The High Expression of p53 Is Predictive of Poor Survival Rather TP53 Mutation in Esophageal Squamous Cell Carcinoma

TP53 is a well-known tumor suppressor gene and one of the most common genetic alterations in human cancers. However, the role of p53 as a prognostic marker of esophageal squamous cell carcinoma (ESCC) is controversial in the association between TP53 alterations and clinical outcomes. To address this issue, we evaluated TP53 mutations, p53 protein expression, clinicopathological parameters, and survivals rates in a large scale of patients with ESCC. Two cohorts were included in this study: TP53 mutations were detected by next-generation sequencing in 316 ESCC patients, and p53 protein expression was tested by immunohistochemistry in 6,028 ESCC patients. Survival analysis was performed using the Kaplan–Meier curve and the Cox proportional hazards model. TP53 mutations were found in ESCC patients from 241 of 316 (76.3%), and the rate of positive expression of p53 protein was 59.1% in 6,028 ESCC patients (including 1819 with high expression of p53 protein), respectively. Most mutations were missense, which has a high expression of p53 protein. Compared with wild-typeTP53, TP53 gene mutations were not significantly associated with survival time (p=0.083). In multivariate analysis, the p53 protein expression was an independent prognostic factor for ESCC. The high-expression group of p53 protein has poor survival (p < 0.001) compared to low-expression group in patients with ESCC. The high expression of the p53 protein, not the TP53 mutation, is predictive of poor survival in patients with ESCC, and p53 protein expression might have the potential to be a prognosis biomarker and therapy target in ESCC.


Introduction
Esophageal carcinoma is one of the most aggressive cancers and the sixth leading cause of cancer death [1]. Esophageal cancer (EC) is the fourth most common malignancy associated with cancer-related death, and esophageal squamous cell carcinoma (ESCC) is the most common pathological subtypes (>90%) in China [2][3][4][5]. With limited early clinical diagnosis approaches and few targeted therapies, although the fve-year survival rate has improved during past decades, it still remains dismal, only at 10%-30% in most countries [6]. Te most efective way to improve the survival rate of ESCC patients is through early detection and treatment. Terefore, the discovery of molecular markers for early screening, prognosis, and efcacy of evaluation to provide personalized treatment for ESCC patients is one of the entry points to reduce the incidence and improve the survival rate of patients with ESCC.
TP53 is one of the most frequently mutated genes in human cancers and has occurred in more than 50% of all human cancers [7,8]. A recent study analyzed the TP53 mutational spectra of 7,525 pan-cancer tissues and found TP53 mutations in 35% of all 30 tumor-type samples, of which the most mutated cohorts are more than 80% and the lowest percentage of TP53 mutation was less than 1% [9]. Even in the same tumor, the heterogeneity including differences in histology, molecular subtype, pathogenic factors, tumor stage, and degrees of diferentiation can also afect the evaluation of the frequency of TP53 mutations [10]. Majority mutations of TP53 cause the loss of function (LOF) of wild-typeTP53 and abrogate their ability to bind on specifc DNA motif and perform its tumor-suppressive function. Gain of function (GOF), dominant negative efect on the wild-typeTP53 allele, the loss of heterozygosity of TP53, and interactions with viral proteins also result in the activity loss of wild-typeTP53 [11,12].
Recent studies using whole-genome and exome sequencing in patients with EC revealed that the most mutated gene is TP53, which means that at least TP53 has a signifcant infuence on EC pathogenicity [13,14]. In general, TP53 mutations often cause changes on the amino acid sequence of the p53 protein, thus disrupting the function of p53 for tumor inhibition. Under normal conditions, it is difcult to detect the expression of wildtype p53 protein due to its short half-life. Terefore, we usually detect the expression level of p53 protein (mutanttype) using immunohistochemistry (IHC) in clinical diagnosis, but in ESCC, the positive rate of p53 protein expression varies greatly in diferent reports, ranging from 27 to 75% [15]. At present, the role of p53 as a prognostic marker of ESCC is controversial. To date, the mechanism and correlation between the expression of the p53 protein and mutation of the TP53 gene or prognostic efect in ESCC are limited and contradictory. Te clinical value of TP53 mutation and p53 protein expression is worth further exploration.
Herein, we conducted an analysis of TP53 mutation and p53 protein expression in two-large ESCC cohorts, using a series of methods including whole-genome sequencing (WGS), whole exon sequencing (WES), regionally targeted sequencing (TRS) and IHC, to further clarify the association between TP53 mutation status/ the expression of p53 protein and clinicopathologic phenotype and prognosis. Tis study provided efective and reliable evidence showing TP53 alterations as a biomarker for clinical diagnosis and treatment in ESCC.

Patients and Samples.
All of the patients and samples were selected from the tissue bank and database of about 500,000 esophageal and gastric cardia carcinomas , established by the State Key Laboratory of Esophageal Cancer Prevention & Treatment and the Henan Key Laboratory for Esophageal Cancer Research of Te First Afliated Hospital, Zhengzhou University [16][17][18]. Te database was reviewed to select the present study cohort. All medical records were collected, including detailed clinical, pathological, and survival information. Carcinoma and adjacent noncancerous tissues of 335 patients, which included frozen tissue samples and parafn tissue samples (for using WGS, WES, and TRS) were collected after surgery. Postoperative tumor parafn-embedded tissues from 6,252 patients were used for producing tissue microarray (TMA). During the cases of screening, patients were excluded from the database according to the following standard: non-ESCC patients, lacking T, N, M stage, preoperative treatment, incomplete following-up, without tissue for IHC staining in TMA, or failed staining. Te detailed fow chart for patient enrollment is shown in Figure 1.
For the frst cohort, a total of 316 surgically resected ESCC tissue specimens (paired primary malignant and adjacent normal) were collected. Te tissues were analyzed on a diferent platform: 316 paired primary malignant and adjacent normal tissues, including 19 cases with frozen tissues that had WGS data available; 92 samples (51 cases of frozen tissue and 41 cases of parafn-embedded tissues) that had WES data available; and 205 samples (203 cases of frozen tissue and 2 cases of parafn-embedded tissues) had TRS data available. Among the 316 patients, 276 were used to make TMA. Te remaining 40 had no tumor specimen available (owing to a lack of specimen or insufcient tumor cells). Te second cohort that contains 6,028 from the 6,252 patients with ESCC TMA was included in the fnal statistical analysis.
All the patients enrolled for this study were staged using the Union for International Cancer Control (UICC) staging standards, 6th (2002), for esophageal cancer. All patients were followed up after diagnosis until the date of death or December 2019.

Genomic DNA Extraction.
Before DNA extraction, tissue was stained with hematoxylin and eosin (H&E) in order to assess the accurate histopathology for each case. Te samples confrmed by a pathologist and in which tumor cells accounted for ≥50%, were chosen for DNA extraction. DNA was isolated from frozen tissue and formalin-fxed, parafnembedded (FFPE) tissue. Frozen tissue specimens were collected during surgery, snap-frozen in liquid nitrogen, and stored at −80°C. DNA was extracted from frozen tissue using the phenol-chloroform protocol [19]. For FFPE tissue, DNA was extracted from 8 sections with a 10 μm thickness of each parafn block using the QIAamp DNA FFPE tissue kit (Qiagen, Hilden, Germany).

Immunohistochemical Staining for p53 Protein
Expression. IHC staining of 4 μm TMA sections was performed by a 2-step protocol using a p53 antibody (1 : 100 dilution) and DAB detection kit (both from Wuhan Servicebio Technology Co., Wuhan, China). In each experiment, both positive and negative controls were included. All images were captured by CaseViewer 2.2 for Windows (3DHISTECH, Budapest, Hungary, Figure 2).

Establishment of Scoring Criterion for p53
Immunohistochemical Staining. Te scoring of IHC staining was completed independently by two experienced pathologists. Staining location, intensity, and patterns were reviewed. Tumor cells having a dark brown precipitate in their nuclei were the criterion for a positive reaction. Te intensity of staining was grouped into four grades: 0 � entirely negative; 1 � weak; 2 � moderate; 3 � strong. Te immunostaining patterns were divided into four terms: 0 � entirely negative; 1 � scattered, meaning only some isolates were positive cells; 2 � focal, where clusters of positive cells were seen in some areas; 3 � difuse, in which the sheets of positive cells were found throughout most of the areas (Figure 2) [20]. Te fnal results were multiplied by the scores of the immunostaining patterns and staining intensity. Patients were categorized as "high expression (>4)" or "low expression (0∼4)" by the use of IHC scoring criteria.
2.6. Statistical Analyses. Statistical analysis was processed by SPSS for Windows, version 25.0. Te T-test and chi-square test or Fisher exact test were used to compare the association of categorical and continuous variables between diferent groups, respectively. Te Kaplan-Meier method analyzed survival tendency and used the log-rank test to compare the survival curves. Cox proportional-hazard models were used for the univariate and multivariate analyses to estimate the hazard ratio of each clinicopathological feature for overall survival (OS). All predictors with p value <0.1 in univariate Cox were selected in multivariate Cox analysis. p values were 2-tailed and considered statistically signifcant with less than 0.05.

Te Clinicopathological Distributions of ESCC Patients
Results. To discover the genetic alteration of TP53 in ESCC, we collected tumor samples to set up two cohorts, including 316 patients for sequencing analysis (276 patients out of 316 patients performed IHC to detect the expression of p53 protein), and 6,028 patients for the expression of p53 protein in TMA, for investigation in this study ( Figure 1). Detailed clinicopathological data are listed in Figure 3 and Table 1. Te follow-up period of all those patients ranged from 0.08 years to 30.87 years, and the median survival time was 2.98 years.  Figure 1: Flow chart of patient selection. Patients who had TP53 mutations by sequencing (a) and p53 protein expression by IHC (b) were included on the basis of clinical and histopathological characteristics and survival status. * the TP53 mutation analysis was included in 316 patients. # the p53 protein expression analysis was included in 276 patients.     (Tables 2 and 3). TP53 mutations were mainly located in the exon 5-8 (79.4%), meaning that most mutations occurred in the DBD, and very few mutations occurred in the AD1, AD2, and TET domains. Mutations were mostly clustered in exon 5 (23.6%) and exon 8 (23.6%), followed by exon 6 (19.2%) (Figure 4(a)).

Correlation of p53 Protein Expression with TA Class and Align GVGD Classifcations.
To speculate on the efect of the protein function of the TP53 missense mutation sites, we submitted queries to the IARC TP53 Database (https://p53. iarc.fr) [23][24][25]. According to TA classifcation, 145 TP53 missense mutation sites were divided into three categories: functional (4/145, 2.7%), partially functional (11/145, 7.6%), and not-functional (130/145, 89.7%) based on the TA class of the protein function (Table 4). We further observed that four of the 145 sites (in 135 patients) with missense mutations didn't afect the function of the wild-p53 protein. However, only one of the four sites was a single mutation, and its p53 protein expression was negative. Te other three sites were multiple mutations, and their p53 protein expression was positively detected (two were high, and one was low). Among the 130 sites (126 patients) with not-functional (108 patients with p53 protein expression tests), 105 patients were positive p53 protein expression (105/108, 97.2%), including 74 patients with p53 protein high expression (74/108, 68.5%). Te p53 protein expression level is consistent with the predicted function based on the mutation sites. In the Align GVGD classifcation, we did not observe a signifcant diference in the positive rate of p53 protein in each group.

Association between TP53 Mutation and Protein Expression with Clinicopathological Parameters and Survival in ESCC.
In the frst cohort, we did not fnd any remarkable association between the TP53 mutation and these clinicopathological characteristics in ESCC (Table 2). Although there was no statistically signifcant diference in survival time when comparing the patients with and without TP53 mutations ( Figure 5(a)), surprisingly, the results showed that the patients who had high protein expression of p53 exhibited signifcantly worse survival than those with low expression when the population narrow down to 276 patients (p � 0.002, Figure 5(b)). Te median survival time for 108 patients with high protein expression of p53 and for 168 patients who had low expression was 2.79 ± 0.63 and 5.27 ± 0.52 years, respectively. In addition, no signifcant diference was observed in the survival time of diferent mutation types of TP53 (p > 0.05, Figure 5(c)), and no signifcant diference was observed in survival time between the hotspot and nonhotspot in TP53 missense mutations patients (p > 0.05, Figure 5(d)).
To better understand the factors which contribute to the association between TP53 mutations/high p53 expression in patients and survival, we compared the mutation types of p53 patients with ESCC. Te high expression of the p53 protein was 66.9% among the 121 patients with TP53 gene missense mutations and only 5.6% in the 36 cases with nonsense mutations. Other types of mutations in the 55 patients (including frame shift, silent, and splice) were 3.6%. Apparently, missense mutation was most likely the cause of p53 protein mutation, leading to the high expression of p53 protein. Other mutation types were possible causes of low expression of the p53 protein. Surprisingly, in the 64 cases of TP53wild-type, high expression of p53 protein accounted for 35.9%, while low expression was 64.1% (41 patients, of which 26 were not expressed at all) (Figure 4(e) and Table 5).  (Table 5). Furthermore, the total survival time of the four groups was signifcantly diferent (p < 0.01), among which the TP53 mutation/p53 high expression group had the worst survival time, followed by the TP53 wild-type/p53 high expression group ( Figure 5(e)).

Association between p53 Protein Expression and Clinicopathological Changes in Cohort with Large-Scale ESCC
Patients. To further validate our discovery of the association between the p53 expression-related prognosis and clinicopathological changes, we assessed additional cohorts to make an analysis (Table 6). Te positive expression of p53 protein was observed in 3,562 (59.1%) ESCC patients, of which 1,819 patients (51.1%) showed high expression of p53 protein. Moreover, the high expression of p53 protein in high-incidence areas was more common than in lowincidence areas with a rate and p value of 31.5% vs. 27.7%, p � 0.003. High expression of p53 protein was closely correlated with poor tumor diferentiation (p < 0.001). Te frequency of high expression of p53 protein in the early ESCC (0 + I stage) was 1.27-fold higher than that in the advanced ESCC (38.2% vs. 30.0%, p � 0.034). Furthermore, the rate of positive cancer embolus in high-expression groups was 1.31-fold higher than in low-expression groups (5.1% vs. 3.9%, p � 0.028).

Independent Prognosis Marker Role of p53 Protein on Survival Analysis in 6,028 ESCC Patients.
To determine whether a high expression level of the p53 protein could be used as a prognosis marker in patients with ESCC, we conducted the analysis using the p53 expression in 6,028 ESCC patients. In this cohort, the analysis showed median OS time in patients with p53 high expression and with low was 2.71 ± 0.09 and 3.08 ± 0.06 years, respectively. Te presence of high expression of p53 protein was signifcantly associated with decreased OS (p < 0.001, Figure 5(f )).

Cox Univariate and Multivariate Regression Analyses.
Cox proportional-hazard models were chosen for the univariate and multivariate analyses. Te univariate Cox regression analysis demonstrated that the sex, age, high/low incidence area, cigarette smoking, alcohol consumption, location, diferentiation, T stage, N stage, M stage, UICC stage, cancer embolus, and p53 protein high/low expression were dependent prognostic factors (p < 0.05, Figures 6(a), 6(b), and 7). In multivariate Cox regression analysis, as compared with the low expression of p53 protein, the patients with high expression (HR � 1.134, 95% CI 1.065 to 1.207) were associated with decreased survival after adjustment for the age, high/low incidence area, cigarette smoking, diferentiation, and T and N stages as independent prognostic factors (p < 0.05, Figure 6(c)).

Stratifed Survival Analysis by within Independent Prognostic Factors.
We performed a stratifed analysis to eliminate the infuence of the independent prognostic factors on evaluating the expression of the p53 protein on prognosis. When patients were stratifed according to high/ low incidence area, cigarette smoking, and N stages, the  Note. When a missense mutation occurs, patients with multiple mutations are preferentially included in the missense mutation group. When a nonsense mutation occurs, the remaining patients are preferentially included in the nonsense mutation group.    consistent trend of OS was observed in diferent stratifcation (Figures 8(a)-8(f)). When stratifed according to patients' age at diagnosis, degree of diferentiation, and T stage, we only found that patients the high expression of p53 protein in >60 (p < 0.001, Figure 8(g)), medium and low diferentiation (p < 0.001, Figures 8(h) and 8(i)), and T3 (p < 0.001, Figure 8(j)) had shorter OS than those with low expression.

Discussion
We have conducted a two-large ESCC cohort analysis to evaluate the association between TP53 mutation statuses/the expression of the p53 protein and clinicopathological features and prognosis in ESCC patients. Most importantly, our results further confrmed that the expression of the p53 protein can refect the prognosis of ESCC patients by using a large number of tissue samples, and high p53 expression indicates a poor prognosis. However, the mutation of the TP53 gene has no obvious correlation with prognosis, which is controversial in previous reports.
With the rapid development of sequencing technology, the prognostic value of TP53 has been confrmed in a variety of tumors, and the mutation of TP53 indicates a poor prognosis across diferent types of human cancers. In this study, despite the high frequency of TP53 mutations, there is no obvious evidence showing its association with clinicopathological features and prognosis in patients with ESCC. Previously, other colleagues showed that TP53 status was correlated with tumor invasion depth, TNM stage, lymph node metastasis, distant metastasis, and diferentiation degree [15]. Meanwhile, the survival time of TP53 gene mutation and p53 protein overexpression was shorter than the control group in ESCC [15]. However, Zhao et al. showed that the high expression of p53 protein was an independent prognostic factor for survival, while the mutation of the TP53 gene was unrelated to prognosis [26]. Te conficting results might be caused by the following reasons: of the limited number of patients, insufcient clinical follow-up, diferent experimental techniques, mutation sites on different target exons, and variable factors.
We further found that the frequency of the TP53 mutation was 76.3% in ESCC, which was between 66.7% and 82.7% of the frequency of the TP53 mutation in EC by NGS, as previously reported [27][28][29][30]. Cui et al. published WGS results for 508 cases of ESCC, which also showed a similar TP53 mutation frequency of 74.8% [30]. Te cause might be that the study population and histopathological type were diferent. Our study confrmed that the main mutation type of TP53 in ESCC patients is a missense mutation, which may play a pivotal role in tumorigenesis because the p53 protein usually has positive expression, or even higher expression, owing to the accumulation of a nonfunctional protein that loss activity as a tumor suppressor, and some of which exert trans-dominant repression over the wild-type counterpart [31,32]. C > T/G > A Transition is the predominant mutation of TP53 gene mutations in our study, and it is a typical marker associated with betel quid chewing, tobacco use, and alcohol drinking in oral squamous cell carcinoma in Taiwanese [33]. C > A/G > T transversion is a typical feature of carcinogen exposure associated with cigarette smoking, which is the most risk factor for ESCC and lung cancer. Tis transversion occurred at the sites of adduct formation for the metabolites of benzo (a) pyrene, a major tobacco carcinogen (codons 157, 248, and 273) [34,35].
To our knowledge, this is the largest sample-scale study that has examined p53 protein expression by IHC. Our study herein showed that the expression of p53 protein was associated with high/low incidence area, degree of diferentiation, and cancer embolus, and revealed that high expression of p53 protein is an independent prognostic factor rather than TP53 gene mutation, with 1.134-fold mortality risk. In previous studies, the correlation between p53 protein expression, clinicopathological features, and prognosis was signifcantly diferent [36][37][38][39][40]. To date, only 13 of the 30 articles have suggested that p53 protein was an adverse factor for prognosis, and the rest of the reports suggested that p53 expression had no efect on prognosis [39]. Wang et al. revealed that a more advanced TNM stage, positive lymph node metastasis, and distant metastasis were associated with p53 protein expression [41]. Although there were no signifcant diferences between p53 expression and T stage, N stage, and M stage in our fnding, our further validation of 6,028 patients demonstrated a signifcantly strong correlation between the expression of p53 protein and clinical phenotype and prognosis. Te judgment for the p53 protein expression is diferent because of the variety of using antibodies and experimental methods used in IHC. In addition to the source of the tumors and the existence of tumor heterogeneity, the diference in evaluation criteria is the most critical factors leading to inconsistent conclusions in previous studies [36,37,40].
Our study showed that the expression of p53 protein in the high-incidence areas of EC was signifcantly higher than in low. Te existence of high/low incidence areas of EC is one of the prominent epidemiological features of ESCC, suggesting that environmental factors play an important role in the pathogenesis of ESCC. In high-incidence areas of EC, exposure to environmental carcinogens (such as nitrite and mold-contaminated foods) may lead to carcinogenesis of the esophageal squamous epithelium. TP53 is one of the most susceptible genes to environmental carcinogens in the process of tumor induction. Afatoxin B1 (AFB1) are potent carcinogens, which induces an arginine to serine at a single base substitution at the third base of codon 249 in TP53 (G > T transversion) [42,43]. N-nitroso compounds (NOC) are an alkylating agent and cause guanine alkylation to generate O 6 -alkylguanine, which results in C > T/G > A transition during DNA replication when paired with thymine [44][45][46]. Te DNA repair protein O 6 -alkylguanine-DNA alkyltransferase (AGT) specifcally repairs O 6alkylguanine adducts in DNA. However, AGT mutations occurred more frequently in patients in high-incidence areas of esophageal cancer than in the normal population [47].
In ESCC, it is not feasible to determine the presence of TP53 gene mutations through the loss of p53 protein expression. When the TP53 gene is mutated, it will cause diferent changes in proteins. Nonsense, frameshift, silent, and splice mutations usually cause protein truncation and 30 Journal of Oncology  make the synthesis of normal p53 protein interfere in order to cause LOF, and p53 protein is not expressed either [21]. Te situation became complicated when a missense mutation occurred because the p53 protein can be caused by various missense mutations. Some will still have the function of wild-type p53 protein, and some will only retain part of the function and acquire gain GOF, which is associated with malignancy, invasion, and metastasis [48]. Due to the broad existence of LOF and GOF of the TP53 mutation, falsepositive or false-negative levels of p53 protein will be detected by using IHC in clinical applications, which might not be a reliable method for evaluating the function of the TP53 mutation [49]. Our observation confrmed that immunohistochemical detection of p53 protein expression could more efectively refect the relationship with prognosis and clinicopathologic features than the mutation status of the TP53 gene in ESCC, which was further supported by studies with a larger sample. Te limitation of this study is the representativeness and accuracy of the application of TMA-based IHC showing the expression of p53 protein with extreme heterogeneity. To conclude, laboratory techniques need to be used to detect p53 protein in TMAs, and the criterion of IHC evaluation must be constant. Limited tissue sampling cannot efectively refect the real status of the whole section because of tumor heterogeneity. Our research simultaneously detected p53 protein in a part of the original pathological sections made TMAs to reduce potential limits, and used the same scoring criteria for judgment. Te results are no diferent from those in TMAs.
In conclusion, TP53 is the most mutated gene in ESCC, showing great potential to be a diagnostic biomarker in treating cancer. However, the role of TP53 mutation and expression of p53 in ESCC still remains unclear. Our study, involving two large-scale cohorts, was the frst to conduct genomic profling and expression of the p53 protein, while demonstrating the association with clinical phenotypes and prognosis. Our fndings show that the expression of p53 protein is more efective in predicting clinicopathological features and prognosis, is a valid biomarker of an unfavorable prognosis, and will contribute to clinical diagnosis, prognosis prediction, and targeted therapy.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon reasonable request.

Ethical Approval
Te study protocol was approved by the Research Ethics Committee of Te First Afliated Hospital of Zhengzhou University.

Consent
All patients provided written informed consent to participate.  Figure 8: Kaplan-Meier curves survival analysis of ESCC patients with high/low p53 protein expression after stratifcation based on independent prognostic factors. OS of patients with p53 protein high vs. low expression in patients with high-incidence area (a) and lowincidence area (b). OS of patients with p53 protein high vs. low expression in patients with (c) and without cigarette smoking (d). OS of patients with p53 protein high vs. low expression in patients with N0 (e) and N1 stage (f ). OS of patients with p53 protein high vs. low expression in patients >60 years (g). OS of patients with p53 protein high vs. low expression in patients with moderate (h) and poor diferentiation (i). OS of patients with p53 protein high vs. low expression in patients with T3 stage (j).