Dynamic Changes of ctDNA‐Based MRD and Genomic Mutations Predict Prognosis in ESCC Treated With Definitive Radiotherapy

ABSTRACT

This study characterized the genomic profile of esophageal squamous cell carcinoma (ESCC) via circulating tumor DNA (ctDNA) to identify predictive biomarkers for efficacy of definitive radiotherapy combined with systemic therapy. In this prospective cohort of 40 ESCC patients, 7 tumor tissues and 71 serial plasma samples were analyzed using a 340‐gene panel. TP53 (45%) and PIK3CA (12.5%) were the most frequent SNVs, followed by NOTCH1/2 (7.5%). The 11q13.2–13.4 amplification (including CCND1/FGF19) occurred in 12.5% of cases. TP53 mutation rate was significantly higher in cT4 than cT2–3 stages (60% vs. 20%, p = 0.010). Notably, patients with persistent TP53 positivity posttreatment had significantly shorter PFS than those converting to negativity (HR = 5.090, p = 0.035); the 2‐year PFS rates were 33.3% vs. 75%, respectively, with OS showing a borderline difference (p = 0.053). Posttreatment ctDNA‐based minimal residual disease (MRD) positivity was 18.8%, correlating with a trend toward worse PFS (HR = 3.027, p = 0.067). Among the 31 patients with paired pre‐ and post‐treatment plasma samples, the positive detection rate of nucleosome‐positioning (left‐shifted profile) decreased from 67.9% at baseline to 35.7% after treatment (χ ² = 5.793, p = 0.016). This work based on plasma ctDNA profiling demonstrated that posttreatment clearance of TP53 mutations may represent a favorable prognostic biomarker in ESCC treated with definitive radiotherapy and systemic therapy. ctDNA‐derived MRD status and nucleosome positioning constitute promising non‐invasive biomarkers for treatment response monitoring and risk stratification for ESCC.

Trial Registration:This study was registered with the Chinese Clinical Trial Registry (ChiCTR) on December 6, 2019 (Registration No. ChiCTR1900027936)

In patients with esophageal squamous cell carcinoma receiving definitive radiotherapy combined with systemic therapy, posttreatment TP53 mutation clearance and ctDNA‐MRD negativity predict favorable outcomes, whereas persistent ctDNA signals indicate elevated relapse risk, supporting nucleosome‐ and MRD‐informed surveillance and risk‐stratification.

Abbreviations

AUC

area under the curve

CCRT

concurrent chemoradiotherapy

cfDNA

cell‐free DNA

CI

confidence interval

CNV

copy number variation

CR

complete response

CT

computed tomography

CTCAE

common terminology criteria for adverse events

ctDNA

circulating tumor DNA

ECOG

Eastern Cooperative Oncology Group

EGFR

epidermal growth factor receptor

ESCC

esophageal squamous cell carcinoma

FFPE

formalin‐fixed, paraffin‐embedded

HR

Hazard Ratio

IFI

involved‐field irradiation

IMRT

intensity‐modulated radiotherapy

MRD

minimal residual disease

ORR

objective response rate

OS

overall survival

PD

progressive disease

PFS

progression‐free survival

PR

partial response

PTV

planning target volume

RECIST

Response Evaluation Criteria in Solid Tumors

SD

stable disease

SIB

simultaneous integrated boost

SNV

single nucleotide variant

1. Introduction

For unresectable locally advanced esophageal cancer, definitive concurrent chemoradiotherapy (CCRT) represents the standard treatment modality [1, 2, 3]. Although its 5‐year overall survival (OS) rate has improved from 26% [2] to 33.7% ~ 46.2% [4] in recent studies, the overall therapeutic efficacy remains far from clinically satisfactory. In the immunotherapy era, despite numerous studies [5, 6, 7, 8, 9] exploring chemoradiotherapy plus immunotherapy in locally advanced esophageal cancer, immune checkpoint inhibitors remain unestablished as standard therapy. Even in advanced disease, single‐agent immunotherapy yields an objective response rate (ORR) of only around 20% [10, 11, 12, 13]. Regarding targeted therapy exploration, only anti‐angiogenic agents (anlotinib) are currently recommended for the treatment of ESCC, indicated for second‐line and beyond therapy in patients with advanced disease. Furthermore, recent reports [14, 15, 16] have demonstrated that the anti‐EGFR antibody nimotuzumab combined with chemoradiotherapy can provide therapeutic benefits in patients with locally advanced or advanced esophageal cancer. Nevertheless, overall, the clinical benefits of both immunotherapy and targeted therapy remain relatively limited in unresectable locally advanced esophageal cancer. Urgent clinical issues to be addressed include: (1) further exploration of optimized combination therapeutic strategies; and (2) identification of robust biomarkers to select patients who may derive superior clinical benefit.

Accordingly, as an exploratory analysis of a previous single‐arm Phase II clinical trial, we performed dynamic monitoring of circulating tumor DNA (ctDNA) before and after definitive radiotherapy in 40 consecutive patients with ESCC enrolled consecutively between April 2020 and January 2022 at the Fourth Hospital of Hebei Medical University. This study aimed to investigate potential genetic susceptibility genes and oncogenic signaling pathway genes involved in the carcinogenesis and treatment of ESCC, identify biomarkers associated with treatment sensitivity and therapeutic response prediction, and guide the formulation of clinical multimodal treatment strategies and agent selection.

2. Materials and Methods

2.1. Design

This study is mainly based on a prospective single‐center Phase II clinical trial conducted by the team. The study was conducted according to the Declaration of Helsinki, and all patients signed informed consent before enrollment. The research protocol has been approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (Number: 2019040). Clinical trial registration number: ChiCTR1900027936 (Date of registration: December 6, 2019).

2.2. Patient Inclusion and Exclusion Criteria

Inclusion criteria: (1) Pathologically confirmed squamous cell carcinoma; (2) Patients with locally advanced esophageal cancer or stage IVb with only extra‐regional metastasis were enrolled. (3) Patients with surgically unresectable disease; (4) ECOG score 0–2; (5) Patients voluntarily signed a consent form.

Exclusion criteria: malignant tumor history or serious underlying disease that may affect completion of treatment.

2.3. Treatment

All patients underwent definitive radiotherapy with involved‐field irradiation (IFI) at a prescription dose of ≥ 50 Gy using intensity‐modulated radiotherapy (IMRT). Conventional fractionation was delivered as 1.8~2.0 Gy per fraction to the PTV‐_C, once daily, 5 days per week. Alternatively, simultaneous integrated boost IMRT (SIB‐IMRT) was administered: PTV‐_C received 50.4 Gy in 28 fractions (1.8 Gy per fraction, once daily, 5 days per week), while PTV‐_G was simultaneously boosted to 63 Gy in 28 fractions (2.25 Gy per fraction, once daily, 5 days per week).

The systemic treatment included chemotherapy or nimotuzumab therapy. All chemotherapy regimens adopted paclitaxel combined with a platinum‐based regimen. Paclitaxel was administered at 135 mg/m² via intravenous infusion on Day 1 (d1). Platinum‐based agents included cisplatin or carboplatin: cisplatin was given at 75 mg/m² via intravenous infusion from d1 to d3, while carboplatin was dosed at AUC = 5 via intravenous infusion on d2. All chemotherapy cycles were repeated every 21 days and administered either concurrently or sequentially with radiotherapy. Nimotuzumab was applied concurrently with radiotherapy, administered via intravenous infusion on the first day of each week during the radiotherapy course (200 mg, once a week, 6 weeks in total, the cumulative dose was 1200 mg).

2.4. Evaluation

Treatment‐related toxicity was evaluated according to CTCAE V4.03. The definition of short‐term efficacy referred to RECIST 1.1, which is mainly based on the imaging results.

2.5. Next‐Generation Sequencing‐Based ctDNA Assay and MRD Determination

Tumor tissue and/or blood samples were collected at baseline before treatment initiation, and plasma samples were obtained within 1 week after completion of all radiotherapy and concurrent systemic therapy. This study adopted a tumor‐naïve strategy. Plasma cell‐free DNA (cfDNA) from all patients was profiled using a fixed 340‐gene panel; variant types interrogated by this panel included single‐nucleotide variants (SNV), small insertions and deletions (indel), structural variants (SV), and copy‐number variations (CNV). Although tumor tissue sequencing data were available for seven patients, these tissue‐derived results were not utilized to select patient‐specific tracking loci or define plasma assay positivity; therefore, the assay applied in this study was not tumor‐informed.

In our study, plasma cfDNA testing was performed with paired peripheral‐blood leukocyte DNA profiling. Plasma and leukocyte samples were sequenced using the corresponding panel. Variants identified in leukocyte DNA were used to recognize and filter out germline variants as well as putative variants derived from clonal hematopoiesis. For patients with evaluable posttreatment plasma samples, posttreatment ctDNA‐MRD was classified as positive if at least one driver variant meeting laboratory reporting criteria was detected in plasma cfDNA following filtering by paired leukocyte DNA; ctDNA‐MRD was deemed negative when no variants fulfilling the above criteria were observed (For somatic mutations from cfDNA, we classified mutations by COSMIC and OncoKB into driver and passenger mutations. We applied criteria of VAF > 0.1%, and read depth > 3 for clear driver mutations, and VAF > 1%, read depth > 10 for passenger mutations).

2.6. Definitions of Key Terms

Driver mutation: Somatic genetic alterations conferring selective growth advantage to tumor cells and promoting oncogenesis, annotated according to the COSMIC and OncoKB databases in this study.

SNV‐driven mutation: A subtype of driver mutation represented by single‐nucleotide variant alterations.

Passenger mutation: Somatic variants without known oncogenic functions that do not confer a growth‐selective advantage to tumor cells.

MRD positivity: Detection of at least one qualifying driver variant in posttreatment plasma ctDNA after filtering germline and clonal‐hematopoiesis‐derived variants using paired leukocyte DNA. TP53 positivity: Detection of qualifying somatic TP53 variant (s) in plasma ctDNA at a given time‐point (baseline or post‐treatment).

2.7. Nucleosome‐Positioning Profiling

In the sequencing workflow of this study, somatic mutations that passed the QC criteria, as well as germline variants on clear tumor suppressors or oncogenes that were classified as variants of uncertain significance (VUS), likely pathogenic (LP), or pathogenic (PAT) according to ACMG standard, were included in the nucleosome analysis. Nucleosome‐positive status was defined as the detection of a left‐shift in fragment‐length distribution; samples exhibiting no left‐shift were classified as nucleosome‐negative. Detection assay and validation: extracting all sequenced fragments covering specific genomic loci and measuring the length distribution of wild‐type and mutant alleles. Tumor‐derived cfDNA tends to be shortened compared to normal cell derived cfDNA. In the case of somatic mutation that only present in tumor, we expect that 100% of mutant fragments originate from tumor and (1‐K*TF/N)*100% of wild type fragment are originated from normal tissue, where K = the copy number of mutant alleles in the tumor genome, N = the total copy number of the loci in the tumor genome, and TF = the tumor fraction in the cell‐free DNA. In the case of germline mutation, we consider that putative pathogenic mutation might be selected in the tumor into loss‐of‐heterozygosity condition. Therefore, we expect that (K*TF/N)/2 * 100% of mutant fragments, and ((N‐K)*TF/N)/2*100% of wild type fragments are derived from the tumor, where N‐K < K. In both of these cases, we see that compared with wild‐type fragments, mutant fragments are more likely to originate from tumor and therefore should be shorter. We performed K‐S test on the length distribution between (1) wild type and mutant allele fragments from the patient and (2) wild type fragments from a 36‐healthy‐donor cohort that sequenced prior, and mutant allele fragments from the patient. For the variants that showed significantly (p < 0.05) differences in both tests, we further manually inspected whether the distribution was left (short) shifted. The assay is calibrated by diluting cultured tumor cells (A549) supernatant into healthy donor serum and obtained a limit‐of‐detection (LOD) for 0.1%. For patients with weak nucleosome‐profile signals that precluded valid analysis, these cases were incorporated into the negative cohort for subsequent analyses.

2.8. Follow‐Up and Statistical Analysis

Patients were followed up every 3 months within the first 2 years after treatment, and every 6 months thereafter. Patients lost to follow‐up were censored at the time of the last available follow‐up. Statistical analyses were performed using SPSS 22.0 software. Progression‐free survival (PFS) and OS were estimated using the Kaplan–Meier method. The distribution of driver gene mutations across subgroups with different clinicopathological characteristics was compared using the chi‐square test for R × C tables. Univariate prognostic analysis was conducted using Cox proportional hazards regression.

3. Result

3.1. Clinical Characteristics and Treatment of Patients

Between April 2020 and January 2022, a total of 40 patients with pathologically confirmed ESCC were enrolled in the entire cohort. Detailed clinical characteristics are summarized in Table 1. The cohort included 10 patients with supraclavicular lymph node metastasis and 1 patient with celiac lymph node metastasis, who were classified as stage M1 according to the 8th edition of the AJCC cancer staging system. Among all patients, 11 were staged N0, 14 were N1, and 15 were N2–3. All patients received comprehensive treatment mainly by radiotherapy, including 28 patients who received simultaneous integrated boost intensity‐modulated radiotherapy (SIB‐IMRT) combined with nimotuzumab, and 12 patients who received sequential or concurrent chemoradiotherapy. The prescribed doses of patients receiving SIB‐IMRT were: PTV_‑C 50.4Gy/28 fractions; PTV_‑G concurrent dose, 63Gy/28 fractions; conventional fractionated radiotherapy, 50.4‐60Gy/28‐33 fractions.

TABLE 1.

Baseline characteristics of patients.

Characteristics	n (%)
Mean age (years) ± SD	66 ± 7.1
Sex
Male	27 (67.5%)
Female	13 (32.5%)
ECOG PS
0 ~ 1	31 (77.5%)
2	9 (22.5%)
Tobacco smoking	15 (37.5%)
Chronic obstructive pulmonary disease	4 (10%)
Family history	9 (22.5%)
Lesion length (by barium meal, cm)	5.6 ± 2.2
Tumor location
Cervical	1 (2.5%)
Upper	11 (27.5%)
Middle	22 (55%)
Lower	6 (15%)
cT stage
T2	1 (2.5%)
T3	14 (35%)
T4	25 (62.5%)
cN stage
N0	11 (27.5%)
N1	14 (35%)
N2‐3	15 (37.5%)
cM stage
M0	29 (72.5%)
M1	11 (27.5%)
cTNM stage
II	5 (12.5%)
III	6 (15%)
IVA	18 (45%)
IVB	11 (27.5%)
Treatment
Concurrent/sequential chemoradiotherapy	12 (30%)
Radiotherapy + nimotuzumab	28 (70%)

3.2. Response and Efficacy

Treatment response was comprehensively evaluated after treatment using computed tomography (CT), esophagography, and ultrasound. Complete response (CR) was achieved in 4 patients (10%), partial response (PR) in 32 patients (80%), stable disease (SD) in 3 patients (7.5%), and progressive disease (PD) in 1 patient (2.5%). With a median follow‐up of 27 months (IQR 14–34), the 1‐, 2‐, and 3‐year PFS rates were 67.5%, 54.9%, and 51.0%, respectively, and the 1‐, 2‐, and 3‐year OS rates were 80.0%, 57.5%, and 54.1%, respectively. Median PFS and OS were not reached (Figure 1).

FIGURE 1.

Response and survival of the whole cohort (A) Pie Chart of Treatment Response; (B) Swimmer Plot of Treatment Response and Survival Outcomes.

3.3. Tumor Tissue and ctDNA Test Results

The tumor tissue or ctDNA detection results are shown in Figure 2. Peripheral‐blood ctDNA testing was performed in 39 patients at pre‐treatment (baseline), among whom 7 patients also received tumor‐tissue sequencing. Posttreatment peripheral‐blood ctDNA testing was conducted in 32 patients, and paired pre‐ and post‐treatment plasma specimens were available for 31 patients. The specific detection and mutation status are shown in Figure 3. The driver gene mutation was detected in 20 patients (50%), the passenger mutation was detected in 7 patients (17.5%), and 12 patients (30%) had no mutation. The somatic mutations mainly consisted of Single Nucleotide Variation‐SNV and Copy‐Number Variation‐CNV. Among SNV‐driven mutations, TP53 mutation had the highest mutation frequency, accounting for 45% (18/40), followed by PIK3CA, NOTCH1/2 and CDKN2A mutations (The mutation frequency was 12.5%, 7.5% and 5% respectively), while CUL3, NFE2L2, and BRCA2 mutations were rare (all 2.5%). The most common mutation was 11q13.2‐11q13.4amp (12.5%), corresponding to CCND1 FGF19 FGF4 FGF3, followed by EGFR and CDKN2A/CDKN2B mutations (both 5%). FGFR1, MDM2, SOX2, and YAP1 mutations were rare (2.5%).

FIGURE 2.

Flow chart of study design and patient enrollment. ESCC, esophageal squamous cell carcinoma; NGS, next‐generation sequencing; ctDNA, circulating tumor DNA.

FIGURE 3.

Summary of patients' somatic mutations in tumor tissues, pretreatment and posttreatment ctDNA detecting status of the 40 evaluable ESCC. (A) DNA/ctDNA detecting status before and after treatment. ctDNA, circulating tumor DNA. (B) Nucleosome‐positioning profiling before and after treatment. (C) SNV mutations of the patients. SNV, single nucleotide variation. (D) CNV mutations of the patients. CNV, copy‐number variation. OS status of the patients. OS, overall status.

3.4. Correlation Between Driver Gene Mutation and Clinical Factors Before Treatment and Prognosis

The mutation of the driver gene and TP53 in each clinical subgroup before treatment is shown in Table 2. No significant difference in SNV driver mutation was found in the analysis of factors such as gender, age, smoking history, family history, and tumor site. Among them, the TP53 mutation rate of cT4 patients was significantly higher than that of cT2‐3 patients. It was 60% vs. 20%, χ ² = 6.709, p = 0.010, TP53 mutation was more distributed in cT4 patients, and there was a correlation between them by Spearman correlation test (r = −0.415, p = 0.009). Kaplan–Meier method was used to plot survival curves for patients with baseline SNV‐driver mutations versus those with no detectable mutations or only passenger‐type mutations. Univariate Cox proportional‐hazards regression prognostic analysis showed no significant differences in OS and PFS between these two subgroups, as shown in Table 3 and Figure 4A,E. For the TP53 mutant group and the non‐mutant group, there was no significant difference in OS and PFS between the two groups, as shown in Table 3 and Figure 4B,F.

TABLE 2.

Pretreatment driver gene mutations by clinicopathological subgroup.

Characteristics	Driver mutant group (n = 20)	No mutant or passenger mutant group (n = 19)	Χ 2	p	TP53 mutant group (n = 18)	No TP53 mutant group (n = 21)	Χ 2	p
Gender			2.235	0.135			1.146	0.284
Male	16	11			14	13
Female	4	8			4	8
Age			0.641	0.423			0.140	0.708
< 70 years old	15	12			13	14
≥ 70 years old	5	7			5	7
Smoking history			0.041	0.839			0.371	0.542
Yes	8	7			6	9
No	12	12			12	12
Family history			0.719	0.396			0.248	0.618
Yes	3	6			3	6
No	17	13			15	15
Tumor site			0.345	0.557			1.035	0.309
Neck + upper chest	7	5			7	5
Middle chest + lower chest	13	14			11	16
cT stage			3.143	0.076			6.709	0.010
T2‐3	5	10			3	12
T4	15	9			15	9
cN			0.981	0.612			0.238	0.888
N0	4	6			4	6
N1	7	7			7	7
N2‐3	9	6			7	8
cM stage			2.820	0.093			0.434	0.510
M0	12	16			12	16
M1	8	3			6	5
cTNM stage			3.535	0.063			2.198	0.138
II‐III	3	8			3	8
IV	17	11			15	13

Note: Bold values represent p‐values showing statistically significant differences or trends toward statistical significance.

TABLE 3.

Driver gene mutation and prognosis before and after treatment.

Group	PFS (%, m)				HR	95% CI	p	OS (%, m)				HR	95% CI	p
	1‐y	2‐y	3‐y	mPFS				1‐y	2‐y	3‐y	mOS
Driver mutation before treatment
Negative/passenger group	73.7%	57.9%	50.7%	—	1	(Reference)	(Reference)	78.9%	57.9%	52.6%	—	1	(Reference)	(Reference)
Positive group	65%	55%	55%	—	1.031	0.409–2.601	0.948	80%	60%	60%	—	0.826	0.319–2.144	0.695
TP53 mutation before treatment
Negative group	66.7%	52.4%	45.8%	32	1	(Reference)	(Reference)	76.2%	52.4%	52.4%	—	1	(Reference)	(Reference)
Positive group	72.2%	61.1%	61.1%	—	0.681	0.268–1.733	0.421	83.3%	61.1%	61.1%	—	0.744	0.288–1.921	0.541
MRD detection after treatment
Negative group	76.9%	65.2%	60.1%	—	1	(Reference)	(Reference)	84.6%	65.4%	60.7%	—	1	(Reference)	(Reference)
Positive group	50%	33.3%	—	4	3.027	0.925–9.902	0.067	66.7%	33.3%	—	13	2.642	0.821–8.503	0.103
TP53 mutation changes before and after treatment
Positive–negative group	83.3%	75.0%	75.0%	—	1	(Reference)	(Reference)	91.7%	75%	75%	—	1	(Reference)	(Reference)
Positive–positive group	50%	33.3%	—	4	5.090	1.126–23.014	0.035	66.7%	33.3%	—	13	4.394	0.979–19.726	0.053
Negative/passenger group	63.6%	50%	43.8%	15	2.716	0.765–9.644	0.122	77.3%	54.5%	49.1%	31	2.408	0.671–8.643	0.178

Note: Bold values represent p‐values showing statistically significant differences or trends toward statistical significance.

FIGURE 4.

Survival analysis of different driver gene mutation subgroups before and after treatment. (A) PFS of the driver mutation group and the non‐mutation/passenger mutation group before treatment; (B) PFS in TP53 mutant group and non‐mutant group before treatment; (C) PFS of MRD (−) group and MRD (+) group based on ctDNA after treatment; (D) PFS of different mutation subgroups of TP53 before and after treatment; (E) The OS of the driver mutation group and the non‐mutation/passenger mutation group before treatment; (F) OS in TP53 mutant group and non‐mutant group before treatment; (G) OS of MRD (−) group and MRD (+) group based on ctDNA after treatment; (H) OS in different mutation subgroups of TP53 before and after treatment.

3.5. The ctDNA‐Based MRD Results and Prognostic Correlation After Treatment

Posttreatment genomic profiling of peripheral blood was performed in 32 patients. The detection rate of ctDNA‐based minimal residual disease (MRD) was 18.8% (6/32), and all MRD‐positive cases harbored TP53 mutations. Among the entire cohort, 12 patients with driver mutations at baseline exhibited conversion to passenger mutations or undetectable mutations after treatment. For PFS, the hazard ratio (HR) of the MRD (+) group versus the MRD (−) group was 3.027 (95% confidence interval [CI]: 0.925–9.902, p = 0.067). For OS, the HR was 2.642 (95% CI: 0.821–8.503, p = 0.103; Table 3, Figure 4C,G). Survival analyses according to TP53 mutation status before and after treatment revealed that patients with baseline TP53 mutation that converted to negativity after treatment had significantly more favorable survival outcomes. PFS was significantly higher in these patients than in those with persistent TP53 mutation (HR = 5.090, 95% CI: 1.126–23.014, p = 0.035). Univariate analysis for OS showed a borderline significant difference between the two groups (HR = 4.394, 95% CI: 0.979–19.726, p = 0.053). Patients with no driver mutations or only passenger mutations throughout treatment showed intermediate survival curves (Table 3, Figure 4D,H).

In the multivariate Cox regression analysis, the covariates included patient sex (male vs. female), age (< 70 years vs. ≥ 70 years), tumor location (cervical plus upper thoracic esophagus vs. Middle and lower thoracic esophagus), cT stage (cT2‐3 vs. cT4), cN stage (N0 vs. N1 vs. N2‐3), cM stage (M0 vs. M1), therapeutic effect (CR vs. PR vs. SD vs. PD), pre‐treatment TP53 mutation status (negative vs. positive), MRD status (negative vs. positive), and dynamic changes in TP53 status before and after therapy (pretreatment positive/posttreatment negative vs. persistently positive before and after treatment vs. no mutation group). Multivariate analysis identified M1 stage as an independent adverse prognostic factor for survival compared with M0 stage (HR = 5.402, 95% CI: 1.786–16.343, p = 0.003). Persistently positive TP53 status throughout treatment showed a borderline significant association with poorer survival relative to patients with posttreatment TP53 mutation clearance (HR = 12.556, 95% CI: 0.986–159.810, p = 0.051).

3.6. Exploratory Analysis: ctDNA‐Based Nucleosome‐Positioning Profiling

At baseline, 22 patients had positive nucleosome‐positioning profiling results, 9 patients had negative results, and Nuke‐based nucleosome‐positioning analysis was not feasible in 9 cases (including one patient without baseline ctDNA testing). After treatment, 11 patients exhibited positive nucleosome‐positioning profiling results, 19 exhibited negative results, and Nuke‐based nucleosome‐positioning analysis was not feasible in 10 cases (including eight patients without posttreatment ctDNA testing). The positive detection rates of nucleosome‐positioning profiling before and after treatment were 56.4% (22/39) and 34.4% (11/32), respectively, χ ² = 3.431, p = 0.064. Among the 31 patients with paired pre‐ and posttreatment plasma samples, the objective response rate ((CR + PR) / total evaluable patients × 100%) was 90.3% (28/31). For patients who achieved an objective response, the positive detection rates of nucleosome‐positioning (left‐shifted profile) before and after treatment were 67.9% (19/28) and 35.7% (10/28), respectively. The posttreatment positive rate of nucleosome‐positioning was significantly decreased (χ ² = 5.793, p = 0.016). Figure 5 shows the CT images and nucleosome distribution detection of 1 CR patient and 1 PD patient before and after treatment.

FIGURE 5.

CT images and nucleosome distribution detection before and after treatment in 1 CR and 1 PD patients respectively. A 53‐year‐old female with cervical esophageal squamous cell carcinoma (T3N1M0 stage III) had a disease‐free survival of 45 months: Before treatment, CT images showed thickening of the cervical esophageal wall and enlargement of the right high paratracheal lymph nodes. ctDNA test showed that the position of nucleosome distribution shifted significantly to the left (red curve moving direction in the figure), and the nucleosome signal was positive, indicating the detection of tumor signals. Mutation analysis showed that low‐abundance pathogenic somatic mutations were detected before treatment, and the mutation abundance was 9.34%. After the end of radiotherapy, CT images showed no obvious thickening of the esophageal wall at the lesion site, and the enlarged lymph nodes nearly disappeared. The imaging evaluation of CR showed that the nucleosome signal turned negative, and no low‐abundance pathogenic somatic cell mutation was detected by ctDNA, suggesting a good treatment effect. (B) A 64‐year‐old male with lower thoracic esophageal squamous cell carcinoma (T3N3M0IV stage A) with survival of 13 months and death due to disease progression: Before treatment, CT images showed thickening of the esophageal wall in the lower thoracic segment and multiple lymph node enlargement in the mediastinal cardia. ctDNA test showed that some nucleosomes shifted significantly to the left (red curve moving direction in the figure), and nucleosome signals were positive. Mutation analysis detected somatic mutation TP53E339 * with a mutation abundance of 1.28%. After radiotherapy combined with nimotuzumab, CT images showed that the thickness of the esophageal wall of the lesion was reduced compared with that before treatment, and new lesions appeared in the liver. MRI indicated liver metastasis, imaging evaluation PD, and ctDNA detection showed that some new nucleosomes shifted to the left, and some nucleosome signals turned positive. The results of mutation analysis showed that somatic mutation TP53 E339* with a mutation abundance of 3.18% showed an increasing trend after treatment.

4. Discussion

Liquid biopsy, which includes circulating tumor DNA (ctDNA), is the most promising noninvasive method for tumor diagnosis and monitoring in recent years. It has the advantages of convenience, noninvasive [17]. The main peak of the ctDNA fragment is about 167 bp, and the content is very low, accounting for 0.01%–1% of all circulating DNA (cell‐free DNA, cfDNA), and its content is related to tumor type, tumor load and disease stage [18], because it carries the same genetic molecular genetic information as the primary tumor and has a short half‐life. Therefore, it is of great significance in tumor diagnosis and dynamic monitoring of therapeutic effects [19]. Currently commonly used clinical ctDNA detection methods include quantitative PCR, ARMS, and high‐throughput sequencing (NGS) [20], which can detect gene mutation, deletion, insertion, fusion, rearrangement, etc. Among them, NGS has demonstrated great clinical application value with its advantages of high sensitivity and high specificity [21, 22] and has been applied in tumors of multiple systems. For esophageal cancer, the gene mutation characteristics based on tumor tissue have been comprehensively studied, but the gene mutation map of ctDNA needs more in‐depth research. Based on this, we conducted NGS dynamic monitoring (tDNA/ctDNA, 340 panel) before and after treatment in the continuously enrolled esophageal cancer cohort, to observe the genetic mutation characteristics of esophageal squamous cell carcinoma, evaluate the prognostic value of ctDNA‐based MRD, and seek new efficacy‐related biomarkers to guide clinical treatment.

For gene mutation characteristics detected in patients, the most common SNV‐driven mutations were TP53 and PIK3CA mutations, accounting for 45% and 12.5% of the whole group, respectively. Among CNV‐driven mutations, the most common mutations were 11q13.2‐11q13.4amp band mutations (12.5%). The corresponding genes are CCND1, FGF19, FGF4, FGF3, and the EGFR mutation rate is low; it is 5%. This result is similar to the gene mutation characteristics based on tumor tissue [23, 24, 25, 26], but compared with the reported mutation frequency of TP53 up to 58% to 93.1%, the mutation frequency of TP53 in this group seems to be slightly lower, considering that the mutation frequency of TP53 in the sample submitted (blood vs. tissue), detection methods, etc. However, in this group of studies and previous reports, it is relatively consistent to show that TP53 and PIK3CA genes are among the most important driver gene mutations in esophageal cancer. TP53 is considered to be a tumor suppressor gene in previous studies. After mutation, TP53 has oncogene function due to spatial conformation changes affecting transcriptional activation function and the phosphorylation process of the p53 protein. At present, the clinical factors related to its occurrence are not clear, and it has been reported that TP53 mutation seems to be related to smoking and drinking [27, 28]. In this study, no significant differences in the distribution of driver mutations were found in different clinical factor subgroups (gender, age, smoking history, family history, etc.), but in the subgroups with late clinical T, N, and M stages, the proportion of TP53 mutations seemed to increase, especially in patients with cT4 stage. The frequency of TP53 mutations was significantly higher than in patients with earlier cT stage (60% vs. 20%, p = 0.010), and it appears that TP53 mutations are associated with later disease stage.

Although TP53 mutations are considered to have oncogene function and appear to be more prevalent in more advanced patients in this study, we found that after antitumor therapy, this mutation appears to be eliminated, and its mutation frequency is reduced from 45% before treatment to 18.8%. After prognostic analysis, we also found that there was no significant correlation between the driver mutation status of patients before treatment and OS or PFS, but for the subgroup that still had TP53 mutation after treatment (Positive–positive group), The PFS was significantly inferior to the positivity‐negative group with TP53 mutation, and the risk of disease progression in the former group was 5.09 times of that in the latter group. For OS, the difference between the two groups was also close to statistical significance (95% CI 0.979–19.726; HR = 4.394; p = 0.053), suggesting that changes in ctDNA TP53 mutation status seem to predict the efficacy of antitumor therapy. Previous studies by Ryosuke Fujisawa et al. [29] also found that TP53 mutations were cleared after chemotherapy in several esophageal cancer patients, suggesting that ctDNA detection could be used to predict early efficacy.

Minimal residual disease (MRD) refers to the clinical state in which a patient with a malignant tumor still has a small number of residual tumor cells or tiny lesions in the body during or after receiving treatment (surgery, chemotherapy, radiotherapy, targeted therapy, or immunotherapy, etc.), and these latent residual remains are difficult to find by conventional clinical methods. However, it is associated with tumor recurrence and prognosis [30]. For MRD, highly accurate ctDNA detection and quantitative methods can detect residual lesions earlier than standard clinical or imaging methods. Studies have confirmed that patients with positive MRD have a poor prognosis. Due to the short half‐life of ctDNA, detection of molecular recurrence is earlier than clinical recurrence, which is more predictive [31, 32]. At present, the research on MRD of esophageal cancer is not mature, and there are few relevant reports. In a ctDNA analysis based on 45 patients with esophageal cancer, it was found that MRD detected after chemoradiotherapy was associated with higher tumor progression, distant metastasis risk, and shorter disease‐specific survival, and MRD detected by ctDNA after chemoradiotherapy was 2.8 months earlier than the tumor progression indicated by imaging evidence [33]. In our study, the MRD detection rate after treatment was 18.8% (6/32), and the 1‐ and 2‐year PFS and OS rates in the MRD (−) group were numerically higher than those in the MRD (+) group, but there was no statistically significant difference between the two groups based on univariate analysis of COX Regression model, which was considered to be related to the small number of cases. Observation time is not sufficient.

To seek new biomarkers for early prediction of efficacy, we also performed ctDNA‐based nucleosome location distribution detection. Nucleosomes are the basic structural units of chromatin formed by DNA and histone proteins; each nucleosome is formed by 146 bp DNA wrapped around 1.75 circles of histone octamer. The formation of nucleosomes plays an important role in the packaging of DNA and the regulation of gene expression. Through the close arrangement of nucleosomes, long DNA molecules can be compressed in the limited space of the nucleus. Changes in the structure and location of nucleosomes can affect the transcriptional activity of genes and thus play an important role in cell function and disease occurrence. At present, the mechanism of nucleosome action in the development of malignant tumors is not clear, and the relevant reports are few. Our study shows that in patients with treatment response (objective response), the positive detection rate of nucleosome‐positioning (left‐shifted profile) decreased significantly from 67.9% at baseline to 35.7% after treatment. In contrast, sustained positive nucleosome‐positioning was observed in the single patient with progressive disease (PD). Taken together, we suggest that nucleosome‐positioning represents a promising efficacy‐related biomarker warranting further investigation.

Several limitations should be noted. First, this study enrolled 40 patients with ESCC and analyzed 71 plasma ctDNA samples. However, matched tumor tissue sequencing was only available for 7 cases. Due to insufficient matched tissue samples for validation, it cannot be fully confirmed that all genetic variants detected in plasma are derived from primary tumors, which introduces potential bias. Further studies with larger sample sizes and paired sequencing of tumor tissues and plasma are warranted to validate the clinical value of the biomarkers identified in the present study. Second, ctDNA detected within 1 week after multimodal therapy may partly originate from radiation‐related tumor necrosis, which constitutes a potential confounding factor [33, 34]. Third, treatment regimens were heterogeneous within this cohort of patients receiving definitive radiotherapy combined with systemic therapy. Owing to insufficient sample size, we could not conduct subgroup or sensitivity analyses stratified by treatment modality.

TP53 is one of the most predominant SNV driver mutations in ESCC and can be eliminated in some patients following definitive radiotherapy or combined modality therapy. Patients with baseline TP53 mutation positivity that converts to negativity after treatment exhibit better prognosis than those with persistent TP53 positivity. The ctDNA‐based minimal residual disease (MRD) and nucleosome positioning are promising efficacy‐related biomarkers worthy of further investigation. Notably, the genomic profiles characterized in this study are plasma‐derived ctDNA alterations. Given the limited number of tumor tissue sequencing samples, these findings may not fully recapitulate the intrinsic genomic landscape of primary tumor tissues. Collectively, serial plasma ctDNA‐MRD monitoring has promising value for risk stratification and guiding individualized surveillance and adaptive therapy for ESCC patients receiving definitive radiotherapy combined with systemic therapy (Figure 6).

FIGURE 6.

Schematic graphical abstract summarizing the present study. Patients with locally‐advanced esophageal squamous cell carcinoma (ESCC) received definitive radiotherapy combined with systemic therapy (chemotherapy or nimotuzumab‐based targeted therapy). Serial plasma samples were collected and subjected to 340‐gene‐panel circulating tumor DNA (ctDNA) sequencing for minimal residual disease (MRD) detection and nucleosome‐positioning profiling. Posttreatment TP53 mutation clearance together with ctDNA‐MRD‐negative status was associated with favorable survival outcomes. By contrast, persistent TP53 mutation and ctDNA‐MRD positivity indicated elevated risk of disease relapse. This study demonstrates that ctDNA‐derived MRD and nucleosome‐positioning signatures represent promising non‐invasive biomarkers to support risk‐stratified surveillance for ESCC patients treated with definitive radiotherapy plus systemic therapy.

Author Contributions

Shutang Liu: writing – original draft, supervision, project administration. Qi Wang: writing – original draft, data curation, supervision. Chun Han: writing – original draft, supervision. Xiaoning Li: resources, investigation. Rutian Cheng: data curation. Lihong Liu: investigation, resources. Hua Dong: investigation, resources. Xuejiao Ren: investigation, resources. Shuman Zhen: resources, investigation. Lan Wang: conceptualization, supervision, project administration.

Funding

Government‐Funded Provincial Outstanding Medical Talent Program (JCPF [2022]) No. 180; National Key Research and Development Program Project (2018YFC1313203); Hengrui Hebei Medical Innovation and Development Cooperation Program (HR202502045).

Ethics Statement

Institutional Review Board approval: This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (Approval No. 2019040).

Consent

Written informed consent was obtained from all individual participants included in the study.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.