Rethinking surgical strategy for margin-positive only T1 colorectal cancer: a multicenter retrospective cohort study Authors: Jin Hyeong Park, Chang Hwan Kim, Dae Hee Pyo, Jong Kyung Park, Chaedong Lim, Ji Hoon Kim, Bong-Hyeon Kye, Hyung Jin Kim Published: 2026 ABSTRACT PurposeEndoscopic resection is frequently performed for early T1 colorectal cancer (CRC), with additional surgery indicated when histopathological risk factors are present, such as a positive vertical margin, submucosal invasion depth greater than 1000 µm, lymphatic or vascular invasion, poorly differentiated adenocarcinoma, signet-ring cell carcinoma, mucinous carcinoma, or grade 2/3 tumor budding at the deepest invasive site. In patients undergoing radical resection for T1 CRC, the incidence of lymph node metastases(LNM) has been reported to be approximately 10%. The purpose of this study is to identify factors influencing the decision for radical resection in patients with pathologically confirmed T1 CRC following endoscopic resection and to evaluate lymph node metastasis rates according to each risk factor, with a focus on determining whether radical resection might be safely avoided in carefully selected low-risk cases.MethodsThe study was designed as a retrospective multicenter comparative cohort study conducted at three tertiary referral centers in Korea. The study population consisted of patients with pathological T1 colorectal cancer who underwent radical resection following endoscopic resection between April 2019 and December 2024. The main outcome measure was the incidence of lymph node metastasis according to pathological risk factors, with a particular focus on patients with margin-positive lesions without any other risk factors. Univariate analyses were performed, and exact binomial confidence intervals were calculated for the margin-positive subgroups.ResultsAmong 250 patients who underwent radical surgery after endoscopic resection, 25 (10.0%) had lymph node metastasis. Lymphovascular invasion and KRAS mutation were significantly associated with lymph node metastasis. Twenty-two patients with margin positivity without any other risk factors had no LNM (0/22; 95% exact binomial confidence interval, 0–15.4%). Half (11/22) of these patients had no residual tumors on surgical specimens, whereas the remainder showed only submucosal residual tumors without nodal involvement.ConclusionAmong patients who underwent radical resection after endoscopic resection for T1 CRC, margin positivity without other recorded high-risk features was not associated with LNM in this cohort. However, given the small sample size and wide confidence interval, these findings should be considered hypothesis-generating and require further validation [1]. Introduction Colorectal cancer (CRC) poses a significant global health burden and accounts for approximately 10% of all cancer cases worldwide. It is the third most frequently diagnosed malignancy and the second leading cause of cancer-related mortality [2, 3]. In recent years, the widespread implementation and enhanced accessibility of screening modalities such as colonoscopy and stool-based tests have substantially facilitated early detection of colorectal neoplasia. Consequently, the diagnosis of early-stage CRC, including lesions confined to the mucosa or submucosa, such as T1 tumors, has markedly increased, reflecting both heightened awareness and advances in preventive healthcare strategies [4, 5]. Among patients with CRC, those diagnosed at T1 stage account for approximately 10% of all cases [6]. Most patients with T1 CRC can often undergo curative endoscopic local excisions such as endoscopic submucosal dissection (ESD) [7]. Although classified as T1, colorectal cancer exhibits LNM rate of approximately 10%; therefore, based on clinicopathological findings, additional surgical intervention, including lymph node dissection (LND), may be warranted [8]. Additional surgical intervention with LND is advised when any of the following risk factors are present: a positive endoscopic vertical margin, submucosal invasion depth of > 1000 µm, evidence of lymphatic invasion, vascular invasion, poorly differentiated adenocarcinoma, signet-ring cell carcinoma, mucinous carcinoma, or grade 2/3 tumor budding at the site of deepest invasion [9, 10]. Nevertheless, in T1 CRC, the likelihood of LNM varies substantially, ranging from 7.4 to 46.9%, depending on the specific combination of these risk factors [11]. These observations underscore the need for refined risk stratification to identify patients who may safely avoid additional radical surgery. Importantly, positive resection margins represent a heterogeneous category rather than a single uniform entity. Lateral, basal, and indeterminate involvement can arise from true residual tumor, artifact, or technical failure and may therefore have different clinical implications. Based on the risk factors mentioned above, this study aimed to identify and characterize the factors that led to the performance of radical resection in patients who underwent endoscopic resection and whose final pathological diagnosis was pT1 CRC, to evaluate their distribution, and to determine whether there exists a subgroup of patients in whom radical resection could potentially be avoided. Previous investigations have primarily focused on the association between positive resection margins after endoscopic resection and local residual or recurrent disease, whereas data directly linking margin positivity to lymph node metastasis in T1 CRC remain limited and inconclusive [12]. Especially, we sought to clarify whether a positive resection margin—in the absence of other high-risk features—truly constitutes an independent oncologic risk factor for lymph node metastasis. Given the potential to avoid major surgery in this group if the risk of lymph node metastasis is truly negligible, margin positivity represents a clinically important variable that warrants critical re-evaluation. Materials and methods Study design and patient selection Patients who underwent radical resection following endoscopic resection (including ESD, endoscopic mucosal resection [EMR], and polypectomy) for CRC and were finally diagnosed with pathological T1 disease between April 2019 and December 2024 at three institutions—Eunpyeong St. Mary’s Hospital, Incheon St. Mary’s Hospital, and St. Vincent’s Hospital (The Catholic University of Korea)—were included in this multicenter retrospective study. Among 3529 patients who underwent surgery for CRC during the study period, 551 had a final pathological staging of T1. After excluding patients who received preoperative radiotherapy for rectal cancer (n = 48), underwent local excision without lymph node dissection (n = 23), had insufficient data (n = 5), or underwent upfront radical resection without prior endoscopic resection (n = 225), 250 patients who received radical resection following endoscopic resection were included in the analysis (Fig. 1).Fig. 1Study flowchart. Flow diagram showing patient selection for the study cohort. Among 3529 patients who underwent surgery for colorectal cancer, 551 had pathological T1 stage. After excluding patients who received preoperative radiation therapy (n = 48), underwent local excision without lymph node dissection (n = 23), had insufficient data (n = 5), or underwent upfront radical resection without prior endoscopic resection (n = 225), 250 patients who underwent radical resection following endoscopic resection were included. Among these, 225 (90.0%) were pathologic N0 and 25 (10.0%) were pathologic N1. CRC, colorectal cancer; N, number Study flowchart. Flow diagram showing patient selection for the study cohort. Among 3529 patients who underwent surgery for colorectal cancer, 551 had pathological T1 stage. After excluding patients who received preoperative radiation therapy (n = 48), underwent local excision without lymph node dissection (n = 23), had insufficient data (n = 5), or underwent upfront radical resection without prior endoscopic resection (n = 225), 250 patients who underwent radical resection following endoscopic resection were included. Among these, 225 (90.0%) were pathologic N0 and 25 (10.0%) were pathologic N1. CRC, colorectal cancer; N, number Cases were included if the final pathological diagnosis after integrating all available specimens (biopsy, endoscopic resection, and radical resection) was pT1, even when no residual tumor was identified in the surgical specimen. The exclusion criteria were as follows:(i)Patients who had received radiotherapy for a rectal tumor before surgery.(ii)Patients who underwent local excision without LND.(iii)Patients with insufficient clinical data.(iv)Patients who underwent upfront radical resection without prior endoscopic resection.(v)Patients with synchronous invasive carcinoma or distant metastases were also excluded from the study. (i)Patients who had received radiotherapy for a rectal tumor before surgery. Patients who had received radiotherapy for a rectal tumor before surgery. (ii)Patients who underwent local excision without LND. Patients who underwent local excision without LND. (iii)Patients with insufficient clinical data. Patients with insufficient clinical data. (iv)Patients who underwent upfront radical resection without prior endoscopic resection. Patients who underwent upfront radical resection without prior endoscopic resection. (v)Patients with synchronous invasive carcinoma or distant metastases were also excluded from the study. Patients with synchronous invasive carcinoma or distant metastases were also excluded from the study. Data collection Initial laboratory findings, imaging studies, endoscopic findings, and pathological features were analyzed. All histopathological variables commonly used to recommend radical resection after endoscopic resection—including vertical margin status, depth of submucosal invasion, lymphatic and vascular invasion, differentiation, and tumor budding—were retrieved from pathology reports when available. Histopathological risk factors were interpreted according to contemporary T1 colorectal cancer risk-stratification frameworks, particularly the JSCCR 2024 criteria, which recognize lymphovascular invasion, poor histological differentiation, deep submucosal invasion, tumor budding, and margin involvement as core adverse features. Clinicopathological data, including age, sex, body mass index (BMI), baseline carcinoembryonic antigen (CEA) levels, tumor location, pathologic T and N stages, number of harvested lymph nodes, presence of lymphovascular and perineural invasion, microsatellite instability (MSI) status, and BRAF, KRAS, and NRAS mutation status, were retrospectively collected from patients’ medical records and systematically analyzed. Tumor location was categorized as right-sided for those arising from the cecum to the transverse colon and left-sided for those arising from the descending colon to the sigmoid colon, and the rectum was classified separately. Rectal lesions were included in the study cohort as they represent a substantial proportion of T1 CRC in clinical practice. However, we acknowledge that rectal cancer presents unique challenges related to surgical complexity, morbidity, and alternative treatment options such as transanal minimally invasive surgery (TAMIS) or transanal excision. The impact of including rectal lesions on overall findings is discussed in the limitations section. Endoscopic and surgical procedures were not standardized across institutions and were performed according to local practice and operator discretion. Statistical analyses The association of clinicopathological variables and LNM was evaluated using Chi-square test or Mann-Whitney U test for categorical variables and Student’s t-test or Fisher’s exact test for continuous variables. Variables that were significant in the univariate analysis were included in the multivariate logistic regression model. For the margin-positive only subgroup, exact binomial confidence intervals (Clopper-Pearson method) were calculated. The result was considered statistically significant if the 95% confidence interval (CI) for the odds ratio (OR) did not cross 1.0 and if the p-value was < 0.05. All statistical analyses were performed using R Studio version 3.0. Ethics statement This study was approved by the Institutional Review Boards of the participating centers (IRB no. PC25RIDI0082). The requirement for informed consent was waived due to the retrospective design of the study. Results Overall patient characteristics Among 250 patients who underwent radical resection following endoscopic resection for pathological T1 CRC during the study period, 25 (10.0%) had pathological LNMs (Fig. 1). Table 1 shows the clinicopathological features of the node-negative (N = 225) and node-positive (N = 25) patient groups. The two groups were assessed for demographic variables, tumor characteristics, molecular features, and histopathological findings. Table 1Comparison of clinicopathological characteristics between node-negative and node-positive groups (post-endoscopic resection cohort)Node negative (N = 225)Node positive (N = 25)p-valueAge, median (IQR), years62 (56–69)67 (57–72)0.042Sex, n (%)0.074 Male145 (64.4%)11 (44.0%) Female80 (35.6%)14 (56.0%)BMI, median (IQR), kg/m225.2 (22.7–27.7)25.5(23.6–27.9)0.561CEA, median (IQR), ng/mL1.78 (1.11–2.66)1.68 (0.84–2.31)0.570Tumor location, n (%)0.636 Right63 (28.0%)6 (24.0%) Left113 (50.2%)15 (60.0%) Rectum49 (21.8%)4 (16.0%)Number of harvested LNs, median (IQR), n (%)17 (14–21)15 (13–18)0.587 Lymphovascular invasion, n (%)5 (2.2%)3 (12.0%)0.035 Perineural invasion, n (%)0 (%)1 (4.0%)0.100 MSI-H, n (%)7 (3.1%)2 (8.0%)0.224BRAF mutation, n (%)0 (0%)0 (0.0%)1.000KRAS mutation, n (%)27 (12.0%)9 (36.0%)0.004NRAS mutation, n (%)2 (0.9%)0 (0.0%)1.000IQR interquartile range, BMI body mass index, CEA carcinoembryonic antigen, LN lymph node, MSI-H microsatellite instability–high Table 1Comparison of clinicopathological characteristics between node-negative and node-positive groups (post-endoscopic resection cohort)Node negative (N = 225)Node positive (N = 25)p-valueAge, median (IQR), years62 (56–69)67 (57–72)0.042Sex, n (%)0.074 Male145 (64.4%)11 (44.0%) Female80 (35.6%)14 (56.0%)BMI, median (IQR), kg/m225.2 (22.7–27.7)25.5(23.6–27.9)0.561CEA, median (IQR), ng/mL1.78 (1.11–2.66)1.68 (0.84–2.31)0.570Tumor location, n (%)0.636 Right63 (28.0%)6 (24.0%) Left113 (50.2%)15 (60.0%) Rectum49 (21.8%)4 (16.0%)Number of harvested LNs, median (IQR), n (%)17 (14–21)15 (13–18)0.587 Lymphovascular invasion, n (%)5 (2.2%)3 (12.0%)0.035 Perineural invasion, n (%)0 (%)1 (4.0%)0.100 MSI-H, n (%)7 (3.1%)2 (8.0%)0.224BRAF mutation, n (%)0 (0%)0 (0.0%)1.000KRAS mutation, n (%)27 (12.0%)9 (36.0%)0.004NRAS mutation, n (%)2 (0.9%)0 (0.0%)1.000IQR interquartile range, BMI body mass index, CEA carcinoembryonic antigen, LN lymph node, MSI-H microsatellite instability–high Comparison of clinicopathological characteristics between node-negative and node-positive groups (post-endoscopic resection cohort) IQR interquartile range, BMI body mass index, CEA carcinoembryonic antigen, LN lymph node, MSI-H microsatellite instability–high The median age was slightly higher in the node-positive group (67 years; IQR, 57–72) than in the node-negative group (median age, 62 years; IQR, 56–69), with a statistically significant difference (p = 0.042). Although there was a higher proportion of women with positive nodes (56.0%) than males with positive nodes (44.0%), this difference did not reach statistical significance (p = 0.074). Histopathologically, no significant differences were found in the distribution of residual tumor on surgical specimen between the groups. The number of harvested lymph nodes did not significantly differ between the groups (median, 17 vs. 16; p = 0.587). However, lymphovascular and perineural invasion were significantly more frequent in the node-positive group (lymphovascular invasion, p < 0.001; perineural invasion, p < 0.001). Lymphovascular invasion was significantly more frequent in the node-positive group (12.0% vs. 2.2%; p = 0.035). Molecular analysis revealed no significant differences between the groups regarding the proportion of MSI-high status (p = 0.224) or NRAS mutations (p = 1.000). No BRAF mutations were detected in either group. However, KRAS mutations were significantly more frequent in the node-positive group (36.0% vs. 12.0%; p = 0.004). Indications for radical resection following endoscopic resection Table 2 outlines the indications for radical resection in patients who underwent endoscopic resection before surgery, stratified by lymph node status. Table 2Indications for radical resection in patients who underwent endoscopic resection before surgeryTotal* (N = 250)Node negative (N = 225)Node positive (N = 25)Margin involvement, n (%)114103 (90.4)11 (9.6) Lateral margin, n (%)4542 (93.3)3 (6.7) Basal margin, n (%)6153 (86.9)8 (13.1) Undetermined margin, n (%)2928 (96.6)1 (3.4)Lymphovascular invasion, n (%)7764 (83.1)13 (16.9)Tumor budding high grade (> 2), n (%)1111 (100.0)0 (0.0)Poor differentiation, n (%)33 (100.0)0 (0.0)Deep submucosal invasion, n (%)198175 (88.4)23 (11.6)*It should be noted that the subcategories are not mutually exclusive. All data are presented as numbers (percentages) Table 2Indications for radical resection in patients who underwent endoscopic resection before surgeryTotal* (N = 250)Node negative (N = 225)Node positive (N = 25)Margin involvement, n (%)114103 (90.4)11 (9.6) Lateral margin, n (%)4542 (93.3)3 (6.7) Basal margin, n (%)6153 (86.9)8 (13.1) Undetermined margin, n (%)2928 (96.6)1 (3.4)Lymphovascular invasion, n (%)7764 (83.1)13 (16.9)Tumor budding high grade (> 2), n (%)1111 (100.0)0 (0.0)Poor differentiation, n (%)33 (100.0)0 (0.0)Deep submucosal invasion, n (%)198175 (88.4)23 (11.6)*It should be noted that the subcategories are not mutually exclusive. All data are presented as numbers (percentages) Indications for radical resection in patients who underwent endoscopic resection before surgery *It should be noted that the subcategories are not mutually exclusive. All data are presented as numbers (percentages) Lymphovascular invasion was observed in 77 patients. Among them, the majority were node-negative (83.1%), whereas a notable proportion (16.9%) were node-positive, representing the highest rate observed among all risk factors. These findings suggest a strong association between lymphovascular invasion and nodal metastases. Tumor budding was classified as positive when high-grade tumor budding (≥ 2) was present. Among the 250 patients, tumor budding status was available for only 127 cases, of which 11 were positive and 116 were negative. Only three cases were characterized by poor differentiation, all of which were node-negative. A significant proportion of patients (n = 198, 79.2%) demonstrated deep submucosal invasion, an adverse histopathological feature that was predominantly found in node-negative cases (88.4%), but also in 23 node-positive cases (11.6%). Margin-positive only T1 CRC without any other risk factors Table 3 presents the clinicopathological characteristics and surgical outcomes of patients who demonstrated margin positivity in endoscopic resection specimens without any other documented high-risk histopathological features. The table details the 22 patients, including their demographic data, margin status, histopathological findings on subsequent surgical specimens, and lymph node evaluations. Table 3Clinicopathological characteristics of patients who showed only margin positivity without any other documented risk factorsNoAgeSexHistopathological findings on ESD specimensHistopathological findings on surgical specimensMargin positivityTumor sizeNumber of Harvested LNsMetastatic LNsLateralBasalUndetermined165M(−)(−)(+)No residual tumor13N0278M(−)(−)(+)3 mm22N0366F(−)(−)(+)5 mm17N0466F(+)(+)(−)No residual tumor14N0565F(+)(+)(−)2 mm44N0673M(+)(−)(−)5 mm17N0760M(−)(−)(+)No residual tumor37N0859M(−)(−)(+)5 mm12N0966M(−)(−)(+)No residual tumor12N01042M(−)(−)(+)No residual tumor30N01166F(−)(−)(+)No residual tumor20N01272M(−)(−)(+)No residual tumor18N01370F(−)(−)(+)2 mm25N01454F(−)(−)(+)No residual tumor15N01559M(−)(−)(+)No residual tumor31N01670M(−)(−)(+)20 mm13N01764M(−)(+)(−)No residual tumor17N01861M(−)(−)(+)No residual tumor21N01981F(+)(+)(−)8 mm18N02063M(−)(−)(+)17 mm19N02188F(−)(+)(−)14 mm22N02260M(+)(+)(−)10 mm18N0ESD endoscopic submucosal dissection, LN lymph node, M male, F female Table 3Clinicopathological characteristics of patients who showed only margin positivity without any other documented risk factorsNoAgeSexHistopathological findings on ESD specimensHistopathological findings on surgical specimensMargin positivityTumor sizeNumber of Harvested LNsMetastatic LNsLateralBasalUndetermined165M(−)(−)(+)No residual tumor13N0278M(−)(−)(+)3 mm22N0366F(−)(−)(+)5 mm17N0466F(+)(+)(−)No residual tumor14N0565F(+)(+)(−)2 mm44N0673M(+)(−)(−)5 mm17N0760M(−)(−)(+)No residual tumor37N0859M(−)(−)(+)5 mm12N0966M(−)(−)(+)No residual tumor12N01042M(−)(−)(+)No residual tumor30N01166F(−)(−)(+)No residual tumor20N01272M(−)(−)(+)No residual tumor18N01370F(−)(−)(+)2 mm25N01454F(−)(−)(+)No residual tumor15N01559M(−)(−)(+)No residual tumor31N01670M(−)(−)(+)20 mm13N01764M(−)(+)(−)No residual tumor17N01861M(−)(−)(+)No residual tumor21N01981F(+)(+)(−)8 mm18N02063M(−)(−)(+)17 mm19N02188F(−)(+)(−)14 mm22N02260M(+)(+)(−)10 mm18N0ESD endoscopic submucosal dissection, LN lymph node, M male, F female Clinicopathological characteristics of patients who showed only margin positivity without any other documented risk factors ESD endoscopic submucosal dissection, LN lymph node, M male, F female Among these cases, undetermined positive resection margins were identified most frequently (n = 15), followed by involvement of the basal margin (n = 6) and lateral margin positivity (n = 5). Histopathological assessment of the surgical specimens revealed that half of the patients (n = 11) showed no residual tumors, while the remaining patients (n = 11) had residual tumors confined to the submucosa (pT1) with varying tumor sizes (ranging from 2 to 20 mm). The number of harvested lymph nodes per patient ranged from 12 to 44. Notably, none of the patients with positive resection margins on initial endoscopic resection specimens, in the absence of any other documented histopathological risk factors, had lymph node metastasis (0/22; 95% exact binomial confidence interval, 0–15.4%). Notably, none of the patients with positive resection margins on initial endoscopic specimens, in the absence of any other documented high-risk histopathological features, had lymph node metastasis (0/22; 95% exact binomial confidence interval, 0–15.4%). Discussion In this multicenter retrospective cohort study focusing on patients who underwent radical resection following endoscopic resection for T1 CRC, we confirmed the distribution and significance of established risk factors for LNM. Notably, our data demonstrated that positive resection margins—in the absence of other documented histopathological risk factors—may confer minimal or no risk of LNMs. Because of the small subgroup size and wide confidence interval, however, these findings cannot establish the safety of omitting radical surgery and should instead be regarded as hypothesis-generating. Radical resection is recommended when high-risk histopathological features are identified after endoscopic resection for T1 CRC. The presence of these factors significantly increases the risk of LNMs and residual disease, which cannot be adequately addressed by local excision alone. Radical surgery, including LND, offers the potential for complete oncologic clearance, accurate staging, and enhanced long-term survival, thereby justifying its use in patients at increased risk based on their pathological profile [13]. In contrast to the margin-positive-only subgroup analysis, lymphovascular invasion remained the most consistent pathological feature associated with nodal metastasis in this cohort, in line with established oncologic principles. However, radical resection in patients with CRC is a relatively complex procedure and may be accompanied by significant complications, with reported morbidity rates reaching up to 35% and considerable variability among centers [14]. By contrast, the risk of LNMs in T1 CRC is relatively low, at approximately 10%. This raises the question of whether all patients, specifically the 80–90% without LNMs, must undergo radical resections. Tumor involvement at the resection margin does not necessarily indicate the presence of residual tumor cells with metastatic potential. As normal mucosa and neoplastic tissue can be readily differentiated, ESD allows for ongoing visual monitoring while pre-cutting the lateral margins and dissecting the submucosal layer. Additionally, a positive resection margin may result from coagulation effects or tangential sectioning during histopathological examination rather than true tumor spread [15]. A recent study confirmed that positive resection margins after ESD for noninvasive colorectal lesions did not affect the recurrence rate. Although LNMs could not be evaluated, all four patients with positive margins showed no evidence of residual tumor [12]. In our study, among the 250 patients who underwent surgery following endoscopic resection, 22 had positive resection margins as the sole documented risk factor for LNMs, without any overlapping additional risk factors. Notably, none of these 22 patients showed evidence of LNM on final histopathological examination (95% exact binomial CI, 0–15.4%). However, the upper bound of this confidence interval remained 15.4%, which does not exclude a clinically meaningful risk. This finding suggests that a positive margin—in the absence of other high-risk features—may be more appropriately considered as a technical factor than as a true independent oncological risk factor for nodal metastasis. Because tumor budding and detailed margin subtype were frequently undocumented, however, the pathological characterization of the margin-positive subgroup was incomplete, which limits the robustness of our risk stratification. Accordingly, larger prospective studies with standardized pathology are needed to confirm and refine these observations. An interesting finding in our study was the significant association between KRAS mutation and lymph node metastasis (p = 0.004), with 36.0% of node-positive patients harboring KRAS mutations compared to 12.0% in node-negative patients. While KRAS mutation status is well established as a prognostic marker in advanced CRC, its role in predicting lymph node metastasis in T1 disease has not been extensively studied. Although KRAS mutation was more frequent in node-positive patients in this cohort, this observation should be regarded as exploratory and insufficient for current clinical risk stratification. This finding warrants further investigation and may have implications for future risk stratification models. This margin-positive only subset accounted for 22 out of 250 patients (8.8%) in our post-endoscopic resection cohort, representing a clinically relevant proportion. Especially in older adult patients, those with significant comorbidities, or those at increased risk from general anesthesia and major surgery, a careful, individualized approach is warranted. Treatment decisions should involve thorough shared decision-making between the surgeon and patient, taking into account the overall risks and benefits, patient preferences, tumor location, and operator expertise [16]. It should be emphasized that the decision to proceed with or avoid radical resection is highly nuanced and multifactorial. Patient comorbidities, anatomical considerations (particularly in rectal cancers where less invasive transanal approaches may be available), patient values, and informed preferences all play critical roles. Our study captures only those patients who proceeded to surgery; outcomes in patients who declined or were deemed unsuitable for surgery are not represented, which introduces an inherent selection bias. Although our results must be interpreted as hypothesis-generating and are insufficient to prove the safety of omitting radical surgery, our intention is not to suggest that surgery can simply be forgone in margin-positive-only cases. Rather, these findings support the concept that additional local treatment—such as repeat endoscopic resection or local excision—may be used as an intermediate step to refine pathological risk assessment. If high-risk features for lymph node metastasis are identified in the locally resected specimen, radical resection with lymph node dissection can still be pursued, whereas in the absence of such features, immediate radical resection may not always be necessary after adequate local treatment. If the absence of LNM risk in margin-positive only cases can be confirmed in larger prospective cohorts, repeat endoscopic resection or local excision—rather than immediate radical resection—may be considered as reasonable alternatives to reduce surgical morbidity in carefully selected patients [17]. These data argue more for considering repeat local treatment than for surveillance alone and should not be taken as definitive evidence that surgery can be safely omitted. However, given the wide confidence interval and the hypothesis-generating nature of this study, such an approach should be adopted only after rigorous validation. Limitation Several limitations should be acknowledged. First, this study is subject to substantial selection bias. The cohort included only patients who proceeded to radical resection after endoscopic resection, whereas patients managed non-operatively (such as those under surveillance, repeat endoscopic treatment, or local excision), those considered unfit for surgery, and those upgraded to pT2 or higher were excluded. However, the primary aim of this investigation was to explore whether there exists a subgroup of T1 colorectal cancer patients in whom radical resection might potentially be avoided, and our findings should be interpreted within this context of hypothesis-generating subgroup identification rather than as a description of outcomes for all patients with margin-positive T1 disease. Second, pathology capture was incomplete for several variables that are directly relevant to risk assessment in T1 colorectal cancer. Tumor budding was not consistently assessed or documented across institutions, particularly in cases where it was absent, leading to a reduced number of patients classified as having high-grade tumor budding. In the remaining 123 cases (49.2%), tumor budding status was recorded as unknown due to incomplete documentation, which represents a significant limitation in our dataset. Furthermore, in 15 of the 22 cases with margin positivity, the histopathology reports did not specify whether the positive margin was lateral or basal, thereby restricting direct comparisons based on the margin location. In many patients with margin positivity, the pathology report did not specify whether the involved margin was lateral, basal, or indeterminate. This incomplete documentation limits the precision with which “margin‑positive only” disease can be defined and may have attenuated or obscured true associations. The inability to precisely determine margin location in a substantial proportion of cases reflects “real‑world” challenges inherent to retrospective pathological review and represents an important limitation in interpreting these results. Moreover, there is a risk that patients with basal margin positivity who were ultimately diagnosed with T2 or higher stage disease may have been excluded from the study. Therefore, it should be recognized that there may be a risk of residual tumor and deeper invasion (pT2 or higher), warranting re-endoscopic resection or local resection when clinically appropriate. Confirmation in larger, prospective cohorts from high-volume centers with standardized pathological evaluation is essential before these findings can be translated into practice. Third, lymphovascular invasion (LVI) and histological differentiation were assessed from routine pathology reports, and several aspects of their evaluation could not be standardized. Because LVI was abstracted from routine pathology reports and the use of immunohistochemical or special stains was neither standardized nor fully documented, some patients categorized as having positive margins only may have harbored occult LVI that was not detected on routine assessment. Similarly, the specific rule used to classify histological differentiation (predominant vs highest-grade/worst differentiation) could not be uniformly verified, which may have affected pathological risk categorization. Fourth, the small sample size of the margin-positive only subgroup limited statistical power and precluded definitive conclusions. As shown in Table 3, only 22 patients presented with margin positivity alone, with no LNM observed; the 95% exact binomial confidence interval extended to 15.4%, indicating that a clinically meaningful risk cannot be entirely excluded. While this finding may suggest a subset of patients in whom radical resection could potentially be omitted, these findings should be regarded as hypothesis-generating. Further validation and generalization will require larger, multicenter prospective studies specifically targeting margin-positive patients and including diverse populations from both Asia and Western countries to account for ethnic and geographic variability in tumor biology and treatment outcomes. Fifth, the underlying causes of marginal positivity were not examined. The available data did not permit evaluation of whether margin involvement was attributable to tumor size, operator expertise, technical challenges, or other factors. Finally, our study was limited to a single ethnic cohort (Korean patients), which may limit generalizability to Western populations where tumor biology and treatment patterns may differ. Conclusion Among patients with T1 CRC who underwent radical resection following endoscopic resection, those with positive resection margins but no other documented high-risk histopathological features did not experience LNM in this cohort. In this context, our findings should be interpreted as hypothesis-generating rather than practice-changing. While they suggest that repeat endoscopic resection or local excision may merit consideration as organ-preserving alternatives to immediate radical resection in some margin-positive-only cases, the limited sample size and statistical power preclude any definitive conclusions about the safety of omitting surgery, and management should remain individualized until larger prospective data are available. In this context, additional local treatment can be viewed as a risk-refining step that preserves the option of subsequent radical resection when high-risk features are identified, while potentially allowing some carefully selected patients to avoid major surgery when such features are not present.