Complications of one-step button percutaneous endoscopic gastrostomy in children

To assess the complications of one-step button percutaneous endoscopic gastrostomy (B-PEG) and determine risk factors for developing stomal infections or gastropexy complications. A retrospective study of 679 children who underwent a B-PEG procedure in a single tertiary care center over a 10-year period to December 2020 was conducted. Patient characteristics, early complications (occurring ≤ 7 days after the procedure), late complications (> 7 days after the procedure), and outcomes were collected from medical records. A list of potential risk factors, including age at procedure, prematurity, underlying neurological disease, and undernutrition, was determined a priori. At least 1 year of follow-up was available for 513 patients. Median follow-up duration was 2.8 years (interquartile range 1.0–4.9 years). Major complications were rare (< 2%), and no death was related to B-PEG. Early complications affected 15.9% of the study population, and 78.0% of children presented late complications. Development of granulation tissue was the most common complication followed in frequency by tube dislodgment and T-fastener complications. Only 24 patients (3.5%) presented stomal infections. Young age at the time of PEG placement (odds ratio (OR) 2.34 [1.03–5.30], p = .042) was a risk factor for developing peristomal infection. T-fastener migration occurred in 17.3% of children, and we found underlying neurological disease was a protective factor (OR 0.59 [0.37–0.92], p = .019). Conclusion: B-PEG is a safe method and associated with a low rate of local infection. However, T-fasteners are associated with significant morbidity and require particular attention in young and premature infants. What is Known: • Percutaneous endoscopic gastrostomy (PEG) is the preferred method to provide long-term enteral nutrition in children to prevent malnutrition. The Pull-PEG method is still the most commonly used with complications , such as stomal infection. Since its description, only a few studies have reported postoperative complications of one-step button PEG (B-PEG). What is New: • T-fastener complications were not rare, and underlying neurologic disease was a protective factor. A very low rate of stomal infection was described, and young age at the time of PEG placement was a risk factor. The B-PEG is a safe method with fewer major complications than P-PEG in children. What is Known: • Percutaneous endoscopic gastrostomy (PEG) is the preferred method to provide long-term enteral nutrition in children to prevent malnutrition. The Pull-PEG method is still the most commonly used with complications , such as stomal infection. Since its description, only a few studies have reported postoperative complications of one-step button PEG (B-PEG). What is New: • T-fastener complications were not rare, and underlying neurologic disease was a protective factor. A very low rate of stomal infection was described, and young age at the time of PEG placement was a risk factor. The B-PEG is a safe method with fewer major complications than P-PEG in children.


Introduction
Percutaneous endoscopic gastrostomy (PEG) is the preferred method to provide long-term enteral nutrition in children to prevent malnutrition in severe chronic diseases [1]. Since its first use in 1980, various PEG techniques have been developed [2,3].
The first PEG procedure used is a pull technique (P-PEG) that remains the most commonly used. At the start of the procedure, a gastroscope is inserted by the mouth into the stomach to make air insufflation to provide gastric distension. The lighted gastroscope is externally visualized by transillumination to guide the placement. A small abdominal incision is then performed, and a guide wire is passed, snared by the gastroscope, and pulled out from the stomach and through the esophagus and the mouth. A PEG tube is then tied to the wire and pulled back down through esophagus, stomach, and through the abdominal incision, affixing the gastric and abdominal walls together. An external bumper is applied to secure the tube against the skin [2,4]. The initial PEG non balloon feeding tube cannot support long-term feeding and must be removed by endoscopy under a second general anesthesia to place a button [5]. Complication rates for PEG reported in the literature vary from 4% to almost 50%. The most frequent early complication with P-PEG is stomal infection [4,6]. Major complications are specific to P-PEG such as esophageal perforation (made by the PEG tube during the tract through the esophagus) or a buried bumper syndrome (when the bumper of the gastrostomy migrates into the stomach or abdominal wall, provoking resistance during the enteral nutrition, pain, or peritubular leakage) [7]. Most major complications occur within the first year after PEG placement and younger age at placement, and the presence of a neurological disorder appeared to be protective factors in most studies [4,6,8,9].
The one-step button PEG (B-PEG) push technique has been first described in the early 90 s [10] but was fully available in 2008 after the development of the MIC-KEY Introducer Kit (Kimberly-Clark, Roswell, GA).
Compared to P-PEG, the one-step PEG procedure uses a push method of button placement over a guide wire through an abdominal incision, after 3 gastropexies are placed. With this procedure, a gastrostomy button is deployed in the stomach, and its intragastric position is confirmed by endoscopy [11]. Its major advantage, compared with P-PEG, is that it requires only 1 general anesthesia, therefore reducing the number of hospitalizations and costs, despite the higher cost of the devices [8,9]. This technique was immediately acceptable to children and their families. Only a few studies have reported postoperative complications of B-PEG in children, which seem to occur with a similar frequency to those of P-PEG [5,[12][13][14][15][16][17]. A previous study from our group comparing the frequency of complications in P-PEG and B-PEG found an associated marked reduction of stomal infection (29.0% and 10.6%, respectively) [12]. We hypothesized this was due to the tube not being exposed to oral/esophageal and gastric bacterial flora during insertion [12]. Based on this early experience, one-step B-PEG has become the preferred technique in our center since 2009.
The primary aim of our present study was to describe the real-life complications and outcomes in children after B-PEG. The secondary aim was to determine risk factors for these complications, focusing on peristomal infection and gastropexy complications.

Materials and methods
We conducted a retrospective study including every patient aged < 18 years who underwent a B-PEG placement in our single tertiary care center between January 2009 and December 2020.
All the patients included benefited from the same onestep placement protocol. A MIC-KEY Introducer Kit (Kimberly-Clark, Roswell, GA) was used, and the button was placed according to the procedure recommended by the manufacturer. In order to form a stoma tract, the gastric and abdominal walls must be fixed together with an effective gastropexy. In this procedure, the gastrostomy was realized using 3 T-fasteners (Saf-T-Pexy T-fastener, Kimberly-Clark Corporation), which are preloaded gastrointestinal suture anchor system which incorporates resorbable sutures and external suture locks to allow swelling of underlying tissue and to prevent postprocedural pain. The anchors were left in place until the suture resorb and slough with the internal component passing through the gastrointestinal tract. All procedures were performed under general anesthesia with an intravenous prophylactic injection of cefamandole (Kefandol, Panpharma, La-Selle-en-Luitré, France). The standard button was a 16-Fr low-profile balloon gastrostomy feeding button (MIC-KEY button Kimberly-Clark). Enteral nutrition was started within 4-6 h following the procedure. Patients were systematically reassessed by a pediatric gastroenterologist 8 weeks later. Only if the T-fastener sutures were persistent after 8 weeks were they removed and classified as a late drop of T-fastener. The first button change was performed 4-6 months after the procedure, under medical supervision in our center.
Patient characteristics (demographic data and history, endoscopic procedure, postoperative complications, and long-term outcome) were collected retrospectively from medical records. Patients' comorbidities were categorized as follows: neurological, gastroenterological, metabolic/ genetic, pneumological/cardiological, oncological, oropharyngeal abnormalities, and prematurity (gestational age < 37 weeks). The indication for the PEG tube (nutritional support and aspiration prevention) was recorded. Undernutrition was defined as a body mass index < 2 standard deviations (SD) for age and sex. The presence of potential risk factors for complications, such as scoliosis, previous abdominal surgery, prematurity, and proton pump inhibitor (PPI) treatment at the time of the procedure, was also recorded.
Postoperative complications were classified as early when they occurred within the first 7 days after the B-PEG placement and as late when they occurred > 7 days after the placement. Non-gastrostomy-related complications were excluded (i.e., those directly linked to the underlying disease). Patients with < 12 months of follow-up were excluded from the analysis of late complications.
Early complications included stomal infection defined by the need for antibiotic treatment. Indeed, since redness and pain are frequent postoperatively, we limited the definition of local infection when redness or pain was associated to fever and/or abscess at the site of gastrostomy requiring antibiotic treatment. The loss of at least 2 T-fasteners within the first 3 days was classified as an early drop of T-fastener. Postoperative pain was assessed using the pain scales, routinely used in our pediatric service, to assess and adjust the grade of analgesics (which follow a uniform protocol). Severe pain was defined as pain resistant to firstand second-class analgesic and the need for a third-class analgesic. Ileus was defined by the necessity for abdominal aspiration for > 12 h responsible for a delay in initiating enteral nutrition. Pneumoperitoneum was diagnosed radiologically and considered noteworthy if associated with pain and/or ileus and/or the impossibility to initiate enteral nutrition within the first 12 h.
Late complications such as the development of stomal granulation tissue, tube dislodgment, excessive leakage or ulceration, and stomal infection including candidosis were recorded. Any T-fastener complications were also recorded and classified as granulation tissue after the development of a granuloma at one of the gastropexy sites; infection at the site of the external suture lock defined by the presence of an abscess at one of the T-fastener suture points requiring an oral antibiotic; transcutaneous migration of a T-bar defined by the exteriorization of one of the T-bars with a natural fall or the necessity to remove it; and a late drop of a T-fastener defined by a T-fastener suture persistent for > 8 weeks.
Major complications were defined as any complication requiring hospitalization for intravenous antibiotics, surgical treatment, or transfusion, or leading to death.

Ethics
The present research was conducted in accordance with protocols, good clinical practice, and the relevant laws and regulations in France. An information letter and an opposition form were completed by the patient's parents or legal guardian. The study was declared to the French Data Protection Authority (Commission Nationale Informatique & Libertés). All data were anonymized.

Statistical analysis
Categorical variables are expressed in terms of frequency and percentage. Quantitative variables are expressed as means ± SD in the case of normal distribution or medians (and interquartile range [IQR]) otherwise. Normality of distributions was checked graphically and using a Shapiro-Wilk test. Risk factors for early complications were identified using a logistic regression model. Risk factors for late complications were identified using a logistic regression model adjusted on follow-up time; factors significant at the level of 0.20 were introduced into a multivariable model. Odds ratio (OR) and their 95% confidence intervals (95%CI) were derived from models as effect sizes. Evolution of complication rate over time was evaluated using a logistic regression model adjusted on follow-up time. Statistical testing was conducted at the two-tailed α-level of 0.05. Data were analyzed using the SAS software (version 9.4; SAS Institute, Cary, NC).

B-PEG early complications
Early complications were rare and mostly benign, affecting 108 (15.9%) patients. Very few major complications were observed, and there was no death. A single patient could present more than 1 complication (Table 1).

B-PEG late complications
We included 513 (75.6%) patients in the late complication analysis. Late complications affected 400 (78.0%) patients ( Table 2). The 3 most frequent late complications were development of granulation tissue, tube dislodgment, and T-fastener complications (Fig. 1). No late death related to the B-PEG was reported.

Risk factors for complications
Prematurity and age < 1 year at the time of gastrostomy placement were risk factors for the early complication of stomal infection. By contrast, age < 1 year was a protective factor for the early drop of a T-fastener. No other risk factor was significantly associated with any other early complication studied ( Table 3). Because of the low frequency of early complications, multivariate analysis was not possible.
Prematurity was the only risk factor for the late complication of tube dislodgment (Table 4). An underlying neurological disease was apparently a protective factor for developing a T-fastener complication. No other significant risk factors were found.

Outcome
Among the 513 patients with > 1 year of follow-up, the median follow-up duration was 3.7 years (IQR 2.3-5.4). Among those, 149 (29.0%) died. Death was related to neither the gastrostomy placement nor any complication but to underlying disease or an accident. Enteral nutrition was discontinued in 154 (38.0%) patients, with a mean enteral nutrition duration of 33.5 months (range 3-123 months). The gastrostomy button in 106 patients was removed. Among those 106 patients, 34 (24.5%) had a persistent gastrocutaneous fistula that required surgical closure.

Discussion
To our knowledge, our series is the largest study of B-PEG and the first to report long-term complications and outcomes. The present study highlights that B-PEG is a safe method: major complications are rare (< 2% of the study population). The present results are consistent with those of  2 previous studies reporting major complications in 1% and 7% of their study population [5,14]. Notably, these rates are lower than those reported for P-PEG reaching 10-15% [6,9]. Although most complications of both techniques are similar, some major complications such as esophageal perforation or buried bumper syndrome are specific to P-PEG, and our findings suggest that B-PEG is a good technique whereby to avoid them [1,4]. Development of stomal granulation tissue is a common complication in B-PEG with variable rates from 4 to 50% [5,14,16,18], which seems to be more frequent than in P-PEG and could be explained by the deeper abdominal incision required for B-PEG than P-PEG [5].
In the P-PEG method, peristomal infection is one of the most frequent complications and is reported in up to 30% of patients [5,9,12,13,16]. The present study shows a very low rate of wound infection (3.5%) with B-PEG, consistent with the low rate of 5% found by Göthberg and Björnsson [5] and confirms our early experience showing a marked difference in peristomal infection between P-PEG (29%) and B-PEG (10.6%) [12]. This marked difference is mostly explained by the push B-PEG technique, which avoids bacteria in the oral and upper gastrointestinal tract during the placement of the gastrostomy button. In the present study, we also show that with increasing experience, the rate of infection decreased over time. We found that stomal infection is more frequent in young infants and those born prematurely. We have no clear explanation for these findings but speculate that an immature immune system favors such a complication. In any case, special attention and reinforcement of preventive measures (careful skin disinfection and perioperative antibiotic prophylaxis) are recommended for this group of patients to avoid peristomal infection.  Some complications are specific to B-PEG (i.e., first button placement and gastropexy). In our previous experience with B-PEG from 2009 to 2010, we found a high rate of premature loss of the button (35%) within the first 6 months after placement. We therefore decided to replace the balloon button systematically 4 months after the B-PEG procedure [12]. The present study shows that 16.6% of our population still experienced tube dislodgment before the first change, suggesting that 3-4 months are a reasonable time for the first replacement. In the present study population, only 1.0% experienced early tube dislodgment during the first 7 days, consistent with findings by Dahlseng et al. [17].
Surprisingly, many of our patients (20.0%) had a T-fastener complication, which is a specific complication of push technique PEG. This complication has been poorly reported in children [14,19]. Kvello et al. found that 13% of their population experienced T-fastener problems and suggested that thinner tissue layers could be a reason for greater T-fastener migration in children than in adults [14]. We found that patients with neurological impairment apparently present less T-fastener complications, which could be explained by the immobility of most of these patients, and therefore, potentially less tension and traction on the gastropexy strings and T-bars. Although not severe, avoiding such complications is an important goal because they cause discomfort and pain and require medical care. Some investigators, such as Göthber and Björnsson, removed the suture lock systematically 2-3 weeks after the procedure to prevent T-fastener complications [5], but this measure has not been fully evaluated. Dhalseng et al. decided to cut the sutures 2 weeks after the procedure, and the rate of T-fastener complications decreased significantly from 13 to 1% [17]. We caution against an early cut of the suture because the gastrostomy tract needs time to be fixed, and a lack of gastropexy may increase the risk of intraperitoneal migration of the button in the first days after B-PEG placement, as we have observed in 1 patient. We did not find any study in the literature reporting risk factors for complications of B-PEG in children, only factors for P-PEG complications. Here, we found prematurity was associated with some B-PEG complications, most probably because of the fragility and thinner tissue layers in this population who had a higher risk of tube dislodgment.
We found that younger patients, aged < 12 months, had greater risks of wound infection and late complications such as T-fastener complications. In studies of P-PEG, young age was not associated with complication rate and was even described as a protective factor for developing major complications [6,20]. By contrast, late complications such as T-fastener complications were significantly less frequent in patients with a neurological disorder. The apparent protective effect of neurological impairment is already reported by McSweeney et al. and was explained by the more frequent hospitalization of this group and thus more intensive medical supervision, which may prevent the occurrence of some complications [6,8]. We confirmed that neither scoliosis nor surgical history is related to complication rate [20].
An important finding of our study is the fragility of the population in whom B-PEG is performed. Indeed, even if no B-PEG-related death was recorded, many children died because of their underlying disease progression. B-PEG offers the advantage of having fewer major complications than P-PEG and seems adapted to this population with comorbidities, who require enteral nutrition to support their growth. Moreover, we also show that some of these patients can be weaned from enteral nutrition, while 25.0% of them will present a persistent gastrocutaneous fistula, which may be related to the longer duration of their gastrostomy feeding tube [21].
The retrospective design of our study may underestimate the prevalence of minor complications, as some patients were referred for B-PEG placement from another hospital.
However, since these are complex patients referred to our ward for performing PEG, we organized the follow-up in a standardized and centralized manner. As previously reported by McSweeney et al. [8], most major complications seem to occur in the first year of follow-up. In the present study, patients with < 12 months of follow-up were excluded from the analysis of late complications, and we adjusted the results on the follow-up time. The main strength of our study is the large sample size, which to our knowledge is the biggest cohort reported to date, and the standardized B-PEG placement, which was performed by a specialized team. The present study allows not only a wide overview of complications of B-PEG but also a long follow-up and extensive experience of these patients who benefited from a B-PEG placement.
Authors' contributions All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Charlotte Jean-Bart, Frédéric Gottrand, and Emeline Cailliau. The first draft of the manuscript was written by Charlotte Jean-Bart, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Declarations
Ethics approval and consent to participate This is an observation study. The present research was conducted in accordance with protocols, good clinical practice, and the relevant laws and regulations in France. An information letter and an opposition form were completed by the patient's parents or legal guardian. The study was declared to the French Data Protection Authority (Commission Nationale Informatique & Libertés). All data were anonymized.

Competing interests
The authors declare no competing interests.