table I). Compared to P53− tumors, the P53+ tumors were associated with distal location (pp<0.001) and advanced tumor stage (p=0.009). There was no association with gender, age of patient at presentation, pathological T stage, vascular/lymphatic invasion or lymph node recovery.
The hazard ratio for DFS for the P53+ group versus the P53− group was 1.08 (95% CI=0.79-1.47; p=0.634). The 5-year DFS rates were 70.9% (95% CI=63.6-77.0%) for the P53− group and 65.0% (95% CI=59.5-70.0%) for the P53+ group (Figure 2). The hazard ratio for OS for the P53+ group versus the P53− group was 1.12 (95% CI=0.77-1.62, p=0.55). The 5-year OS rates were 81.0 (95% CI=75.2-85.6%) for P53− and 77.6% (95% CI=73.2-81.3%) for P53+ (Figure 2). The patients with P53+ tumors were more likely to have received chemotherapy and radiotherapy than those with P53− tumors (pp<0.001) which probably reflects their clinical staging and distal location. However, analysis according to chemotherapy sub-group did not reveal any significant difference in outcome for those with P53+ tumors in either the chemotherapy-treated or chemotherapy-naive groups.
Discussion
The present study sought to investigate the clinical utility of aberrant P53 expression as a prognostic and predictive biomarker in CRC. We had a large study population (n=824) which was drawn only from the VICTOR trial. We ensured technical excellence by performing the IHC in a fully accredited diagnostic lab and having two reviewers for the immunostaining. Under these conditions we failed to find any prognostic or predictive value for aberrant P53 expression.
The published data regarding the effect of P53 in CRC are heterogeneous (2, 15-22). In this study we defined two patterns of aberrant expression which are associated with TP53 mutation and grouped these patterns together as P53+. As the “absent” pattern has only recently been described, it has not been included in previous studies. Since both the absent pattern and the pattern of strong over-expression are associated with TP53 mutation, we felt is was justified to group the two patterns together. Tumors with wild-type pattern of P53 expression were designated as P53−. Consistent with many studies, our data showed that 65% of the tumors were P53+. We found that aberrant P53 expression was associated with advanced stage (pp<0.001). This finding runs contrary to the Fearon and Vogelstein model which postulates that TP53 mutation presages the acquisition of invasive tendencies in an adenoma (23). However, other studies have also reported this association (24-26), possibly suggesting that a diathesis for lymph node metastasis may follow TP53 mutation. However despite this association, analysis of DFS and OS showed no difference between the P53+ and the P53− tumors either in the group overall or in stage-specific analysis. Furthermore, there was no effect of aberrant P53 expression on outcome of patients who received either chemotherapy or radiotherapy.
In summary, this was a meticulous study of the prognostic and predictive value of aberrant P53 expression conducted in the context of the VICTOR clinical trial for stage II and III CRC. Whilst many of the previously reported clinicopathological associations were confirmed, there was no prognostic or predictive information to be derived from evaluation of this biomarker.
Acknowledgements
The Authors would like to thank Neil Hand, Anne Kane, Dr Andrew Green, Haitao Wang and Joy Winter for their contribution to sample handling, tissue processing, immunostaining and slide scanning.
Footnotes
-
↵* These Authors contributed equally to this study.
- Received November 12, 2014.
- Revision received December 15, 2014.
- Accepted December 17, 2014.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved