Scoring system supporting suture decision-making for duodenal submucosal tumors

Abstract

BACKGROUND

In recent years, endoscopic resection (ER) has been employed for the excision of submucosal tumors (SMTs). Nonetheless, ER in the duodenum is linked to elevated risks of both immediate and delayed hemorrhagic complications and perforations. Satisfactory suturing is crucial for reducing the occurrence of complications.

AIM

To establish a clinical score model for supporting suture decision-making of duodenal SMTs.

METHODS

This study included 137 individuals diagnosed with duodenal SMTs who underwent ER. Participants were evenly divided into two groups: A training cohort (TC) comprising 95 cases and an internal validation cohort (VC) with 42 cases. Subsequently, a scoring system was formulated utilizing multivariate logistic regression analysis within the TC, which was then subjected to evaluation in the VC.

RESULTS

The clinical scoring system incorporated two key factors: Extraluminal growth, which was assigned 2 points, and endoscopic full-thickness resection, which was given 3 points. This model demonstrated strong predictive accuracy, as evidenced by the area under the receiver operating characteristic curve of 0.900 (95% confidence interval: 0.823-0.976). Additionally, the model’s goodness-of-fit was validated by the Hosmer-Lemeshow test (P = 0.404). The probability of purse-string suturing in low (score 0-2) and high (score > 3) categories were 3.0% and 64.3% in the TC, and 6.1% and 88.9% in the VC, respectively.

CONCLUSION

This scoring system may function as a beneficial instrumentality for medical practitioners, facilitating the decision-making process concerning suture techniques in the context of duodenal SMTs.

Key Words: Endoscopic resection; Duodenal submucosal tumors; Clinical score model; Suturing technique; Submucosal tumors

Core Tip: Endoscopic resection of duodenal submucosal tumors carries high risks of hemorrhage and perforation, highlighting the importance of effective suturing. Thus, we established a clinical score model for supporting suture decision-making of duodenal submucosal tumors. The clinical score comprised extraluminal growth (2 points) and endoscopic full-thickness resection (3 points) with good discriminatory power. The scoring system could provide endoscopists the references for supporting suture decision-making of duodenal submucosal tumors.

INTRODUCTION

Surgical interventions, specifically open or laparoscopic wedge resection, have traditionally been the treatment of choice for the majority of patients presenting with submucosal tumors (SMTs) within the gastrointestinal tract. Yet, such surgical interventions could potentially impact the patient’s quality of life. Endoscopic approaches have demonstrated their viability, safety, and efficacy in managing these digestive SMTs. However, the duodenum poses unique challenges for endoscopic resection (ER) due to its thin wall, limited luminal space, abundant blood supply, and its adjacency to critical organs, as well as its exposure to harsh fluids like gastric acid, bile, and pancreatic enzymes. Consequently, ER within the duodenum is associated with an increased propensity for significant adverse outcomes, with a particular emphasis on the risks of perforation and delayed bleeding[1,2].

Several studies have indicated that ER can be a minimally invasive approach for duodenal SMTs, maintaining curative potential when performed by skilled endoscopists[3,4]. Throughout the excision procedure, proper suturing is essential for minimizing complications. However, there is a scarcity of research on suturing techniques specifically for duodenal SMTs. We retrospectively conducted an in-depth analysis of the medical records pertaining to a sizable case series of individuals afflicted with duodenal SMTs who had undergone ER. Subsequently, we formulated a clinical scoring model designed to guide clinical decision-making with respect to the selection of suture techniques for these neoplasms. The objective was to guide less experienced endoscopists in making tailored suturing decisions, thereby enhancing clinical outcomes.

MATERIALS AND METHODS

Inclusion criteria

We conducted a single-institution retrospective analysis involving 137 successive patients with a diagnosis of duodenal SMTs who received ER at Zhongshan Hospital, Fudan University in Shanghai, China, from December 2016 to August 2022. The study included only those patients for whom comprehensive demographic and clinical data, as well as follow-up information, were accessible. Prior to surgery, patients underwent assessments using endoscopy, computed tomography, or endoscopic ultrasonography. We scrutinized lesion attributes, endoscopic techniques, complications, the rate of en-bloc resection, the rate of complete resection, and the incidence of local recurrence for every patient. Ethical approval for this study was granted by the Zhongshan Hospital Ethics Committee, aligning with the principles of the Declaration of Helsinki (No. B-2018-222). Informed written consent was secured from all participants involved in the study.

ER techniques and perioperative management

For the dissection of the tumor, a standard endoscope equipped with a transparent cap, along with specialized instruments such as a hook knife, a dual knife, as well as an IT knife, were utilized. Supplementary equipment comprised an argon plasma coagulation device, thermal biopsy forceps, and a polypectomy snare for procedural needs. Patients were subjected to general anesthesia with endotracheal intubation to safeguard their well-being and alleviate discomfort throughout the surgical intervention. As a precautionary measure, intravenous antibiotics were administered to the patients 30 minutes prior to the operation to prevent infection. CO₂ insufflation was employed throughout the procedure to mitigate the risk of gas-related complications. Furthermore, A 20-gauge needle preloaded with isotonic saline solution was percutaneously inserted into the right lower quadrant of the abdominal cavity, serving to facilitate the decompression of intraoperatively induced pneumoperitoneum as needed.

The determination of ER approach for duodenal SMTs was predicated upon the morphological characteristics of the tumor as ascertained through endoscopic examination. For polypoid or superficial SMTs in the duodenum, endoscopic mucosal resection or electrosurgical cutting is typically employed. In cases where the SMT is deeply situated, endoscopic submucosal dissection or endoscopic full-thickness resection (EFTR) is chosen, depending on the lesion’s depth. For lesions that are too extensive to be excised in one-piece, endoscopic piecemeal mucosal resection is considered. Following ER, a nasogastric tube was introduced to enable gastric decompression and to permit surveillance for any manifestations of hemorrhagic complications, specifically delayed bleeding, at the site of resection. Postoperatively, patients are closely monitored for any symptoms that may indicate complications. Should a patient present with persistent fever, vomiting of blood (hematemesis), black tarry stools (melena), or severe pain, an urgent endoscopic examination and computed tomography scan would be conducted. Additionally, patients are treated with proton pump inhibitors to reduce gastric acid production, antibiotics to prevent infection, and hemocoagulase injections to aid in blood clotting and control bleeding.

Outcome definitions

The key outcomes measured in this study encompassed: (1) The length of time for the surgical procedure and the patients’ hospitalization; (2) The execution of en-bloc resection (removing the tumor as a single, undivided mass) and the achievement of complete resection (which involves removing the tumor with no visible remnants at the resection site under macroscopic review and with clear margins confirmed through pathological analysis); (3) The methods of suturing employed (either the use of metallic clips or a purse-string suture technique); and (4) The incidence of complications (including bleeding, perforation, stricture, as well as pancreatitis) and the rate of local tumor recurrence.

Statistical analysis

Continuous variables were expressed as the average values plus or minus the standard deviation (SD), while categorical variables were depicted as counts (with their respective percentages). Baseline patient characteristics were compared utilizing Student’s t-tests or χ²-tests. Univariate analysis was conducted to identify potential predictive factors. Subsequently, parameters exhibiting a P value below the threshold of 0.05 were selected for inclusion in the multivariate logistic regression analysis (MLRA) (Figure 1). The outcomes of the logistic regression analyses were articulated with the presentation of odds ratios, beta coefficients, 95% confidence intervals (CIs), as well as their corresponding P values.

Figure 1 Variable selection process. Potential predictors were analyzed using univariate analysis. Then, covariates with P < 0.05 were selected for the multivariate logistic regression model.

The beta coefficients derived from the logistic regression analysis were employed to devise a scoring system for the prediction of suturing techniques, with the scores being rounded to the nearest whole number corresponding to the absolute value of the beta coefficient. Only beta coefficients associated with P values less than 0.05 were incorporated. Ultimately, the scoring system was utilized to forecast suturing techniques, allocating two points for extraluminal growth and three points for EFTR. Each subject’s total score was determined by summing the individual variable scores, and then categorizing them into low and high risk groups. The rates of purse-string suturing were also computed for each category.

The discriminatory power of the model was assessed utilizing receiver operating characteristic curves. The model’s calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, which is a statistical method employed to determine the fit of the model to the observed data. The logistic regression analysis was conducted employing the ‘caret’ R package, with the refinement of model performance parameters accomplished through a ten-fold cross-validation process. All statistical analyses were carried out utilizing SPSS 26.0 and R programming language 4.0.2.

RESULTS

Clinical characteristics of the patients

Patients with duodenal SMTs (n = 137), consisting of 67 females and 70 males with a mean age of 53.1 years (SD: 11.6 years), were consecutively included and underwent ER at Zhongshan Hospital, Fudan University, Shanghai, China. It is noteworthy that none of these subjects reported a history of endoscopic/surgical procedures on the duodenum. The predominant locations affected were the duodenal bulb, accounting for 68 cases (49.6%), and the descending portion, with 19 cases (13.9%) and 49 cases (35.8%) for SMTs in proximity to and away from the papilla, respectively. An extraluminal growth pattern was observed in 12.4% of the duodenal SMTs. Regarding the shape, 8.0% exhibited irregularity, and 13.9% displayed an anabrotic or ulcerated mucosal surface. The largest dimension of the tumors varied, with an average of 13.2 mm and a SD of 7.0 mm. The submucosal layer was identified as the most commonly implicated site, present in 86 cases (62.8%), followed by the muscularis propria, which was involved in 51 cases (37.2%) (Supplementary Table 1).

The 137 patients who participated all underwent ER, which included electric cutting for 3 patients (2.2%), endoscopic mucosal resection for 28 patients (20.4%), endoscopic submucosal dissection for 65 patients (47.4%), EFTR for 37 patients (27.0%), and endoscopic piecemeal mucosal resection for 4 patients (2.9%). The array of endoscopic tools utilized during the excision encompassed the hook knife for 39 patients (28.5%), dual knife for 10 patients (7.3%), IT knife for 49 patients (35.8%), and a snare for 61 patients (44.5%). For suturing, metal clips were applied in 107 cases (78.1%), and purse-string suturing was used in 30 cases (21.9%). The average procedure time was 36.7 minutes with a SD of 29.8 minutes. Submucosal fibrosis was observed in 12 cases (8.8%) during the operations. A nasogastric tube was used in 93 patients (67.9%). In total, en-bloc and complete resection was successfully performed on 107 duodenal SMTs, accounting for 78.1% of the cases (Supplementary Table 1).

The mean length of hospitalization for the patients was 6.0 days, with a SD of 9.7 days. The pathological analysis revealed the following diagnoses: 29 cases (21.2%) of ectopic pancreas, 14 cases (10.2%) of lipoma, 17 cases (12.4%) of neuroendocrine tumors, 29 cases (21.2%) of gastrointestinal stromal tumors (GISTs), and 37 cases (27.0%) of Brunner’s gland adenomas. Complications arose in 8 patients (5.8%): three had delayed bleeding which was managed effectively with endoscopic hemostasis. Three experienced delayed perforations, with two recovering with conservative treatment and one requiring laparoscopic repair. One patient experienced a perforation accompanied by superficial ulcer bleeding in proximity to the surgical site, which was subsequently resolved following further surgical intervention. Additionally, one patient developed severe acute pancreatitis, which showed improvement with symptomatic management. No recurrences were noted during the course of this study (Supplementary Table 1).

Comparisons between two different suturing techniques

We further compared two different suturing techniques and the univariate analysis showed that the duodenal SMTs in the group of purse-string suturing tended to exhibit extraluminal growth (50% vs 1.9%, P < 0.001), often located within the muscularis propria (83.3% vs 24.3%, P < 0.001), predominantly diagnosed as GIST (56.7% vs 11.2%, P < 0.001). EFTR (86.7% vs 10.3%, P < 0.001) was commonly employed, accompanied by gastric tube insertion (96.7% vs 59.8%, P < 0.001). Moreover, procedures within this group typically had a longer duration (60.9 ± 32.3 vs 30.2 ± 25.5, P < 0.001) (Table 1).

Table 1 Demographic information, lesion characteristics and procedural outcomes of the two different suturing techniques, n (%).

	Metal clip (n = 107)	Purse-string suturing (n = 30)	P value
Demographic information
Male	55 (51.4)	15 (50.0)	0.892
Age (years), mean ± SD	53.8 ± 11.8	50.8 ± 10.8	0.215
Hypertension	29 (27.1)	5 (16.7)	0.242
Diabetes mellitus	5 (4.7)	0 (0.0)	0.512
Lesion characteristics
Growth pattern			< 0.001
Intraluminal growth	105 (98.1)	15 (50.0)
Extraluminal growth	2 (1.9)	15 (50.0)
Morphology			0.490
Regular	97 (90.7)	29 (96.7)
Irregular	10 (9.3)	1 (3.3)
Mucosa			0.321
Smooth	90 (84.1)	28 (93.3)
Anabrotic or ulcerative	17 (15.9)	2 (6.7)
Location			0.269
Duodenal bulb	49 (45.8)	19 (63.3)
Descending part (near the papilla)	17 (15.9)	2 (6.7)
Descending part (not near the papilla)	40 (37.4)	9 (30.0)
Horizontal part	1 (0.9)	0 (0.0)
Infiltration depth			< 0.001
Submucosa	81 (75.7)	5 (16.7)
Muscularis propria	26 (24.3)	25 (83.3)
Max diameter (mm), mean ± SD	12.9 ± 7.0	14.6 ± 6.8	0.227
Procedural outcomes
Endoscopic methods
Electric cutting	3 (2.8)	0 (0.0)	0.220
EMR	28 (26.2)	0 (0.0)	0.002
ESD	62 (57.9)	3 (10.0)	< 0.001
EFTR	11 (10.3)	26 (86.7)	< 0.001
EPMR	3 (2.8)	1 (3.3)	0.881
Intraoperative endoscopic instruments
Hook knife	30 (28.0)	9 (30.0)	0.833
Dual knife	8 (7.5)	2 (6.7)	1.000
IT knife	40 (37.4)	9 (30.0)	0.456
Snare	45 (42.1)	16 (53.3)	0.272
Histopathologic evaluation
Ectopic pancreas	24 (22.4)	5 (16.7)	0.495
Lipoma	14 (13.1)	0 (0.0)	0.080
NET	16 (15.0)	1 (3.3)	0.164
GIST	12 (11.2)	17 (56.7)	< 0.001
Brunner’s gland adenoma	33 (30.8)	4 (13.3)	0.056
Others	8 (7.5)	3 (10.0)	0.945
Submucosal fibrosis	10 (9.3)	2 (6.7)	0.926
En-bloc resection	84 (78.5)	23 (76.7)	0.830
Complete resection	84 (78.5)	23 (76.7)	0.830
Stomach tube	64 (59.8)	29 (96.7)	< 0.001
Surgery time (minute), mean ± SD	30.2 ± 25.5	60.9 ± 32.3	< 0.001
Hospital stay (day), mean ± SD	5.1 ± 9.4	8.9 ± 10.5	0.059
Complications	6 (5.6)	2 (6.7)	1.000

EMR: Endoscopic mucosal resection; ESD: Endoscopic submucosal dissection; EFTR: Endoscopic full-thickness resection; EPMR: Endoscopic piecemeal mucosal resection; NET: Neuroendocrine tumor; GIST: Gastrointestinal Stromal Tumor.

Demographic and clinical features alongside procedural outcomes of the training cohort and validation cohort

The demographic and clinical features, as well as the outcomes of the procedures, for both the training cohort (TC) and the validation cohort (VC) are detailed in Supplementary Table 2. Encompassing a total of 137 patients, the study comprised 95 participants in the TC and 42 in the VC. It was observed that the clinical profiles and procedural results were well-matched between the two cohorts.

Risk factors predicting suturing technique

Four key variables demonstrated significant differences based on the two distinct suturing techniques: These were growth pattern, infiltration depth, endoscopic methods, and histopathologic evaluation (Figure 1). These variables were consequently incorporated into the MLRA. Upon conducting the multivariate analysis with 10-fold cross-validation, it was determined that growth pattern and endoscopic methods were the model’s predictors with statistical significance (Table 2). Figure 2 illustrates the relative importance of each variable as determined by the sensitivity analysis.

Figure 2 Importance of risk factors for technical difficulty of esophageal endoscopic submucosal dissection. EFTR: Endoscopic full-thickness resection; GIST: Gastrointestinal stromal tumor.

Table 2 Multivariate logistic regression analysis for predicting suturing methods of duodenal submucosal tumors.

Factors	OR (95%CI)	β coefficient	P value	Point assigned
Growth pattern
Intraluminal growth	1
Extraluminal growth	6.694 (1.212-55.684)	1.901	0.044	2
Infiltration depth
Submucosa	1
Muscularis propria	1.237 (0.049-13.673)	0.213	0.873	NA
Endoscopic methods
Non-full-thickness resection	1
EFTR	30.518 (3.251-788.983)	3.418	0.009	3
Histopathologic evaluation
GIST	0.606 (0.095-3.371)	-0.502	0.575	NA
Non-GIST	1

OR: Odds ratio; CI: Confidence interval; NA: Not available; EFTR: Endoscopic full-thickness resection; GIST: Gastrointestinal Stromal Tumor.

Scoring system to predict suturing technique

A predictive scoring model for suturing techniques was developed utilizing variables that exhibited significant statistical disparities within the demographic and lesion feature analysis. This scoring model was formulated utilizing MLRA within the TC and subsequently validated within the VC. We applied a scoring system for predicting suturing technique by assigning two points to extraluminal growth and three points to EFTR (Table 2). For each case within the TC, the cumulative score was determined by summing up the points allocated to the respective variables (Table 3). Higher scores indicated a greater probability of utilizing purse-string suturing for wound management. The probability of purse-string suturing in low (score = 0-2) and high (score > 3) categories were 3.0% and 64.3% in the TC, and 6.1% and 88.9% in the VC, respectively (Table 4, Figure 3). The area under the receiver operating characteristic curve for the TC and VC were 0.908 and 0.891, respectively (Supplementary Figure 1). The scoring model presented good discriminatory power (area under the curve: 0.900, 95%CI: 0.823-0.976) (Figure 4) as well as goodness-of-fit using the Hosmer-Lemeshow test (P = 0.404).

Figure 3  The purse-string suturing rate of the two categories in the training and validation cohorts.

Figure 4 The receiver operating characteristic curve for the scoring system. ROC: Receiver operating characteristic.

Table 3 Distribution of scores for predicting suturing methods of duodenal submucosal tumors in the training and validation cohorts.

Total points	TC, patients (n = 95)	TC, combination (n = 20)	TC, combination rate (%)	VC, patients (n = 42)	VC, combination (n = 10)	VC, combination rate (%)
0	67	2	3.0	33	2	6.1
3	15	7	46.7	5	4	80.0
5	13	11	84.6	4	4	100.0

TC: Training cohort; VC: Validation cohort.

Table 4 Classification for predicting suturing methods of duodenal submucosal tumors in the training and validation cohorts.

Category	Total points	TC, patients (n = 95)	TC, purse-string suturing (n = 20)	TC, rate (%)	VC, patients (n = 42)	VC, purse-string suturing (n = 10)	VC, pate (%)
Low	67	2	3.0	33	2	6.1
High	≥ 3	28	18	64.3	9	8	88.9

TC: Training cohort; VC: Validation cohort.

DISCUSSION

Previous studies has indicated that ER for duodenal SMTs is linked to a significant incidence of complications, as referenced in studies[5,6]. Therefore, the application of ER for these tumors remains a contentious issue. However, our study demonstrated that ER for duodenal SMTs yielded favorable clinical results, with no observed recurrences. We posit that effective suturing could mitigate the likelihood of complications, and there is a dearth of literature on suturing techniques in the context of duodenal ER. Thus, our research aimed to identify the key factors influencing the selection of various suturing techniques for duodenal ER. The purse-string suturing technique, a procedural modality within the realm of ER, involves the utilization of nylon sutures in conjunction with metallic clips to secure mucosal defects and facilitate tissue retraction, ultimately effecting wound apposition post-ER[7]. This technique, along with the use of metal clip closures, has been shown to decrease the risk of postoperative complications, and the specific conditions of the wound necessitate the selection of appropriate suturing methods[8].

In our center, purse-string suture closure was predominantly utilized for tumors that exhibited extraluminal growth, infiltrated the muscularis propria, or were diagnosed as GIST, as well as for those patients who had undergone EFTR. The sensitivity analysis highlighted that the tumor’s growth pattern and the endoscopic techniques employed were significant determinants in the selection of suturing approaches. Compared to lesions growing intraluminally, those growing extraluminally were situated deeper, necessitating a more secure purse-string suture for post-resection closure. Besides, the wounds following EFTR were often deeper compared with non-full-thickness resection, necessitating purse-string suturing for secure closure.

Based on the analyses, our research has developed and confirmed a new, straightforward scoring system designed to forecast the suturing technique for duodenal SMTs. This scoring system had two factors: Growth pattern and endoscopic methods. When the score ranged from zero to two, metal clips were indicated for suturing the wound. Conversely, a score exceeding three suggested that purse-string suturing was more appropriate. Our scoring system had a number of advantages. Firstly, it was founded on readily accessible clinical characteristics, making it practical for use in clinical practice and potentially aiding in the prediction of suturing methods. Moreover, the incorporation of ten-fold cross-validation in our model parameter tuning resulted in a scoring system with strong discriminative power. This system was capable of classifying duodenal SMTs into various risk groups and identifying appropriate suturing techniques, thus acting as a valuable guide for novice endoscopists.

Our study, while informative, was not without its constraints. Firstly, we conducted only internal validation for our scoring system, suggesting that its robustness needs to be further tested with cases from external sources to enhance the system’s applicability across different settings. Secondly, given that this was a retrospective analysis, there was a possibility of inherent bias. Therefore, it is imperative for future prospective studies to address these limitations and to strengthen the validity of our findings.

CONCLUSION

To sum up, the scoring system we developed might offer clinicians a valuable tool for predicting the appropriate suturing techniques for duodenal SMTs.