Quantitative TP53 circulating tumor DNA predicts tumor response and survival outcomes in esophageal squamous cell carcinoma

Abstract

BACKGROUND

Reliable biomarkers for esophageal squamous cell carcinoma (ESCC) are critically needed. While circulating tumor DNA (ctDNA) shows promise, its utility in ESCC requires further validation.

AIM

To evaluate ctDNA as a dynamic biomarker for real-time tumor burden assessment and prognostic stratification in ESCC.

METHODS

Matched plasma, leukocyte, and tumor tissues from 63 ESCC patients were analyzed. We employed a tumor-informed, TP53-specific ctDNA strategy, focusing on quantitative TP53 assessment to correlate postoperative ctDNA with clinical outcomes.

RESULTS

TP53 ranked as the gene most commonly found to be mutated. Preoperative TP53 ctDNA levels correlated with advanced T, N, and tumor-node-metastasis stages. Detection of TP53 ctDNA after surgery was associated with markedly poorer disease-free and overall survival. In neoadjuvant chemotherapy patients, post-treatment ctDNA status perfectly associated with pathological response. Multivariate analysis confirmed high preoperative TP53 ctDNA as an independent risk factor for recurrence. A tumor-agnostic analysis showed no prognostic value.

CONCLUSION

Quantitative TP53 ctDNA is a promising biomarker for real-time tumor response assessment and prognostic stratification in ESCC.

Key Words: Circulating tumor DNA; TP53; Esophageal squamous cell carcinoma; Tumor response; Variant allele frequency

Core Tip: Based on a study of 63 esophageal squamous cell carcinoma patients, quantitative analysis of TP53 circulating tumor DNA may help predict postoperative recurrence, assess response to neoadjuvant chemotherapy, and stratify patient survival, suggesting its potential value as a noninvasive biomarker for longitudinal monitoring.

INTRODUCTION

Globally, esophageal cancer ranks among the most common malignancies, with high incidence and mortality rates[1]. In China, esophageal squamous cell carcinoma (ESCC) accounts for approximately 90% of esophageal cancer cases, which is a distinct epidemiological feature from the adenocarcinoma-predominant pattern observed in Western populations[2]. For locally advanced ESCC, current standard-of-care includes neoadjuvant chemotherapy (NCT) or neoadjuvant chemoradiotherapy followed by surgery[3,4]. Although the CROSS trial demonstrated that neoadjuvant chemoradiotherapy improves R0 resection rates and reduces local recurrence, it did not reduce the rate of pure distant metastasis[5]. This finding suggests that a subset of patients may harbor intrinsically aggressive disease or exhibit resistance to current treatment regimens, leading to unfavorable outcomes regardless of aggressive multimodal therapy. These limitations underscore the urgent need for biomarkers that can accurately stratify patients, predict treatment response, and enable real-time monitoring throughout the perioperative period.

Currently, ESCC management relies primarily on imaging, endoscopy, and serum tumor markers. However, these modalities have inherent limitations across the perioperative continuum. For preoperative staging, computed tomography (CT) demonstrates suboptimal sensitivity and specificity for detecting lymph node metastases and assessing depth of tumor invasion[6]. During neoadjuvant therapy, conventional imaging using response evaluation criteria in solid tumors (RECIST) is often inadequate for accurately assessing pathological response[7], as it fails to distinguish between viable tumor, fibrosis, and inflammation. Postoperatively, conventional serum tumor markers such as squamous cell carcinoma antigen[8] and carcinoembryonic antigen lack sufficient sensitivity for early detection of residual disease and recurrence. These gaps highlight the critical need for dynamic, non-invasive biomarkers capable of guiding clinical decisions from diagnosis through long-term follow-up.

Circulating tumor DNA (ctDNA), secreted from tumor cells[9], carries genetic alterations consistent with primary tumors[10], which correlate with the tumor type, load, and degree of progression of the disease[11], has a short half-life, allows for real-time monitoring[12], and can predict cancer occurrence better than traditional methods[13]. Currently, the clinical application of ctDNA detection technologies, which include polymerase chain reaction, next-generation sequencing, and whole-genome sequencing[14], involves a trade-off between sequencing efficacy and cost. Furthermore, despite emerging evidence of ctDNA in solid tumors[15-21], its clinical utility in ESCC-specific therapeutic monitoring remains underexplored.

The objective of this study was to evaluate whether ctDNA can serve as a dynamic biomarker for real-time monitoring of tumor burden and for stratifying prognosis in ESCC patients.

MATERIALS AND METHODS

Patients

This was a prospective observational study conducted at the Department of Thoracic Surgery between August 2015 and June 2020, enrolling 63 consecutive patients with ESCC. The inclusion criteria were as follows: (1) Histologically confirmed ESCC; (2) Surgical resection; (3) No prior antitumor therapy before initial blood sampling; (4) Absence of other malignant tumor history; and (5) Eastern Cooperative Oncology Group Performance Status score of 0-1. The exclusion criteria were as follows: (1) Insufficient plasma or tissue sample quality for sequencing; (2) Loss to follow-up immediately after surgery; (3) Refusal to participate in follow-up assessments; and (4) Diagnosis of other synchronous malignancies. This was an exploratory prospective cohort study, and the sample size was determined based on the number of consecutive eligible patients who underwent surgery during the study period. No formal sample size calculation was performed prior to enrollment. Of the 63 enrolled patients, 2 were excluded due to immediate loss to follow-up, leaving 61 patients for 5-year survival analysis. Loss to follow-up was treated as censoring in Kaplan-Meier and Cox regression analyses. The study protocol received approval from the Medical Ethics Committee of Shanghai Changhai Hospital, approval No. CHEC2020-021. Prior to enrollment, all participants were fully informed about the study’s objectives, methods, potential risks, anticipated benefits, and their rights, and subsequently gave written informed consent. The investigation adhered to the principles of the Declaration of Helsinki and applicable ethical regulations. Baseline clinicopathological characteristics including age, sex, tumor-node-metastasis (TNM) stage (American Joint Committee on Cancer 8^th edition), tumor differentiation grade, and pretreatment serum tumor markers were systematically recorded for all participants. Patients underwent regular follow-up assessments every 3 months during the first 2 years post-treatment and, every 6 months thereafter. Follow-up evaluations included physical examination, contrast-enhanced CT imaging, and serum biomarker profiling. Disease progression was documented according to the RECIST 1.1 criteria.

Next-generation sequencing

Blood was collected at multiple time points, and plasma cell-free DNA (cfDNA), leukocyte genomic DNA, and tumor tissue DNA (tDNA) were sequenced across a panel of 61 tumor-associated class I and class II genes (Supplementary Table 1) using capture-based targeted next-generation sequencing technology. All sequencing procedures were conducted at the Shanghai Yunsheng Medical Laboratory, using established protocols. For tDNA analysis, samples were fragmented by ultrasonication to 200 bp fragments. Libraries were prepared from 100 ng of input DNA using the KAPA HyperPrep Kit (KAPA Biosystems, United States), according to the manufacturer's specifications. Target enrichment was performed using a custom 61-gene oncology panel, followed by 101 bp paired-end sequencing on an Illumina HiSeq 2500 platform (Illumina, United States). Analysis of ctDNA was performed according to our previously reported methods[22].

Statistical analysis

Categorical variables were presented as n (%). Continuous variables with a normal distribution were expressed as mean ± SD, while non-normal data were reported as median [interquartile range (IQR)]. Comparisons of measures between groups were made using the independent samples t-test if they were normally distributed and the variance was homogeneous and the non-parametric rank sum test for non-normally distributed data. Comparisons of counts between groups were made using Fisher’s exact test or Kruskal-Wallis test. Survival analyses were performed using Kaplan-Meier curves (log-rank test) and Cox proportional hazards regression for multivariate adjustment. All statistical analyses and variant allele frequency (VAF) visualizations were generated with X-tile (Yale University, New Haven, CT, United States), GraphPad Prism 9 (GraphPad Software, San Diego, CA, United States) and R 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided, and differences were considered statistically significant when the P value was < 0.05.

Bias control

To minimize potential bias, laboratory personnel performing next-generation sequencing were blinded to clinical outcome data. Plasma and tissue samples were processed using standardized protocols, and ctDNA analysis was conducted without knowledge of patients’ pathological staging or treatment response. All survival endpoints were predefined before statistical analysis.

RESULTS

Clinical and pathological characteristics of the study cohort

The study workflow, detailing patient enrollment, sample collection, and the specific analytic subsets, is outlined in Supplementary Figure 1. The cohort comprised 63 ESCC patients (median age 65 years, IQR 60.5-70.0; 81% male) with median body mass index 23.7 kg/m² (IQR 22.1-24.9). Risk factor distribution included: 41% with both smoking/alcohol history, 17% smokers only, 5% drinkers only, and 37% with neither habit. Tumor characteristics revealed: 63% mid-thoracic location, median tumor size 3.5 cm (IQR 2.4-4.5), 70% moderately differentiated, with 22% showing angiolymphatic and 21% neural invasion. Treatment approaches included Ivor-Lewis surgery (78%) and NCT (14%, taxol-platin regimen) (Supplementary Table 2).

Characterization of mutations in tumor tissue and ctDNA

Using a 61-gene panel, we performed parallel sequencing of plasma cfDNA and matched leukocyte genomic DNA in 63 patients at two timepoints: Preoperatively and 1-week postoperatively. Additional analyses included: (1) Plasma cfDNA from NCT patients prior to treatment initiation; and (2) tDNA from non-NCT patients. TP53 emerged as the most frequently mutated gene across all timepoints. All postoperatively collected ctDNA samples showed no additional genetic mutations compared to their preoperative counterparts. Among the 54 treatment-naïve patients, 39 patients with validated tDNA results (Supplementary Figure 2). Of these, 31 (79.5%) showed perfect concordance between preoperative ctDNA mutations and corresponding tumor tissue variants, demonstrating high sensitivity for tumor-derived mutations. The remaining 8 cases exhibited discordant mutations between preoperative ctDNA and matched tDNA (Supplementary Table 3). We postulate that the tumor tissue analyzed in these cases might not have represented the predominant tumor clone, potentially due to intratumoral heterogeneity or sampling bias during tissue acquisition.

Tumor-informed and tumor-agnostic ctDNA in ESCC prognosis and burden

Association between ctDNA and the prognosis of ESCC: Among the 31 ESCC cases with matched preoperative ctDNA-tDNA profiles, a tumor-informed approach was implemented for subsequent analysis. Patients with undetectable postoperative ctDNA demonstrated significantly prolonged median disease-free survival (DFS) compared to ctDNA-positive cases [33.96 months vs 5.98 months; hazard ratio (HR) = 3.046, 95% confidence interval (CI): 1.245-7.453; P = 0.010]. Although a trend toward improved overall survival (OS) was observed (45.11 months vs 20.86 months), this difference did not reach statistical significance (P = 0.075) (Figure 1A and B).

Figure 1 Kaplan-Meier survival analyses comparing circulating tumor DNA status in patients. A: Disease-free survival stratified by postoperative tumor-informed circulating tumor DNA (ctDNA) status; B: Overall survival stratified by postoperative tumor-informed ctDNA status; C: Disease-free survival according to tumor-agnostic postoperative ctDNA status; D: Overall survival based on tumor-agnostic ctDNA detection. Log-rank tests were used for all comparisons. HR: Hazard ratio; CI: Confidence interval; ctDNA: Circulating tumor DNA.

In our tumor-agnostic analysis of 61 patients with complete 5-year follow-up, classification based on class I/II variants identified 27 postoperative ctDNA-positive cases. However, neither 5-year OS nor DFS rates differed significantly between ctDNA-positive and negative groups (P > 0.05) (Figure 1C and D). These findings suggest that in ESCC, not all ctDNA-detected variants identified through tumor-agnostic approaches exert clinically significant prognostic impact.

Association between tumor-agnostic ctDNA and the burden of ESCC: Under this analytical strategy, analysis of preoperative ctDNA VAF in 63 patients revealed no significant associations with pathological stages (P > 0.05 for all), indicating that tumor-agnostic ctDNA burden was not correlated with disease stage in this cohort (Supplementary Figure 3).

TP53 ctDNA in ESCC burden

TP53 ctDNA VAF before and after surgery: Quantitative analysis of TP53 ctDNA in the 54 patients without NCT patients revealed significant reductions in VAF following surgical resection (median preoperative: 0.11% vs postoperative: 0.00%; P < 0.001) (Figure 2A). In the NCT subgroup (n = 9), median variant allele frequency (VAF) were 0.23% (pre-NCT), 0.00% (post-NCT), and 0.00% (postoperative), with no statistically significant differences observed (P = 0.395) (Figure 2B), likely due to the small sample size.

Figure 2 Dynamics of TP53 circulating tumor DNA variant allele frequency in esophageal squamous cell carcinoma patients. A: Postoperative TP53 ctDNA variant allele frequency was significantly reduced compared to preoperative levels; B: In the neoadjuvant chemotherapy (NCT) subgroup, no significant differences were observed in TP53 ctDNA levels across treatment timepoints (pre-NCT vs post-NCT vs post-surgery); ctDNA not detected: Shown as 0.01% on log scale. VAF: Variant allele frequency; ctDNA: Circulating tumor DNA.

Association of TP53 ctDNA with the pathological stage: Stratified analysis of preoperative TP53 ctDNA VAF in 63 patients revealed significant associations with pathological stages (including post-neoadjuvant therapy stages). T-stage stratification showed median VAF of 0.00% (T1, T2) and 1.09% (T3) (P = 0.002) (Figure 3A). Similarly, N-stage classification demonstrated progressively higher VAF: 0.05% (N0) and 0.54% (N+) (P = 0.034) (Figure 3B). TNM staging analysis confirmed this trend with 0.05% (I, II) and 0.54% (III, IV) (P = 0.034) (Figure 3C), indicating significant correlation between TP53 ctDNA burden and advanced disease stages.

Figure 3 Association between preoperative TP53 circulating tumor DNA variant allele frequency and pathological staging in esophageal squamous cell carcinoma patients. A: T-stage stratification: Median TP53 circulating tumor DNA variant allele frequency was 0.00% (T1, T2) and 1.09% (T3); B: N-stage stratification: Progressive increase in median mutation abundance from 0.05% (N0) to 0.54% (N+); C: Tumor-node-metastasis stage stratification: Significant elevation in median mutation abundance with advancing stage: 0.05% (stage I-II) and 0.54% (stage III-IV); ctDNA not detected: Shown as 0.01% on log scale. VAF: Variant allele frequency; ctDNA: Circulating tumor DNA.

TP53 ctDNA for lymph node prediction: To explore a biomarker for informing neoadjuvant therapy decisions, we tested whether preoperative TP53 ctDNA VAF could predict lymph node metastasis, a primary indication for such treatment. The receiver operating characteristic analysis in 54 non-NCT ESCC patients showed an area under the curve of 0.704 (Figure 4). The optimal discriminatory cut-off VAF was 0.4%, providing a sensitivity of 89.3% and a specificity of 57.7%, with a positive predictive value of 69.4% and a negative predictive value of 83.3%.

Figure 4 Receiver operating characteristic analysis of preoperative TP53 circulating tumor DNA for lymph node metastasis prediction. The analysis yields an area under the curve of 0.704, indicating the potential of TP53 circulating tumor DNA variant allele frequency as a predictive biomarker for nodal metastasis. AUC: Area under the curve; CI: Confidence interval.

Association of TP53 ctDNA with the tumor regression grade: Pathological response analysis in the NCT cohort (n = 9) revealed significant association between post-NCT TP53 ctDNA status and tumor regression grade (TRG) per College of American Pathologists criteria (P = 0.012). All TP53 ctDNA-negative patients achieved major pathological response (TRG 1-2), while positive cases were exclusively classified as TRG 3 (no response) (Table 1).

Table 1 Correlation between the tumor regression grade and post-neoadjuvant chemotherapy TP53 circulating tumor DNA status, n (%).

Variables	Total (n = 9)	TRG 1 (n = 2)	TRG 2 (n = 4)	TRG 3 (n = 3)	P value
Post-NCT TP53 ctDNA					0.012
-	6 (66.67)	2 (100.00)	4 (100.00)	0 (0.00)
+	3 (33.33)	0 (0.00)	0 (0.00)	3 (100.00)

TRG: Tumor regression grade; NCT: Neoadjuvant chemotherapy; ctDNA: Circulating tumor DNA.

TP53 ctDNA in ESCC prognosis

Association between postoperative TP53 ctDNA and the prognosis of ESCC: In the 5-year follow-up cohort (n = 61), TP53 ctDNA-positive status remained a strong prognostic marker. The positive group demonstrated significantly shorter median DFS (6.00 months vs 31.73 months; HR = 2.918, 95%CI: 1.333-6.387; P = 0.005) and OS (20.33 months vs 40.31 months; HR = 2.842, 95%CI: 1.294-6.240; P = 0.006) compared to negative patients (Figure 5A and B).

Figure 5 Prognostic impact of TP53 circulating tumor DNA in esophageal squamous cell carcinoma patients. A: Disease-free survival according to postoperative TP53 circulating tumor DNA (ctDNA) status (positive vs negative); B: Overall survival based on postoperative TP53 ctDNA detection; C: Disease-free survival according to preoperative TP53 ctDNA variant allele frequency (≤ 0.32% vs > 0.32%); D: Overall survival according to preoperative TP53 ctDNA variant allele frequency (≤ 0.32% vs > 0.32%). All comparisons were performed using log-rank tests. VAF: Variant allele frequency; ctDNA: Circulating tumor DNA; CI: Confidence interval.

Association between preoperative TP53 ctDNA and the prognosis of ESCC: The optimal cut-off for preoperative TP53 ctDNA VAF in relation to DFS was determined by X-tile software in the 54 non-NCT patients. Using this cut-off of 0.32%, patients were stratified into two groups. Kaplan-Meier analysis revealed that the high VAF group (> 0.32%) had significantly inferior DFS (7.87 months vs 51.48 months; HR = 3.878, 95%CI: 1.964-7.657; P < 0.001) and OS (21.67 months vs not reached; HR = 3.441, 95%CI: 1.718-6.892; P < 0.001) compared to the low VAF group (≤ 0.32%) (Figure 5C and D).

Cox regression analysis of prognostic factors: In ESCC patients without NCT, a comprehensive Cox proportional hazards analysis was performed, evaluating both conventional and molecular prognostic markers. The final model incorporated TNM staging, preoperative TP53 ctDNA, and key demographic variables (age, gender, smoking history, alcohol consumption).

Advanced TNM stage (stage pIII-IV vs pI-II) was significantly associated with both disease relapse and overall mortality, demonstrating a 2.48-fold increased recurrence risk (95%CI: 1.06-5.82; P = 0.036) and a 2.61-fold higher mortality risk (95%CI: 1.12-6.13; P = 0.027). High preoperative TP53 ctDNA VAF emerged as an independent prognostic factor for DFS (HR = 2.49, 95%CI: 1.11-5.59; P = 0.027), whereas its association with OS showed a strong trend but did not reach statistical significance (HR = 2.23, 95%CI: 0.99-5.01; P = 0.052). In contrast, conventional clinical parameters including age, gender, smoking status, and alcohol consumption showed no significant associations with outcomes (all P > 0.2) (Figure 6).

Figure 6 Multivariable Cox regression analysis of prognostic factors. Multivariate analysis identified high preoperative TP53 circulating tumor DNA variant allele frequency as an independent factor for disease recurrence, alongside advanced tumor-node-metastasis stage. A: Univariate analysis of disease-free survival predictors; B: Multivariate analysis of disease-free survival predictors; C: Univariate analysis of overall survival predictors; D: Multivariate analysis of overall survival predictors. HR: Hazard ratio; CI: Confidence interval; TNM: Tumor-node-metastasis; DFS: Disease-free survival; OS: Overall survival; VAF: Variant allele frequency; ctDNA: Circulating tumor DNA.

DISCUSSION

This study employed a 61-gene panel to analyze plasma cfDNA and leukocytes from 63 ESCC patients, with matched tumor tissue analysis performed in 39 treatment-naïve cases who had validated tDNA results. TP53 emerged as the most frequently mutated gene.

Firstly, we evaluated postoperative ctDNA using both tumor-informed and tumor-agnostic strategies. Our tumor-informed approach, which utilized postoperative tissue instead of custom panels due to constraints, revealed significant tumor heterogeneity. Discordant mutations in 8 patients between preoperative ctDNA and tumor DNA underscored the limitations of tissue sampling and highlighted the complementary role of ctDNA[23]. Their exclusion resulted in a 79.5% (31/39) eligibility rate for the tumor-informed analysis. Although this refined cohort demonstrated the prognostic value of ctDNA clearance for DFS, the strategy itself has critical limitations. It provides no information for preoperative staging and, with a concordance rate of only 79.5% even using surgical specimens, would likely perform worse with smaller diagnostic biopsies, thereby precluding reliable preoperative assessment. In stark contrast, the tumor-agnostic approach, applied to all 61 patients who completed follow-up, showed no significant prognostic association based on simple ctDNA positivity[24,25]. Moreover, preoperative ctDNA levels measured by this method were devoid of a significant correlation with tumor burden. These findings support the use of a more focused, biologically informed approach.

Given the high prevalence of TP53 mutations, we further analyzed TP53 ctDNA. Consequently, patients with postoperative TP53 ctDNA had significantly worse OS and DFS than TP53-negative cases - consistent with prior reports in breast cancer[26], which also corroborates our previous findings[22]. Notably, the prognostic discrimination achieved by TP53-specific analysis - as reflected in the highly significant P values for both OS and DFS - was statistically stronger than that observed with the broader tumor-informed approach.

Beyond postoperative monitoring, preoperative TP53 ctDNA levels correlated significantly with T and N stages, suggesting a role in refining staging accuracy and guiding initial treatment strategies[27,28]. Notably, elevated preoperative TP53 VAF appeared to be an independent prognostic factor for survival, suggesting its potential as a non-invasive tool at diagnosis. Patients with values exceeding the established cutoff may thus benefit from intensified treatment and enhanced surveillance. Furthermore, preoperative TP53 ctDNA demonstrated promising accuracy (area under the curve = 0.7) in predicting lymph node metastasis. Current indications for neoadjuvant therapy in ESCC rely primarily on imaging-based lymph node positivity, which lacks pathological confirmation and is prone to false positives (e.g., reactive inflammatory enlargement) and false negatives (e.g., subcentimeter metastatic nodes). The integration of TP53 ctDNA assessment could significantly improve the precision of patient selection for neoadjuvant therapy. Postoperatively, TP53 ctDNA levels decreased markedly, supporting its utility for dynamic response monitoring, consistent with findings for epidermal growth factor receptor ctDNA in lung cancer[29].

Accurate response assessment remains challenging in ESCC treated with NCT. While RECIST criteria are often suboptimal in this setting[30,31], pathological TRG - especially when integrated with ypN stage - provides more reliable prognostic stratification[32,33]. The emergence of watch-and-wait strategies[34,35] further underscores the need for sensitive, non-invasive biomarkers. Our analysis revealed that all patients with detectable TP53 ctDNA after NCT exhibited TRG 3. Notably, two patients who developed new TP53 mutations post-NCT also presented with TRG 3, suggesting the potential emergence of resistant subclones (Supplementary Table 4). While these findings are promising, they are derived from a small subgroup (n = 9) and should be considered preliminary and hypothesis-generating. Future studies should focus on longitudinal ctDNA monitoring during NCT, integrated with endoscopic ultrasound and imaging, to dynamically detect resistance early and guide personalized treatment adaptation.

Our study has several limitations. The sample size was relatively small, and the 61-gene panel did not cover several ESCC-related genes (e.g., KMT2D, FAT1) or all TP53 exons, potentially increasing false-negative rates. Infrequent blood sampling (e.g., single postoperative time point) limited the resolution of ctDNA dynamics, as seen in other cancers with continuous monitoring[36]. This is a particular constraint for claims of “real-time” monitoring, which would ideally require a denser sampling schedule. Additionally, the low NCT rate in our cohort may affect the generalizability of NCT response findings. It is also important to note that our tumor-informed strategy, relying on postoperative tissue, is not a true tumor-informed approach and is unsuitable for preoperative staging.

In summary, this study is the first to quantitatively demonstrate that TP53 ctDNA levels robustly predict tumor response and prognosis in ESCC, supporting its utility as a primary non-invasive biomarker. We plan to expand our cohort to refine risk stratification and leverage our prior findings to develop a cost-effective, high-accuracy ctDNA panel and predictive model for precise ESCC prognostication.

CONCLUSION

Our findings position TP53 ctDNA as a promising, versatile biomarker in ESCC. Postoperatively, its detection independently predicted inferior survival, while its dynamic reduction reflected real-time tumor burden decrease. Preoperatively, TP53 ctDNA levels correlated with TNM stage, and patients with VAF above the defined cutoff experienced significantly worse outcomes, highlighting its utility in refining preoperative staging and risk stratification. Furthermore, in neoadjuvant settings, post-NCT TP53 ctDNA positivity was associated with patients with poor pathological response (TRG 3). Collectively, these data suggest that TP53 ctDNA has the potential to enable integrated prognostic stratification and therapy response monitoring throughout the ESCC management continuum. Further prospective validation in larger cohorts is warranted.