Low Risk for Marginal Ulcers in Duodenal Switch and Gastric Bypass in a Well-Defined Cohort of 472 Patients

Data for age, BMI, and weight are shown as mean ± standard deviation, unless otherwise noted. Categorical data were analyzed by χ2 test, while two-tailed Student’s t test was used for quantitative data. p < 0.05 was regarded as statistically significant. For both data sets, the years at risk were calculated subtracting the date of surgery from the time of survey or last clinical follow-up in SOReg, respectively. IR of MU per patient per year was calculated by dividing the total number of MU by number of years at risk. To identify predictors for development of MU, we performed univariate and multivariate analysis including clinically relevant baseline and perioperative data and BMI reduction at 2 years, as well as smoking habits and ongoing PPI treatment. To test equality of two incidence rates for MU between survey and SOReg, we used exact tests due to the small numbers of cases and report mid p values due to the discreteness of the p value. SPSS (IBM, Armonk, NY, USA) was used for all analysis, except the analysis between the two IR which was performed in using STATA (Stata/SE version 15.1 for Windows).

In total, 472 patients (64%) answered the questionnaire, whereof 65% were women and 42% had a DS. As demonstrated in Table 1, DS patients had somewhat higher preoperative BMI (56 vs 51) and longer operative time (149 vs 93 min).

The cohort had a mean follow-up of 6.1 ± 2.7 years. In total, 41 (8.6%) of the patients reported in their questionnaire to have had an MU during the study period, thus resulting in an incidence rate of 1.4% (DS 1.3% and RYGBP 1.5%), without significant difference between the two procedures (p = 0.721). As demonstrated in Table 2, gastrointestinal bleeding, abdominal pain, and dysphagia were the most common symptoms leading to endoscopy and the subsequent diagnosis of MU. All 23 ulcers were found on the small bowel side of the anastomosis, and in 4 RYPGBs, a stenosis in the gastrojejunostomy was also diagnosed.

In both the univariate and multivariate analysis, there was a significant association between a former MU and ongoing PPI treatment (OR 11.2 [3.6–34.7], p < 0.001). No other associative factors were found Table 3.

For SOReg-data, the incidence rate for MU in our DS and RYGBP was 1.1 and 0.7%, respectively, with no significant differences between the two procedures (p = 0.458). Compared with the results from the present survey, there was a trend towards a lower overall risk for MU (0.9% vs 1.4%, p = 0.084) as well as in RYGBP (0.7% vs 1.5%, p = 0.067).

The present incidence rate of MU was just above 1%, i.e., only one out of 100 operated patients will suffer an MU annually. As mentioned earlier, Hess et al. described an incidence of 0.3% under a follow-up period of over 10 years in DS, whereas Pata and Scopinaro demonstrated a higher incidence of MU in classic biliopancreatic diversion, 2% and 3.2%, respectively [28, 29]. The difference could perhaps depend on the larger gastric remnant in biliopancreatic diversion, resulting in a higher amount of gastric acid passing through the upper anastomosis. This theory is supported by an earlier study by our group on RYGBP, in which we were able to show that the time with pH < 4 in the gastrojejunostomy was longer in patients with MU (69%) compared with in non-MUs (20%, p < 0.01) [14]. Interestingly, these absolute numbers are very similar to the time with pH < 4 in non-symptomatic DS patients, 59% in the acid-producing sleeve and 14% in the buffered duodenal bulb, as mentioned in the introduction. It would therefore be of great interest to measure the acidity in the present patients with MU. In the background population, the lifetime prevalence of a peptic ulcer is estimated to be 5–10% with an incidence of 0.1–0.3% per year [30, 31]. Thus, the risk for having an ulcer is ten-fold in patients with a history of bariatric surgery. The majority of MU presented within the first 2 years, which is in line with several other studies [17, 19, 32]. Furthermore, none of our patients suffered a repeated MU nor a perforated MU, which otherwise is reported in 0.25 to 1% of cases [33].

Compared with SOReg-data, the present survey had a trend towards higher overall incidence rate for MU (1.4% vs 0.9%, p = 0.084) as well as MU in RYGBPs (1.5% vs 0.7%, p = 0.067). This is probably due to more detailed data obtained by our questionnaire and the fact that SOReg-registrations are made in routine clinical care. Moreover, the diminishing follow-up rate of our patients in SOReg (38% and 21% at 2 and 5 years, respectively) reduces the possibility of having a registered MU.

PPI treatment leads to complete resolution in all MUs. Persisting PPI treatment was seen in almost three quarters of our former MU patients, thus highly associated to MU (OR 11.2 [3.6–34.7], p < 0.001). This high number indicates either an increased vulnerability to acid-related conditions or a concern among patients and physicians of having a repeated ulcer. Furthermore, a quarter of the patients in the present cohort had ongoing PPI treatment, which is higher than in an earlier national study by our group on gastrointestinal symptoms 5 years after bariatric surgery. In that study, we were able to verify the good anti-acidic effect in RYGBP as the use of PPI was significantly reduced postoperatively, from 9.9 to 6.4% of the patients [34]. However, in general practice (n = 40,000 patients), it was found that a surprisingly large proportion, 14.7%, were PPI users and the author concluded that a substantial number of individuals did not meet the correct indications for prescription [35]. Finally, in minimizing the risk for MU development in bariatric surgery, there is an increasing trend to administer PPI as a postoperative prophylaxis. However, neither a clear consensus on the duration of such treatment (30 or 90 days) nor solid evidence of its effect in the long term exists [20, 22, 36, 37].

The well-defined cohort and thorough review of patients’ charts is a strength of the present study, thus adding to the limited number of studies on MU in DS. Although the number of patients (n = 472) is a limitation, the present work provides a more accurate incidence rate of MU than in former registry studies. The response rate was not optimal (64%); however, in a sensitivity analysis, responders only differed from the remaining patients by being older (41 vs 34 years) and more often having comorbidities (65% vs 55%, p < 0.05 for both). Moreover, we lack knowledge of peptic ulcer history as well as preoperative PPI use and smoking habits at time of MU diagnosis, three factors demonstrated to be of importance in gastric bypass patients [38]. Finally, the low number of endoscopically verified MU is a limitation; however, in all patients, a formal diagnosis of MU was set in medical charts.

In conclusion, the incidence rate of marginal ulcers after duodenal switch and gastric bypass was low, around 1%, and did not differ between the two procedures. We believe that this is reassuring for future bariatric surgery candidates as well as bariatric surgeons.