Surgical management of invasive gastric cancer

INTRODUCTION — Although gastric cancer remains a lethal disease, the prognosis has improved modestly over the last two decades due to advances in surgical treatment, postoperative care, and multimodality therapy. The high mortality rate reflects the prevalence of advanced disease at presentation and relatively aggressive biology. Early lesions are usually asymptomatic and infrequently detected outside the realm of a screening program. (See "Clinical features, diagnosis, and staging of gastric cancer".)

An additional contributing factor to the persistently high mortality rate is the change in the distribution of cancers from the body and antrum to the proximal stomach during the past 20 years. Cancers involving the proximal stomach and esophagogastric junction (EGJ) have increased steadily at a rate exceeding that of any other cancer except melanoma and lung cancer. Aside from a correlation with increasing obesity, the reasons for this are unclear. Proximal lesions are biologically more aggressive and have a worse prognosis, stage for stage, than do distal gastric cancers. This suggests that their pathogenesis differs from cancers arising in other parts of the stomach [1]. (See "Epidemiology of gastric cancer", section on 'Incidence' and "Epidemiology of gastric cancer", section on 'Changes in histologic pattern' and "Gastric cancer: Pathology and molecular pathogenesis" and "Epidemiology and pathobiology of esophageal cancer", section on 'Epidemiology' and "Epidemiology and pathobiology of esophageal cancer".)

The classification and management of cancers arising at the EGJ have evolved over time. In the latest, eighth edition of the tumor, node, metastasis (TNM) staging manual [2], tumors arising at the EGJ with the tumor epicenter no more than 2 cm into the proximal stomach are staged as esophageal cancers (the so-called Siewert I/II EGJ tumors [3]). However, by contrast, EGJ tumors with their epicenter located more than 2 cm into the proximal stomach are staged as stomach cancers, as are all cardia cancers not involving the EGJ. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer", section on 'TNM staging criteria' and "Clinical features, diagnosis, and staging of gastric cancer", section on 'TNM staging criteria' and "Multimodality approaches to potentially resectable esophagogastric junction and gastric cardia adenocarcinomas", section on 'Siewert classification'.)

This topic review will focus on issues related to surgery for cancers arising in the stomach. Surgery may be required as a component of the staging evaluation, for potentially curative treatment of localized disease, or for palliation in cases of advanced disease. Adjuvant and neoadjuvant therapy for noncardia gastric cancers, multimodality approaches for EGJ cancers, and the management of advanced disease are discussed in detail elsewhere. (See "Adjuvant and neoadjuvant treatment of gastric cancer" and "Multimodality approaches to potentially resectable esophagogastric junction and gastric cardia adenocarcinomas" and "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus" and "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer" and "Local palliation for advanced gastric cancer".)

EARLY DIAGNOSIS OF GASTRIC CANCER — Resection offers the best chance for long-term survival for patients with localized gastric cancer, particularly in combination with adjuvant or perioperative chemotherapy or chemoradiotherapy [4,5]. A major problem is the identification of patients at a time when they are potentially curable. In the United States, two-thirds of patients present with stage III or IV disease, while only 10 percent have stage I disease (table 1) [6].

Mass screening programs have been effective at detecting early gastric cancer in high incidence areas such as Japan where over one-half of patients have early tumors [7]. The relatively low incidence of gastric cancer in many other regions including the United States makes this strategy prohibitively costly and unwarranted. (See 'Adjuvant and neoadjuvant therapy' below and "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging".)

Nevertheless, some individuals at high risk for gastric cancer may be appropriate candidates for screening. Specific risk factors for cancer of the stomach or gastroesophageal junction include a genetic predisposition (eg, hereditary nonpolyposis colorectal cancer, familial polyposis coli, and hereditary diffuse gastric cancer), dietary factors, acid hyposecretory conditions, infection with Helicobacter pylori, and Barrett's esophagus. (See "Gastric cancer screening" and "Risk factors for gastric cancer".)

STAGING EVALUATION — Optimal therapy depends on accurate staging of the extent of disease. The preoperative evaluation permits the assessment of clinical stage, which is important in guiding decisions about preoperative or perioperative therapy. Pathologic staging depends on the findings at subsequent surgical exploration and examination of the pathologic specimen, facilitating consideration of adjuvant therapy and assessment of response to preoperative therapy.

Complete clinical or preoperative staging of patients with gastric cancer includes the following:

●Physical examination, including evaluation of appropriate nodal areas (especially left supraclavicular nodes) as well as the abdomen.

●Computed tomography (CT) scans of abdomen and pelvis, and for proximal lesions, CT scan of the chest. Although CT is not very accurate for assessing the depth of tumor invasion of the stomach wall or regional nodal involvement, it may detect distant nodal or visceral metastases, ascites, or carcinomatosis. Preoperative CT scans often underestimate the extent of disease, principally because of radiographically undetectable metastases involving the liver and peritoneum [8]. (See "Clinical features, diagnosis, and staging of gastric cancer", section on 'Computed tomography scan in all patients'.)

●Endoscopic ultrasound (EUS) may provide more accurate staging evaluation of the tumor (T) and nodal (N) stage than CT and also allows for preoperative biopsies. Its principal roles are in the evaluation of a patient being considered for neoadjuvant therapy and to identify tumors that may be amenable to endoscopic mucosal resection. If it has not been performed recently, patients being considered for surgical resection should undergo full esophagogastroduodenoscopy prior to the procedure; we find this to be optimally performed at the time of another endoscopic procedure, such as EUS. (See "Early gastric cancer: Epidemiology, clinical manifestations, diagnosis, and staging" and "Overview of endoscopic resection of gastrointestinal tumors" and "Clinical features, diagnosis, and staging of gastric cancer".)

Staging laparoscopy — Staging laparoscopy, while more invasive than CT or EUS, has the advantage of directly visualizing the liver surface, peritoneum, and local lymph nodes, and permitting biopsy of any suspicious lesions. Laparoscopy may identify radiographically occult metastatic disease, thus avoiding unnecessary exploratory laparotomy. In addition, peritoneal cytology should be obtained during staging laparoscopy. (See "Clinical features, diagnosis, and staging of gastric cancer", section on 'Staging laparoscopy'.)

The selection of patients who need staging laparoscopy is controversial. Some experts suggest restricting staging laparoscopy for patients with EUS stage T3/4 disease [9]. Guidelines from the National Comprehensive Cancer Network suggest only that laparoscopy be "considered" for patients who appear to have locoregional disease (other than stage IV, Tis or T1a stage) after conventional radiographic and EUS staging [10]. However, we believe it can sometimes be difficult to differentiate T2 and T3 lesions on EUS; our practice is to use preoperative staging laparoscopy for any medically fit patient who appears to have more than a T1 lesion on EUS, no histologic confirmation of stage IV disease, and who would not otherwise require a palliative gastrectomy because of symptoms. (See "Clinical features, diagnosis, and staging of gastric cancer", section on 'Staging laparoscopy'.)

Diagnostic laparoscopy should also be undertaken in patients who are being considered for neoadjuvant therapy trials (eg, those with positive peritoneal cytology).

Significance of positive peritoneal cytology — Positive peritoneal cytology in the absence of metastatic disease elsewhere identifies a group of patients who are at particularly high risk for peritoneal failure [11,12]. Survival is dismal when gastrectomy is the first line of therapy [13-16].

However, opinion differs as to the optimal way to manage such patients, in particular the role of gastrectomy in such patients. At some institutions, such patients are offered a clinical trial of peritoneum-directed therapy. However, those who decline entry into the trials may be referred for four to six months of upfront chemotherapy, then reassessed. If there is no evidence of extragastric progression, they then receive five weeks of chemoradiotherapy, followed by a five-week rest period and a formal restaging evaluation (including a repeat laparoscopy with peritoneal washings). Patients with no evidence of metastatic disease who convert to negative cytology are considered eligible for potentially curative surgery since a small number can achieve long-term disease-specific survival [17-20]. (See 'Neoadjuvant chemotherapy and chemoradiotherapy' below and "Adjuvant and neoadjuvant treatment of gastric cancer".)

This is a controversial area, however, and others disagree with this approach. There are no randomized trials proving the benefit of neoadjuvant therapy in this setting. At some (predominantly but not exclusively Asian) institutions, patients with positive peritoneal cytology are considered suitable candidates for peritoneum-directed therapy [21-26], and at others, these patients are offered only palliative systemic chemotherapy. One phase II trial examining the role of laparoscopic hyperthermic intraperitoneal chemotherapy in conjunction with neoadjuvant systemic chemotherapy prior to consideration of gastrectomy in 20 patients with either isolated positive peritoneal cytology or peritoneal carcinomatosis without other distant metastases, reported a three-year overall survival rate of 28 percent [26]. There were four patients still alive without disease recurrence from 32 to 49 months after the diagnosis of metastatic disease.

Additional clinical trials are needed to define the best treatment option for this highly select group of patients.

Staging system — The tumor, node, metastasis (TNM) staging system of the combined American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) is most frequently used; the current version (eighth edition, 2017) is outlined in the table (table 1) [2]. (See "Clinical features, diagnosis, and staging of gastric cancer", section on 'Staging systems'.)

One of the prominent changes compared with the earlier 2010 TNM classification is the redefining of the boundary between esophageal and gastric cancer. In the 2017 revision, tumors involving the esophagogastric junction (EGJ) with the tumor epicenter no more than 2 cm into the proximal stomach are staged as esophageal cancers (table 2) [2]. EGJ tumors with their epicenter located more than 2 cm into the proximal stomach are staged as stomach cancers, even if they involve the EGJ, as are all cardia cancers not involving the EGJ.

Nodal (N) status — The presence and number of involved lymph nodes is one of the most important prognostic indicators. However, the inadequacy of lymphadenectomy in contemporary North American series of gastric cancer resection is notable, even among patients treated predominantly at academic institutions [4,27-29]. (See 'Extent of lymph node dissection' below and 'Adjuvant chemotherapy and chemoradiotherapy' below.)

The optimal extent of regional node dissection during gastrectomy for gastric cancer is intensely debated. At the core of the debate is the relationship between prognosis and the number of nodes in the surgical specimen. (See 'Extent of lymph node dissection' below.)

The number of examined nodes influences the accuracy of staging, and influences survival [30-32]. The relationship between number of nodes and outcomes was illustrated in a series of 1038 patients undergoing potentially curative resection for gastric cancer [31]. The location of positive nodes did not significantly affect median survival, while the number of positive lymph nodes had a profound influence on survival provided that at least 15 lymph nodes were available for analysis. The AJCC staging guidelines recommend that at least 16 regional nodes be assessed pathologically but that removal/evaluation of more nodes (30 or more) is desirable [2]. The same guidance applies to patients treated with neoadjuvant therapy [33].

Regional nodes for tumors involving different parts of the stomach are depicted in the figure (figure 1). Involvement of other intraabdominal nodal groups (ie, pancreatoduodenal, retropancreatic, peripancreatic, superior mesenteric, middle colic, paraaortic, and retroperitoneal) is classified as distant metastases [2]. Such results can only be achieved with a solid, common-purpose relationship between the surgeons and pathologists.

SURGICAL TREATMENT FOR LOCALIZED DISEASE — Complete surgical removal of a gastric cancer with resection of adjacent lymph nodes represents the best chance for long-term survival. Abdominal exploration (preferably by laparoscopy, either as a separate procedure or at the onset of planned resection, as recovery from laparoscopy is much easier for patients than from laparotomy) with curative intent should be undertaken unless there is unequivocal evidence of disseminated disease, there is major vascular invasion, a neoadjuvant approach is considered, or there are medical contraindications to surgery.

Indicators of unresectability — The only widely accepted criteria of unresectability for gastric cancer are the presence of distant metastases, invasion of a major vascular structure, such as the aorta, or disease encasement or occlusion of the hepatic artery or celiac axis/proximal splenic artery. Distal splenic artery involvement is not an indicator of unresectability; the vessel can be resected en bloc with a left upper quadrant exenteration: stomach, spleen, and distal pancreas.

The lymphatics around the stomach are rich, and the presence of locoregional lymph node metastases that are located geographically distant from the tumor (eg, celiac nodes with a primary tumor on the greater curvature of the stomach) should not necessarily be considered an indicator of unresectability (see 'Nodal (N) status' above). However:

●Patients who have bulky adenopathy fixed to the pancreatic head that might indicate the need for a Whipple procedure are at a high risk for occult metastatic disease. In these cases, staging laparoscopy and/or upfront chemotherapy should be strongly considered rather than surgery initially. Performance of a Whipple for gastric cancer is an extremely rare occurrence. (See 'Staging laparoscopy' above.)

●Lymph nodes behind or inferior to the pancreas, aorto-caval region, into the mediastinum, or in the porta hepatis are typically considered outside of the surgical field and thus evidence of unresectability. These nodes would fall into areas that would be defined as third or fourth echelon nodes in the Japanese nomenclature.

Since resection of the primary lesion also offers the most effective means of symptom palliation, exploration also is occasionally considered in patients with known metastatic disease if the severity of symptoms so dictates. The choice of operation for gastric cancer depends on the location of the tumor within the stomach, the clinical stage, and the histologic type. The major surgical considerations include the extent of stomach resection (with total gastrectomy employed far less frequently than subtotal gastrectomy), and a more limited lymph node dissection is typically employed. (See "Total gastrectomy and gastrointestinal reconstruction" and "Partial gastrectomy and gastrointestinal reconstruction".)

Linitis plastica — In approximately 5 percent of primary gastric cancers, a broad region of the gastric wall or even the entire stomach is extensively infiltrated by malignancy, resulting in a rigid thickened stomach, termed linitis plastica. The prevalence may be higher in younger individuals [6,34]. Although most commonly due to poorly differentiated (diffuse-type) infiltrating gastric cancers, this pattern can, in rare circumstances, represent metastatic spread from lobular cancer of the breast. (See "Gastric cancer: Pathology and molecular pathogenesis" and "Pathology of breast cancer".)

Linitis plastica has an extremely poor prognosis [35-39], and this is attributed to the potential for early spread and advanced stage at diagnosis, as well as the frequent presence of microscopic disease at the surgical margins because of its diffuse nature. Fifty to 60 percent of all patients have metastatic disease (mainly within the peritoneal cavity) at diagnosis [39,40]. Nodal involvement is frequent, and extensive surgery may be required for complete excision [37,41]. Many surgeons consider the presence of linitis plastica to be a contraindication to potentially curative resection [37]. However, others report that long-term survival in selected patients who undergo optimal resection (negative margins guided by intraoperative frozen section analysis and a D2/D3 lymphadenectomy) is comparable with that of optimally resected patients without linitis plastica [42]. In our view, all of these patients should undergo some form of preoperative therapy and diagnostic laparoscopy prior to resection to identify the subset of patients who might benefit from resection. Even with this approach, long-term survival in these patients is uncommon. (See "Adjuvant and neoadjuvant treatment of gastric cancer" and "Clinical features, diagnosis, and staging of gastric cancer", section on 'Staging laparoscopy' and "Diagnostic staging laparoscopy for digestive system cancers".)

Total versus partial gastrectomy — Gastrectomy is the most widely used approach for therapy of invasive gastric cancer, although superficial cancers can sometimes be treated endoscopically. Total gastrectomy (figure 2), which removes the entire stomach, is usually performed for lesions in the proximal (upper third) of the stomach, while partial gastrectomy (distal gastrectomy, subtotal gastrectomy) (figure 3) with resection of adjacent lymph nodes appears to be sufficient for lesions in the distal (lower two-thirds) of the stomach. Patients with large midgastric lesions or infiltrative disease (eg, linitis plastica) may require total gastrectomy. In most series, quality of life after partial gastrectomy is superior to that after a total gastrectomy, at least in the short term [43,44]. Data on quality of life in long-term survivors of total gastrectomy, including those with a jejunal reservoir reconstruction, are limited [45,46]. (See "Total gastrectomy and gastrointestinal reconstruction" and "Partial gastrectomy and gastrointestinal reconstruction".)

Distal tumors — At least two significant trials show no added survival benefit for total compared with partial gastrectomy for patients with distal tumors [47,48].

In one of these reports, 169 patients undergoing potentially curative resection for antral adenocarcinoma were randomly assigned to total or partial gastrectomy [48]. The overall complication and perioperative mortality rates were 32 and 1.3 percent, respectively, with total gastrectomy, and 34 and 3.2 percent with partial gastrectomy, respectively. Five-year survival was similar in both groups. Similar results were noted in a trial of the Italian Gastrointestinal Tumor Study Group, in which the five-year survival rates were 65 and 62 percent in the partial and total gastrectomy groups, respectively [47].

Proximal and esophagogastric junction tumors — The issues guiding the extent of surgical resection are more complex for tumors of the proximal stomach. Tumors of the proximal stomach that do not invade the esophagogastric junction (EGJ) can technically be approached with either total gastrectomy or proximal subtotal gastrectomy. Currently, total gastrectomy remains the preferred treatment for proximal gastric cancer, while proximal gastrectomy may be suitable for select patients and offer nutritional benefits [49]. Total gastrectomy is preferred by most surgeons for the following two reasons:

●The Roux-en-Y reconstruction performed during total gastrectomy is associated with an extremely low incidence of reflux esophagitis (2 percent). In comparison, approximately one-third of patients develop reflux esophagitis after a proximal subtotal gastrectomy [50].

●Proximal subtotal gastrectomy may leave behind lymph nodes along the lesser curvature of the stomach, which is the most common site of nodal metastases.

The best evidence comparing the two approaches came from a meta-analysis of two randomized trials and nine retrospective studies of patients with proximal gastric cancer [51]. Compared with total gastrectomy, proximal subtotal gastrectomy resulted in similar five-year survival (61 versus 64 percent) but more cancer recurrences (39 versus 24 percent). There were also more complications associated with a proximal subtotal gastrectomy, including anastomotic stenosis (27 versus 7 percent) and reflux esophagitis (20 versus 2 percent).

There is increasing interest in the so called "double track" reconstruction associated with proximal gastrectomy. This approach seeks to maintain the distal stomach while eliminating reflux by essentially performing a gastrojejunostomy to the Roux limb between the esophagojejunostomy and the jejunojejunostomy (figure 4) [52]. The KLASS-05 study which randomized patients between proximal gastrectomy with double track reconstruction and total gastrectomy with Roux-en-Y reconstruction reported similar short-term outcomes between the two approaches, including operative time, blood loss, morbidities, and gastrointestinal symptoms at two weeks postoperatively [53]. Longer term functional and oncologic outcomes are not yet known.

Surgical management of tumors of the EGJ is discussed in detail elsewhere. (See "Surgical management of resectable esophageal and esophagogastric junction cancers", section on 'Esophagogastric junction cancer resection'.)

Extent of lymph node dissection — Given the apparent impact of D2 lymphadenectomy on disease-specific survival, most major cancer centers are performing a D2, as compared with a D1, dissection. Treatment guidelines published by the National Comprehensive Cancer Network (NCCN) [10] recommend that gastric cancer resection include the regional lymphatics, including perigastric (D1) nodes as well as those along the left gastric artery, common hepatic artery, celiac artery, splenic hilum, and splenic artery (D2 lymph nodes), with the goal of examining 15 or more lymph nodes. (See 'Nodal (N) status' above.)

As long as it is carried out in specialized, high-volume centers with appropriate surgical expertise and postoperative care, a pancreas- and spleen-preserving D2 lymphadenectomy provides superior staging information and may provide a survival benefit while avoiding excess morbidity. By contrast, there is no evidence that a D3 (paraaortic) lymphadenectomy confers a survival benefit over a D2 dissection, and it is associated with greater perioperative mortality. We recommend that a D3 dissection not be considered for surgical treatment of gastric cancer. Furthermore, routine splenectomy or pancreatectomy during gastric resection should be discouraged as it increases morbidity and mortality without improving survival. It should only be considered if there is evidence of direct tumor extension to the pancreas or spleen.

The optimal extent of lymph node dissection, however, remains an area of significant discussion in surgical circles. Japanese and Korean surgeons routinely perform extended lymphadenectomy, a practice that some suggest at least partially accounts for the better survival rates in Asian as compared with Western series [54]. The term "extended lymphadenectomy" variably refers to either a D2 or a D3 lymph node dissection. However, even these surgeons are increasingly moving away from the most aggressive dissections. (See 'Prognosis' below.)

The draining lymph node basins for the stomach have been meticulously divided into 16 stations by Japanese surgeons (table 3): stations 1 to 6 are perigastric, and the remaining 10 are located adjacent to major vessels, behind the pancreas, and along the aorta (figure 5) [55].

●D1 lymphadenectomy refers to a limited dissection of only the perigastric lymph nodes (stations 1 to 7). In the Japanese literature, a D1+ lymphadenectomy refers to a D1 lymphadenectomy plus stages 8a, 9, and 11p [56].

●D2 lymphadenectomy is an extended lymph node dissection, entailing removal of nodes along the hepatic, left gastric, celiac, and splenic arteries, as well as those in the splenic hilum (stations 1 to 12a) [56].

●D3 dissection is a superextended lymphadenectomy. The term has been used by some to describe a D2 lymphadenectomy plus the removal of nodes within the porta hepatis and periaortic regions (stations 1 to 16), while others use the term to denote a D2 lymphadenectomy plus periaortic nodal dissection (PAND) alone [57]. Most Western surgeons (and the American Joint Committee on Cancer [AJCC]/Union for International Cancer Control [UICC] tumor, node, metastasis [TNM] staging classification [58]) classify disease in these regions as distant metastasis and do not recommend removing nodes in these areas during a potentially curative gastrectomy.

The arguments in favor of extended lymphadenectomy (ie, D2 or D3 versus D1) are that removing a larger number of nodes more accurately stages disease extent and that failure to remove these nodes leaves behind disease (which would be a potentially fatal event) in as many as one-third of patients [59-61]. A consequence of more accurate staging is to minimize stage migration (the "Okie phenomenon," as described by Will Rogers) [61,62]. The resulting improvement in stage-specific survival may explain, in part, the better results in Asian patients.

The influence of total lymph node count on stage-specific survival was studied in a series of 3814 patients undergoing gastrectomy for T1-3N0-1 (classified according to the 1997 AJCC gastric cancer staging system and reported to the Surveillance, Epidemiology, and End Results [SEER] database between 1973 and 2000) [63]. For every stage subgroup (T1/2N0, T1/2N1, T3N0, T3N1), survival was significantly better as more nodes were examined. Although cut point analysis revealed the greatest survival difference when 10 lymph nodes were examined, there were significant survival differences for cut points up to 40 nodes examined, always in favor of a greater number of nodes in the specimen.

There are two main arguments against the routine use of an extended lymphadenectomy: the higher associated morbidity and mortality (particularly if splenectomy is performed in order to achieve extended lymphadenectomy) and the lack of a survival benefit for extended lymphadenectomy in most large randomized trials.

Randomized trials and meta-analyses — Multiple prospective randomized trials both in Asian and Western populations have failed to show an overall survival benefit with D2 versus D1 lymphadenectomy [64-71] or with D3 compared with D2 lymphadenectomy [57,72-75]. Two [64,69] of three separate meta-analyses comparing D1 with D2 node dissection have concluded that there is no survival benefit to D2 lymphadenectomy, although the most recent and inclusive analysis of five randomized controlled trials did find a modest advantage for D2 lymphadenectomy in terms of disease-specific survival, which was counterbalanced by a twofold higher rate of postoperative mortality [76]. Both of the meta-analyses studying the benefit of D2 versus more extended lymphadenectomy have concluded that there is no significant survival benefit from a more extended lymphadenectomy than D2 [64,76].

D1 versus D2 dissection — As noted above, most of the multiple randomized trials have failed to demonstrate a survival benefit for D2 versus D1 nodal dissection in patients with invasive gastric cancer. However, many clinicians consider that the two largest trials (the Dutch and the Medical Research Council [MRC] trials [66,67,77]) were flawed and underpowered to show a survival advantage to D2 lymphadenectomy:

●MRC trial – The MRC randomly assigned 400 patients undergoing potentially curative resection to a D1 or a D2 lymphadenectomy [66]. Postoperative morbidity was significantly greater in the D2 group (46 versus 28 percent), as was operative mortality (13 versus 6 percent). The excess morbidity and mortality were clearly associated with the liberal use of splenectomy and distal pancreatectomy to achieve complete node dissection. With long-term follow-up, survival rates were comparable despite the doubling of operative mortality (five-year survival rates 33 versus 35 percent) [78].

●Dutch trial – The largest randomized trial came from the Dutch Gastric Cancer Group and compared D1 with D2 lymphadenectomy in 711 patients who were treated with curative intent [67,77]. This trial relied heavily upon input from a Japanese surgeon, who trained 11 Dutch surgeons in the technique of radical lymph node dissection and monitored the operative procedures. Despite these efforts to maintain quality control of the surgical procedures, both under removal and over removal of required nodal stations occurred, somewhat blurring the distinction between the groups.

Both postoperative morbidity (43 versus 25 percent) and mortality (10 versus 4 percent) were higher in the D2 group, and a statistically significant survival advantage in the radical dissection group was not observed, either in the initial report [67] or with longer follow-up [70,77], despite a significantly lower risk of recurrence. This was attributed to the detrimental impact of increased operative mortality in this group [79].

The design of the Dutch trial was based on the assumption that radical lymph node dissection would increase the survival rate from 20 to 32 percent, likely an overestimation of benefit. Furthermore, 40 percent of enrolled patients had early gastric cancer, an unexpectedly high proportion that was not anticipated when the trial was designed.

The Dutch trial has been updated with 15-year follow-up [70]. The survival curves have continued to separate, although the difference in overall survival is still not statistically significant (22 versus 28 percent in the D1 and D2 arms, respectively, p = 0.34). However, the gastric cancer-related death rate is significantly higher in the D1 arm (48 versus 37 percent), while death rates due to other causes were not different. This supports the concept that if the D2 dissection can be done with low operative mortality, similar to that of a D1 dissection (as occurs in high volume centers), survival will be positively impacted.

●Cochrane analysis –The most recent meta-analysis included five randomized trials of D1 versus D2 dissection [66,68,70,71,80]. While there was no significant advantage for D2 lymphadenectomy for overall survival (n = 5 trials, hazard ratio [HR] 0.91, 95% CI 0.71-1.17) or disease-free survival (DFS; n = 3 trials, HR 0.95, 95% CI 0.84-1.07), there was a significant difference in disease-specific survival, favoring D2 lymphadenectomy (HR 0.81, 95% CI 0.71-0.92). The quality of the evidence was judged as moderate. However, D2 lymphadenectomy was also associated with a high postoperative mortality rate (relative risk [RR] 2.02, 95% CI 1.34-3.04).

In our view and that of others, including the NCCN [10] and European Society for Medical Oncology (ESMO) [81], based on the most recent analysis of the Dutch trial and the 2015 Cochrane review, D2 dissection is preferred in patients with potentially curable gastric cancer when it can be done without increasing operative mortality.

D2 versus D3 dissection

●JCOG trial 9501 – The multicenter Japan Clinical Oncology Group (JCOG) study 9501 randomly assigned 523 patients to D2 versus D3 (D2 + PAND) dissection. The overall perioperative complication rate in the D3 group was significantly higher (28.1 versus 20.9 percent), although there were no differences in major complications (anastomotic leak, pancreatic fistula, abdominal abscess, pneumonia) and perioperative mortality was very low (0.8 percent) in both groups [72]. Five-year recurrence-free (approximately 63 percent in both groups) and overall survival (70 versus 69 percent) were no better after extended lymphadenectomy [57].

One of the confounding issues with the JCOG trial is that in subgroup analysis, patients with node-negative disease fared significantly better with the more aggressive D3 operation than with D2 lymphadenectomy. Conversely, patients who were node-positive fared significantly better with a D2 lymphadenectomy than with more aggressive D3 surgery [57]. The reasons for these counterintuitive results are unclear.

Nevertheless, the high survival rate in both groups is notable in view of the fact that over 60 percent of both groups had positive nodes. These data underscore the marked differences in outcome between gastric cancers arising in Western and Asian populations. (See 'Prognosis' below.)

Data from the JCOG trial as well as those from other groups [72,82] suggest that a D2 dissection can be performed safely with a perioperative mortality rate that is under 2 percent.

●Meta-analysis – The 2015 Cochrane analysis of the JCOG trial and two other smaller randomized trials of D2 versus D3 (with PAND) dissection [74,75] concluded that resection of the paraaortic nodes did not provide any significant survival benefit [76]. Furthermore, while there was no significant difference in perioperative mortality with extended lymphadenectomy, the 95 percent confidence intervals were very wide (RR 1.67, 95% CI 0.41-6.73) and included the possibility of a nearly sevenfold increase in the risk of perioperative mortality. Thus, paraaortic lymphadenectomy should not be considered a routine practice for surgical treatment of gastric cancer.

The importance of surgeon and institutional expertise — The first United States study to assess outcome using Japanese lymphadenectomy criteria was performed using patients in a randomized trial of adjuvant therapy and provided a sobering view of current surgical practice in patients with resectable gastric cancer [83]. In this randomized trial (Intergroup trial 0116) examining the utility of adjuvant chemoradiotherapy, in 556 patients with potentially resectable gastric cancer, 54 percent underwent less than a D1 lymphadenectomy, while D1 or ≥D2 procedures were performed in 36 and 10 percent, respectively. Others have shown that fewer than one-third of American patients undergoing gastric cancer surgery had 15 or more lymph nodes removed/examined, even in academic and high-volume hospitals [84-86]. (See "Adjuvant and neoadjuvant treatment of gastric cancer".)

The importance of extent of lymphadenectomy on outcome was demonstrated when patients were stratified according to the Murayama Index [MI], a computer-based model that predicts the likelihood of disease in the regional nodal stations left undissected by the surgeon [87,88]. Median survival had not been reached in patients with a low MI, whereas it was only 27 months for cases with MI ≥5.

In view of these data, we believe that D2 nodal dissection is indicated, but it should only be performed in selected centers where surgeons have demonstrated acceptably low operative morbidity and mortality rates. Supporting this view is a systematic review and meta-analysis that identified 28 articles reporting the relationship between hospital and surgeon factors with procedure-related morbidity, procedure-related mortality, and five-year survival for gastric cancer surgery [89]. Higher hospital volume was associated with lower, unadjusted procedure-related mortality. Surgeon volume, level of training, hospital volume, and specialization were also associated with procedure-related morbidity, procedure-related mortality, and five-year survival, but the effect was not consistent across all studies.

Our own experience, as well as that at Memorial Sloan Kettering and in Japan, suggests that mortality rates under 2 percent should be expected at centers with higher patient volume [72,82]. Unfortunately, surgical expertise with more extensive nodal dissection is limited (though increasing with more specialty fellowship training) in the United States, and data from the above meta-analysis and the American College of Surgeons suggest that procedure-related mortality is significantly higher in American series, ranging from 5.5 to 13 percent [6,89].

An examination of mortality rates for gastrectomy in the United States has demonstrated that among patients with gastric tumors who presented to more than 700 hospitals between 1982 and 1987, the average perioperative mortality rate was 7.2 percent. Not surprisingly, variability in perioperative mortality appears dependent on the volume of gastrectomies at individual institutions:

●In one series, perioperative mortality rates ranged from 8.7 to 13 percent at very high volume (over 21 procedures per year) and very low volume hospitals (fewer than 5 procedures per year), respectively [90]. The adjusted odds ratio for death at the highest compared with the lowest volume institutions was 0.72 (95% CI 0.63-0.83).

●In a study conducted in the state of Texas over a three-year period (1999 to 2001) that examined gastrectomies performed at all nonfederal hospitals, over 1800 resections were performed in 214 hospitals, with high-volume centers performing more than 15 resections per year and low-volume centers performing fewer than 3 [91]. Hospital mortality ranged from 5.2 to 6.2 percent in the low and intermediate hospitals, respectively, and it was 1.1 percent at the high-volume centers. Even with a relatively low cutoff of only 15 resections per year for high-volume centers, there were only two high-volume centers in the state, underscoring the need for regionalization of this specialized surgery. The impact of hospital volume on long-term DFS or overall survival is less clear [92].

●In a follow-up study of the Texas experience, hospital demographics appear to be changing over time for the better. In an analysis of changes to these patterns of high-, intermediate-, and low-volume centers, using data from 2010 to 2015 investigators found fewer overall hospitals performing gastric resections (from 214 down to 193); while there was an increase in both the number of high-volume centers (5 versus 2) and low-volume centers (142 versus 134) over this time period, there was a substantial decrease in the number of intermediate-volume centers (46 versus 78) [93]. There was a more than doubling of the percentage of patients who were treated at high-volume centers (28.7 versus 12.4 percent), while the percentage treated at low-volume centers remained stable. The lower mortality rate at high-volume centers compared with intermediate- and low-volume centers persisted, but all groups saw a decrease in both operative mortality and perioperative complications.

The biggest differences in mortality between the high-volume centers and others were in the patients with significant comorbidities. Among patients with no comorbidities, the mortality rates were 2.38 versus 2.31 and 0.91 percent for low-, intermediate-, and high-volume centers, respectively, and 5.60 versus 5.17 and 2.61 percent for those patients with comorbidities at the same centers, respectively. Undoubtedly these improvements are multifactorial, and include better patient selection, intraoperative management, and perioperative management.

Open versus laparoscopic resection — Open gastrectomy remains the preferred surgical treatment for gastric cancer worldwide. In high-volume, experienced centers, however, laparoscopic gastric resection (and more recently robotic gastrectomy though data with this are more limited) provides an alternative that offers patients a faster recovery and fewer complications while recovering a similar number of lymph nodes compared with open surgery [94-101]. (See "Laparoscopic gastrectomy for cancer".)

In a 2016 Cochrane review of 13 randomized trials, laparoscopic and open gastrectomy resulted in similar short- and long-term outcomes, including mortality rate, recurrence rate, positive margin rate, complication rate, length of hospital stay, and requirement of blood transfusion, although the quality of the evidence was rated low or very low [102].

Long-term oncologic data from several prospective randomized trials are not expected for another several years and are required before laparoscopic surgery can be considered oncologically equivalent to open surgery in treating invasive gastric cancer. (See "Laparoscopic gastrectomy for cancer", section on 'Eastern experience'.)

Techniques of open and laparoscopic gastrectomy are discussed in separate topic reviews. (See "Partial gastrectomy and gastrointestinal reconstruction" and "Total gastrectomy and gastrointestinal reconstruction".)

Sentinel lymph node biopsy for early gastric cancer — For patients with early stage gastric cancer, the risk of lymph node metastasis is low (2 to 18 percent for T1, 20 percent for T2). Sentinel lymph node (SLN) mapping has been proposed in this setting to identify those patients undergoing surgical rather than endoscopic management who need more extensive surgery [103,104]. At least in theory, patients with positive SLNs should undergo partial or total gastrectomy with requisite D2 lymphadenectomy while those with a negative SLN biopsy are recommended for a more limited gastric (eg, wedge or segmental) resection with a D1 or no formal lymphadenectomy [105]. However, the available data on the accuracy of SLN mapping for early gastric cancer are conflicting, and in our view, the use of SLN mapping for early gastric cancer remains investigational and should be limited to clinical trials. (See "Early gastric cancer: Treatment, natural history, and prognosis", section on 'Gastrectomy'.)

In selected hospitals in Asia (primarily Japan), SLN biopsy is offered to patients with small (<4 cm) clinical-stage T1 or 2, N0 (cT1/2N0) gastric cancer. However, questions remain as to the accuracy of determination of lymph node status, and whether SLN detection rates are acceptable. Conflicting data have been reported by two prospective trials comparing SLN biopsy against standard lymphadenectomy:

●In the SENORITA trial, 258 patients with early gastric cancer (cT1, N0, M0; tumor <3 cm) underwent laparoscopic sentinel node navigation surgery (LSNNS), which consists of laparoscopic sentinel basin dissection followed by stomach-preserving surgery if intraoperative frozen section of the sentinel nodes revealed no metastasis [106]. Compared with standard laparoscopic gastrectomy with lymphadenectomy (269 patients), LSNNS resulted in similar rates and severity of 30-day morbidity. Oncologic outcomes are not yet available.

●In another Korean trial, 580 patients with stage IA (table 1) gastric cancer (all tumor ≤3 cm) underwent LSNNS or standard laparoscopic gastrectomy [107]. Stomach-preserving surgery (endoscopic resection, wedge resection, or segmental resection) was performed in 81 percent of the patients in the LSNNS group, all of whom had negative SLN biopsy. At three years, the disease-free survival (DFS) rates in the standard and LSNNS groups were 95.5 and 91.8 percent, respectively (difference 3.7 percent; 95% CI –0.6 to 8.1). Three patients with recurrence and five with metachronous tumors in the LSNNS group required "rescue standard surgery," and afterward, and the overall three-year disease-specific and overall survival rates in the standard and LSNNS groups were 99.5 versus 99.1 percent and 99.2 versus 97.6 percent, respectively. The LSNNS group showed better physical function, fewer symptoms, and improved nutrition than the standard group.

●In a prospective study of 397 patients with cT1/2 gastric cancer <4 cm, all of whom underwent SLN biopsy followed by standard lymphadenectomy, SLN biopsy was technically successful in 98 percent [108]. Of the 57 patients with lymph node metastasis, 53 (93 percent) had a positive SLN, resulting in a false negative rate of 7 percent.

●Less favorable results were reported in another prospective study of 440 patients with cT1 gastric cancer, who also underwent SLN biopsy followed by gastrectomy with standard lymphadenectomy [109]. Overall, 13 of 28 patients with lymph node metastasis were not identified by SLN biopsy, resulting in a false negative rate of 46 percent. Seven of these 13 patients had nodal metastases outside the lymphatic basin that was resected after SLN mapping (skip metastasis).

Two earlier meta-analyses also cautioned against routine use of SLN biopsy in treating early gastric cancer because of its relatively low sensitivity for detection of LN metastases (77 to 88 percent) and high levels of heterogeneity among surgeons [110,111]. In addition, long-term data on disease recurrence and survival are not available after SLN biopsy for gastric cancer [112]. Finally, there is a profound difference between gastric cancer and two other cancers for which SLN has become an accepted strategy (breast cancer, melanoma), in that if the final pathology differs from that seen on intraoperative frozen section analysis, reoperation is a much bigger operation for gastric cancer and is associated with greater risks to the patient. For all of these reasons the use of SLN mapping for early gastric cancer remains investigational and should be limited to clinical trials [113,114]. (See "Overview of sentinel lymph node biopsy in breast cancer" and "Evaluation and management of regional nodes in primary cutaneous melanoma".)

ADJUVANT AND NEOADJUVANT THERAPY — While complete resection is the single strongest predictor of long-term survival, poor results with surgery alone, especially in patients with nodal metastases, provide the rationale for adjuvant and neoadjuvant approaches using chemotherapy, radiation therapy, or a combination of the two.

Adjuvant chemotherapy and chemoradiotherapy — The benefit of postoperative adjuvant combined modality therapy using contemporary radiation therapy techniques and leucovorin-modulated fluorouracil (FU) was shown in a United States Intergroup study (INT-0116) [4]. The details of the adjuvant regimen are provided elsewhere.

Three-year overall and disease-free survival were significantly better for patients receiving chemoradiotherapy in addition to postoperative chemotherapy. These results changed the standard of care in the United States, although there has been controversy because of the inadequacy of surgical treatment. Although D2 nodal dissection was recommended, it was only performed in 10 percent of enrollees, and 54 percent did not even have clearance/examination of the D1 (perigastric) nodes. This noncompliance likely contributed to inferior survival and a 64 percent relapse rate in the surgery alone arm. (See 'Extent of lymph node dissection' above.)

Subsequent trials in which patients received more extended lymphadenectomy have failed to confirm the enhanced survival from the addition of chemoradiotherapy compared with perioperative chemotherapy alone [115,116]. As a result, updated guidelines from the National Comprehensive Cancer Network [10] recommend adding chemoradiotherapy to postoperative chemotherapy for patients with resected T2-4 or node-positive gastric cancer only if less than a D2 lymph node resection has been carried out. (See "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Adjuvant chemoradiotherapy'.)

By contrast, outside of the United States, chemotherapy alone (either following surgery or combined preoperative and postoperative [perioperative] administration, based on randomized trials demonstrating a survival benefit compared with surgery alone) is more often recommended. (See 'Neoadjuvant chemotherapy and chemoradiotherapy' below and "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Adjuvant chemotherapy'.)

Neoadjuvant chemotherapy and chemoradiotherapy — The goals of preoperative therapy are to increase the resectability rate, reduce the rate of local and distant recurrence, avoid resection in patients with highly aggressive disease who will not benefit from resection, and ultimately, improve survival. Although postoperative chemoradiotherapy is more commonly used in the United States, results from several well-conducted international trials have led to the adoption of perioperative (neoadjuvant plus postoperative) chemotherapy approach to treatment of potentially resectable gastric and esophagogastric junction (EGJ) cancer in much of Europe and other parts of the world. (See "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Neoadjuvant/perioperative chemotherapy'.)

Data from several uncontrolled series suggest that some patients with initially locally advanced unresectable disease may respond to chemotherapy or chemoradiotherapy sufficiently that they are able to undergo potentially curative surgery. Perioperative morbidity and mortality rates are acceptable. The risk to benefit ratio may be less favorable in obese patients and those over the age of 60 [117], although this has not been seen in all studies [118].

However, this approach has not been widely adopted, largely due to the lack of randomized trials defining benefit, particularly relative to postoperative chemoradiotherapy. This topic is discussed in detail elsewhere. (See "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Initially locally unresectable nonmetastatic disease'.)

An increasing number of clinicians who treat gastric cancer have a bias towards neoadjuvant therapy approaches for two major reasons:

●First, the ability to deliver adequate postoperative therapy may be compromised by complications of the surgery. As seen in the United States Intergroup 0116 trial of surgery with or without chemoradiotherapy [4], up to one-third of patients do not recover quickly enough to complete the full course of postoperative treatment. Similarly, in the MAGIC trial of induction chemotherapy versus surgery alone, only 66 percent of the patients who underwent surgery were able to begin postoperative chemotherapy and approximately 40 percent of patients completed all therapy (despite a permissible delay of up to 12 weeks) [5]. (See "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Neoadjuvant/perioperative chemotherapy'.)

●Second, some patients have aggressive disease and develop metastases within a short period of time, despite having an adequate operation. These patients do not benefit from surgery, and the delay in operative intervention with induction therapy may have permitted their identification prior to exploratory laparotomy, thus sparing them unnecessary surgery.

As noted above, one clinical scenario in which a neoadjuvant systemic therapy approach is favored by the author, in the absence of a clinical trial of peritoneal-based therapy, is for patients who otherwise appear to have potentially resectable disease but are found to have positive peritoneal cytology at the time of staging laparoscopy (but no overt intraperitoneal metastatic deposits). While relatively few patients do not progress and thus go to resection, this is a controversial area, and others disagree with this approach. (See 'Significance of positive peritoneal cytology' above.)

PROGNOSIS — Prognosis after resection varies according to the pathologic extent of disease, the tumor location, and the population studied. As noted above, Asian populations do better than Western populations. Although stage distribution accounts for some of this difference [119], outcomes differ even when stratified by stage [6,120-123]. In general, proximal tumors (cardia, esophagogastric junction) have a worse prognosis than do distal cancers [124].

Western populations — The most important prognostic factors after resection of gastric cancer are stage (reflecting disease extent), histologic type, status of the resection margins, and in some reports, age and sex.

Limited data are available on observed survival stratified by stage using the 2017 eighth edition tumor, node, metastasis (TNM) staging classification (table 1) [2]. Stratification in overall survival according to pathologic stage in the absence of neoadjuvant therapy (figure 6) [125] and following neoadjuvant therapy (figure 7) [2] are depicted in the figures.

The influence of margin status, age, and sex can be illustrated by an analysis of 208 patients who underwent radical resection of a gastric cancer without preoperative or postoperative chemotherapy or radiation therapy at a single institution [126]. In multivariate analysis, patients with microscopically positive (R1 resection) or macroscopically positive (R2 resection) margins were three times more likely to die of their cancer (hazard ratio [HR] for cancer-specific death 2.97, 95% CI 1.94-4.55). In addition, men had a significantly greater likelihood of dying from gastric cancer than did women, and overall survival (but not cancer-specific survival) was significantly worse among patients over the age of 59 compared with younger individuals.

Outcomes may be improving over time. Some of the most impressive contemporary results for treatment of high-risk gastric cancer in Western populations were reported from a multicenter Italian randomized trial comparing two different adjuvant chemotherapy regimens (leucovorin-modulated FU versus weekly intensive therapy with cisplatin, epirubicin, plus leucovorin-modulated FU) in 400 patients with either T3 or node-positive gastric cancer [127]. While there was no evidence of a survival benefit for the more intensive chemotherapy regimen, the five-year overall survival rates in both groups were higher than expected (50 and 52 percent in both arms), despite the fact that fewer than one-half of either group completed the entire course of adjuvant therapy. Notably, approximately 80 percent of all enrolled patients underwent either a D1 or D2 lymphadenectomy, and 80 percent had more than 15 nodes examined. (See 'Extent of lymph node dissection' above and "Adjuvant and neoadjuvant treatment of gastric cancer".)

Among patients undergoing neoadjuvant therapy, outcomes are dictated by the post-treatment (yp) stage rather than the pretreatment clinical stage. Patients who have substantial residual disease at resection after neoadjuvant therapy (particularly nodal metastases [128-130]) have a relatively poor prognosis. The best way to manage these patients is not established. (See "Adjuvant and neoadjuvant treatment of gastric cancer", section on 'Postoperative management of poor responders'.)

Asian populations — The better outcomes seen in many (but not all [131]) studies of Asian populations can be illustrated by a carefully conducted randomized trial from the Japan Clinical Oncology Group (JCOG) in which 523 patients, approximately two-thirds with node-positive (stage III) disease, were assigned to D2 versus D3 lymphadenectomy [57]. The overall five-year disease-free survival rates in both groups were approximately 62 percent, and the operative mortality rates in both groups were less than 1 percent. These patients were carefully selected for good overall health, and they had documented negative peritoneal cytology. Nevertheless, these are far better outcomes than would be expected in even highly selected Western populations. The study also revealed paradoxical results on subgroup analysis; patients with negative nodes did better with a more extensive operation while those with node-positive disease had a better outcome with a less extensive operation. The reasons for this are unknown.

Other analyses have also demonstrated superior outcomes in Asian as compared with non-Asian populations [132,133].

Several hypotheses have been put forth to explain the differences between Western and Asian patients, including stage migration, differences in treatment (particularly surgical techniques), and race-related differences in tumor biology, behavior, and host-tumor interactions. Whether treatment factors predominate over biologic factors is unclear; the data are conflicting:

●The importance of treatment-related factors was suggested in a stage-stratified comparison of outcomes after gastric cancer resection among 3176 Japanese patients treated in Tokyo according to Japanese techniques and 279 Japanese immigrants to Hawaii who were treated using Western surgical techniques between 1974 and 1985 [134]. For every stage of disease (using the Western American Joint Committee on Cancer [AJCC] TNM staging system from 2002), five-year survival rates were superior for the nonimmigrant Japanese patients:

•Stage I – 96 versus 86 percent

•Stage II – 77 versus 69 percent

•Stage III – 49 versus 21 percent

•Stage IV – 14 versus 4 percent

Incomplete nodal staging in the population treated by western surgical techniques in this older series may also have contributed to some of the differences observed, particularly in stage I and stage II patients.

●Interestingly, a survival advantage for Asian ethnicity was shown in another analysis that included only patients treated in the United States. This may suggest perhaps some underlying genetic basis for the differences. In this analysis of 12,773 patients undergoing gastrectomy for treatment of adenocarcinoma and reported to the United States Surveillance, Epidemiology, and End Results (SEER) database, the independent prognostic effect of ethnicity was evaluated by adjusting for other known prognostic factors [135]. Survival was better for the patients of Asian ethnicity than for the White patients overall (five-year survival 41 versus 29 percent), and the difference in outcomes was most pronounced for those with early stage (ie, IA to IIB (table 1)) disease.

Prognostic models — Nomograms have been developed and validated to predict disease-specific survival following complete (R0) resection in both Asian and Western populations, which take into account the patient's age and sex, the primary tumor site and Lauren histologic subtype (ie, diffuse, intestinal, mixed), the depth of invasion, and the number of both positive and negative resected nodes [136-143]; one such nomogram is illustrated in the figure (figure 8) [139]. An important point is that nomograms such as these do not take into account adjuvant or perioperative treatments, and thus may underpredict prognosis in certain groups of patients [144].

Nomograms have also been developed to predict conditional survival probability (ie, prognosis for patients surviving a specified period of time) after an R0 resection [145]. Risk stratification models are also under development using both clinicopathologic and molecular factors [126].

Sites of disease recurrence — Treatment failure in patients with resected gastric carcinoma can be broken into local recurrence and distant metastases. In most series, distant failures predominate, even in patients with a positive surgical margin [6,146-148]. In data from the American College of Surgeons, recurrence following attempted curative resection was local or regional in 40 percent and systemic in 60 percent [6]. However, locoregional recurrences may be more frequent in patients treated with surgery alone or surgery plus postoperative chemotherapy without radiation therapy, and among those who have a fewer number of negative resected lymph nodes in the operative specimen [149]. Sites of locoregional failure include the luminal margins, the resection bed, and the regional nodal basins [31].

The predominant sites of systemic recurrence in the era of preoperative therapy are the liver and peritoneum [150]. With the more widespread use of preoperative therapy, there seems to be a shift of recurrences to later time periods. Metastatic disease beyond the abdomen is uncommonly the first site of recurrence aside from the supraclavicular nodes.

POST-TREATMENT SURVEILLANCE — There are no randomized trials to guide the postoperative surveillance strategy. As more patients appear to be surviving longer, perhaps the most important aspect of post-treatment monitoring is for nutritional sequela. Consensus-based guidelines from the National Comprehensive Cancer Network [10] suggest the following:

●History and physical examination every three to six months for one to two years, then every 6 to 12 months for three to five years, then annually.

●Complete blood count and chemistry profile as clinically indicated.

●Monitor for iron and vitamin B12 deficiency as well as bone health (Ca++, vitamin D, and consider bone density studies earlier, especially in female patients) in surgically treated patients (especially after total or proximal gastrectomy), and treat as indicated.

●It is important to recognize that drug absorption is altered after gastrectomy, with some agents seeing more rapid absorption, while others, such as birth control pills, see decreased absorption. Patients should be encouraged to discuss these issues with their pharmacist.

●For patients who had a partial or subtotal gastrectomy, upper gastrointestinal endoscopy as clinically indicated.

●Guidelines for radiologic imaging differ according to pathologic disease stage, although it is again important to stress that there are no hard data to support these recommendations:

•For resected pathologic stage I (T1aN0, T1bN0) treated with surgical resection or T1a treated with endoscopic resection, computed tomography (CT) of the chest/abdomen/pelvis with oral and intravenous contrast as clinically indicated.

•For resected pathologic stage II/III (treated with neoadjuvant and/or adjuvant therapy), CT of the chest/abdomen/pelvis with oral and intravenous contrast is recommended every 6 to 12 months for the first two years, then annually up to five years, and/or positron emission tomography (PET)/CT can be considered as clinically indicated.

Similarly, updated guidelines from the European Society for Medical Oncology suggest regular post-treatment follow-up with dietary support (assessment for vitamin and mineral deficiencies), and periodic radiologic investigation (CT of the thorax and abdomen) in patients who are candidates for further cancer-specific therapies [81].

Some clinicians check tumor markers (particularly carcinoembryonic antigen [CEA]) with each follow-up visit and perform restaging CT scans every three to six months, at least for the first two years, which is controversial. Although early detection can facilitate treatment before the development of symptomatic bulky disease that may be difficult to manage (eg, bowel entrapment from peritoneal disease recurrence), there are no data to support the view that early detection of an asymptomatic recurrence by tumor marker elevation or radiologic imaging improves quality of life or prolongs survival [151-153]. Given this there is an increasing consideration for more limited imaging when one balances paucity of defined benefit with the inconvenience, increased cost, and recognition of the risks from exposure to medical radiation. (See "Radiation-related risks of imaging".)

For patients who have had a partial gastrectomy, we perform surveillance endoscopy every 6 to 12 months for the first two years, since survivors are at higher risk for a remnant or second primary gastric cancer, but these patients may likely have a lower risk of gastric cancer for many subsequent years. Some have suggested annual endoscopic follow-up for up to 20 years after the primary operation to allow for early detection of remnant gastric cancers [154], but this is controversial. (See 'Gastric remnant carcinoma' below.)

We supplement all patients who have undergone total or partial gastrectomy with vitamin B12 (once monthly intramuscularly or a sublingual B-complex vitamin daily is generally the most preferred method) and monitor levels of both vitamin B12 and D. We are becoming increasingly aware of the critical role vitamin D plays and the alarming frequency of low levels in patients. (See "Clinical manifestations and diagnosis of vitamin B12 and folate deficiency" and "Vitamin D deficiency in adults: Definition, clinical manifestations, and treatment".)

In addition, patients may have persistent anemia due to pretreatment surgical blood loss and decreased long-term absorption of iron with subsequent iron deficiency. Treatment of iron deficiency anemia is addressed separately. (See "Treatment of iron deficiency anemia in adults".)

RECURRENT AND METASTATIC DISEASE

Palliative gastrectomy — For patients with locally advanced or metastatic disease, surgical intervention may provide effective palliation of symptoms such as pain, nausea, persistent bleeding, or obstruction. However, anemia in a stable patient should not be considered an indication for a palliative gastrectomy. Radical gastrectomy in the above settings provides no survival benefit, but palliative gastrectomy can provide symptomatic relief and a possible improvement in survival in selected patients, although this is controversial. The criteria for selection of patients who may benefit from palliative gastrectomy as compared with other palliative procedures (including radiation therapy, endoscopic intervention, and surgical bypass) are not firmly established. These issues are discussed in detail elsewhere. (See "Local palliation for advanced gastric cancer".)

Locally recurrent disease — Rather than surgery, most patients with recurrent disease are offered systemic chemotherapy for palliation of symptoms. Occasionally, local palliative procedures, such as radiation therapy (in patients not previously having received radiation therapy at the site of recurrence), are needed for symptom control. The treatment of advanced gastric cancer is discussed in detail elsewhere. (See "Initial systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer" and "Local palliation for advanced gastric cancer", section on 'Radiation therapy'.)

Curative resection is not often attempted in patients with locally recurrent disease (which is defined as a tumor of similar histology to the original, which typically develops at the anastomotic site), although there are a few case series that describe long-term survivors [155-161]. In a series of 60 patients (seen over a thirty-year period) who underwent an attempted resection for recurrent gastric or gastroesophageal cancer, 29 were resectable at laparotomy, 14 of whom required adjacent organ resection, and 6, interposition grafting [155]. Median overall survival for patients undergoing resection was 26 months, compared with six months for unresectable patients.

While these results suggest that salvage surgery may contribute to prolonged survival, the opportunity to resect locally recurrent gastric tumors is infrequent at best. This approach should be restricted to highly selected, fit patients with an isolated local recurrence and no evidence of distant spread.

Gastric remnant carcinoma — A small proportion of patients undergoing subtotal gastrectomy for gastric cancer or peptic ulcer disease may develop a gastric remnant carcinoma [158,162-165]. In a series of 541 patients who underwent subtotal gastrectomy for early gastric cancer and were followed for at least 15 years, a gastric remnant carcinoma was diagnosed in 16 (2.9 percent), and the 20-year cumulative risk was 4 percent [162]. In another series of 6365 gastrectomies, 85 (1.3 percent) developed gastric remnant carcinoma [154].

The early diagnosis of a gastric remnant carcinoma relies on endoscopic examination, which is recommended by some at least annually for up to 20 years [154]. It is important to look at this also from an alternative perspective; for each remnant cancer found, 500 to 1500 upper endoscopies will likely need to be performed. The diagnosis of a metachronous gastric cancer versus recurrence may be difficult, but a different histologic type compared with the original tumor and location that is nonanastomotic favors a metachronous cancer in the gastric remnant.

The usual treatment is completion gastrectomy. Endoscopic management might be an option for early gastric remnant carcinomas (ie, those limited to the mucosa or submucosa), although the available data on the long-term success of this approach are limited [166].

Results of surgical resection may be more favorable when a gastric remnant adenocarcinoma develops at a nonanastomotic site. In one series of 52 such patients who underwent a second operation with curative intent because of a remnant gastric cancer, the overall five-year survival rates were significantly higher for those with a nonanastomotic recurrence (96 versus 37 percent) [158].

Surgical management of metastatic disease

Isolated peritoneal metastases — For patients with apparently isolated peritoneal metastases, the role of cytoreductive surgery (CRS) with or without heated intraperitoneal chemotherapy (HIPEC) remains controversial, and it is not widely considered a standard approach [21,24,167,168]. The phase III CYTO-CHIP study analyzed the impact of CRS alone versus CRS plus HIPEC in highly selected patients with apparently isolated peritoneal metastases from gastric cancer, with data derived from prospective databases of 19 French cancer treatment centers [24]. All patients had to have complete CRS with curative intent (no residual nodules >2.5 mm). The group receiving HIPEC in addition to CRS had better median survival (19 versus 12 months) and five-year recurrence-free survival (17 versus 4 percent), but whether these results are better than could have been achieved with conventional systemic chemotherapy is unclear. A significant proportion of patients in both groups had positive cytology only. It should also be noted that these patients were accumulated over a 25-year period and that these 277 total patients (or less than one patient per center per year) represent a very highly selected population. All clinicians who care for significant numbers of patients with gastric cancer have occasional patients who have been long-term survivors despite the presence of stage IV disease by cytology or histology and without HIPEC. These data, while demonstrating a statistical significance, should not be overinterpreted.

A Chicago consensus document has been published addressing CRS and HIPEC for a number of diseases including gastric cancer [168]. In the text of the manuscript the authors correctly note that for gastric cancer a prospective randomized trial is needed to prove benefit, and that patients with peritoneal metastases from gastric cancer should be considered for clinical trials. Unfortunately, in the algorithm that accompanies this section, patients are directed toward having CRS and/or HIPEC without mention of the need for a clinical trial. At our institution, patients are only considered for CRS and/or HIPEC within the context of a clinical trial.

Metastasectomy

Hepatic — Hepatic metastasectomy has been performed for the rare patient with isolated gastric cancer liver metastases. The major reason this is such a rare event is that the rate of isolated liver metastases in gastric cancer is only 0.5 percent (and this amongst Asian populations) [169,170]. (See "Potentially resectable colorectal cancer liver metastases: Integration of surgery and chemotherapy".)

Despite extremely stringent selection, long-term survival is extremely rare following hepatic resection of apparently isolated gastric cancer liver metastases [169-174]. There is a lack of consensus as to indications for hepatectomy or appropriate patient selection criteria, but overall, it is quite unlikely that an individual patient will benefit from resection of gastric cancer liver metastases.

Other nonsurgical methods for local tumor ablation of isolated liver metastases are discussed elsewhere. (See "Nonsurgical local treatment strategies for colorectal cancer liver metastases".)

Pulmonary — Pulmonary metastasectomy for metastatic gastric cancer can potentially result in long-term survival in a highly selected group of patients. However, this approach should be reserved for those patients who present with small isolated lesions after a prolonged disease-free interval. (See "Surgical resection of pulmonary metastases: Outcomes by histology", section on 'Gastric cancer'.)

SPECIAL CONSIDERATIONS DURING THE COVID-19 PANDEMIC — The COVID-19 pandemic has increased the complexity of cancer care. Important issues include balancing the risk from delaying cancer treatment versus harm from COVID-19, minimizing the number of clinic and hospital visits to reduce exposure whenever possible, mitigating the negative impacts of social distancing on delivery of care, and appropriately and fairly allocating limited health care resources. Specific guidance for decision-making for cancer surgery on a disease-by-disease basis is available from the American College of Surgeons, from the Society for Surgical Oncology, and from others. These and other recommendations for cancer care during active phases of the COVID-19 pandemic are discussed separately. (See "COVID-19: Considerations in patients with cancer".)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Gastric cancer" and "Society guideline links: Gastric surgery for cancer".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topic (see "Patient education: Stomach cancer (The Basics)")

SUMMARY AND RECOMMENDATIONS

●Indications for surgical management of gastric cancer – Surgery may be required as a component of the staging evaluation for gastric cancer, for potentially curative treatment of localized disease, or for palliation in cases of advanced disease. Complete surgical eradication of a gastric tumor with resection of adjacent lymph nodes represents the best chance for long-term survival. Unless there is unequivocal evidence of disseminated disease, there is major vascular invasion, or there are other contraindications to surgery, either abdominal exploration (preferably with an initial laparoscopic approach) with curative-intent resection should be undertaken or a neoadjuvant approach should be considered. (See 'Introduction' above.)

●Surgical techniques – The choice of operation for gastric cancer depends on the location of the tumor within the stomach, the clinical stage, and the histologic type. The major surgical considerations include the extent of stomach resection (total versus partial gastrectomy) and the extent of lymph node dissection.

•Total gastrectomy (figure 2) is usually performed for lesions in the proximal (upper third) stomach, while distal gastrectomy (with resection of adjacent lymph nodes) appears to be sufficient for lesions in the distal (lower two-thirds) stomach. Patients with large midgastric lesions or infiltrative disease (eg, linitis plastica) may require total gastrectomy. (See 'Total versus partial gastrectomy' above.)

•Tumors of the proximal stomach that do not invade the esophagogastric junction (EGJ) can be approached with either total gastrectomy or proximal subtotal gastrectomy. Total gastrectomy is preferred by most surgeons. (See 'Proximal and esophagogastric junction tumors' above.)

•Surgical management of tumors involving the EGJ is discussed elsewhere. (See "Surgical management of resectable esophageal and esophagogastric junction cancers", section on 'Esophagogastric junction cancer resection'.)

•The optimal extent of lymphadenectomy is debated. While several randomized trials have failed to show an overall survival benefit from a D2 compared with a D1 resection, excess morbidity and mortality were clearly associated with the use of splenectomy and distal pancreatectomy to achieve complete node dissection. The most recent report of the Dutch trial and a 2015 Cochrane meta-analysis suggest that cancer-specific mortality rates are significantly lower in patients who undergo a D2 rather than a D1 lymphadenectomy. (See 'D1 versus D2 dissection' above.)

As a result, we suggest a D2 rather than a D1 resection for patients who are undergoing potentially curative resection of a gastric cancer (Grade 2B). However, more extensive nodal dissection should only be performed in selected centers where surgeons have demonstrated acceptably low operative morbidity and mortality rates. Perioperative mortality rates under 2 percent should be expected at centers with higher patient volume.

By contrast, there is no evidence that a D3 resection (paraaortic lymphadenectomy) confers a survival benefit over a D2 dissection, and it is associated with greater perioperative mortality. We recommend that a D3 dissection not be considered for surgical treatment of gastric cancer (Grade 1A). (See 'D2 versus D3 dissection' above.)

•While open gastrectomy remains the standard surgical treatment for gastric cancer worldwide, laparoscopic gastrectomy is performed with increasing frequency in high-volume centers with the requisite expertise (ie, primarily Asian countries). Laparoscopic gastric cancer surgery is most commonly performed for early gastric cancers that are not amenable to endoscopic resection. Laparoscopic gastrectomy for more advanced gastric cancers has been shown to be feasible in Asia, but further validation as to its long-term outcomes compared with open surgery is needed in Western populations before it can be considered a standard approach. (See 'Open versus laparoscopic resection' above.)

●Role of neoadjuvant therapy – Data from several uncontrolled series suggest that some patients with initially locally advanced unresectable disease may respond to chemotherapy or chemoradiotherapy sufficiently that they are able to undergo potentially curative surgery. However, this approach should ideally be considered in the context of a clinical trial. (See 'Neoadjuvant chemotherapy and chemoradiotherapy' above.)

●Palliative gastrectomy – In patients with locally advanced or metastatic disease, surgical intervention may provide effective palliation of symptoms such as pain, nausea, bleeding, or obstruction. The criteria for selection of patients who may benefit from palliative gastrectomy as compared with other palliative procedures (including radiation therapy, endoscopic intervention, and surgical bypass) are not firmly established. (See 'Palliative gastrectomy' above.)

●Metastasectomy – Hepatic metastasectomy for isolated lesions is not associated with long-term survival. Pulmonary metastasectomy for isolated lesions can potentially result in long-term survival in rare, highly selected patients. Cytoreductive surgery and heated intraperitoneal chemotherapy for patients with gastric cancer and peritoneal metastases or positive peritoneal cytology in the absence of overt metastatic disease should only be considered in the context of a clinical trial. (See 'Metastasectomy' above.)

●Postsurgical surveillance – There are no randomized trials to inform the optimal frequency or components of post-treatment surveillance. We follow published guidelines from the National Comprehensive Cancer Network [10]. (See 'Post-treatment surveillance' above.)