Introduction

Intestinal duplication (ID) is a rare congenital malformation that can involve any segment of the gastrointestinal tract [1, 2]. ID is classically defined using three main characteristics: a well-developed layer of smooth muscle, an alimentary tract epithelial lining, and, frequently, an intimate attachment to some portion of the gastrointestinal tract [3]. ID occurs in 1 out of 4500 live-births, and most of the patients are diagnosed under the age of 2, or even prenatally [3]. The exact etiology of ID remains unknown, but several embryologic theories have been described [2, 4]. ID can either be asymptomatic and discovered incidentally in pre- or postnatal imaging or it can lead to symptoms, such as abdominal pain, vomiting, chronic constipation or a palpable abdominal mass. Moreover, ID can lead to acute complications, such as intestinal obstruction due to segmental volvulus or intussusception, gastrointestinal bleeding, and perforation.

Taking into account the potential risks related to mechanical complications, as well as the frequent presence of heterotopic gastric mucosa, ID requires complete surgical resection [5, 6]. Two main surgical techniques have been described: duplication enucleation (DE) and intestinal resection-anastomosis (IRA) [6]. Enucleation was initially described for locations on the gastro-intestinal tract where a resection and anastomosis should be avoided, such as the esophagus, stomach, duodenum, and ileocecal region [3]. Recently, several studies [7, 8] have advocated for the use of enucleation for small bowel or colonic duplications.

The aim of our study was to compare the outcomes of the two main surgical techniques for ID in children, namely duplication enucleation (DE) and intestinal resection anastomosis (IRA). Moreover, to determine whether DE is potentially feasible in all cases of ID, we aimed to assess whether a muscular layer was present using pathology review of cases of duplications treated by IRA.

Materials and methods

A retrospective monocentric study was conducted, including all children (0–18 years of age) operated for an ID between 2005 and 2023, at our institution. Patients with a pre- or postnatal diagnosis of small bowel or colonic duplication were included. Patients with duplications located on the esophagus, stomach or duodenum were excluded. All patient received perioperative antibioprophylaxis (cefotaxime and metronidazole).

Approval from the local ethical committee was obtained, as well as parental consent.

In our department, when ID was suspected prenatally and confirmed postnatally using radiological imaging, a surgery was systematically scheduled 2–3 months after birth. ID was classically approached using a laparotomy, and an IRA or DE could be performed, depending on the surgeon’s preference and intra-operative feasibility. The surgeons were from the same surgical team.

The laparoscopic approach for ID has been progressively introduced since 2014 at our institution. Surgery could be performed using a totally laparoscopic approach or laparoscopy-assisted (initial laparoscopy followed by a conversion to a laparotomy for ID resection).

For enucleation, ID was removed by incision of the muscular layer of the ID, followed by cautious dissection in the muscular layer between the intestinal duplication and the intestinal wall, taking care to avoid a mucosal breach on the intestinal side. In the event of a mucosal perforation, it would be sutured using an absorbable suture. The muscular layer would then be closed using single stitches with an absorbable suture to cover the ID bed and avoid bowel stricture. The procedure followed the same principles, regardless of the surgical approach (laparoscopy or laparotomy). For laparoscopy, a 5 mm optical and two 3 mm ports were used.

For IRA, resection and anastomosis were carried out using systematically a small laparotomy, even after an initial laparoscopic approach. After exteriorization of the duplication, bowel resection was performed centered on the duplication, after coagulation of the vessels. The end-to-end anastomosis was then carried out manually. We chose not to perform intestinal resection and anastomosis using a totally laparoscopic approach because of the high risk of peritoneal contamination and infection upon bowel incision. Laparoscopy was used as a first step in order to localize the ID and facilitate its exteriorization through a mini-laparotomy around the umbilicus.

Medical charts were retrospectively reviewed and collected data included demographics, symptoms, surgical findings, and postoperative outcomes. Acute onset symptoms were defined as the occurrence of acute symptoms related to ID and requiring emergent surgery, such as intussusception, volvulus, and intestinal obstruction. Primary outcome was length of hospital stay, and secondary outcomes included delay to first feeds, which was determined by surgeon’s preference and based on progressively introduced enhanced recovery after surgery (ERAS) items and post-operative complications. Postoperative complications were assessed according to the Clavien–Dindo classification [9]. Patients were compared depending on the surgical technique, either DE or IRA.

Pathology review

We retrospectively reviewed the operative specimens with our local pathologist and assessed the presence of a muscular layer between the ID and healthy intestinal segment to determine the feasibility of DE in cases operated by IRA. We constituted three histopathological groups: without a muscular layer (Fig. 1A)/with a common muscular layer (Fig. 1B)/with two separate muscular layers.

Fig. 1
figure 1

Histopathological analysis of intestinal duplications. A Ileo-cecal duplication with a common muscular layer measuring 4 mm (blue arrow). B Ileo-cecal duplication without muscular layer between both mucosae (yellow arrow)

Full size image

We also described the integrity of the intestinal wall in cases of IRA to determine whether the resection was due to damage of the intestinal wall.

Statistical analysis

Descriptive data were presented using median and interquartile range, or number and percentage. Both groups (DE and IRA) were compared using t tests (or Mann–Whitney test when appropriate) for quantitative variables and contingency tests for qualitative variables. Statistical significance was determined using a p < 0.05.

Results

Baseline characteristics/preoperative data

Overall, 51 patients underwent surgery for ID between 1995 and 2023, including 25 (49%) girls. Whereas 27 patients underwent DE (53%), 24 underwent IRA (47%). Median age at surgery was 5 months for DE (2.9–18.2) and 4 months for IRA (0.4–21.3) (p = 0.31). Both groups were comparable in terms of gender, gestational age, and birth weight (Table 1).

Table 1 Patient characteristics
Full size table

Preoperative symptoms were present in 49% of the cases (n = 25), with acute onset in 33% of patients (n = 17). Acute onset included intussusception in 18% (n = 9), segmental volvulus in 6% (n = 3), and intestinal obstruction without intussusception or volvulus in 10% (n = 4). Median age at acute onset was 6.7 months (IQR: 0.3–14 months). Other symptoms included abdominal pain in 14% (n = 7), and diarrhea in 2% (n = 1).

A cystic image was detected prenatally with ultrasound in 30 patients (59%), at a median age of 26 weeks of gestation. Patients that underwent DE were more likely to have had a prenatal diagnosis compared to patients that underwent IRA, although this did not reach statistical significance (70% vs 46%, p = 0.09).

When ID was prenatally diagnosed (n = 30), preoperative symptoms were present in 23% (n = 7), with an acute onset in 13% (n = 4), including 1 volvulus and 3 intestinal obstruction without volvulus or intussusception. Median age at acute onset in patients with a prenatal diagnosis was 0.4 months (range: 0.1–3.5 months). Children without a prenatal diagnosis were more likely to present with preoperative symptoms (86%, p < 0.0001) and with an acute onset (62%, p = 0.0006).

Patients that underwent IRA were more likely to have presented with preoperative symptoms compared to patients that underwent DE (71% vs 30%, p = 0.005), and were also more likely to present with acute onset (54% vs 15%, p = 0.007).

Surgery

In 45% of the cases (n = 23), a laparoscopic approach was initially attempted, followed by conversion to laparotomy in 70% (n = 16) (Fig. 2). The first case of totally laparoscopic DE was performed in 2015 and since then, a laparoscopic-assisted DE was successfully performed in 7 cases (14%). After 2015, there was a significant increase in the use of laparoscopy, with an initial laparoscopic approach in 87% of the cases (n = 20/23), compared to 11% between 2005 and 2014 (n = 3/28, p < 0.0001) (Fig. 3).

Fig. 2
figure 2

Flow-chart of patients with intestinal duplications, depending on surgical approach

Full size image
Fig. 3
figure 3

Surgical approach for intestinal duplications between 2005 and 2022

Full size image

When a laparotomy was used, an IRA was performed in 55% (n = 24/44) and a DE in 45% (n = 20/44). When a patient was scheduled for an elective laparoscopy for ID (n = 19), 63% required conversion to laparotomy (n = 12). After conversion, 50% finally underwent IRA (n = 6) and 50% DE (n = 6) (Fig. 4).

Fig. 4
figure 4

Flowchart of patients undergoing surgery for intestinal duplications, depending on surgical indication

Full size image

Median operative time was 93 min (IQR 74–119 min), and was similar for DE and IRA (90 vs 95 min, p = 0.22) (Table 1).

Intraoperative findings revealed that the main locations of ID were the ileum (n = 38, 75%) and ileocecal junction (n = 11, 24%) (Table 2). There were no differences in terms of location of ID between children that underwent DE or IRA.

Table 2 Location of intestinal duplication
Full size table

Outcomes

First feeds were initiated earlier after DE, on postoperative day (POD)-1, compared to IRA (POD-3, p < 0.0001). Length of stay was shorter after DE, with a median of 4 days, compared to IRA (6 days, p < 0.0004) (Table 1). Sub-analysis on surgical approach revealed that length of stay after DE was shorter with laparoscopy compared to laparotomy (1 vs 4 days, p < 0.0001) (Table 3). Moreover, when including only patients that underwent an open approach, DE was still associated with a shorter hospital stay compared to IRA (4 vs 6 days, p = 0.02) (Table 3). Delay to first feed was not modified by surgical approach.

Table 3 Length of stay depending on surgical approach and technique
Full size table

Finally, after excluding patients with acute onset, length of stay remained longer after IRA (6 days) compared to DE (3 days, p = 0.01).

Overall length of stay in our cohort progressively decreased throughout our study period with a median of 7 days before 2015 and 3 days after 2015.

Postoperative complications occurred in four patients (8%), including two patients in each group (Table 4). None of the patients required reintervention after surgery for ID.

Table 4 Post-operative complications
Full size table

All the children were asymptomatic and thriving, with a median follow up of 8 months.

Pathology

Out of the 24 patients that underwent IRA, 22 specimens were retrieved and reviewed. No muscular layer between the ID and intestine was identified in 7 patients (32%) while a muscular layer was identified in 15 patients (68%): 7 (32%) had a common muscular layer and 8 (36%) had two separate muscular layers. The median thickness of the muscular layer was 3 mm (1–8).

Moreover, we found an intact intestinal wall in 91% of the cases (n = 20/22).

Discussion

To the best of our knowledge, this study is the first to compare two surgical techniques for ID, and demonstrates that DE is associated with shorter time to first feeds and postoperative stay, without increasing operative time or postoperative complications.

The rate (59%) of prenatal diagnosis in our study is slightly higher compared to the multicentric French study published in 2012 (54%), which highlights continuous progress in prenatal diagnosis [10]. Preoperative symptoms occur in half of the patients, with acute onset in 33% of patients. Symptoms, including acute onset, were found in a higher proportion of patients that underwent IRA compared to DE, which suggests the need for prenatal diagnosis and early surgical management. Indeed, when acute onset occurs, such as intussusception or volvulus, enucleation may not be feasible as resection of damaged intestine may be required.

Overall, an initial laparoscopic approach was attempted in only 45%. This can be explained by our study period, which started in 2005, whereas the first totally laparoscopic enucleation in our institution was performed in 2014. Since then, a laparoscopic approach has been attempted in the majority of the cases (87%), with a significant increase in the use of laparoscopy after 2014. The main explanation for this increase is the relatively recent development of minimally invasive surgery for most pediatric conditions at our institution. The rate of conversion to laparotomy reached 70%, which is slightly lower than the 80% rate reported in the previous French study from 2012 [10]. The benefits of laparoscopy compared to open surgery have already been widely described for many pediatric conditions and include decreased length of stay and complications rates [11]. In our study, the use of laparoscopy is significantly associated with a decrease in hospital stay, regardless of the surgical technique used for ID. To the best of our knowledge, this is the first study to compare the outcomes between laparoscopy and open surgery for ID. Moreover, we reported no complications requiring surgical reintervention after laparoscopic enucleation, whereas a recent study reported a 15% complication rate, that the authors attributed to the use of monopolar coagulation [8]. In order to further facilitate enucleation, Jin et al. have recently advocated for the use of robotic-assisted surgery, that was associated with a decreased length of stay without increasing operative time [7]. In the light of these results, we suggest that an initial laparoscopic approach should be attempted in all cases of ID without complication. In cases with acute onset and occlusion, the laparoscopic approach should be discussed depending on the feasibility of pneumoperitoneum insufflation, and cardiorespiratory instability.

In our study, DE allows earlier post-operative feeds and earlier discharge compared to IRA. The median length of stay is 4 days after DE, but may be further decreased, in order to achieve same-day discharge. The recent development of enhanced recovery after surgery (ERAS) protocols in pediatric minimally invasive surgery has shown benefits, aiming to optimize perioperative care by reducing surgical stress to facilitate a quick return to baseline [12]. Postoperative length of stay during our period decreased after 2015, concomitantly with the development of laparoscopy in our department. The decreased time to first feed and discharge after DE may be due to the absence of bowel opening and anastomosis, as well as the reduced proportion of open surgeries. In our cohort, patients that underwent laparoscopic DE had a median length of stay of 1 day, which is shorter than reported by Shchapov et al. in a cohort of 20 patients with laparoscopic DE resection that had a median post-operative stay of 15 days, including a period of 3 to 5 days of exclusive parenteral nutrition [8].

The main ID locations in our study were ileum and ileocecal junction, respectively 75% and 11%, which is in line with previous studies [10]. Unlike other studies, we decided to focus on intestinal duplication from jejunum to rectum and we excluded gastric, esophageal, and duodenal duplication, for which enucleation is usually performed.

Pathology review represents a major contribution to our study. We specifically researched and measured the muscular layer. The presence of a muscular layer in two-thirds of IRA cases strengthens the feasibility of DE for these patients. However, it should be highlighted that the median thickness of the muscular layer is measured as 3 mm, which may not always be macroscopically identified intra-operatively. The magnified vision during laparoscopic surgery may facilitate the visualization and dissection of the muscular layer, and may be further improved using three-dimensional laparoscopy [13, 14].

The fact that the integrity of the intestinal wall was preserved in more than 90% of the cases of IRA suggests that an intestinal enucleation should be attempted in the majority of the cases.

This study has several limits, due to its retrospective and monocentric nature. Although all the surgeons were from the same team, the type and approach of each procedure depended on the surgeon’s preference. In addition, initiation of first feeds was based on the surgeon's preference that can constitute a bias.

Moreover, we chose to use a large study period, and surgical practices have evolved since the beginning of the period. Finally, this study presents with selection bias, due to its retrospective nature, and the increased rate of patients with acute onset in the IRA group compared to DE.

Conclusions

This is the first study to compare the two surgical procedures for ID. Compared to intestinal resection with anastomosis, duplication enucleation is associated with decreased postoperative length of stay and delay to first feeds without increasing post-operative complications. These results highlight the necessity and the feasibility of DE procedures for the majority of patients with intestinal duplication. Moreover, with the development and known benefits of minimally invasive surgery in children, we believe that DE should be the procedure of choice.

A muscular layer was identified in 68% of intestinal resection specimens highlighting the theoretical feasibility of enucleation in the majority of the cases.