LAUSR

To the Editor: I read with great interest the excellent article ‘‘The Impact of Different Surgical Techniques on Outcomes in Laparoscopic Sleeve Gastrectomies’’ from Berger et al, published in the September 2016 issue of Annals of Surgery. The role of the impact of different technical details on the outcome of sleeve gastrectomies is particularly relevant, and several aspects pointed out by the authors certainly deserve the maximum attention of the international bariatric surgical community. The set of data reported in the study is impressive and very informative. However, some concerns need to be raised because they constitute excellent examples of the strength and limitations of a data registry observational retrospective study, even if large and very well conducted. One of the main conclusion of the article is related to the increased risk of a leak associated with buttressing. Certainly, the authors clarify that the literature on staple line reinforcement has conflicting results. They also acknowledge the lack of specific information about the buttressing products used in their series. This limitation is, indeed, very relevant as it has been shown that there are significant differences between them. Several systematic reviews, with or without meta-analysis, either show the absence of significant differences between reinforced and nonreinforced patients or a reduction in the leak rates when buttressing is adopted. In particular, Parikh et al’s meta-analysis included 112 studies (8922 patients) and showed a lower leak rate when buttressing was used (2.1% vs 3.2%) but the result did not reach statistical significance. They already noted a difference in outcome if bioabsorbable or nonabsorbable buttressing material was used. Gagner and Buchwald included, then, 88 articles in their quantitative analysis (8920 patients). They demonstrated that the pooled comparison of multiple types of buttress material might be misleading: compared to ‘‘no reinforcement,’’ the leak rate after the use of absorbable synthetic polymer (polyglycolic acid: trimethylene carbonate— PGA:TMC) membrane was significantly lower, while after bovine pericardium was three times higher. The complete discernment of the data presented in Berger et al’s study implies the knowledge of the mix of reinforcement materials used in the series, but unfortunately, those details were not collected. Furthermore, in the first part of the study are likely to be included patients whose sleeve gastrectomies were reinforced with a particular device that, withdrawn for the thoracic surgery indications in January 2012, was still available for abdominal surgery until its definitive recall in August 2012. Such heterogeneity is inherent in the registry-based observational retrospective study design and does not undermine the outstanding value of the report. However, it must be carefully taken into account if we want to translate its finding into our clinical practice. Another critical aspect is the absence of information about the type of stapler and cartridges used. The authors propose the hypothesis that the layers of buttress material could increase ischemia or decrease the relative staple heights. They also suggest that the 4 buttress layers, overlapped at the notches, could predispose to leaks. However, leaks occur mostly (89%) at the esophagogastric junction and not at the staple notches. Actually, the hypothesis brings up the very relevant question of the relationship between the staple height and stomach wall. Surely, the buttress thickness, even if minimal, must be taken into account when choosing the stapler cartridge, to avoid the hazard of a staple height too small to accommodate both the gastric wall and the reinforcement material. Of course, the risk to ascribe to the reinforcement the consequences of an inaccurate cartridge choice is real. The evaluation of such a bias cannot prescind from the knowledge of the devices adopted in the series but, again, this information is not included in the registry to limit the data burden of the participating centers. It also strikes that the overall leak rate reported by Berger et al on such a broad sample (0.90%, 1632 patients, according to the table, and 0.94%, 1703 patients, according to the text) is much lower than in the rest of the literature: Parikh et al’s systematic review found it to be 2.2%, Gagner and Buckwald 2.14% (it ranged from 1.1% in patients reinforced with absorbable polymer to 3,3% in those reinforced with bovine pericardium); similar results are generally reported. The same applies for bleeding. This difference is not explained by the authors; underreporting or selective reporting to the registry could account for it. Of course, this is another well-known pitfall of national data collections. A complete critical appraisal of the article would require much more space. These few remarks are meant to draw our attention to the role that this kind of studies should have in the decision-making of our clinical practices. A large observational retrospective study could represent the reality of what is done, but its conclusions may be compromised by unmeasured confounders, the absence of relevant data or their heterogeneity.