Introduction

In the USA, the prevalence of obesity (a body mass index (BMI) of >30) is around 30 % in the adult population [1]. The incidence is increasing, and the World Health Organization predicts that in 2025 there will be 300 million obese people worldwide [2].

Obesity is associated with a range of comorbidities such as metabolic syndrome, early osteoarthrosis, obstructive sleep apnea, and a high risk of cardiovascular disease [3]. At present, bariatric surgery is the only long-term effective treatment for morbid obesity (BMI of >40). It aims at inducing weight loss by reducing the gastric volume and/or absorptive capacity of the intestines. A wide variety of bariatric procedures have been developed such as (laparoscopic) adjustable gastric banding, (laparoscopic) gastric sleeve resections and laparoscopic Roux-en-Y gastric bypass (LRYGB). LRYGB is considered the gold standard because of the superior results. Compared to gastric banding, LRYGB produces sustained weight loss and higher resolution of obesity-associated morbidities. The laparoscopic approach is associated with faster recovery, shorter length of stay, higher success rate, and lower morbidity and mortality compared to the open procedure (RYGB) [4–7].

At present, bariatric surgery is mainly performed in high-volume centers. The LRYGB is a major operation with potentially severe early and late complications. The majority of the complications occur during the procedure or in the early phase afterwards. Due to a more sufficient follow up and increasing performance of the procedures, a higher number of late complications are identified [8].

One of these late complications is marginal ulceration. A marginal ulcer is defined as an ulcer at or near the gastrojejunostomy (GJS). In medical literature, at least three synonyms are used: marginal, ischemic, and anastomotic ulcer. In this text, we will use the term marginal ulcer (MU). As the number of LRYGB performed worldwide rises, the number of patients with MU will subsequently increase [4, 9, 10].

The incidence of MU is unclear, and reports vary from 0.6 to 25 %. MU is associated with, sometimes severe, morbidity and can be potentially lethal [11–13]. Patients may present with perforation or massive bleeding after an asymptomatic onset. Other, less acute symptoms are epigastric burn and/or vomiting [14–17].

This systematic review analyses literature published about MU. The main focus will be the incidence and risk factors for development of MU. The evidence for preoperative testing and treatment of Helicobacter pylori, standard prescription of proton pump inhibitors (PPIs) prophylaxis and symptoms at presentation were also assessed. To the best of our knowledge, no review of the available literature has been published yet.

Materials and Methods

Literature Search

The Cochrane Database of systematic reviews, the Cochrane central register of controlled trials and the PubMed database were independently searched by two separate investigators (UKC, ABG) using the keywords ((Peptic ulcer disease OR marginal ulceration OR anastomotic ulcer OR ischemic ulcer OR ulcers OR ulcera*)) AND (((((“Bariatric Surgery”[Mesh:noexp]) OR “Gastric Bypass”[Mesh])) OR (gastric bypass*[tiab])) OR (bariatric[tiab]) in order to identify studies published until the first of October 2012. MeSH terms and free text words were combined to avoid exclusion of recent articles that had not been given a MeSH label yet. Only full texts published in English were included. Electronic links to related articles and references were cross-checked.

Study Selection and Data Extraction

The PRISMA statement for systematic reviews and meta-analysis was used for study selection and data extraction [18]. From the potentially eligible publications, only studies that reported on ulcer disease around the GJS were included. A clear definition of study objectives, description of data collection, and a minimum of four patients were required for inclusion.

Exclusion criteria were less than four cases, full text in a language other than English, words used in a different context, pathology in the remnant gastrointestinal tract, gastrogastric fistulae, or radiologic diagnosis of MU. Studies about the reoperative management but not about incidence or pathophysiology of MU were left out of the analysis but included in the text for additional information. Data was retrieved from the articles only. No attempt was made to obtain missing/additional data from the authors or institutions.

Data Synthesis

Each of the selected studies was thoroughly analyzed by two investigators (UKC and ABG). The data was extracted from the original articles by using a preformatted sheet as proposed by the Cochrane Collaboration. Study period; study design (randomization, prospective, or retrospective consecutive data collection); comparability of study groups; adequate follow up; and presence of performance, selection, attrition, or detection bias were assessed. In cases of retrospective analysis of data collected from a prospective consecutive database, the study was qualified as being prospective. Any differences of opinion between the two investigators were discussed and resolved during a consensus meeting.

Results

Included Studies

Search process and study selection are displayed in a flowchart (Fig. 1). With the above search terms, 394 publications were retrieved. Three hundred eighteen contained the search items in a different context and were therefore deemed irrelevant. A total of 76 articles were selected for closer reading. Forty-one were excluded based on the abstracts. Of the 36 remaining articles, one was not written in English and therefore discarded. References were cross-checked and six additional articles were found. A total of 41 articles were scrutinized and examined for data. Other articles were kept for additional information [9, 11, 12, 14, 15, 17, 19–41]. The additional Table 1 contains the included studies and rates their quality.

Fig. 1
figure 1

Flowchart of a systematic review about the incidence and symptoms of marginal ulceration after Roux-en-Y gastric bypass

Full size image
Table 1 Quality of included studies (additional)
Full size table

Patients

All patients met the criteria for morbid obesity and a total of 16,987 patients (mainly female patients; with age ranging from 16 to 72) underwent a LRYGB and were included in the present review. During follow-up, 787 (4.6 %) patients developed MU. The time between surgery and presentation with MU varied between 1 month and 6 years [22, 31, 32, 42].

In three studies, standard screening was performed, both asymptomatic and symptomatic patients with MU were traced [22, 23, 25]. In the other studies, only symptomatic patients were analyzed.

Age and weight were normally distributed in most research groups and did not predispose for the development of MU. Male sex seemed to increase the risk for MU but not significantly [15, 32, 43].

Risk Factors

The incidence of marginal ulceration (ranging from 0.6 to 25 %) is listed in Table 2 together with the use of prophylactic PPIs, the technique by which the anastomosis is created and the symptoms.

Table 2 Baseline characteristics, incidence of MU, operative characteristics, evaluation of patients, and proton pump inhibitor prophylaxis
Full size table

Type of Procedure and Suture Material

Thirty-two articles mentioned surgical technique; 78.3 % RYGB was performed laparoscopically. No difference in ulcerogenic potential was found between open and laparoscopic procedure [12, 21] (Table 2). Capella et al. showed that the use of staples results in a higher incidence of MU compared to absorbable suture materials. In the study by Rasmussen et al., 32 % of the ulcer beds showed remnants of suture material at esophagogastroduodenoscopy [19–21, 32, 34, 44]. Local ischemia seems to enlarge the risk for MU [45].

Position of the Pouch and the Role of Gastric Acid

Historically, the first focus of interest was the position and size of the pouch. The concentration of the parietal cell mass in the stomach is divided into areas [46]. Most parietal cells are situated in the antrum, whereas proximal in the stomach almost no cells are present [20]. Patients with a large, less proximal pouch have a higher risk for MU because a part of the antrum is included. In biliopancreatic diversion, the pouch is more obliquely orientated, containing more parietal cells and there is a higher incidence of ulcers. A small proximal pouch, limited to the cardia, reduced the occurrence of MU from 5.2 to 0.01 % in 1 year [43]. In LRYGB with a micropouch, the incidence of MU is also lower [46, 47]. The technique for pouch creation in RYGB is now standardized [48].

H. pylori

The incidence of infection with H. pylori in patients who are screened for bariatric surgery differs between 22 and 67 % [26, 41, 49–51]. In this review, 12 articles tested the presence of H. pylori at the MU site. In 10.5 %, the test was positive for infection [12, 23, 26, 28, 32, 33, 37, 41, 52–54].

Two studies found an association between preoperative infection and eradication of H. pylori in relation to MU and other gastrointestinal complications [26, 32, 55]. The recent published study of Rawlins et al. did not show a significant difference in the rate of complications between patients who were infected preoperatively with H. pylori or not [56].

Suggs et al. published a study of 23 patients with MU after surgery who all tested negative for H. pylori with the Campylob acter-like organism test [28, 31, 37, 41, 52]. Marano et al. and D'Hondt et al. were also unable to demonstrate a relationship between infection with H. pylori and MU [30, 52].

NSAIDs, Smoking, Diabetes Mellitus, Cardiovascular Disease, and Other Patient Demographic Risk Factors

The use of nonsteroidal inflammatory drugs (NSAIDs) increase the incidence of peptic ulcer disease (PUD) significantly [57, 58]. Another risk factor for PUD is the abuse of tobacco [59]. Wilson et al. performed a uni- and multivariate analysis on the use of NSAIDs and tobacco after LRYGB. Both factors independently predicted formation of MU. Protection against MU was achieved when PPIs were simultaneously used with NSAIDS [60]. In this review, 19 of the included articles scored the use of NSAIDs in the patients with MU. Of the 365 patients, 98 used NSAIDs at the time of presentation [61–64]. The use of NSAIDS is not only related to the formation of MU, they also inhibit healing of ulcer disease [65].

Ten articles mentioned smoking. A mean of 35.8 % of the patients smoked while developing MU. Smoking is a risk factor, particular for perforated MU. After healing, Patel et al. present three patients who developed recurrent ulceration—all heavy smokers. Another study showed a significant difference in the formation of MU as well as in healing capacities between smokers and nonsmokers [14, 31, 35, 67].

Seven studies mentioned patient's comorbidities. Two studies focused on the influence of diabetes mellitus (DM) and cardiovascular disease on MU. One found an increased risk for MU in patients who suffered from DM. The other study did not [12, 15, 32–35, 52]. None of the studies found a correlation between the use of alcohol and the presence of MU [67].

Symptoms

Csendes and Garrido found that of all patients with MU, 28–100 % does not have “typical” symptoms as epigastric or abdominal pain, nausea, and/or vomiting. Some patients have no symptoms at all (Table 3) [22, 25].

Table 3 Characteristics of symptomatology of MU
Full size table

A total of 30 articles (777 patients) described symptoms at presentation. Of the 777 patients, 441 (56.8 %) experienced epigastric burn. In 117 (15.1 %) patients, bleeding was the main symptom. Patients with perforated MU will present with signs of acute abdomen at the emergency room [14, 15, 17, 24, 26–29, 31, 34, 35, 37, 40, 43, 68].

Suggs et al. described 23 patients who developed MU, only seven had the classical, non-acute symptoms such as abdominal pain. Ten presented with melena (four also had hematemesis) and eight required blood transfusion. Of the patients, 17 out of 23 were readmitted.

Perforation

The incidence of perforated MU after LRYGB is around 1–2 % in the total population, which means that around 20 % of the patients with MU present with perforation [14, 24, 26, 28, 29, 35, 40]. Felix et al. described that 69 % of the patients with a perforated MU had identifiable risk factors including smoking, use of NSAIDs or steroids. Although 31 % had no identifiable risk factor, roughly a third of this group had a history of treatment for MU. Twenty percent of the patients had no warning signs prior to perforation [14].

Ulcer Treatment: Pharmaceutical Treatment, Reoperation, and Upper Endoscopy

Medical treatment of MU consists of PPIs, H2 antagonists, Sulcrafate®, or a combination of these medications. Thirty-one articles mentioned a form of treatment. Of the 801 (including patients with perforation) patients, 67.9 % could be sufficiently treated with medication alone [9, 11, 13, 22–25, 27, 30, 33–35, 37, 43, 52–54, 56, 67–70]. Endoscopy confirmed the healing properties of PPIs in late MU. Other patients were treated by radiologic or endoscopic interventions. Around 23 % of all patients needed one or more reoperations for complete healing [13, 15, 27, 31, 43, 52–54, 67].

Most of the patients in need of surgery are those with perforation, dilated pouch, retractable marginal ulcer, or gastrogastric fistulae. The majority of data about revisional surgery for marginal ulceration reflects the open operation technique which is known for its greater complication rate including leakage, wound infection, blood loss, and higher mortality rate. At present, laparoscopic revisions are effective and safe also after open primary procedure [31, 35, 45, 71–75]. All patients who presented with perforation needed reoperation or at least radiology-assisted drainage [14, 26, 28, 35, 40].

Patel et al. presented a case series of reoperation for marginal ulceration with a success rate of 87 % [31]. In some studies, an attempt was made to enhance the healing process by removal of the foreign material with upper endoscopy [15, 32, 34, 67, 70].

Proton Pump Inhibitors as Prophylaxis

In the last few years, the prophylactic prescription of PPIs after RYGB has become standard procedure. However, no consensus exists about the duration of usage (Table 2). In the literature, the time of postoperative PPI administration differs between 30 days to 2 years, some authors argue for lifelong usage [19, 24, 25, 28, 29, 33, 40, 52].

D'Hondt et al. found that there was no statistical significance in the incidence of MU between patients who did or did not receive PPIs postoperatively. The incidence of MU in this study was 10.7 % with a minimal follow up of 6 months [52].

As previous described, NSAIDS increase the risk on ulcer formation. However PPIs provide significant protection used simultaneously with NSAIDS [60].

Discussion

The performance of Roux-en-Y gastric bypass (both primary and as revisional procedure) is globally increasing to enormous numbers with a subsequent rise in its associated complications such as marginal ulcer. Most articles in this systematic review on MU are retrospective.

The majority of studies examined symptomatic patients. The two studies examining a consecutive group of patients show that the incidence of MU is underestimated. One of the studies had a follow-up period of 2 months after surgery. It is likely that the ulcers were still superficial due to the early detection and therefore less prone to cause symptoms.

As soon as the importance of the position of the pouch became known, the operation was internationally standardized. Introduction of the laparoscopic technique further contributed to a standard pouch formation procedure [12, 48]. A dilated pouch may predispose to late ulceration because of the increasing number of parietal cells after dilatation [12, 76]. Some authors advocate a vagotomy in an attempt to reduce the secretion of gastric acid [12, 15, 43, 76, 77]. Acid secretion is also partially regulated by gastrin levels. Because of the negative feedback mechanism, acid secretion rises when pouch pH is high [78–80]. A decrease in pH increases the development of MU. In most patients, treatment with PPIs alone is adequate to treat and prevent MU. This supports the role of gastric acid in the formation of MU [27, 68, 70, 76].

The protective mechanism of stomach evacuation is probably due to the subsequent absence of acid production by the remnant stomach, caused by the hormonal feedback mechanism. The formation of fistulae between the remnant stomach and the pouch and gastrogastric fistulae enhance the development of MU because they increase the amount of gastric acid. The vulnerable jejunal mucosa is exposed to the harmful acid [11, 21, 32, 76, 81].

Various ways to create the gastrojejunostomy are described [82]. Evidence supports the use of absorbable suture material, as foreign materials are found in a third of the MUs [21].

The incidence of H. pylori infection found at the preoperative screening in patients undergoing bariatric surgery ranges from 22.4 to 61.3 % depending on the patients region of origin [50, 51, 83, 84]. Some authors suggest that H. pylori increases a variety of gastrointestinal symptoms after gastric bypass and therefore advise standard eradication therapy even without testing in patients prior to surgery. In perforated ulcer disease, Hartin et al. hypothesize that preoperative detection and eradication of H. pylori infection may decrease the incidence and/or severity of peptic ulcer-related problems but this is not scientifically supported [26, 50].

Some studies advocate the opposite and in the literature only some of the patients presenting with MU tested positive for H. pylori. In this review, the mean incidence of H. pylori infection in MU is 12 % (range, 0–33). The percentage of H. pylori infection in normal gastric and/or duodenal ulcers is between 70 and 97 % [64, 85–89]. Although patients with perforated MU were not included in the total group of patients to prevent bias, analysis of these patients is important because early identification of MU can prevent this serious complication [14]. All patients described needed reoperation or drain placement. After Sasse et al. adopted a two-step approach to ulcer prevention, no new cases of perforation occurred. This protocol included a 12-week empirical treatment with PPI direct postoperative and a zero-tolerance policy towards the use of NSAIDs [35].

The ulcerogenic potential of NSAIDs has been extensively studied in the general population. NSAIDs achieve the anti-inflammatory effect by inhibiting the cyclo-oxygenase (COX)-2 pathway. COX 2 is responsible for the tissue prostaglandin production. They also interfere with the COX-1 and thereby the production of the PGE2 prostaglandin responsible for the gastric mucous barrier [61–63]. The exact significance of NSAIDs as a factor in MU is unknown because quantification of usage is difficult to assess. Most patients describe over-the-counter usage of NSAIDs on an as needed basis [34]. The same principal applies to smoking. Although the percentage of smokers is given in the affected population, the percentage in control groups is unknown. Only one study mentioned the use of alcohol, it was not significantly related to the development of MU.

Prophylactic PPI administration was introduced in some research groups after evaluation. However, the variety of duration in administration, the small number of patients, and the lack of follow up made it impossible to provide solid evidence concerning the benefits of this protocol; a positive effect does seem to exist [19, 24, 28, 29, 52].

This review did not focus on the treatment of MU. Most patients respond well on PPIs and lifestyle adjustments alone [32–34]. In order to achieve healing, NSAIDs should be stopped; patients who are smokers must be motivated to quit smoking and anticoagulation therapy should be antagonized in case of hemorrhagic presentation. Revisional bariatric surgery is technically challenging and has been associated with high morbidity rates and can be potentially lethal. However, the laparoscopic approach has shown to be safe with good results [31, 35, 45, 71–73].

Conclusion

This systematic review represents the best available evidence to date. The incidence of MU ranges from 0.6 to 25 % and no methodological high-quality studies are available for identification of the risk factors.

The pathophysiology of MU remains unclear. The only evidence-based consensus is that the risk of MU can be diminished with proximal pouch orientation and the use of absorbable suture material. Risk factors seem to be NSAID usage without PPIs, smoking [14, 17, 28, 31], and use of non-absorbable suture material [19, 21].

It can be concluded that the pathogenesis of MU formation after RYGB is different compared to PUD. Various factors contribute to this complication [28, 31, 32, 37].

Symptoms at presentation such as epigastric burn, vomiting, hematemesis, or melena merit diagnostics for MU. An acute abdomen, weeks to months and even years after RYGB, may indicate perforation. MU can be treated with PPIs, sometimes with Ulcogant®. When perforated, reoperation or (percutaneous) drainage is often required.

A trend is noticed in favor of postoperative prophylactic PPI administration to prevent MU. With the increasing number of LRYGB and the consequent rise in MU, many of which are asymptomatic, more knowledge about the pathophysiology, prevention, and treatment is required.