Introduction: Leaks from intrathoracic esophagogastric anastomosis are thought to be associated with higher rates of morbidity and mortality than leaks from cervical anastomosis. We challenge this assumption and hypothesize that there is no significant difference in mortality based on the location of the esophagogastric anastomosis.
Methods: A systematic literature search was conducted using PubMed and Embase databases on all studies published from January 2000 to June 2015, comparing transthoracic (TTE) and transhiatal (THE) esophagectomies. Studies using jejunal or colonic interposition were excluded. Outcomes analyzed were leak rate, leak-associated mortality, overall 30-day mortality, and overall morbidity. Meta-analyses were performed using Mantel-Haenszel statistical analyses on studies reporting leak rates of both approaches. Nominal data are presented as frequency and interquartile range (IQR); measures of the association between treatments and outcomes are presented as odds ratio (OR) with 95 % confidence interval.
Results: Twenty-one studies (3 randomized controlled trials) were analyzed comprising of 7167 patients (54 % TTE). TTE approach yields a lower anastomotic leak rate (9.8 %; IQR 6.0-12.2 %) than THE (12 %; IQR 11.6-22.1 %; OR 0.56 [0.34-0.92]), without any significant difference in leak associated mortality (7.1 % TTE vs. 4.6 % THE: OR 1.83 [0.39-8.52]). There was no difference in overall 30-day mortality (3.9 % TTE vs. 4.3 % THE; OR 0.86 [0.66-1.13]) and morbidity (59.0 % TTE vs. 66.6 % THE; OR 0.76 [0.37-1.59]).
Discussion: Based on meta-analysis, TTE is associated with a lower leak rate and does not result in higher morbidity or mortality than THE. The previously assumed higher rate of transthoracic anastomotic leak-associated mortality is overstated, thus supporting surgeon discretion and other factors to influence the choice of thoracic versus cervical anastomosis.