Candy cane revision after Roux-en-Y gastric bypass

From our prospectively collated bariatric surgery database, we selected all cases of CC surgically revised between 2010 and 2017 at our high case volume bariatric centre of excellence. All operations were performed laparoscopically and involved a diagnostic laparoscopy to exclude any other pathologies, including internal herniae. Following that, resection of the CC segment was performed using a linear stapler with a 34-Fr orogastric tube passed through the gastrojejunostomy in order to avoid narrowing it. No revision of gastrojejunostomy was required. Length of the CC was assessed intraoperatively with reference to the laparoscopic instruments, which is a standard technique at our centre. In all cases where mesenteric and/or Petersen’s spaces were identified as open, laparoscopic closure was performed. Cases where another simultaneous procedure was performed (such as cholecystectomy, internal hernia repair, major adhesiolysis or a minimiser ring placement) were excluded from the study in order to avoid confounding factors interfering with CC revision outcomes review. All barium contrast swallow tests and CT scans were reported by radiologists, and diagnosis of CC was made by assessing the redundant blind-end loop and measuring its length. All upper gastrointestinal endoscopies were performed and reported by experienced gastroenterologists. Data were collected from patient’s medical notes and where necessary were supplemented by telephone follow-up with patients directly.

Variables analysed included demographic data, pre- and post-CC revision weight, Body Mass Index (BMI), indications for CC revision, perioperative and follow-up data, symptom resolution and complications. Suboptimal weight loss was defined as failure to achieve 20% total body weight loss (TBWL) following primary RYGB or weight regain of 25% or more. Complications were classified according to the Clavien–Dindo classification [11].

Statistical analysis with SPSS v.20 was performed. Data where values were normally distributed are presented as mean ± SD as well as ranges and percentages. Descriptive statistics were computed for all variables. Chi-square test was performed for comparison between the CC size and different symptom categories (e.g. pain, regurgitation/vomiting). Student’s t test was used for analysis of continuous data. All p values are two-sided. In order to test the accuracy of the diagnostic test in correctly detecting CC, the area under the ROC Curve was calculated. The values can range from 0.5 (no diagnostic ability) to 1.0 (perfect diagnostic ability), where 0.7 to 0.8 is considered acceptable and over 0.8 is considered excellent [12].

Study was performed as part of routine clinical practice and audit of hospital outcomes; therefore, ethics committee approval was not required. Data collection was performed in compliance with the Declaration of Helsinki.

Although RYGB is a safe and effective operation, a proportion of patients will later undergo revision surgery [13]. Indications for revision surgery fall broadly into two camps: operation failure—predominantly patients who have experienced suboptimal weight loss or weight regain—and intractable symptoms, such as abdominal pain or vomiting. Such revision surgery is successful for some, but many patients will experience no benefit [14, 15].

Candy cane or hockey stick after RYGB refers to an excessively long-blind afferent Roux limb at the gastrojejunostomy site. This anatomical phenomenon, which may be detected on preoperative investigations or during subsequent surgery, has been attributed as a cause of both RYGB failure and a wide range of common postoperative symptoms and is in some centres considered an indication for revision surgery [8, 9]. Despite this, there are very little data regarding the efficacy of surgical treatment for the condition as well as diagnostic accuracy of preoperative investigations to reveal CC.

Our study is the largest series of surgically treated candy canes to date. Our data support previous case series which describe heterogeneous symptomatology of CC [8, 9]. A proportion of patients post-RYGB present with diverse symptoms and/or evidence of surgical failure: our data support the commonly held clinical belief that CC should be considered among the differential diagnoses in such cases. Our results show, however, that widely employed investigation methods have poor sensitivity for the diagnosis of CC. Therefore, failure to diagnose a candy cane preoperatively does not exclude the diagnosis. Barium swallow, which had an overall true positive rate of 63%, does become more sensitive when the size of the CC is 2.5 cm and above (> 90% true positive rate). Nonetheless, two or even three diagnostic investigations may be necessary in order to reveal CC in symptomatic patients after RYGB.

In our series of an unselected cohort of patients with CC, the vast majority experienced complete or partial symptom resolution (73%) following CC revision. The rate of symptom resolution reported by us is lower than 94% presented by Aryaie et al. in their series of 19 patients with preoperatively defined ‘CC syndrome’ [9]. This could be due to the fact that patients with very specific symptom profile, i.e. postprandial epigastric pain or nausea and vomiting were selected in that study, whereas we included patients with all presenting syndromes and their combinations. The patients who benefited most from the CC surgery in our study were those presenting with regurgitation and/or vomiting (67% symptom resolution rate). More than half of the patients complaining of abdominal pain or reflux benefited from the CC revision (57% and 60% symptom resolution rate, respectively). It has been hypothesised that these symptoms may occur due to decreased small bowel motility in the blind-end loop with associated dilatation from food residual and possibly a co-existing bacterial overgrowth [9]. Therefore, resection of the CC is expected to alleviate these symptoms early postoperatively.

It has been debated whether patients suffering from poor weight loss or post-RYGB weight regain would benefit from the CC revision. Our data showed weight loss in 3 of 10 patients who had presented with suboptimal weight lost post-RYGB and in further 4 patients who had not complained of poor weight loss or weight regain prior to the CC revision. However, 3 patients from the latter group underwent the revisional surgery within 1.5 year from the primary RYGB and further 1 patient was later diagnosed with a pharyngeal pouch. These factors could have contributed to the observed weight loss, which could have happened regardless of the revision surgery. Furthermore, the weight loss following the CC surgery could be influenced by the intensified multidisciplinary follow-up which these patients would have received. Other theories regarding weight loss after the CC revision are the relative narrowing of the gastrojejunostomy following the revision and the removal of the ‘food reservoir’ [9], but the influence of those is difficult to assess in this study setting.

All patients who did not report any improvement in their symptoms after the CC revision continued to be investigated and treated by the bariatric multidisciplinary team. In 5 cases, other diagnoses (such as small-intestine bacterial overgrowth) were found and further 2 patients received ongoing support for pain management. It is therefore not only important to investigate further patients in whom CC revision does not result in symptom resolution but also to consider additional preoperative investigations such as routine testing for small-intestine bacterial overgrowth (which is currently a routine practice in our department).

The rate of complications seen in this study is significant but it is comparable to those of other studies of revisional bariatric surgery [14‐16]. Therefore, as with other revisional surgery, it is recommended that they are performed at large volume bariatric centres to minimise the risk.

We acknowledge that this study has limitations. Firstly, the cohort of patients studied was small and heterogeneous, especially with regard to the symptom profile, size of the CC and length of the postoperative follow-up. Nonetheless, this is the largest series to date. Secondly, we did not have a comparator group in this study. It is plausible that some patients may have experienced improvement in symptoms over time without the surgical intervention and enhanced multidisciplinary follow-up after revisional surgery may have played role in improved outcomes. Moreover, the number of patients in our series may have been too small to detect relationships between presenting symptomatology and the likelihood of symptom resolution post-revision. There was no standardised scoring for symptoms assessment pre- and post-CC revision which could have affected reported symptom resolution rate as it was based on patient’s report at the follow-up. It is also worth noting that due to the design of our study, we cannot make any conclusions regarding the specificity of the investigations assessed for CC detection. Lastly, with just over half of the patients undergoing a prophylactic closure of mesenteric and/or Petersen’s defects, it is possible that some might have had an undiagnosed internal hernia, and this potentially could be a confounding factor when interpreting postoperative symptom resolution.

Nonetheless, as the largest series of CC revisions to date, this study adds in a meaningful way to our understanding of this post-RYGB complication. Our data would suggest that even in the situation where a CC is newly diagnosed during a diagnostic laparoscopy, revision should be performed in symptomatic patients, giving them a high chance of at least partial symptomatic improvement.

CC remains a significant challenge in terms of diagnosis and management. It is likely to be responsible for some cases of both primary RYGB failure as well as intractable symptoms post-RYGB. More than one diagnostic tool (barium swallow ± upper gastrointestinal endoscopy ± CT) is likely to be required when investigating symptomatic patients after RYGB. Revision of CC is associated with good rates of symptom resolution (especially regurgitation and/or vomiting) in the majority of patients.