State-of-the-art curative-intent gastrectomy requires R0 resection and should be accompanied by an extended lymphadenectomy (D2 dissection) [27, 28]. Whether open or minimally invasive surgical (MIS) techniques are utilized appears to be of lesser consequence oncologically, as long as principles of complete local resection and regional dissection are adhered to [29–32]. The following operative components are based on open gastrectomy standards, but seem to be equally relevant for a MIS approach. For early GC (T1N0), endoscopic mucosal resection (EMR) or submucosal dissection (ESD) may suffice as definitive therapy [33]; both require proper specialty skills and currently appear to be limited to few centers within the United States with appropriate technical and clinical expertise. EMR and ESD techniques will not be described in further detail within this chapter. For all more advanced stages of nonmetastatic gastric adenocarcinoma, complete locoregional resection is the central component for curative-intent therapy. In the operating room, general endotracheal anesthesia is introduced, and the patient is usually placed in a supine position for a planed open celiotomy; planned laparoscopic resection may favor different positions based on the operating surgeon’s preference. It may be helpful to consider a short repeat upper gastrointestinal endoscopy after induction of anesthesia prior to resection or later for anastomotic assessment [34]; the author has used this liberally to verify tumor location and extent and to assess the mucosal appearance of gastric or esophageal components to be used in the reconstruction or for anastomotic sites. In addition, laparoscopy should be performed now unless already done in a separate setting. In up to 20 % of cases, laparoscopic confirmation of intra-abdominal metastases will still provide the opportunity to avoid an otherwise noncurative gastrectomy in this setting.

A transabdominal approach will be sufficient for most complete resections, but incision placement for open gastrectomy is not standardized and follows personal preferences. While many surgeons choose upper midline incisions, the author prefers a bilateral subcostal margin incision approach. Rarely is there benefit to a combined left thoracoabdominal incision, but for high, large gastric tumors in obese patients, this can generate much superior exposure if needed. A routine thoracoabdominal approach for GC resection is not beneficial compared to the transabdominal-only access and thus not recommended [35, 36]. With proper exposure and resectability established, the main resective objectives are R0 resection and lymphadenectomy. Total gastrectomy out of principle is not necessary; lesser extent resections, especially for distally located tumors, have shown comparable survival results, with fewer morbidity and functional challenges [37, 38]. Appropriate macro- and microscopically negative margins should be obtained as feasible at duodenal and esophageal resection sites. In challenging scenarios of advanced disease burden, it can be acceptable to leave a positive margin at these sites, as long as parameters such as serosal involvement or significant nodal burden imply a minimal curative potential. Intragastric margins of 5 cm are traditionally recommended for subtotal gastrectomy, at least for intestinal-type disease [39, 40]; diffuse-type lesions may require wider margins. A healthy tissue esophageal margin length of 2 cm seems to be sufficient for resection of Siewert type II and III lesions treated with gastrectomy [41]. The choice of gastrectomy extent (and of lymphatic dissection extent) will not only depend on location and extent of the primary tumor but also on potential reconstruction needs and options (Fig. 7.1). In general, for distal lesions a subtotal gastrectomy is adequate. For lesions in the middle third of the stomach, the decision between total or near-total gastrectomy depends on the proximal margin status and considerations for possibly safer reconstruction (gastrojejunostomy leak rates have been described as occurring half as often as those after esophagojejunostomy [42]). Proximal third lesions will essentially always require either total gastrectomy or proximal gastrectomy with a special reconstruction such as small bowel interposition [43]. Proximal gastrectomy with subsequent esophagogastrostomy is not recommended, especially after pyloroplasty, for concerns of significant reflux. Avoiding any pyloromyotomy or pyloroplasty in this setting is recommended, but does not completely preempt reflux-related problems; distal gastric emptying problems that require endoscopic or even operative management may occur in 5–15 % of cases. Lesions at the GE junction require special operative planning based on the lesions’ epicenter and, more importantly, the proximal disease extent. Siewert type I lesions require a transthoracic or transhiatal esophagectomy and should not be approached with an attempt to perform a gastrectomy [44]. Siewert type II lesions are located at the gastric cardia; these can either be approached via esophagogastrectomy with retrogastric LND analogous to type I lesions or through an extended gastrectomy as long as not more than 3 cm of distal esophageal involvement exists and proximal negative margins (of 2 cm or greater) can be obtained [45]. Siewert type III lesions are in biologic terms proximal gastric cancers, and a transabdominal approach should be fully sufficient as long as no more extensive submucosal esophageal involvement exists [44, 45]. It appears permissible to decide upon the best resection extent for proximal gastric cancers close to the GEJ intraoperatively through esophageal transection and frozen section analysis, as long as the surgeon is experienced with performing an esophagectomy in this setting and prepared to do so if necessary and as long as right gastric and gastroepiploic vasculature is initially preserved for a gastric tube reconstruction in case an esophageal resection becomes necessary.

Total or near-total omentectomy is frequently performed en bloc with a gastrectomy for cancer and represents a good way to initiate the dissection. Omental bursectomy has been widely applied as a means to accomplish more complete resection of posterior wall lesions; it includes removal of the anterior peritoneal leaf of the mesocolon in an attempt to not enter the lesser sac and completely remove this retrogastric structure. While it appears less sensible from an oncologic standpoint, especially for transmural tumors with serosal involvement and progression risks [46, 47], it nevertheless appears to be a useful technique to identify the relevant retroperitoneal plains above the pancreas for identifying lymph nodes at hepatic and splenic arteries. Careful attention is applied to not injure the pancreas parenchyma in the process. For all gastric tumors except those close to the GE junction, the proximal duodenum is freed and prepared for transection; in this process, dissection of gastroepiploic LNs off the underlying pancreas and deep ligation and transection of gastroepiploic vessels will keep the inferior parapyloric and gastroepiploic (level 6) LNs on the specimen and will allow for easy access to the duodenum. The dissection is now carried from distal to proximal, with division of lesser omentum and mobilization of paragastric tissues at the lesser curvature up to the diaphragmatic crus. If the extended LND is to be performed en bloc, common hepatic artery LNs are now mobilized and kept with the specimen. The origin of the left gastric artery should always be identified and divided for cancer resections; splenic artery nodes are dissected away from pancreas and artery, and short gastric vessels are divided close to the spleen. The spleen can most frequently be preserved unless direct tumor involvement or a large hilar LN burden requires splenectomy. Splenic hilar LN involvement is rare for tumors not located at the fundus or proximal two thirds of the greater curvature. Even when splenic hilar dissection is desired in fundus or greater curvature primaries, spleen-preserving hilar LN dissection has been applied, since spleen preservation may have important benefits for reduced postoperative morbidity [48–51]. The proximal transection is now determined based on anticipated margin needs. This is either at the level of the distal esophagus or transgastric with preservation of the proximal stomach if feasible. In the latter scenario, the lesser curvature transection should extend close to the GE junction without narrowing the esophagogastric passage, primarily to support a complete left gastric artery LND, while more length can be preserved toward the greater curvature if possible. This then shall allow for an easier reconstruction, with a subsequent anastomosis close to the greater curvature transection site. Completion of the retroperitoneal dissection with celiac lymphadenectomy and clearance of tissues to the diaphragmatic crural tissue completes the gastrectomy. For locally advanced tumors, multivisceral resections are occasionally necessary and indicated when resulting in a R0 resection that still offers curative potential. In this situation, the surgeon ought to be prepared to perform an en bloc segmental hepatectomy, diaphragmatic resection, pancreatosplenectomy, left adrenalectomy, or colectomy as required.

The propensity of gastric adenocarcinomas to involve lymph nodes (LNs) is high. Although actively debated over the past decade, lymphadenectomy at the time of curative-intent gastrectomy has shown benefits to staging accuracy and to cancer control and has thus become standard of care [28, 52]. Resection of the appropriate paragastric and of second echelon (left gastric, common hepatic, splenic, celiac artery) LNs (D2 dissection) is generally sufficient; wider dissections have not shown superior results [35]. This procedure should yield at least 15 or more LNs for the pathologic evaluation, but greater total LN counts have been associated with better survival outcomes [53–55]. A long-term survival or disease-specific control benefit to extended LN dissection (ELND) has now been demonstrated in at least two randomized controlled trials, despite a greater early morbidity and mortality in the Dutch trial after D2 dissection [28, 52, 56]. These were related to an increased rate of pancreatosplenectomy with D2 dissection [57], but this survival hazard has been superseded by a long-term overall survival benefit due to greater disease control. As discussed earlier, splenectomy and distal pancreatectomy are strongly discouraged unless deemed necessary based on tumor involvement [58, 59].

ELND can be performed en bloc with the gastrectomy as described above, or in a separate specimen. The paragastric nodes (i.e., paracardial, lesser and greater curvature, right gastric artery, and gastroepiploic artery LNs) are always best removed with the adjacent stomach portion. Since the LN group to be removed is variable based on the tumor location, a good strategy is to remove any paragastric LNs adjacent to stomach that is also to be removed. Dissection of the named artery LNs will then complete a sensible D2 dissection. If these left gastric, common hepatic, splenic, and celiac artery LNs do not appear grossly abnormal, the author has divided the left gastric artery pedicle to facilitate gastrectomy as initial step, to be followed by the retroperitoneal dissection of these structures as second step (Fig. 7.2). This allows not only for better exposure but also improved pathologic identification of relevant retrogastric LN involvement. The left gastric artery should generally be divided in cancer resections, in part for better nodal clearance; occasionally, an accessory left hepatic artery is encountered that can be preserved, as LNs can be dissected around the proximal left gastric artery, and the gastric branch can be divided after separating from the hepatic branch. In most Western patients, it is not possible to identify all LNs of interest visually during the dissection. The goal is therefore to free the relevant and named arterial vasculature of all surrounding lympho-areolar and adipose tissue, rather than obtain specific LNs or a certain total number of LNs. LN counts are determined by the pathologist and do not only reflect radicality of dissection, but also quality of the specimen pathologic examination, and other clinicopathologic factors including preoperative therapy effects and nutritional implications. A median total LN count between 20 and 30 appears to be an acceptable standard [27, 53, 54]. In some Asian centers, limiting the LND in patients with low likelihood for LN involvement is being explored, such as through laparoscopic sentinel LN biopsy for early GCs [60, 61], but these techniques are not yet accepted as proven standards.

Most gastric resections are followed by Roux-en-Y jejunal reconstruction, either as esophagojejunostomy or gastrojejunostomy (Fig. 7.3). The jejunal limb is best created with a length of around 45 cm to achieve the lowest degree of both Roux-stasis and of dumping problems postoperatively [62]. Billroth 1 and 2 reconstructions have been described after distal gastrectomy, but appear acceptable regarding appropriate oncologic dissection extent and functional outcomes only for very distally located tumors [63, 64]. A potentially challenging scenario for either reconstruction technique is that of a small proximal gastric reservoir with uncontrolled access of biliary small bowel contents and the related bile reflux risk [65, 66]. Similarly, after proximal gastrectomy, a small distal reservoir too and biliary reflux have to be avoided. A Merendino small bowel interposition between the esophagus and distal gastric reservoir and the avoidance of pyloric manipulation if possible present acceptable options, as shown in (Fig. 7.2) [43]. As a general important aspect, reconstruction preferences should not compromise the resection extent. Pouch reconstructions are rarely performed in the United States as there has been no convincing evidence of postoperative nutritional superiority; some reports describe a potential long-term quality of life benefit [67, 68].