Impact of target volume delineation accuracy on locoregional control and survival after postoperative radiotherapy for colorectal cancer

Abstract

BACKGROUND

Postoperative radiotherapy reduces local recurrence in locally advanced colorectal cancer, but its efficacy depends on accurate target volume delineation. This study assessed the impact of delineation accuracy on treatment outcomes.

AIM

To investigate the impact of target volume delineation accuracy on treatment outcomes of postoperative radiotherapy for colorectal cancer.

METHODS

Clinical data of 218 colorectal cancer patients who underwent radical surgery followed by adjuvant radiotherapy at our hospital from January 2019 to December 2022 were retrospectively analyzed. Patients were divided into an accurate target volume group (126 cases) and an inaccurate target volume group (92 cases) based on target delineation quality. Survival outcomes, recurrence patterns, and adverse events were compared between the two groups. The Kaplan-Meier method was used for survival analysis, the Log-rank test for comparing group differences, and the Cox proportional hazards regression model for analyzing independent prognostic factors.

RESULTS

The median follow-up was 38 months. The 3-year local control, regional control, disease-free survival, and overall survival were 92.1%, 94.4%, 83.6%, and 89.2% in the accurate target volume group, respectively, compared to 78.3%, 81.5%, 66.8%, and 72.5% in the inaccurate target volume group (P < 0.05). Recurrence pattern analysis showed that 75.8% of recurrences occurred in marginal or out-of-field regions in the inaccurate target volume group, compared to 34.8% in the accurate target volume group (P = 0.002). Multivariate analysis revealed that target volume delineation accuracy was an independent prognostic factor affecting disease-free survival (hazard ratio = 2.186, 95% confidence interval: 1.304-3.665, P = 0.003) and overall survival (hazard ratio = 2.398, 95% confidence interval: 1.334-4.311, P = 0.004). The incidence of grade ≥ 2 late intestinal adverse events was 8.7% in the accurate target volume group, lower than 18.5% in the inaccurate target volume group (P = 0.033).

CONCLUSION

Target volume delineation accuracy is an independent prognostic factor affecting the efficacy and safety of postoperative radiotherapy for colorectal cancer. Accurate target volume delineation can improve locoregional control and survival outcomes, reduce marginal and out-of-field recurrence, and decrease late adverse events. Clinical practice should strengthen quality control of target volume delineation, follow standardized guidelines, adopt multidisciplinary collaboration models, and ensure accuracy of target volume delineation.

Key Words: Colorectal cancer; Postoperative radiotherapy; Target volume delineation; Survival prognosis; Recurrence pattern; Adverse events

Core Tip: Accurate target volume delineation is essential for effective postoperative radiotherapy in colorectal cancer. This study demonstrates that inaccurate target definition leads to markedly higher marginal and out-of-field recurrence, inferior disease-free and overall survival, and increased late intestinal toxicity. Target accuracy emerged as an independent prognostic factor comparable to nodal stage and margin status. These findings underscore the need for standardized delineation guidelines, multidisciplinary review, and advanced imaging integration to ensure optimal tumor coverage and minimize treatment failures.

INTRODUCTION

Colorectal cancer is one of the most common malignancies worldwide, with both incidence and mortality rates ranking among the highest[1]. Although surgical resection remains the cornerstone of curative treatment for colorectal cancer, the local recurrence rate after surgery alone can still reach 15%-30% for locally advanced patients (pT3-4 and/or N+)[2]. Postoperative adjuvant radiotherapy has been proven to significantly reduce local recurrence risk and improve patient survival prognosis, especially for rectal cancer and high-risk colon cancer patients[3]. With the widespread application of precise radiotherapy techniques such as intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT), the conformality and uniformity of radiation dose distribution have been significantly improved, effectively reducing radiation dose to surrounding normal tissues while ensuring tumor target volume dose coverage.

However, the application of precise radiotherapy techniques places higher demands on the accuracy of target volume delineation. Target volume delineation is the primary step in radiotherapy planning and directly determines the precision and effectiveness of radiotherapy[4]. Target volume delineation for postoperative radiotherapy of colorectal cancer is relatively complex, requiring comprehensive consideration of multiple factors, including primary tumor location, anatomical structural changes caused by surgical procedures, regional lymphatic drainage pathways, and high-risk recurrence areas. Although multiple international consensus guidelines for target volume delineation have been published, there remains considerable variation in the accuracy of target volume delineation in clinical practice, which may be related to factors such as radiation oncologist experience level, degree of guideline understanding, and individual patient anatomical variations[5]. Inaccurate target volume delineation may lead to omission of high-risk areas or excessive irradiation of normal tissues, thereby affecting local control (LC) rates and treatment-related adverse events.

Currently, research on the impact of target volume delineation accuracy on the efficacy of postoperative radiotherapy for colorectal cancer is relatively limited, mostly consisting of small-sample retrospective analyses or target volume delineation quality assessment studies[6]. There is a lack of large-sample, systematic clinical studies to quantitatively assess the relationship between target volume delineation accuracy and patient survival prognosis and recurrence patterns[7]. Therefore, this study retrospectively analyzed clinical data from 218 colorectal cancer patients who received postoperative adjuvant radiotherapy, comparing survival outcomes, recurrence patterns, and adverse events between accurate and inaccurate target volume groups, aiming to elucidate the impact of target volume delineation accuracy on treatment outcomes and provide clinical evidence for optimizing quality control of target volume delineation in postoperative radiotherapy for colorectal cancer.

MATERIALS AND METHODS

Study population

Clinical data of 218 colorectal cancer patients who underwent radical surgery followed by adjuvant radiotherapy at our hospital from January 2019 to December 2022 were retrospectively analyzed. Inclusion criteria: (1) Pathologically confirmed colorectal adenocarcinoma; (2) Underwent radical surgical treatment (R0 resection); (3) Postoperative pathological staging of pT3-4 and/or N+; (4) received IMRT or VMAT after surgery; and (5) Complete clinical data and available follow-up information. Exclusion criteria: (1) Distant metastasis confirmed by preoperative imaging examination or postoperative pathology; (2) History of previous pelvic radiotherapy; (3) Concurrent other malignancies; (4) Radiotherapy completion rate < 90%; and (5) Severe cardiac, hepatic, or renal dysfunction. This study was approved by the hospital ethics committee and complied with the principles of the Declaration of Helsinki. As this was a retrospective analysis that did not affect patient treatment decisions, informed consent was waived with ethics committee approval.

Clinical data collection

General information and clinicopathological characteristics of patients were collected, including: Gender, age, performance status score [Eastern Cooperative Oncology Group (ECOG) score], primary tumor location (rectum or colon), pathological type, degree of differentiation, pT stage, pN stage, number of lymph node metastases, vascular invasion status, neural invasion status, margin status, preoperative carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 levels, surgical approach (open or laparoscopic), whether neoadjuvant therapy was received, postoperative chemotherapy regimen and number of cycles. All pathological data were reviewed and confirmed by 2 senior pathologists.

Target volume delineation and quality assessment

All patients were positioned supine and immobilized with thermoplastic body masks. Computed tomography (CT) simulation positioning scanning was performed with 3 mm slice thickness, and the scanning range was determined according to the primary tumor location: For rectal cancer and left colon cancer patients, from the 5^th lumbar vertebra to 5 cm below the ischial tuberosity; for right colon cancer and transverse colon cancer patients, from the 12^th thoracic vertebra to 5 cm below the ischial tuberosity, to include the para-aortic lymphatic drainage area. Images were transmitted to the radiotherapy planning system (Eclipse or Pinnacle). Target volume delineation was performed according to the Radiation Therapy Oncology Group (RTOG) 0822 consensus guidelines for rectal cancer and relevant domestic expert consensus for colon cancer, with explicit anatomical boundaries defined for mesorectal, presacral, and regional nodal volumes based on standardized lymphatic drainage patterns. Clinical target volume (CTV) delineation included: Tumor bed, anastomosis, and regional lymphatic drainage areas. For rectal cancer patients, CTV included pelvic lymphatic drainage areas (presacral, obturator, internal iliac, external iliac lymph node regions), while for colon cancer patients, corresponding regional lymph node drainage areas were included based on primary location. Planning target volume was created by expanding CTV by 7-10 mm (considering setup error and organ motion).

This study employed a retrospective assessment method to evaluate the quality of target volume delineation for patients who had completed treatment. All delineations were independently reviewed by three senior radiation oncologists with specialized expertise in gastrointestinal malignancies (each with > 10 years of experience), using consensus-based assessment with predefined anatomical landmarks and lymphatic drainage patterns according to the RTOG 0822 guidelines. Based on target volume delineation quality, patients were divided into accurate and inaccurate target volume groups. The accurate target volume group (n = 126) was defined as target volume delineation meeting guideline requirements, including all high-risk areas, with clear and reasonable boundaries, approved by senior radiation oncologists (> 10 years of work experience). The inaccurate target volume group (n = 92) was defined as target volume delineation having one or more of the following issues: (1) Omission of high-risk lymphatic drainage areas; (2) Inappropriate target volume boundaries (excessive reduction or expansion > 1 cm in a single dimension); (3) Tumor bed localization deviation > 5 mm; and (4) Insufficient consideration of surgical factors (such as anatomical structural changes).

Radiotherapy regimen

6 MV X-ray IMRT or VMAT technique was used. Prescription dose: Postoperative adjuvant radiotherapy 50 Gy/25 fractions, 5 fractions/week, with simultaneous integrated boost to 55-56 Gy/25 fractions for areas with high-risk factors (positive margins or < 1 mm from margins, extracapsular lymph node invasion, intraoperative tumor rupture). Organ at risk dose constraints followed RTOG standards: Small bowel V45 < 195 cc, bladder V50 < 50%, femoral head V50 < 5%, bone marrow V40 < 40%. Based on patient performance status and pathological characteristics, patients with ECOG score 0-1 and no obvious contraindications to chemotherapy received a concurrent oxaliplatin plus capecitabine regimen, modified FOLFOX6 (folinic acid, fluorouracil, and oxaliplatin) regimen regimen for concurrent chemoradiotherapy; patients aged > 70 years or ECOG score 2 or with significant underlying diseases received single-agent capecitabine chemotherapy (825 mg/m², bid, oral on radiotherapy days). All patients received concurrent capecitabine during the 5-week radiotherapy course, followed by adjuvant chemotherapy with the FOLFOX (folinic acid, fluorouracil, and oxaliplatin) or CAPOX (capecitabine, oxaliplatin) regimen for a total treatment duration of six months (including the concurrent phase), with chemotherapy cycles initiated within 4 weeks after radiotherapy completion.

Observation indicators

Primary efficacy evaluation indicators included LC rate, regional control rate (RC), disease-free survival rate (DFS), and overall survival rate (OS). LC rate was defined as the proportion of patients with no local recurrence confirmed by imaging, and RC rate was defined as the proportion of patients with no regional lymph node recurrence confirmed by imaging. DFS time was defined as the time from radiotherapy start to recurrence at any site, second primary tumor, or death, and OS time was defined as the time from radiotherapy start to death from any cause or last follow-up. Recurrence diagnosis was based on imaging examination [CT, magnetic resonance imaging (MRI), or positron emission tomography-CT] combined with elevated tumor markers, with pathological biopsy confirmation when necessary.

Recurrence pattern analysis was classified according to the spatial relationship between the recurrence site and the original radiotherapy target volume: In-field recurrence was defined as 95% of the recurrence volume located within the 95% isodose line of the prescription dose; marginal recurrence was defined as 20%-95% of the recurrence volume located within the 95% isodose line of the prescription dose; out-of-field recurrence was defined as < 20% of the recurrence volume located within the 95% isodose line of the prescription dose. Recurrence patterns were jointly determined by 2 radiation oncologists and 1 radiologist, with registration and analysis of recurrence imaging with the original radiotherapy plan. Acute radiotherapy adverse events were assessed weekly during radiotherapy using Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 to evaluate gastrointestinal reactions (nausea, vomiting, diarrhea), urinary system reactions, hematologic toxicity, and skin reactions. Late adverse events were defined as adverse events occurring 3 months after radiotherapy completion and were continuously assessed during follow-up.

Follow-up

Patients were followed up every 3 months for 2 years after radiotherapy completion; thereafter every 6 months until 5 years; and annually after 5 years. Standardized surveillance included contrast-enhanced CT imaging of the chest, abdomen, and pelvis at each follow-up visit for the first two years, then annually thereafter, combined with serum CEA monitoring and colonoscopy at 1 year and every 2-3 years subsequently, ensuring reliable recurrence detection and pattern classification. The follow-up cutoff date was June 30, 2025.

Statistical analysis

SPSS version 26.0 statistical software was used for data analysis. Continuous variables were expressed as mean ± SD or median (range), with normality tested using the Shapiro-Wilk test and homogeneity of variance tested using Levene’s test. An independent samples t-test was used for comparison between groups for continuous variables meeting normality and homogeneity of variance; otherwise Mann-Whitney U test was used. Categorical variables were expressed as n (%), with group comparisons using χ² test or Fisher’s exact test (when theoretical frequency < 5). Survival analysis used the Kaplan-Meier method to plot survival curves, with the Log-rank test for comparing group differences. Cox proportional hazards regression model was used for univariate and multivariate analysis, with variables having P < 0.10 in univariate analysis included in multivariate analysis to screen independent prognostic factors. The proportional hazards assumption of Cox regression was tested using Schoenfeld residuals. P < 0.05 (two-sided test) was considered statistically significant.

RESULTS

Patient baseline characteristics

Among 218 patients, there were 132 males and 86 females, with a median age of 59 years (range 32-78 years). There were 126 cases in the accurate target volume group and 92 cases in the inaccurate target volume group. There were no statistically significant differences between the two groups in baseline characteristics including gender, age, ECOG score, primary tumor location, pathological type, pathological differentiation, pT stage, pN stage, vascular invasion, neural invasion, margin status, preoperative CEA level, preoperative carbohydrate antigen 19-9 level, surgical approach, neoadjuvant therapy, chemotherapy regimen, and number of chemotherapy cycles (P > 0.05), indicating comparability (Table 1).

Table 1 Comparison of clinicopathological characteristics between the two groups, n (%).

Characteristic	Accurate target volume group (n = 126)	Inaccurate target volume group (n = 92)	χ2/Z value	P value
Gender			0.428	0.513
Male	78 (61.9)	54 (58.7)
Female	48 (38.1)	38 (41.3)
Age			0.195	0.659
≤ 60 years	68 (54.0)	52 (56.5)
> 60 years	58 (46.0)	40 (43.5)
Age [years, M (range)]	58 (32-78)	60 (35-77)	-0.864	0.388
ECOG score			1.203	0.548
Score 0	52 (41.3)	34 (37.0)
Score 1	62 (49.2)	48 (52.2)
Score 2	12 (9.5)	10 (10.9)
Primary tumor location			0.015	0.903
Rectum	76 (60.3)	56 (60.9)
Colon	50 (39.7)	36 (39.1)
Pathological type			0.003	0.958
Adenocarcinoma	118 (93.7)	86 (93.5)
Mucinous adenocarcinoma	8 (6.3)	6 (6.5)
Pathological differentiation			0.782	0.676
Well differentiated	18 (14.3)	10 (10.9)
Moderately differentiated	86 (68.3)	66 (71.7)
Poorly differentiated	22 (17.5)	16 (17.4)
pT stage			0.621	0.431
T3	94 (74.6)	65 (70.7)
T4	32 (25.4)	27 (29.3)
pN stage			1.847	0.397
N0	38 (30.2)	24 (26.1)
N1	58 (46.0)	38 (41.3)
N2	30 (23.8)	30 (32.6)
Vascular invasion			0.127	0.722
No	84 (66.7)	59 (64.1)
Yes	42 (33.3)	33 (35.9)
Neural invasion			0.009	0.924
No	98 (77.8)	71 (77.2)
Yes	28 (22.2)	21 (22.8)
Margin status			0.583	0.445
Negative	112 (88.9)	78 (84.8)
Positive or close	14 (11.1)	14 (15.2)
Preoperative CEA level			0.152	0.697
Normal	78 (61.9)	54 (58.7)
Elevated	48 (38.1)	38 (41.3)
Preoperative CA19-9 level			0.341	0.559
Normal	94 (74.6)	65 (70.7)
Elevated	32 (25.4)	27 (29.3)
Surgical approach			0.098	0.754
Open	46 (36.5)	32 (34.8)
Laparoscopic	80 (63.5)	60 (65.2)
Neoadjuvant therapy			0.218	0.641
No	98 (77.8)	74 (80.4)
Yes	28 (22.2)	18 (19.6)
Concurrent chemotherapy regimen			0.446	0.504
mFOLFOX6	88 (69.8)	61 (66.3)
Single-agent capecitabine	38 (30.2)	31 (33.7)
Number of chemotherapy cycles			0.528	0.768
≤ 4 cycles	26 (20.6)	21 (22.8)
4-6 cycles	82 (65.1)	58 (63.0)
> 6 cycles	18 (14.3)	13 (14.1)

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; ECOG: Eastern Cooperative Oncology Group.

Comparison of survival outcomes between two groups

As of June 30, 2025, the median follow-up time was 38 months (range 6-66 months). During follow-up, 63 patients experienced disease progression, including 26 cases (20.6%) in the accurate target volume group and 37 cases (40.2%) in the inaccurate target volume group. A total of 42 patients died, including 15 cases (11.9%) in the accurate target volume group and 27 cases (29.3%) in the inaccurate target volume group. The 3-year LC, RC, DFS, and OS were 92.1% [95% confidence interval (CI): 87.2%-97.0%], 94.4% (95%CI: 90.1%-98.7%), 83.6% (95%CI: 77.1%-90.1%), and 89.2% (95%CI: 83.6%-94.8%) in the accurate target volume group, respectively, compared to 78.3% (95%CI: 69.8%-86.8%), 81.5% (95%CI: 73.5%-89.5%), 66.8% (95%CI: 57.2%-76.4%), and 72.5% (95%CI: 63.4%-81.6%) in the inaccurate target volume group, with statistically significant differences between the two groups (P < 0.05) (Table 2, Figure 1). Kaplan-Meier survival curves showed that the accurate target volume group was significantly superior to the inaccurate target volume group in both DFS and OS (Log-rank test, P = 0.001 and P < 0.001).

Figure 1 Kaplan-Meier survival curves comparing accurate vs inaccurate target volume delineation groups. A: Kaplan-Meier curves for local control; B: Regional control; C: Disease-free survival; D: Overall survival stratified by target volume delineation accuracy. Blue solid lines represent the accurate delineation group, and red dashed lines represent the inaccurate delineation group. The 3-year local control, regional control, disease-free survival, and overall survival rates were 92.1% vs 78.3%, 94.4% vs 81.9%, 83.6% vs 66.8%, and 89.2% vs 72.5% for the accurate and inaccurate groups, respectively. Patients with accurate target volume delineation demonstrated significantly superior outcomes across all four endpoints (all log-rank P < 0.005). LC: Local control; RC: Regional control; DFS: Disease-free survival; OS: Overall survival.

Table 2 Comparison of survival outcomes between the two groups, n (%).

Survival indicator	Accurate target volume group (n = 126)	Inaccurate target volume group (n = 92)	χ2 value	P value
3-year local control rate (95%CI)	92.1 (87.2-97.0)	78.3 (69.8-86.8)	8.742	0.003
3-year regional control rate (95%CI)	94.4 (90.1-98.7)	81.5 (73.5-89.5)	9.328	0.002
3-year disease-free survival rate (95%CI)	83.6 (77.1-90.1)	66.8 (57.2-76.4)	10.156	0.001
3-year overall survival rate (95%CI)	89.2 (83.6-94.8)	72.5 (63.4-81.6)	11.483	< 0.001

CI: Confidence interval.

Recurrence pattern analysis

During follow-up, 63 cases of disease progression occurred, including 56 cases of local or regional recurrence (23 cases in the accurate target volume group, 33 cases in the inaccurate target volume group) and 7 cases of distant metastasis only (3 cases in the accurate target volume group, 4 cases in the inaccurate target volume group). The following analysis focused only on the 56 patients with local or regional recurrence. In the accurate target volume group, there were 15 cases of in-field recurrence (65.2%), 5 cases of marginal recurrence (21.7%), and 3 cases of out-of-field recurrence (13.0%); in the inaccurate target volume group, there were 8 cases of in-field recurrence (24.2%), 14 cases of marginal recurrence (42.4%), and 11 cases of out-of-field recurrence (33.3%). The distribution of recurrence patterns differed statistically significantly between the two groups (χ² = 10.236, P = 0.006) (Table 3). Further analysis showed that 75.8% (25/33) of recurrences in the inaccurate target volume group occurred in marginal or out-of-field regions, compared to only 34.8% (8/23) in the accurate target volume group, with a statistically significant difference (χ² = 9.824, P = 0.002).

Table 3 Comparison of recurrence patterns between the two groups, n (%).

Recurrence pattern	Accurate target volume group (n = 23)	Inaccurate target volume group (n = 33)	χ2 value1	P value
In-field recurrence	15 (65.2)	8 (24.2)	10.236	0.006
Marginal recurrence	5 (21.7)	14 (42.4)
Out-of-field recurrence	3 (13.0)	11 (33.3)

¹χ² test was used for overall comparison (df = 2).

Univariate analysis of factors affecting DFS

Univariate analysis results showed that target volume delineation accuracy (P = 0.002), pT stage (P = 0.015), pN stage (P = 0.008), lymph node metastasis status (P = 0.012), neural invasion (P = 0.034), margin status (P = 0.006), and preoperative CEA level (P = 0.041) were associated with DFS (P < 0.10) (Table 4).

Table 4 Univariate analysis of factors affecting disease-free survival.

Factor	HR	95%CI	P value
Gender (male vs female)	1.153	0.682-1.950	0.596
Age (> 60 years vs ≤ 60 years)	1.284	0.768-2.146	0.339
ECOG score (2 vs 0-1)	1.562	0.785-3.107	0.204
Tumor location (colon vs rectum)	0.894	0.524-1.524	0.681
Differentiation (poor vs moderate-well)	1.418	0.757-2.656	0.275
pT stage (T4 vs T3)a	1.856	1.126-3.059	0.015a
pN stage (N2 vs N0-1)a	1.924	1.187-3.118	0.008a
Lymph node metastasis (N+ vs N0)a	1.832	1.145-2.930	0.012a
Vascular invasion (yes vs no)	1.473	0.886-2.449	0.136
Neural invasion (yes vs no)a	1.726	1.041-2.863	0.034a
Margin status (positive vs negative)a	2.168	1.243-3.781	0.006a
Preoperative CEA (elevated vs normal)a	1.628	1.021-2.596	0.041a
Preoperative CA19-9 (elevated vs normal)	1.384	0.824-2.323	0.218
Surgical approach (laparoscopic vs open)	0.823	0.495-1.368	0.452
Neoadjuvant therapy (yes vs no)	0.765	0.413-1.418	0.398
Chemotherapy regimen (single vs combination)	1.245	0.735-2.109	0.415
Target delineation (inaccurate vs accurate)a	2.254	1.356-3.746	0.002a

^aP < 0.10, included in multivariate analysis.

HR: Hazard ratio; CI: Confidence interval; CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; ECOG: Eastern Cooperative Oncology Group.

Multivariate analysis of factors affecting DFS

Variables with P < 0.10 in univariate analysis were included in Cox multivariate regression analysis. Results showed that target volume delineation accuracy [Hazard ratio (HR) = 2.186, 95%CI: 1.304-3.665, P = 0.003], pN stage (HR = 1.782, 95%CI: 1.089-2.915, P = 0.021), and margin status (HR = 1.928, 95%CI: 1.095-3.396, P = 0.023) were independent prognostic factors affecting DFS (Table 5). The proportional hazards assumption of the Cox regression model was verified by the Schoenfeld residual test (P = 0.156), indicating good model fit.

Table 5 Multivariate analysis of factors affecting disease-free survival.

Factor	HR	95%CI	P value
Target delineation (inaccurate vs accurate)	2.186	1.304-3.665	0.003
pT stage (T4 vs T3)	1.524	0.914-2.541	0.105
pN stage (N2 vs N0-1)	1.782	1.089-2.915	0.021
Lymph node metastasis (N+ vs N0)	1.386	0.832-2.308	0.21
Neural invasion (yes vs no)	1.448	0.864-2.427	0.16
Margin status (positive vs negative)	1.928	1.095-3.396	0.023
Preoperative CEA (elevated vs normal)	1.396	0.871-2.238	0.166

HR: Hazard ratio; CI: Confidence interval; CEA: Carcinoembryonic antigen.

Univariate analysis of factors affecting OS

Univariate analysis results showed that target volume delineation accuracy (P = 0.001), pT stage (P = 0.023), pN stage (P = 0.004), lymph node metastasis status (P = 0.008), neural invasion (P = 0.042), margin status (P = 0.003), preoperative CEA level (P = 0.036), and vascular invasion (P = 0.089) were associated with OS (P < 0.10) (Table 6).

Table 6 Univariate analysis of factors affecting overall survival.

Factor	HR	95%CI	P value
Gender (male vs female)	1.268	0.698-2.304	0.436
Age (> 60 years vs ≤ 60 years)	1.395	0.774-2.516	0.268
ECOG score (2 vs 0-1)	1.684	0.782-3.626	0.182
Tumor location (colon vs rectum)	0.856	0.465-1.576	0.619
Differentiation (poor vs moderate-well)	1.538	0.762-3.104	0.229
pT stage (T4 vs T3)a	1.982	1.096-3.584	0.023a
pN stage (N2 vs N0-1)a	2.156	1.268-3.666	0.004a
Lymph node metastasis (N+ vs N0)a	2.034	1.205-3.434	0.008a
Vascular invasion (yes vs no)a	1.658	0.928-2.964	0.089a
Neural invasion (yes vs no)a	1.812	1.022-3.213	0.042a
Margin status (positive vs negative)a	2.486	1.352-4.572	0.003a
Preoperative CEA (elevated vs normal)a	1.746	1.036-2.944	0.036a
Preoperative CA19-9 (elevated vs normal)	1.462	0.802-2.665	0.214
Surgical approach (laparoscopic vs open)	0.795	0.444-1.424	0.442
Neoadjuvant therapy (yes vs no)	0.728	0.362-1.465	0.374
Chemotherapy regimen (single vs combination)	1.318	0.725-2.397	0.364
Target delineation (inaccurate vs accurate)a	2.542	1.426-4.532	0.001a

^aP < 0.10, included in multivariate analysis.

HR: Hazard ratio; CI: Confidence interval; CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; ECOG: Eastern Cooperative Oncology Group.

Multivariate analysis of factors affecting OS

Variables with P < 0.10 in univariate analysis were included in Cox multivariate regression analysis. Results showed that target volume delineation accuracy (HR = 2.398, 95%CI: 1.334-4.311, P = 0.004), pN stage (HR = 1.946, 95%CI: 1.132-3.345, P = 0.016), and margin status (HR = 2.154, 95%CI: 1.156-4.013, P = 0.016) were independent prognostic factors affecting OS (Table 7). The C-index of the model was 0.712 (95%CI: 0.654-0.770), showing good predictive capability.

Table 7 Multivariate analysis of factors affecting overall survival.

Factor	HR	95%CI	P value
Target delineation (inaccurate vs accurate)	2.398	1.334-4.311	0.004
pT stage (T4 vs T3)	1.586	0.872-2.885	0.132
pN stage (N2 vs N0-1)	1.946	1.132-3.345	0.016
Lymph node metastasis (N+ vs N0)	1.468	0.842-2.560	0.176
Vascular invasion (yes vs no)	1.324	0.736-2.382	0.348
Neural invasion (yes vs no)	1.512	0.846-2.702	0.162
Margin status (positive vs negative)	2.154	1.156-4.013	0.016
Preoperative CEA (elevated vs normal)	1.448	0.852-2.461	0.171

CEA: Carcinoembryonic antigen.

Adverse events

There was no statistically significant difference in the incidence of acute adverse events between the two groups (P > 0.05). The incidence of grade 3-4 acute adverse events was 15.1% in the accurate target volume group and 17.4% in the inaccurate target volume group (P = 0.643). Regarding late adverse events, the incidence of grade ≥ 2 intestinal adverse events was 8.7% in the accurate target volume group, lower than 18.5% in the inaccurate target volume group (P = 0.033) (Table 8). The radiotherapy completion rate was > 95% in both groups, with no significant difference in the proportion of patients who interrupted treatment due to acute toxic reactions (2.4% in the accurate target volume group vs 3.3% in the inaccurate target volume group, P = 0.681).

Table 8 Comparison of adverse events between two groups, n (%).

Adverse event	Accurate target volume group (n = 126)	Inaccurate target volume group (n = 92)	χ2 value	P value
Acute adverse events
Grade 1-2 gastrointestinal reactions	94 (74.6)	72 (78.3)	0.424	0.515
Grade 3-4 gastrointestinal reactions	12 (9.5)	10 (10.9)	0.118	0.731
Grade 1-2 hematologic toxicity	68 (54.0)	54 (58.7)	0.491	0.484
Grade 3-4 hematologic toxicity	7 (5.6)	6 (6.5)	0.083	0.773
Grade 1-2 urinary system reactions	32 (25.4)	26 (28.3)	0.23	0.632
Grade 3-4 urinary system reactions	2 (1.6)	3 (3.3)	0.678	0.41
Grade 1-2 skin reactions	28 (22.2)	24 (26.1)	0.449	0.503
Grade 3-4 skin reactions	1 (0.8)	2 (2.2)	0.799	0.371
Any grade 3-4 acute adverse events1	19 (15.1)	16 (17.4)	0.215	0.643
Late adverse events
Grade ≥ 2 intestinal adverse events	11 (8.7)	17 (18.5)	4.554	0.033
Grade ≥ 2 urinary system adverse events	6 (4.8)	8 (8.7)	1.464	0.226
Intestinal obstruction	3 (2.4)	6 (6.5)	2.245	0.134

¹Any grade 3-4 acute adverse events refer to the total number of patients experiencing any grade 3-4 toxic reaction. In the accurate target volume group, 3 patients experienced two types of grade 3-4 toxicities simultaneously; in the inaccurate target volume group, 5 patients experienced two types of grade 3-4 toxicities simultaneously.

DISCUSSION

This study found that the 3-year LC and RC in the accurate target volume group were 92.1% and 94.4%, respectively, significantly higher than 78.3% and 81.5% in the inaccurate target volume group. This result is basically consistent with previous literature reports[8]. Studies have reported that standardized target volume delineation can reduce the local recurrence rate by approximately 40%[9]. In this study, accurate target volume delineation improved the LC rate by 13.8 percentage points, an improvement of significant clinical importance.

Recurrence pattern analysis provided direct evidence for understanding the mechanism by which target volume delineation accuracy affects efficacy. This study showed that 75.8% of recurrences in the inaccurate target volume group occurred in marginal or out-of-field regions, compared to only 34.8% in the accurate target volume group, indicating that inaccurate target volume delineation led to insufficient dose coverage of high-risk areas. Some scholars have found through spatial registration analysis of recurrent lesions with original radiotherapy plans that approximately 60%-70% of local recurrences are related to inappropriate target volume delineation[10]. The proportion of marginal and out-of-field recurrence in the inaccurate target volume group in this study was even higher, possibly related to factors such as postoperative anatomical structural changes in colorectal cancer, complex lymphatic drainage pathways, and difficulty in identifying high-risk areas[11].

Specific manifestations of inaccurate target volume delineation mainly include omission of high-risk lymphatic drainage areas, inappropriate target volume boundary settings, tumor bed localization deviation, and insufficient consideration of surgical factors. The regional lymphatic drainage pathways of colorectal cancer, especially rectal cancer, are complex, including presacral, obturator, internal iliac, and external iliac lymph node regions, with unclear anatomical landmarks after surgery increasing delineation difficulty[12]. Previous studies have shown significant differences in target volume delineation by different radiation oncologists for the same patient, with target volume coefficient of variation reaching 30%-50%[13]. Standardized target volume delineation guidelines and quality control processes can effectively reduce this variability[14].

Through post-hoc analysis, we identified three primary contributing factors to delineation inaccuracy: Operator inexperience (cases performed by residents without adequate supervision, accounting for 42% of inaccurate cases), inadequate imaging quality (particularly poor soft-tissue contrast on non-contrast CT simulation, 31%), and anatomical complexity in post-surgical settings with altered anatomy (27%). Our preliminary subgroup analysis further revealed that the presacral space and lateral pelvic sidewall nodes were the most vulnerable anatomical regions for delineation errors, consistent with areas where anatomical landmarks are most affected by surgical disruption. These findings emphasize the need for structured training programs, enhanced imaging protocols, including MRI fusion when available, and mandated multidisciplinary review for complex post-surgical anatomy.

Multivariate analysis in this study showed that target volume delineation accuracy was an independent prognostic factor affecting DFS (HR = 2.186) and OS (HR = 2.398), with an impact degree comparable to pN stage and margin status. This finding highlights the central position of target volume delineation in postoperative radiotherapy for colorectal cancer. Studies using propensity score matching analysis have found that standardized target volume delineation can improve the 3-year OS by approximately 15%[15], close to the results of this study (16.7 percentage point improvement).

The mechanism by which target volume delineation accuracy affects survival prognosis is multifaceted. First, accurate target volume delineation ensures that high-risk recurrence areas receive sufficient radiotherapy dose coverage, thereby effectively controlling subclinical lesions[16]. Postoperative microscopic residual disease and regional lymph node micrometastases in colorectal cancer are the main sources of local recurrence, and adequate dose coverage can significantly reduce recurrence risk[17]. Second, accurate target volume delineation can avoid excessive irradiation of normal tissues while ensuring tumor control, maintaining patient treatment compliance, and quality of life[18]. In this study, the incidence of late intestinal adverse events in the accurate target volume group was significantly lower than in the inaccurate target volume group, suggesting that reasonable target volume settings help reduce treatment-related toxicity.

It is noteworthy that this study found that pN stage and margin status were also independent factors affecting prognosis, consistent with literature reports[19]. Patients with positive lymph node metastasis and positive margins still have a higher recurrence risk even after receiving postoperative radiotherapy and require more accurate target volume delineation and possible dose escalation[20]. For such high-risk patients, the accuracy of target volume delineation may be more critical, as any omission of high-risk areas may lead to treatment failure.

This study showed that there was no significant difference in the incidence of acute adverse events between the two groups, but the incidence of late intestinal adverse events in the accurate target volume group was significantly lower than in the inaccurate target volume group (8.7% vs 18.5%). This result suggests that accurate target volume delineation can not only improve tumor control but also reduce treatment-related toxicity, achieving a better therapeutic window. Radiation-induced intestinal injury is the most common late complication of postoperative radiotherapy for colorectal cancer, seriously affecting patient quality of life[21].

The reasons for increased late adverse events in the inaccurate target volume group may be multifaceted. First, inappropriate target volume boundaries leading to excessive expansion increase the irradiation volume and dose to the normal intestine[22]. Studies have shown that for every 50 cc increase in small bowel V45, the risk of grade ≥ 3 intestinal adverse events increases by approximately 30%[23]. Second, inaccurate target volume delineation may lead to difficulty in plan optimization, sacrificing normal tissue protection to meet target volume dose coverage requirements[24]. Additionally, target volume localization deviation may result in important organs receiving doses higher than planned during actual treatment.

Modern precise radiotherapy techniques, such as IMRT and VMAT, can achieve highly conformal dose distributions, but this advantage is based on accurate target volume delineation[25]. If target volume delineation is inaccurate, precise radiotherapy techniques not only fail to deliver their advantages but may even lead to insufficient dose in high-risk areas or excessive dose to normal tissues due to steep dose gradients. Therefore, while promoting the application of precise radiotherapy techniques, quality control of target volume delineation must be strengthened.

Based on the results of this study, improving the accuracy of target volume delineation in postoperative radiotherapy for colorectal cancer has important clinical value. First, professional training of radiation oncologists should be strengthened, particularly systematic learning of pelvic anatomy and lymphatic drainage pathways[26]. Multi-center studies have shown that the consistency of target volume delineation by radiation oncologists who have undergone standardized training can improve by more than 40%[27]. Second, multimodal image fusion technology should be fully utilized, combining preoperative MRI, postoperative CT, and positron emission tomography-CT when necessary to accurately identify the tumor bed and high-risk lymphatic drainage areas[28].

Establishing standardized target volume delineation processes and quality control systems is equally important. It is recommended to adopt a multidisciplinary team model, with radiation oncology, surgery, and radiology physicians jointly discussing and determining target volume ranges[29]. For complex cases, peer review should be conducted, with senior physicians reviewing target volume delineation quality. Additionally, the development of artificial intelligence-assisted target volume delineation technology provides new tools for improving delineation accuracy and consistency[30]. Preliminary studies have shown that deep learning-based automatic delineation systems can achieve accuracy comparable to expert physicians and significantly reduce delineation time[31].

This study has some limitations. First, as a retrospective study, there may be selection bias and confounding factors. Although baseline characteristics of the two groups were balanced, some unmeasured factors, such as radiation oncologist experience and image quality, may affect the results. Second, the assessment of target volume delineation accuracy was mainly based on whether it met guideline requirements, and this dichotomous classification method may not fully reflect the continuous changes in target volume delineation quality. Future studies could adopt more refined quantitative assessment methods, such as target volume coefficient of variation and high-risk area coverage index. Third, the sample size of this study was relatively limited, with insufficient subgroup analysis capability, failing to explore in depth the differences in the impact of factors such as different tumor locations and stages on target volume delineation accuracy.

Future research should conduct prospective randomized controlled trials to further verify the effectiveness of target volume delineation quality control intervention measures. At the same time, a standardized target volume delineation quality assessment system should be established, and objective quantitative indicators developed. The application prospects of artificial intelligence technology in target volume delineation deserve in-depth research, including optimization of automatic delineation algorithms and establishment of individualized target volume prediction models. Additionally, the value of new imaging technologies such as functional MRI and metabolic positron emission tomography in optimizing target volume delineation should be explored.

Future prospective studies should evaluate the impact of atlas-based auto-contouring systems, artificial intelligence-assisted delineation tools, and structured training programs on improving delineation quality and patient outcomes. Additionally, research should focus on identifying and validating quantitative metrics for delineation quality assessment beyond subjective expert review. With the continuous development of precise radiotherapy techniques, the accuracy of target volume delineation will increasingly become a key factor determining treatment success or failure.

CONCLUSION

In conclusion, this study confirmed that target volume delineation accuracy significantly affects the efficacy and safety of postoperative radiotherapy for colorectal cancer. Accurate target volume delineation can improve locoregional control and survival outcomes while reducing late adverse events. Target volume delineation accuracy is an independent prognostic factor, and its clinical importance is no less than that of traditional pathological risk factors. These findings emphasize the necessity of strengthening quality control of target volume delineation in postoperative radiotherapy for colorectal cancer. It is recommended that clinical practice should follow standardized guidelines, adopt multidisciplinary collaboration models, utilize advanced imaging technologies, and establish comprehensive quality assurance systems to ensure that every patient receives optimal radiotherapy target volume design.