Conservative management of delayed perforation after colorectal endoscopic submucosal dissection: a single-center experience

Abstract

Background

Delayed perforation is a rare but serious complication of colorectal endoscopic submucosal dissection (ESD). This study aimed to evaluate the clinical characteristics and outcomes of patients undergoing conservative management for delayed perforation after colorectal ESD.

Methods

We retrospectively reviewed the medical records of 21 patients with delayed perforation after colorectal ESD treated between May 2011 and May 2025. Among them, 19 patients who received initial conservative management were included. These patients were divided into two groups: the Success Group (SG, n = 14), where conservative management succeeded, and the Failure Group (FG, n = 5), which required conversion to surgery. A comparative analysis of clinical, endoscopic, and laboratory findings between the two groups was performed.

Results

Compared with the FG, the SG had lower baseline serum creatinine (median 67.0 vs. 90.0 µmol/L, P = 0.023), higher baseline albumin (mean 41.2 vs. 36.4 g/L, P = 0.042), and a significantly higher rate of complete endoscopic closure (P = 0.013) during the procedure. During conservative management, the SG exhibited a slower rise and a more rapid decline in serum procalcitonin (PCT) levels compared to the FG.

Conclusions

In this small single-center cohort, successful conservative management of delayed perforation after colorectal ESD appeared to be associated with adequate intraoperative defect closure and a relatively better baseline physiological profile, including lower preprocedural serum creatinine and higher preprocedural albumin levels. These findings should be interpreted as exploratory associations rather than definitive predictive indicators. Serial monitoring of serum PCT may provide additional information during conservative management, but treatment escalation should be determined through comprehensive clinical, laboratory, imaging, and multidisciplinary assessment.

Trial registration

Clinical trial number not applicable.

Introduction

Endoscopic submucosal dissection (ESD) has become a mainstream therapeutic modality for colorectal neoplasms, particularly for lesions exceeding 20 mm in diameter [1]. According to current guidelines, the primary indications for colorectal ESD include laterally spreading tumors non-granular type (LST-NG), lesions with a depressed component (Paris classification type 0-IIc), and large mucosal tumors with suspected shallow submucosal invasion, which are difficult to remove en bloc by conventional techniques [2, 3]. The standard ESD procedure consists of marking the lesions, submucosal injection to lift the target tissue, circumferential mucosal incision, and meticulous dissection of the submucosal layer using specialized electrosurgical knives [3, 4]. Compared to conventional endoscopic mucosal resection (EMR), ESD offers superior en bloc and curative resection rates irrespective of lesion dimensions, yields lower long-term recurrence, and provides optimal specimens for accurate histological assessment [5]. Nevertheless, this advanced technique is associated with a steeper learning curve and an increased risk of adverse events [6]. Major procedural complications include bleeding, post-polypectomy electrocoagulation syndrome, and perforation [7–9]. Among these, perforation represents the most critical clinical emergency as the resultant spillage of feces and colonic flora can provoke a more devastating peritonitis compared to upper gastrointestinal perforations [10].

Perforations associated with ESD are generally classified into two categories: intraoperative (occurring during the dissection) and delayed (manifesting after the procedure is completed). While intraoperative perforations occur at an estimated rate of 2% to 14% [11–14], delayed perforations are substantially rarer, with an incidence ranging from 0.1% to 0.4% [15]. However, the insidious onset of delayed perforation often hinders early detection, frequently progressing to severe peritonitis and systemic sepsis [16].

Current therapeutic strategies for bowel perforation include conservative medical management, endoscopic closure, and surgical rescue. For delayed cases, immediate endoscopic clipping is typically unfeasible since the endoscope has already been withdrawn, leaving conservative care or surgery as the primary pathways. Non-operative management primarily consists of fasting, intravenous fluid resuscitation, broad-spectrum antibiotic therapy, and gastrointestinal decompression [17]. There have been reports of delayed perforation being successfully treated with conservative measures alone, thereby avoiding surgery [1]. However, conversion to surgery becomes necessary if a patient’s condition fails to improve with conservative management. For patients who are not suitable for conservative therapy, an initial attempt at conservative management may delay definitive care, and subsequent conversion to surgery could exacerbate trauma [18]. Conversely, proceeding directly to surgery in patients who would otherwise be suitable for conservative treatment may increase postoperative complications, prolong hospital stays, and raise healthcare costs [19].

At present, a consensus on the optimal management of delayed post-ESD perforation is lacking. The absence of specific guidelines and large-scale, prospective randomized trials means that clinical decision-making relies heavily on expert or institutional experience. To address this gap, the present study aims to analyze and compare the clinical characteristics of patients experiencing varied outcomes after undergoing conservative management for delayed colorectal perforation. By exploring the clinical characteristics and associated factors for non-operative treatment, this research seeks to provide valuable clinical experience and an evidence-based reference for identifying ideal candidates for conservative therapy.

Methods

Patients and study design

This retrospective study included patients from the Endoscopic Center of the First Affiliated Hospital of Sun Yat-sen University who underwent colorectal ESD for colorectal lesions between May 2011 and May 2025. This study was approved by the Ethics Committee of the Affiliated Hospital of Sun Yat-sen University (Approval No.: [2025]279) and conducted by the Declaration of Helsinki (2013). The requirement of written informed consent was waived because of the retrospective nature of the study; however, the patients were informed of their right to opt out of the study on the homepage of our institute’s website.

Patients were eligible for inclusion if they met the following criteria: (1) confirmed diagnosis of delayed perforation after colorectal ESD; and (2) an initial treatment strategy of conservative management. Exclusion criteria included patients with intraoperative perforations (detected and treated during the endoscopic procedure) or those who underwent immediate surgery without any prior trial of conservative management.

In our center, a trial of initial conservative management is the preferred first-line strategy for patients with delayed perforation who are hemodynamically stable, driven by the principles of minimally invasive care and often patient/family preference to avoid immediate surgery. However, this initial decision is not arbitrary; it relies on a rigorous multidisciplinary evaluation involving both endoscopists and gastrointestinal surgeons. Specifically, patients were deemed eligible for the initial conservative approach if they met the following specific criteria: (1) General: hemodynamically stable without signs of impending septic shock, presenting with localized abdominal pain rather than severe diffuse peritonitis with board-like rigidity; (2) Imaging: CT confirming localized free air or intra-abdominal fluid collection but without gross fecal spillage; and (3) Laboratory: absence of fulminant systemic inflammatory response syndrome (SIRS) requiring immediate surgical resuscitation. The decision to convert to surgery following failed conservative management was determined by a multidisciplinary evaluation involving endoscopists and surgeons. All colorectal ESD procedures were performed by associate chief physicians who had performed at least 1000 ESD procedures.

Outcome measures and definitions

Primary outcome parameters were incidence and clinical success of conservative treatment. Secondary outcome parameter was the length of hospital stay. Data retrieved from the medical records included: baseline patient characteristics, intra-procedural endoscopic findings (e.g., degree of defect closure), clinical manifestations (time of symptom onset, fever, peritoneal signs), laboratory investigations, specific management methods and follow-up parameters (e.g., serial laboratory markers including WBC, CRP, PCT, and length of hospital stay). The adequacy of prophylactic defect closure was retrospectively determined through a rigorous review of both the explicit textual descriptions in the endoscopic operative reports and the intraoperative endoscopic images. The three-tier classification is illustrated in Fig. 1 and defined as follows: (1) None: no prophylactic clipping was attempted or recorded; (2) Incomplete closure: prophylactic clips were applied, but the mucosal edges could not be entirely approximated, leaving a visually exposed submucosal or muscular defect; and (3) Complete closure: the entire mucosal defect was successfully and closely approximated using clips, with no visible exposed raw surface remaining.

Fig. 1.

Representative endoscopic images of the three-tier classification for prophylactic defect closure following colorectal ESD. A No closure (no prophylactic clips applied). B Incomplete closure (partial approximation with clips, leaving a visually exposed area of the defect). C Complete closure (entire mucosal defect successfully and closely approximated with clips)

Delayed perforation was defined as a perforation not detected during the initial ESD procedure (with no immediate post-procedural symptoms or endoscopic findings), but diagnosed subsequently when the patient developed delayed clinical symptoms (e.g., abdominal pain, fever) that prompted confirmatory imaging (CT or radiography) revealing extraluminal free air.

The standard conservative treatment protocol included strict fasting, intravenous fluid resuscitation, empirical broad-spectrum antibiotics, and the selective use of somatostatin or its analogues. During conservative management, patients were closely monitored with serial abdominal examinations, continuous vital sign monitoring, and dynamic tracking of laboratory parameters (including WBC count, neutrophil ratio, CRP, and PCT) daily or every other day during the acute phase (the first 3–5 days), with further intervals determined by the patient’s dynamic clinical trajectory.

Clinical success of conservative treatment was defined as the complete resolution of peritoneal symptoms, normalization of inflammatory biomarkers, and successful resumption of an oral diet leading to hospital discharge, without the need for any surgical intervention. To characterize the initial clinical severity at the time of delayed perforation diagnosis, we retrospectively assessed the presence of systemic inflammatory response syndrome (SIRS) using available data on body temperature, heart rate, respiratory status when documented, and white blood cell count. Full SOFA and qSOFA scores were not calculated because several required variables, particularly respiratory rate, mental status, and blood pressure at the exact time point of diagnosis, were not consistently recorded in the medical or nursing charts. In addition, some laboratory variables required for complete SOFA scoring were not routinely obtained at the exact diagnostic time point in this emergency setting. To avoid inaccurate retrospective classification or biased imputation, we therefore reported SIRS criteria as the available sepsis-related severity parameter. Clinical failure was defined as the necessity to switch from the initial conservative approach to a surgical procedure due to clinical deterioration during observation. Indications for conversion included uncontrollable progression of diffuse peritonitis, persistent high fever despite antibiotic escalation, or the development of hemodynamic instability. Surgical procedures performed included primary closure, bowel resection, and/or fecal diversion. A patient was considered fully recovered when they achieved all of the following: (1) continuous normalization of body temperature for at least 48 h; (2) complete resolution of abdominal pain and peritoneal irritation signs; (3) a progressive decline or normalization of inflammatory markers (WBC, CRP, PCT); and (4) the ability to tolerate a semi-liquid diet without gastrointestinal discomfort.

Statistical analysis

For continuous variables, data were summarized as mean with standard deviation or median with interquartile ranges (IQR); for categorical variables, data are summarized as frequency with percentage. The student t-test for normally distributed data or the Mann–Whitney U test for non-normally distributed data was used to compare differences in continuous variables between the Success and Failure Groups. The Pearson chi-square test (when all expected cell frequencies were ≥ 5) or Fisher’s exact test (when any expected cell frequency was < 5) was used to compare categorical variables, when appropriate. P values < 0.05 were considered statistically significant. All analyses and visualization were performed using R version 4.5.1 (R Core Team, 2025).

Results

Patient characteristics

From May 2011 to May 2025, a total of 1030 colorectal ESD were performed at our institute and 21 cases (2.0%) of delayed perforation were detected (Fig. 2).

Fig. 2.

Flow chart of the study population

2 patients underwent immediate surgical intervention and were excluded. The remaining 19 patients who received initial conservative management were included in the final analysis. Based on the clinical outcomes, these 19 patients were divided into two groups: the Success Group (SG, n = 14, 73.7%) and the Failure Group (FG, n = 5, 26.3%). Table 1 shows the characteristics of the patients according to the clinical outcome of conservative management.

Table 1.

Baseline characteristics of patients according to clinical outcome of conservative management

Cases, n (%)	Success Group	Failure Group	P-value
	14 (73.7)	5 (26.3)
Age (years), mean (SD)	63.3 (9.2)	65.6 (10.8)	0.648
Sex, n (%)			0.628
Male	6 (42.9)	3 (60)
Female	8 (57.1)	2 (40)
Diabetes, n (%)	4 (28.6)	1 (20)	1
Hypertension, n (%)	5 (35.7)	2 (40)	1
ASA-score, n (%)			1
≤ 2	10 (71.4)	4 (80)
> 2	4 (28.6)	1 (20)
Preprocedural (baseline) laboratory test
Cr (µmol/L), median (IQR)	67 (56.2–80.2)	90 (86–100)	0.023
Alb (g/L), mean (SD)	41.2 (3.7)	36.4 (5.6)	0.042
WBC (×109/L), mean (SD)	5.8 (1.9)	5.7 (1.5)	0.889
NEUT ratio, mean (SD)	0.54 (0.1)	0.6 (0.09)	0.21
Perforation site, n (%)			0.897
Cecum	2 (14.3)	2 (40)
Ascending colon	1 (7.1)	0 (0)
Transverse colon	4 (28.6)	1 (20)
Descending colon	0 (0)	0 (0)
Sigmoid colon	3 (21.4)	2 (40)
Rectum	2 (14.3)	0 (0)
Rectosigmoid	1 (7.1)	0 (0)
Hepatic flexure	1 (7.1)	0 (0)
Length of hospital stay, mean (SD)	10.5 (3.5)	16.6 (4)	0.005

ASA indicates American Society of Anesthesiology, Cr Creatinine, Alb Albumin, WBC White blood cell, NEUT Neutrophil, SD Standard deviation, IQR Interquartile range

The mean age of the patients was 63.3 ± 9.2 years in the SG and 65.6 ± 10.8 years in the FG. Regarding sex distribution, the SG consisted of 6 males (42.9%) and 8 females (57.1%), while the FG included 3 males (60.0%) and 2 females (40.0%). Baseline physical status, assessed by the American Society of Anesthesiologists (ASA) classification, was comparable between the two clinical outcome groups. The majority of patients in both the SG and FG had an ASA score of ≤ 2 (71.4% vs. 80.0%, respectively). There was no statistically significant difference in the proportion of patients with an ASA score > 2 between the two groups (28.6% vs. 20.0%, P = 1.000). None of the patients suffered from inflammatory bowel disease. The site of perforation was the sigmoid colon in 5 cases (26.3%), the transverse colon in 5 cases (26.3%), the cecum in 4 cases (21.1%), the rectum in 2 cases (10.5%), the rectosigmoid in 1 case (5.3%), the hepatic flexure in 1 case (5.3%), and the ascending colon in 1 case (5.3%). The median level of creatinine (Cr) of the 19 patients was 72 (57–89) µmol/L. The mean levels of albumin (Alb), white blood cell (WBC) and neutrophil (NEUT) ratio were 39.7 ± 4.6 g/L, (5.5 ± 1.7)×10⁹/L and 0.55 ± 0.10, respectively. The baseline comorbidities of diabetes and hypertension, laboratory results of WBC and NEUT ratio, and the perforation site did not differ significantly between the 2 groups. Compared to the FG, the SG had lower Cr [SG : 67µmol/L (IQR, 56.2–80.2); FG: 90µmol/L (IQR, 86–100), P = 0.023] and higher Alb (41.2 ± 3.7 vs. 36.4 ± 5.6, P = 0.042). Mean hospital stay for the FG was 16.6 ± 4.0 days, which was significantly longer than 10.5 ± 3.5 days in the SG (P = 0.005).

Characteristics of perforations and clinical manifestation of patients

Table 2 shows the characteristics of perforations and clinical manifestation of patients according to the clinical outcome of conservative management.

Table 2.

Characteristics of perforations and clinical manifestation of patients according to outcome of conservative management

Cases, n (%)	Success Group	Failure Group	P-value
	14 (73.7)	5 (26.3)
Tumor size (cm), median (IQR)	2 (2-3.8)	2 (2-2.5)	0.469
Tumor depth, n (%)			0.155
Lamina propria	13 (92.9)	3 (60)
Submucosal	1 (7.1)	2 (40)
Histology of tumor, n (%)			0.819
Hyperplastic polyp	1 (7.1)	0 (0)
Sessile serrated lesion	1 (7.1)	0 (0)
Tubular adenoma	5 (35.7)	1 (20)
Villous tubular adenoma	5 (35.7)	2 (40)
HGIEN	2 (14.3)	2 (40)
Specimen size (cm), mean (SD)	4 (1.6)	3.1 (2)	0.25
Procedure duration (min)	63.2 (26.8)	63 (33.8)	0.989
Closure of defect, n (%)			0.013
None	2 (14.3)	3 (60)
Incomplete	3 (21.4)	2 (40)
Complete	9 (64.3)	0 (0)
Time of symptom onset, n (%)			0.515
0–12 h	5 (35.7)	1 (20)
12–24 h	5 (35.7)	2 (40)
24–48 h	4 (28.6)	1 (20)
> 48 h	0 (0)	1(20)
Time to diagnosis, n (%)			0.884
0–12 h	1 (7.1)	1 (20)
12–24 h	3 (21.4)	1 (20)
24–48 h	7 (50)	2 (40)
> 48 h	3 (21.4)	1 (20)
Symptoms of peritoneal irritation
Patients meeting SIRS criteria, n (%)	5 (35.7)	4 (80)	0.141
Abdominal pain, n (%)			1
Localized	13 (92.9)	5 (100)
Diffuse	1 (7.1)	0 (0)
Direct tenderness/Rebound tenderness, n (%)			0.754
No	3 (21.4)	1 (20)
Localized	10 (71.4)	3 (60)
Diffuse	1 (7.1)	1 (20)
Fluid collection on CT images, n (%)	7 (50)	4 (80)	0.347
Use of somatostatin or its analogues, n (%)	2 (14.3)	1 (20)	1
Peak postprocedural body temperature (°C), median (IQR)	38.5 (37.7–38.8)	38.9 (38-39.2)	0.307
Postprocedural laboratory test
Peak WBC (×109/L), mean (SD)	15.3 (1.86)	17.4 (1.48)	0.454
Peak NEUT ratio, mean (SD)	0.91 (0.03)	0.93 (0.02)	0.098
Peak CRP (mg/L), mean (SD)	171.8 (76.7)	158.7 (119)	0.796
Peak PCT (ng/mL), median (IQR)	1.4 (0.4–2.4)	1.5 (1.1–3.7)	0.963

HGIEN indicates High grade intraepithelial neoplasia, CT Computed tomography, WBC White blood cell, NEUT Neutrophil, CRP C reactive protein, PCT Procalcitonin, SIRS Systemic Inflammatory Response Syndrome, SD Standard deviation, IQR Interquartile range

The median sizes of the tumor and the resected specimen were 2 (2–3) cm and 3.5 (2.5–5.0) cm, respectively. The depth of tumor invasion included 16 cases with lamina propria involvement and 3 cases with submucosal invasion. The pathology of the lesion included villous tubular adenoma in 7 cases, tubular adenoma in 6 cases, high grade intraepithelial neoplasia in 4 cases, sessile serrated lesion in 1 case and hyperplastic polyp in 1 case. The mean procedure duration was 0.96 ± 0.46 h (h). During the procedure, five patients underwent no defect closure, while partial closure was achieved with endoscopic clips in five patients and complete closure in nine patients. Six patients developed abdominal symptoms within 0–12 h, seven within 12–24 h, five within 24–48 h, and one after 48 h. Two patients were diagnosed by imaging within 0–12 h, four within 12–24 h, nine within 24–48 h, and four after 48 h. At the time of diagnosis, 5 patients in the Success Group (35.7%) and 4 patients in the Failure Group (80.0%) met the SIRS criteria. Although this difference did not reach statistical significance because of the small sample size (P = 0.141), the higher proportion of patients meeting SIRS criteria in the Failure Group suggested a greater degree of initial systemic inflammatory response. Full SOFA and qSOFA scores were not analyzed because the variables required for reliable retrospective calculation were not consistently available at the exact time of diagnosis. All 19 patients had abdominal pain after colorectal ESD, among which 18 patients presented with localized abdominal pain and 1 with diffuse abdominal pain. Four patients showed no direct tenderness or rebound tenderness, 13 exhibited localized direct tenderness and rebound tenderness and two had diffuse direct tenderness and rebound tenderness. Following the onset of peritoneal irritation signs, all 19 patients ultimately underwent confirmatory abdominal CT scans. CT demonstrated intraperitoneal free air in all patients, which served as the major radiological evidence supporting the diagnosis of delayed perforation. No patient showed isolated retroperitoneal emphysema or a predominantly retroperitoneal air-distribution pattern. Concurrent intra-abdominal or pelvic fluid collection was observed in 10 patients (52.6%). However, the initial CT reports did not describe radiologically defined abscess formation, such as a loculated or encapsulated rim-enhancing fluid collection, internal gas bubbles, or an air-fluid level. Therefore, these findings were reported as acute fluid collection rather than abscess formation. The presence of fluid collection did not significantly differ between the Success Group and the Failure Group (50.0% vs. 80.0%, P = 0.347). All patients received routine intravenous broad-spectrum antibiotics after the onset of clinical manifestations and three patients were additionally administered somatostatin or its analogs. During conservative management, the patients’ body temperature was regularly measured, and laboratory parameters including WBC count, NEUT ratio, CRP, and serum procalcitonin (PCT) were rechecked. The peak body temperature and peak values of each laboratory parameter during treatment were recorded during the course of treatment. The median level of the peak body temperature was 38.5 (37.5–38.9) ℃. The mean levels of the peak WBC, NEUT ratio and CRP were (15.0 ± 5.2)×10⁹/L, 0.91 ± 0.03, 140.2 ± 84.3 mg/L, respectively. The median level of the peak PCT was 1.44 (0.44–3.7) ng/mL.

No significant differences were observed between the SG and FG in terms of tumor size, depth of tumor invasion, histopathological type of the tumor, specimen size, procedure duration, time of symptom onset, time to diagnosis, severity of peritoneal irritation signs, postoperative use of somatostatin or its analogs, presence of fluid collection suggested by postoperative CT, or peak postoperative laboratory parameters (including WBC, NEUT ratio, CRP, and PCT). However, a significant difference was observed between the groups regarding the adequacy of intraoperative defect closure (P = 0.013). The success rate of conservative management increased proportionally with the adequacy of prophylactic defect closure. Notably, conservative therapy was uniformly successful (100%) in all 9 patients who received complete intraoperative closure.

During the observation period of conservative management, all 5 patients in the Failure Group experienced clinical deterioration despite the escalation of intravenous antibiotics to meropenem. Specifically, 3 patients exhibited persistent, unremitting abdominal pain with an expanding area of tenderness, indicating a high risk of progressing to diffuse peritonitis. The remaining 2 patients developed persistent high fever (> 39 °C) and clinical signs suggestive of impending septic shock. Notably, one of these two patients presented with paralytic ileus (manifested as the cessation of flatus and defecation) and an episode of coffee-ground emesis (approximately 100 mL); subsequent nasogastric tube decompression yielded a large volume of dark-green bilious drainage. Due to these uncontrollable progressions of intra-abdominal infection and impending physiological decompensation, conservative management was deemed to have failed, and all 5 patients were promptly converted to surgical intervention. 2 patients underwent ileocecal resection with ileo-ascending colon anastomosis, 2 patients underwent laparoscopic sigmoidectomy with colostomy, and 1 patient underwent primary closure of the transverse colon with colostomy. For patients who failed conservative management and were converted to surgery, the postoperative outcomes were highly favorable, reflecting the safety of timely surgical rescue. Following surgery, 2 patients (40.0%) required brief admission to the Intensive Care Unit (ICU) for hemodynamic monitoring and optimization. Notably, there were no cases (0%) of surgical site infection (SSI) or anastomotic leakage. Furthermore, no patients required reoperation (0%), and the overall mortality rate was 0%.

Trajectory of postoperative laboratory parameters

We recorded three postoperative laboratory parameters following colorectal ESD: WBC, NEUT ratio and PCT. Data from different time periods (0–12 h, 12–24 h, 24–36 h, 36–72 h) were plotted as paired box plots. Figures 3, 4 and 5 showed the trajectories of these parameters over time between the two groups.

Fig. 3.

Postoperative white blood cell count trajectory in patients according to conservative treatment outcomes

Fig. 4.

Postoperative neutrophil ratio trajectory in patients according to conservative treatment outcomes

Fig. 5.

Postoperative procalcitonin trajectory in patients according to conservative treatment outcomes

In both groups, the three postoperative laboratory parameters showed a consistent trajectory of initial increase followed by a decrease. The WBC count increased within 0–24 h, peaked at 12–24 h, and then decreased during 24–72 h. Meanwhile, the NEUT ratio and PCT level rose within 0–36 h, reached their peak at 24–36 h, and declined thereafter within 36–72 h. Compared with the Failure Group, the Success Group seemed to show no obvious difference in the trajectories of WBC count and NEUT ratio. However, during the treatment, the PCT level in the successful group exhibited a smaller increase but a greater decrease.

Discussion

Recently, ESD has been performed in many Asian and Western countries [20–22]. Delayed perforation is a rare but potentially life threatening adverse event of colorectal ESD with a reported incidence ranging from 0.4% to 0.7% [10]. Due to its rarity, clinical features and outcomes of patients with delayed perforation have been scarcely reported and its optimal treatment strategy remains uncertain. This single-center retrospective study aimed to explore clinical and procedural characteristics associated with the outcome of conservative management.

In brief, our study identified that the clinical success of initial conservative management for delayed colorectal perforation appeared to be associated with two main factors: robust baseline physiological reserve (indicated by normal-low preprocedural creatinine and higher albumin) and the achievement of complete prophylactic defect closure during the index ESD procedure. The association between lower preprocedural creatinine levels and conservative treatment success is biologically plausible. Serum creatinine is a key indicator of renal function. The potential relevance of preprocedural creatinine in our cohort should be interpreted in the context of previous studies on colonic perforation and emergency abdominal surgery. Horiuchi et al. [23] reported that preoperative serum creatinine level was associated with mortality following colonic perforation. Similarly, Lee et al. [24] found that, among patients undergoing surgery for colonic perforation, non-survivors were more likely to have high creatinine levels and organ failure at presentation. These findings support the biological plausibility that serum creatinine may reflect renal function, systemic physiological reserve, and the severity of acute intra-abdominal infection in patients with bowel perforation. Therefore, the relatively lower preprocedural creatinine observed in our Success Group may be interpreted as a possible marker of better baseline physiological reserve. However, the absolute median creatinine values in both groups largely remained within or near the normal physiological range. Thus, this finding should be interpreted as a relative difference rather than overt renal dysfunction, and it remains an exploratory observation requiring validation in larger cohorts. This baseline profile is characteristic of our institution’s role as a major tertiary referral center, where the patient cohort typically comprises older individuals (mean age ~ 64 years) with a higher burden of concurrent metabolic and cardiovascular comorbidities who are referred for advanced endoscopic management. Similarly, higher serum albumin levels are a well-established indicator of good nutritional status and physiological resilience [25]. Hypoalbuminemia is a known risk factor for increased morbidity and mortality in various surgical and critical care settings, as it is associated with impaired immune function, poorer tissue healing, and a heightened systemic inflammatory response [26–30]. Our findings suggest that these markers of baseline patient robustness may contribute to the capacity to combat localized infection without surgical intervention.

The most striking finding of our study is the 100% success rate of conservative management in patients who achieved complete prophylactic defect closure. This highlights a critical technical threshold. While conservative therapy is often effective for smaller perforations—such as those typically encountered during Endoscopic Mucosal Resection (EMR)—large defects present a much greater challenge. According to studies by Cho et al. and An et al. [31, 32], a perforation size exceeding 10–15 mm is a primary independent predictor of conservative treatment failure. Since colorectal ESD inherently creates large artificial ulcers that far exceed this threshold, the risk of massive fecal leakage is naturally higher. Our results suggest that complete prophylactic closure essentially ‘downsizes’ these high-risk large defects into a sealed, controlled environment, effectively replicating the favorable conditions seen in smaller EMR-related perforations. Several previous studies have also supported the feasibility of non-operative management for iatrogenic endoscopic colorectal perforations in carefully selected patients. Shin et al. [33] reported that 20 of 41 patients with iatrogenic colonoscopic perforation underwent conservative management, with a success rate of 90%. Jovanovic et al. [34] further demonstrated that immediate endoscopic closure of iatrogenic colon perforation was feasible in selected cases, with successful closure achieved in 83% of attempted cases, allowing some patients to avoid surgery. In addition, Magdeburg et al. [35] reported that endoscopic closure was more feasible after therapeutic colonoscopy than after diagnostic colonoscopy, with a lower subsequent need for surgery in therapeutic perforations. These findings are consistent with our observation that adequate defect closure and careful patient selection may create favorable conditions for successful conservative management, although timely surgical conversion remains essential when clinical deterioration occurs.

The clinical dividends of achieving conservative success through such meticulous closure are substantial. In our study, the Success Group had a significantly shorter length of hospital stay than the Failure Group (10.5 ± 3.5 days vs. 16.6 ± 4 days, P = 0.005). This aligns with the findings of Park et al. [36], who demonstrated that successful non-operative management of colonoscopic perforations not only shortens hospitalization but also facilitates earlier resumption of oral intake and reduces the duration of intravenous antibiotic therapy. Notably, the hospital stay for our failure group was still substantially shorter than the conversion-to-surgery cases reported by An et al. [32] (31.0 days) and Cho et al. [31] (28.3 days), further emphasizing the importance of a streamlined, multidisciplinary response when conservative measures fail.

Finally, our study emphasizes the critical need for timely surgical conversion. Identifying the early signs of conservative failure is paramount. As noted by Kim et al. [37] in their analysis of endoscopic perforations, severe or unrelenting abdominal pain requiring repeated analgesics within the first 24 h strongly predicts the need for surgery, likely reflecting ongoing transmural leakage due to incomplete closure. These clinical observations support the rationale for careful monitoring: while an initial conservative approach is potentially feasible for appropriately selected and completely closed patients, vigilant monitoring is indispensable. A failure of inflammatory markers (such as PCT) to plateau, coupled with progressive abdominal pain or signs of peritoneal irritation, should prompt early multidisciplinary reassessment and surgical consultation to avert the catastrophic morbidity associated with delayed operative rescue.

Interestingly, and in contrast to some prior reports, traditional indicators of infection severity—such as the degree of fever, the severity of peritoneal irritation signs, CT findings of fluid collection, and peak levels of inflammatory markers like WBC, CRP, and even PCT—did not significantly differ between the Success and Failure Groups in our cohort. This highlights a unique aspect of delayed perforation management: the decision to convert to surgery may not be solely dependent on the initial severity of presentation, but rather on the patient’s inherent physiological response to therapy and the foundational stability provided by the initial endoscopic repair. Somatostatin or its analogues were utilized in a small subset of our patients (n = 3) as an adjunctive conservative measure. Pharmacologically, somatostatin effectively inhibits diverse gastrointestinal secretions and reduces splanchnic blood flow, which theoretically decreases the volume of intestinal spillage into the peritoneal cavity, thereby facilitating defect healing. However, in our cohort, the use of somatostatin did not show a statistically significant difference between the Success and Failure Groups. This lack of significance is primarily attributable to the extremely limited sample size. Furthermore, this introduces a potential confounding by indication (selection bias): physicians might have selectively administered somatostatin only to patients who presented with a more severe initial systemic inflammatory response. Consequently, any potential protective effect of the medication was likely masked by the higher baseline risk of these specific patients. Therefore, while we currently view somatostatin as a biologically plausible adjunctive therapy, its exact clinical efficacy in treating delayed colorectal perforation warrants further investigation in larger cohorts.

Observation of the dynamic trajectory of PCT after ESD offered a novel perspective. While the peak PCT level itself was not statistically different, the Success Group exhibited a trajectory characterized by a smaller initial rise and a more pronounced subsequent decline. This pattern suggests that the dynamic response of PCT to conservative therapy, rather than its absolute peak value, may be a more sensitive real-time indicator of treatment efficacy. A sharp, unrelenting rise or a persistent plateau in PCT levels, as indicated in our proposed treatment algorithm, should raise a high index of suspicion for failure conservative management and prompt serious consideration for surgical conversion.

Our study had several limitations. The most significant limitation is the small sample size. With only 19 patients in the conservative management cohort and merely 5 in the Failure Group, the statistical power of our study is inevitably curtailed. This precluded us from performing more sophisticated multivariate logistic regression analyses to identify independent predictors while controlling for confounders. The statistical associations found, though significant, require validation in larger cohorts, as the risk of Type I and Type II errors is substantial. Given the small sample size and the absence of multivariable analysis, these variables should be interpreted strictly as exploratory associations reflecting baseline physiological reserve, rather than definitive predictive indicators. These findings cannot establish causality and require further validation in larger, multi-center prospective cohorts before guiding clinical decision-making. Second, the retrospective design carries potential selection and information biases. Treatment decisions of initial conservative approach and timing of conversion were not standardized but were based on individual endoscopist and surgeon judgment, reflecting real-world practice but also introducing heterogeneity. While all endoscopists were highly experienced, variations in closure technique and assessment of “completeness” could exist. Third, our follow-up protocol for laboratory markers, while showing general trends, was not prospectively standardized at uniform time intervals for all patients, which may affect the precision of the trajectory analysis. Finally, the generalizability of our findings may be limited to high-volume tertiary centers with extensive ESD experience and readily available surgical support. Outcomes in lower-volume settings may differ.

In conclusion, this single-center retrospective study suggests that an initial trial of conservative management may be feasible in carefully selected patients with delayed colorectal perforation following ESD, particularly when patients are hemodynamically stable and have undergone adequate intraoperative defect closure. Relatively lower preprocedural creatinine and higher albumin levels were observed in patients with successful conservative management, but these variables should be regarded only as exploratory markers of baseline physiological reserve rather than independent predictors or criteria for treatment selection. During conservative management, serial clinical evaluation combined with laboratory and imaging reassessment remains essential. Dynamic changes in serum PCT may help raise clinical suspicion of conservative treatment failure, but surgical conversion should be based on comprehensive multidisciplinary judgment rather than any single biomarker. Further multicenter studies with larger cohorts are required to validate these preliminary observations and to clarify their clinical applicability.

Authors’ contributions

Jinhong Guan, Wenjie Li: analysis and interpretation of data; drafting of the article; critical revision of the article for important intellectual content; and final approval of the article. Li Li, Xianwen Guo: conception and design, critical revision of the article for important intellectual content, and final approval of the article. Zhen Ding: final approval of the article. The corresponding author had full access to the study data and was responsible for the decision to submit the paper for publication.

Funding

This work was supported by grants from the 100 Top Talents Program of Sun Yat-sen University (Grant No. R03009), the National Natural Science Foundation of China (Grant Nos. 82470670 and 82270675), and the National Key Research and Development Program of China (Grant No. 2023YFC2307004).

Data availability

The datasets generated and analyzed during the current study are not publicly available due to patient privacy but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki. The research protocol was approved by the Institutional Review Board of The First Affiliated Hospital of Sun Yat-sen University (approval no.: [2025]279).

The requirement for written informed consent was waived by the Institutional Review Board of The First Affiliated Hospital of Sun Yat-sen University due to the retrospective nature of this study. An opt-out option was provided to the patients on the institute’s website.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.