FormalPara Key Points

TP53 status may affect the response to anti-epidermal growth factor receptor (EGFR) therapies as a consequence of p53 activity on the EGFR promoter, and TP53 mutation is often associated with its overexpression.

Our analysis suggests a predictive/prognostic role of abnormal p53 expression in patients with metastatic colorectal cancer treated with anti-EGFR therapy.

The finding of p53 overexpression as a resistance mechanism to anti-EGFR therapies represents a further step toward personalised treatment.

1 Introduction

The transmembrane tyrosine kinase epidermal growth factor receptor (EGFR or HER1, ErB1) belongs to the ErbB family along with ErbB-2, ErbB-3 and ErbB-4a and is a key driver of cell proliferation, survival, adhesion, migration and differentiation [1,2,3]. Deregulation of the EGFR pathway was observed in several malignancies such as colorectal, non-small cell lung, breast, ovary, renal, head and neck, pancreatic, prostate, cervical and bladder cancer [1, 4, 5]. In this setting, EGFR overexpression/upregulation and hyperactivation might be responsible for the promotion of tumour cell growth, resistance to apoptosis, synthesis of angiogenic and growth factors and metastatic spread. As a consequence, EGFR blockade plays an established role as an anti-cancer strategy, especially in colorectal cancer (CRC), non-small cell lung cancer, and head and neck tumours [6,7,8,9,10,11,12,13,14]. Presently, the anti-EGFR monoclonal antibodies cetuximab and panitumumab represent the cornerstone of treatment for RAS wild-type (WT) metastatic CRC (mCRC) [15, 16].

Unfortunately, not all patients with RAS WT mCRC are sensitive to EGFR-targeting agents and resistance to these drugs is still an open issue [17,18,19,20,21]. Several potential mechanisms for the lack of efficacy of anti-EGFR have been investigated beyond N-RAS and K-RAS mutational status, such as BRAF V600E mutation [22,23,24,25,26], PI3KCA exon 20 mutation, PTEN loss [22, 27,28,29], Stat3 and Akt phosphorylation [22, 30,31,32], HER2 amplification [22, 33,34,35,36], HER3 expression [37], IGF1R activation [22, 38, 39], MET amplification [22, 40], altered vascular endothelial growth factor/vascular endothelial growth factor receptor expression [22, 41], EGFR gene copy number alteration and EGFR methylation [42, 43], but for most of these no definitive data are available.

Among the potential mechanisms of resistance, data suggest that TP53 mutations might play a role in tumour cell sensitivity to anti-EGFR antibodies. In fact, recent studies have suggested that activation of the EGFR pathway leads to malignant transformation only if the p53 protein is inactivated [44].

The p53 protein, which is encoded by the tumour suppressor gene p53 (TP53) located in chromosome 17p, is one of the most important tumour suppressors. Acting as a zinc-containing transcription factor [45,46,47,48,49], it regulates downstream genes involved in DNA repair, cell-cycle arrest and apoptosis [50,51,52,53,54]. More specifically, in the event of cellular stress signals such as genotoxic damage, oncogene activation, hypoxia or loss of intercellular adhesion, p53 takes on an active tetrameric form and stops cell-cycle progression and induces DNA repair, senescence and cell death through apoptosis to preserve the genomic integrity. Consequently, p53 has a crucial role in the regulation of cell proliferation and cancer development inhibition; for its pivotal role in protection from tumour development, it has been defined as “the guardian of the genome” [54,55,56,57,58,59,60,61].

Mutations or deletions of the TP53 gene can be found in nearly 50% of human cancers and they lead to an impaired tumour suppressor function [62]. Tumours harbouring a TP53 mutation are usually characterised by genomic instability and poor prognosis [54, 63, 64]. Moreover, it was shown that in p53 mutants, the loss of WT p53 function is also associated with the gain of new oncogenic roles promoting cancer, metastasis, inhibition of apoptosis and drug resistance [64,65,66,67].

Mutations in TP53 usually occur in 40–60% of patients with CRC. There are well-known logistical difficulties and resource limitations associated with direct sequencing of the TP53 gene. Therefore, most studies have used immunohistochemistry to detect mutant p53, with the assumption that a mutation is often associated with its overexpression. However, the lack of expression is generally indicative of WT TP53. Only in limited cases, a complete loss of p53 expression is associated with truncating mutations and loss of heterozygosity (LOH). Several trials documented conflicting results regarding the association between p53 over-expression and clinical outcome [68,69,70,71].

The predictive function of TP53 mutations in patients with mCRC treated with targeted therapies has not been established so far. However, some evidence suggests that TP53 status may affect the response to anti-EGFR therapies as a consequence of the activity of p53 on the promoter of the EGFR [72,73,74,75,76,77]. However, findings from previous analyses investigating the role of p53 in this setting were contradictory. Based on these considerations, we designed our study with the aim to assess the predictive/prognostic role of p53 abnormal expression in patients with mCRC treated with anti-EGFR therapy by using a validation cohort.

2 Methods

2.1 Patients and Methods

Patients with RAS and BRAF WT mCRC treated with second-line or third-line irinotecan and cetuximab were included in our retrospective analysis. Between May 2011 and May 2018, subjects treated at the University of Cagliari were included in the exploratory cohort, while subjects treated at Istituto Oncologico Veneto constituted the validation cohort. Tumour samples were retrospectively tested for p53 immunohistochemical (IHC) expression with the aim of evaluating the correlation with clinical outcome in terms of overall survival (OS), progression-free survival (PFS), response rate (RR) and disease control rate (DCR).

The analysis of p53 expression were performed on formalin-fixed and paraffin-embedded tumour samples for all patients using an anti-p53 antibody (DO7, prediluted; Leica Biosystems, Newcastle upon Tyne, UK) and the Bond Polymer Refine Detection Kit (Leica Biosystems) on a BOND-MAX automated IHC stainer (Leica Biosystems), as instructed by the manufacturer. The cut-off point for p53 expression was set as 30%, as previously described. In detail, if 30% or more of the malignant nuclei were positive, the tumour was scored as positive, while if fewer than 30% of the nuclei were stained, the tumour was scored as negative (Fig. 1) [78,79,80]. For study purposes, right-sided and left-sided colorectal primary tumours were defined as proximal or distal to the splenic flexure.

Fig. 1
figure 1

Representative p53 immunohistochemical staining in RAS/BRAF wild-type colorectal adenocarcinomas. Diffuse overexpression in a primary colorectal carcinoma (a) and in a hepatic metastatic tumour (b); note the absence of p53 immunostaining in the surrounding normal colon mucosa and hepatic parenchyma. c A p53 overexpressing mucinous colorectal adenocarcinoma. d, e Heterogeneous and low p53 expression in two cases of primary colorectal adenocarcinoma. f A complete loss of p53 expression; note the presence of positive stromal cells that are used as positive controls of the reaction (original magnifications ×5, ×10 and ×20; scale bars 100 μm)

Full size image

2.2 Statistical Analysis

Statistical analysis was performed with the MedCalc Statistical Software Version 14.10.2 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2014). The association between categorical variables was estimated by the Fisher exact test for categorical binomial variables or by the chi-square test in all other instances. Survival probability over time was estimated by the Kaplan–Meier method. Significant differences in the probability of survival between the strata were evaluated by the log-rank test. The independent role of variables that were statistically significant at a univariate analysis was assessed with a logistic regression analysis.

Overall survival was defined as the time interval between the date of the beginning of cetuximab/irinotecan treatment to death or the last follow-up visit for patients who were lost at follow-up. Progression-free survival was defined as the interval between the date of the beginning of cetuximab/irinotecan treatment to death, first sign of clinical progression or the last follow-up visit for patients who were lost at follow-up.

Response rate was defined as the percentage of patients who achieved a partial or complete response to treatment according to RECIST Version 1.1. Disease control rate was defined as the percentage of patients with stable disease or partial/complete response to treatment.

Based on the results from the 88 patients of the exploratory cohort, we tried to validate the findings in a validation cohort. Then, we identified the validation group sample size according to good/poor prognosis group ratio and survival analysis, from the exploratory cohort. To validate the difference in terms of a 12-month rate among poor prognosis patients (around 15%) and good prognosis patients (around 70%), as we obtained in the exploratory cohort (assuming a probability alpha of 0.05 and a beta of 0.10), with a two group ratio of 0.7, the required sample size would have been 32 patients (13 + 19), using a “comparison of proportion test”.

3 Results

3.1 Patient Characteristics

Globally, 120 patients with RAS/BRAF WT mCRC were included in our analysis, 88 in the exploratory cohort, treated at the University of Cagliari and 32 in the validation cohort, treated at Istituto Oncologico Veneto, between May 2011 and May 2018. Patient baseline characteristics are detailed in Tables 1 and 2.

Table 1 Patient characteristics in the exploratory and validation cohorts
Full size table
Table 2 Patient baseline characteristics according to p53 expression
Full size table

In the exploratory cohort as well as in the validation cohort, p53 IHC low expression was 41% (36/88 and 13/32 patients, respectively). p53 IHC overexpression was found in the 59% of patients in both cohorts (52/88 patients in the exploratory cohort and 19/32 patients in the validation cohort).

In the exploratory cohort, 59% of the patients had a left-sided colorectal cancer, 46% of them had p53 low expression and 54% p53 overexpression. Conversely, among patients with right-sided CRC, 30% had p53 low expression and 70% had p53 overexpression (p = 0.15).

The left colon site was the most common in the validation group with 62% of patients. In this subgroup of patients, 48% had p53 low expression and 52% had p53 overexpression. Conversely, among patients with right-sided CRC, 14% had p53 low expression and 86% had p53 overexpression (p = 0.114).

3.2 Clinical Outcomes

At a median follow-up of 10.5 months (95% confidence interval [CI] 9.0–13.7), in the exploratory cohort, we observed a median PFS (mPFS) of 6 months (95% CI 3.8–18.0) and a median OS (mOS) of 10.0 months (95% CI 9.0–13.6). Median PFS was 8.0 months (95% CI 6.9–18.0) vs 3.0 months (95% CI 2.9–11.1) in patients non-overexpressing p53 and in patients overexpressing p53, respectively (hazard ratio [HR]: 0.39; p < 0.0001). Median OS was 19.5 months (95% CI 17.0–30.5) vs 8 months (95% CI 5.9–9.8) in patients non-overexpressing p53 and in patients overexpressing p53, respectively; (HR: 0.23; p < 0.0001) (Fig. 2). Overall RR was 33%. Response rate was 58% vs 15% in patients with p53 non-overexpressing tumours and in patients with p53 overexpressing tumours, respectively (p < 0.0001). Overall DCR was 59%. Disease control rate was 86% vs 40% in patients with p53 non-overexpressing tumours and in patients with p53 overexpressing tumours, respectively (p < 0.0001) (Table 3).

Fig. 2
figure 2

Kaplan–Meier progression-free survival and overall survival in patients overexpressing p53 (dotted line) and in patients non-overexpressing p53 (continuous line) in the exploratory cohort. a Median progression-free survival was 8.0 months (95% confidence interval [CI] 6.9–18.0) vs 3.0 months (95% CI 2.9–11.1) in patients overexpressing p53 and in patients non-overexpressing p53, respectively (hazard ratio 0.39; p < 0.0001). b Median overall survival was 19.5 months (95% CI 17.0–30.5) vs 8 months (95% CI 5.9–9.8) in patients overexpressing p53 and in patients non-overexpressing p53, respectively (hazard ratio: 0.23; p < 0.0001)

Full size image
Table 3 Response in patients of the exploratory and validation cohorts in overall population, in non-overexpressing p-53 tumours and in overexpressing p53 tumours
Full size table

These results were confirmed in the validation cohort, where we observed an mPFS of 7 months (95% CI 4.6–12.0) and an mOS of 15.8 months (95% CI 9.5–28.0). Response rate was 34%, and DCR was 84%. Response rate was 69% in patients with low p53 and 11% in patients with abnormal p53 (p = 0.0007); DCR was 100% vs 74% for patients with low p53 and abnormal p53 tumours, respectively (p = 0.0475) (Table 3). Median OS was 18 months in patients with p53 low expression (95% CI 13.6–28.0) vs 11.46 months in patients with p53 overexpression (95% CI 8.2–18.0; HR: 0.26; p = 0.0027) and mPFS was 8.1 months in patients with low p53 (95% CI 6.5–12.0) vs 5.8 months in patients with abnormal p53 (95% CI 3.0–11.4; HR: 0.48; p = 0.0399) (Fig. 3).

Fig. 3
figure 3

Kaplan–Meier progression-free survival and overall survival in patients overexpressing p53 (dotted line) and in patients non-overexpressing p53 (continuous line) in the validation cohort. a Median progression-free survival was 8.1 months (95% confidence interval [CI] 6.5–12.0) in patients non-overexpressing p53 vs 5.8 months (95% CI 3.0–11.4) in patients overexpressing p53 (hazard ratio: 0.48; p = 0.0399). b Median overall survival was 18 months (95% CI 13.6–28.0) in patients non-overexpressing p53 vs 11.4 months (95% CI 8.2–18.0) in patients overexpressing p53 (hazard ratio: 0.26; p = 0.0027)

Full size image

Regarding the tumour sidedness, patients with left-sided CRC had better outcomes than patients with left-sided CRC in the exploratory cohort, with an mOS of 12.5 months (95% CI 9.9–19.0) vs 8 months (95% CI 5.9–9.8), respectively (HR: 0.5; p = 0.0232). Finally, while not statistically significant, patients with left-sided CRC had better outcomes than patients with right-sided CRC in the validation cohort with 16.0 months (95% CI 13.5–28) vs 8.9 months (95% CI 3.4–9.0), respectively (HR: 0.3; p = 0.067) (Fig. 4).

Fig. 4
figure 4

Kaplan–Meier overall survival in patients with right-sided colorectal cancer [CRC] (dotted line) and in patients with left-sided CRC (continuous line) in the exploratory and validation cohorts. a In the exploratory cohort, median overall survival was 12.5 months (95% confidence interval [CI] 9.9–19.0) vs 8 months (95% CI 5.9–9.8) in patients with left-sided CRC and in patients with right-sided CRC, respectively (hazard ratio: 0.5; p = 0.0232). b In the validation cohort, median overall survival in patients with right-sided CRC (dotted/green line) and in patients with left-sided CRC (continuous/blue line) is shown. Median overall survival was 16.0 months (95% CI 13.5–28.0) vs 8.9 months (95% CI 3.4–9.0) in patients with left-sided CRC and in patients with right-sided CRC, respectively (hazard ratio: 0.37; p = 0.0670)

Full size image

Therefore, p53 expression and sidedness showed a statistically significant correlation with OS at univariate analysis in both cohorts. At the multivariate analysis, in the exploratory cohort, p53 expression (Exp (b) 4.5750, p < 0.0001) and primary tumour location (Exp (b) 2.0820, p = 0.0078) maintained an independent role. In the validation cohort, only p53 expression (Exp (b) 5.0367, p = 0.0050) maintained an independent role.

Furthermore, to validate the prognostic role and the predictive effect of p53 in patients with RAS/BRAF WT mCRC treated with anti-EGFR, we analysed the survival data from the first-line palliative therapy in both cohorts. At a median follow-up of 25.1 months (95% CI 20.5–28.5), in the exploratory cohort, a significant difference in mOS was found between patients non-overexpressing p53 and patients overexpressing p53 [30.2 months (95% CI 26.9–47.9) vs 17.1 (95% CI 13.8–44.9), p < 0.0001, respectively]. These results were confirmed in the validation cohort with an mOS of 34.7 months (95% CI 25.8–40) vs 22.5 (95% CI 17–34), p = 0.0003 (Fig. 5). Finally, a non-statistically significant trend towards better mPFS was found in patients non-overexpressing p53 vs patients overexpressing p53 in the exploratory cohort [11 months (95% CI 10.6–20.9) vs 8 months (95% CI 7.9–20), p = 0.135) respectively] and the validation cohort [13.1 months (95% CI 11.8–20.4) vs 11.7 months (95% CI 8–18.9), p = 0.1933), respectively] (data not shown).

Fig. 5
figure 5

Kaplan–Meier overall survival from first-line palliative therapy in patients overexpressing p53 (dotted line) and in patients non-overexpressing p53 (continuous line) in the explorative (a) and the validation cohort (b). a Median overall survival was 30.2 months (95% confidence interval [CI] 26.9–47.9) vs 17.1 (95% CI 13.8–44.9) in patients with low p53 expression and in patients overexpressing p53, respectively (hazard ratio: 0.28; p < 0.0001) in the exploratory cohort. b Median overall survival was 34.7 months (95% CI 25.8–40) vs 22.5 (95% CI 17–34) in patients with low p53 expression and in patients overexpressing p53, respectively (hazard ratio: 0.17; p = 0.0003) in the validation cohort

Full size image

4 Discussion and Conclusions

The emergence of drug resistance represents the major limitation to the development and use of molecularly targeted cancer therapies. In this scenario, only a few studies tried to identify the prognostic and the predictive role of p53 in patients treated with anti-EGFR therapies.

Preclinical data suggest that both WT and mutant TP53 can modulate EGFR expression in various tumours. Ludes-Meyers et al. showed that WT and mutant TP53 use different mechanisms to activate EGFR transcription. They identified the binding site for WT TP53 in an EGFR promoter gene-specific sequence and observed that the p53 DNA-binding capacity seemed to be related to this gene transcription activation; furthermore, this sequence appeared to be the same target site of positive and negative regulators of EGFR promoter activity. Conversely, mutant TP53 did not require the WT p53-binding site for EGFR promoter transactivation but the oligomerisation domain after recruitment by other transcription factors. As mutant p53 is often overexpressed in tumour cells, this could lead to strong and persistent activation of growth factor genes, such as high EGFR levels and consequent aberrant cell proliferation [72,73,74,75,76]. A further study demonstrated that the loss of p53 in normal human keratinocytes is responsible for increased EGFR expression by a mechanism involving YY1 and Sp1 and does not require p53 binding to the EGFR promoter [77]. In summary, preclinical findings show that p53 modulates EGFR promoter activity both directly by DNA binding and indirectly through other transcription factors.

Other clinical studies evaluated the co-expression of mutant p53 and EGFR mutations/overexpression in different tumours, suggesting a correlation with poor prognosis and a lack of response to tyrosine kinase inhibitors, chemotherapy and radiotherapy in non-small cell lung cancer, head and neck, and hepatocellular malignancies [81,82,83,84].

However, the clinical role of TP53 mutational status in CRC is still controversial. Several studies showed that patients with a tumour harbouring a TP53 mutation have a significantly poorer outcome than patients with WT TP53 tumours. Moreover, p53 overexpression has a negative prognostic impact. Conversely, other findings failed to demonstrate a correlation between p53 overexpression and clinical outcome in patients with CRC [81,82,83]. Theodoropoulos et al. retrospectively evaluated the relationship between p53 and EGFR expression assessing the correlation with clinical/histological prognostic factors and the impact on prognosis or survival in 164 patients with CRC with at least a 5-year follow-up. Overexpression of EGFR and p53 were significantly associated with an advanced T stage, suggesting that both proteins cause a growth advantage in deep invasion and they are a late event in CRC carcinogenesis [85].

In contrast with previous findings, Shyhmin et al. revealed that WT TP53 may enhance sensitivity to EGFR inhibitors and radiation through cell-cycle arrest, apoptosis and DNA damage repair, in CRC and non-small cell lung cancer [86]. The other two studies showed that TP53 mutations may be predictive of cetuximab sensitivity and TP53 genotyping could be useful to optimise the selection of patients with mCRC who should benefit from cetuximab-based chemotherapy [87, 88].

Of note, most pathologists do not have access to TP53 sequencing, and therefore, they use p53 immunohistochemistry as a surrogate for TP53 mutational analysis. Immunohistochemistry is quick, easy to perform and less expensive than sequencing. Hence, p53 immunohistochemistry represents a feasible biomarker already in use in the clinical setting. On this basis, our study aimed to evaluate whether the p53 abnormal expression could influence the clinical outcome of a patient with RAS/BRAF WT mCRC, treated with anti-EGFR.

Globally, our analysis suggests that p53 overexpression may predict resistance to anti-EGFR treatment in patients with mCRC. In the exploratory cohort, a significant benefit in terms of all clinical outcomes (OS, PFS and RR) was observed in p53 WT patients, compared with those showing p53 over-expression. Furthermore, the p53 low expression status was more frequent in left-sided tumours. These results were confirmed in the validation cohort.

Furthermore, we analysed survival data from the first palliative therapy, which did not contain anti-EGFR antibodies. This analysis showed a significant benefit in terms of OS in p53 low expression patients compared with those showing p53 over-expression, in both cohorts. Conversely, no statistically difference in terms of PFS was observed in patients non-overexpressing p53 vs patients overexpressing p53 in the exploratory and validation cohorts. These data confirmed the prognostic role of p53 expression and suggested the absence of a predictive impact to response to therapies that do not contain anti-EGFR.

In the era of precision medicine, p53 analysis could improve the selection of patients who could benefit from anti-EGFR therapy. The identification of the optimal treatment strategy is successfully provided by a better knowledge of the mCRC heterogeneous nature. Therefore, the finding of p53 overexpression as a resistance mechanism to anti-EGFR therapies represents a further step toward personalised treatment.

The limitations of our study are mainly related to its retrospective nature. In particular, this latter feature and the absence of a control arm without EGFR therapy do not allow us to clearly demonstrate that p53 overexpression is associated with anti-EGFR resistance. Still, our analysis suggests a possible predictive p53 role in this setting. In future studies, it would also be interesting to investigate the association between the TP53 mutation, p53 IHC expression and clinical outcomes to confirm the use of p53 IHC, which is quick, easy to perform and inexpensive, as a surrogate for TP53 mutational analysis.

Although retrospective, we believe that our analysis could provide a relevant addition to the biological picture underlying the mechanism of resistance to anti-EGFR monoclonal antibodies in patients with mCRC. Further prospective studies, with a control arm without anti-EGFR therapy, will be needed to validate the prognostic and predictive effect of p53 expression in this setting, to better refine the molecular profile of patients more likely to benefit from this crucial treatment strategy.