Chapter 45 Gastric/GE Junction Cancer
Epidemiology and Etiology
Epidemiology
In the United States, gastric cancer now ranks 14th in incidence among the major types of malignancy. Over the past 60 years there has been a significant decline in the incidence of gastric cancer among both sexes in Western countries1 (see web-only Fig. 45-1). In the United States, from 1930 to 1980, the incidence decreased from 38 to 10 (per 100,000) for men and from 30 to 5 (per 100,000) for women. In 2010, it was estimated that 21,000 new cases and approximately 10,570 deaths will occur.2 The disease rarely occurs before the age of 40 years, but its incidence increases steadily thereafter and peaks in the seventh decade. African Americans, Hispanic Americans, and Native Americans are twice as likely to develop gastric cancer as are whites.3 There has been a slight improvement in 5-year (OS) among U.S. patients: 11% (1900 to 1963), 15% (1974 to 1976), 18% (1980 to 1982), and 21% (1989 to 1994).3
Although gastric cancer incidence has decreased significantly in the United States, on a worldwide scale gastric cancer has a very high incidence and is still a leading cause of cancer death.2,4 Of the 45 countries in which age-adjusted death rates for gastric cancer were compared for 2000 (see web-only Fig. 45-2), the United States ranked 45th for both males and females.2 Kyrgyzstan ranked first for both males (47.0/100,000) and females (18.9/100,000).
The site of origin within the stomach has changed in frequency in the United States over recent decades, with proximal lesions now being diagnosed and treated much more often than previously. The largest percentage of gastric cancers still arises within the antrum or distal stomach (~40%), are least common in the body of the stomach (~25%), and are of intermediate frequency in the fundus, cardia, and esophagogastric junction (~35%). In 1930, most cases of gastric carcinoma originated in the distal stomach (body and antrum). The reduction in incidence of gastric cancer from 1930 to 1980 primarily is attributable to a decline in distal lesions.5
Since the 1970s there has been a steady rise in the incidence of adenocarcinoma of the gastroesophageal (GE) junction and proximal stomach. For both sites the rate of increase during the 1970s and 1980s surpasses that of any other cancer, including melanoma, lung cancer, and non-Hodgkin’s lymphoma.6 Similar trends in the increased incidence of gastric cardia cancer have also been reported in Denmark, where rates were much higher in men, and in the United Kingdom, where rates were highest among those in higher socioeconomic classes.7
Etiology
Although the precise cause is unknown, environmental factors are likely to be important. The incidence of gastric cancer in first- and second-generation Japanese who have moved to the United States is much lower than in native Japanese citizens.8 Studies of immigrant populations from Japan and Eastern Europe show that the risk of the disease declines markedly in the second and third generations.9 The excess risk of first-generation immigrants is largely restricted to the intestinal type.10 The risk of gastric cancer is inversely associated with socioeconomic status throughout the world,11 and the marked decline in the incidence of gastric cancer in industrialized countries, including the United States, suggests that environmental factors play an important role. These may be related to poor nutrition, inadequate sanitation facilities, and substandard quality of food preservation.
Diets rich in fruits or vegetables are associated with a reduced risk of gastric cancer,12 whereas diets containing abundant quantities of smoked, heavily salted, or poorly preserved foods are associated with an increased risk of this disease.12,13 Excessive dietary salt has been associated with atrophic changes in the gastric mucosa.14 Chronic atrophic gastritis and the associated abnormality, intestinal metaplasia, are most closely linked to the intestinal type of gastric adenocarcinoma. These lesions can progress to dysplasia and carcinoma.15 The prevalence of atrophic gastritis and intestinal metaplasia is highest in regions of the world that have the highest rates of gastric cancer.16
It has been postulated that endogenous formation of N-nitroso compounds (amine or amides) can occur in the stomach when both an amino compound and a nitrating agent such as nitrate are present.13 Anaerobic bacteria that colonize stomachs in which achlorhydria occurs can convert nitrates and nitrites to potentially carcinogenic nitroso compounds.12
Epidemiologic studies have demonstrated an association between infection with Helicobacter pylori and risk of gastric cancer.17 Serologic studies have reported that persons with H. pylori infection have a threefold to sixfold higher risk of distal gastric cancer than those without this infection.18,19 The exact role of H. pylori in gastric carcinogenesis is still to be determined, although infection is associated with the development of chronic atrophic gastritis and decreased acid production.12 However, only a minority of infected patients develop gastric cancer, and data do not yet exist on the effect of treatment of the H. pylori infection on subsequent malignancy. Proximal gastric cancers (gastroesophageal junction, cardia) do not appear to be related to H. pylori infection, atrophic gastritis, or decreased acid production, although it is possible that an inverse relationship exists.
There has been a dramatic increase in the incidence of GE and gastric cardia carcinoma over the past few decades, similar to the increase in adenocarcinomas of the distal esophagus, suggesting that they may have similar causes.20,21 Although the reasons for these changes are unknown, they may be related to the increased incidence of esophageal reflux and Barrett’s esophagus.
In summary, the worldwide decline in distal, predominantly intestinal-type gastric adenocarcinoma may be the result of improved food handling (refrigeration and storage). In contrast, proximal (mostly diffuse-type) gastric cancer, which is equally prevalent in both high-risk and low-risk regions, is probably attributable to other, as yet unrecognized, factors.20 The exact role of H. pylori in gastric carcinogenesis remains to be defined.14
Prevention And Early Detection
Early detection would markedly improve prognosis of gastric cancer, because surgical resection has a high cure rate with lesions limited to the mucosa or submucosa. In the United States, however, the incidence of early gastric cancer is less than 5% in most series. In Japan, the incidence of carcinoma confined to the mucosa or submucosa was only 3.8% in the 1955 to 1956 period but by 1966 the incidence of early lesions had increased to 34.5% because of vigorous screening procedures (5-year OS of 90.9% in this cohort of patients). The value of screening is evidenced by the 5-year OS of 53% for all gastric cancers in Japan versus 21% worldwide.4
Germline mutations in the CDH1 gene, which encodes the E-cadherin protein, have been recognized in families with hereditary diffuse gastric adenocarcinoma. Carriers of these mutations have a 70% lifetime risk of developing gastric cancer. Several reports of prophylactic gastrectomy have demonstrated the routine presence of microscopic intraepithelial carcinomas for patients having regular endoscopic surveillance with multiple random biopsies.22–24 Early total gastrectomy has been recommended for this small patient population because of the lack of effective early tumor detection. Microscopic evaluation of the proximal and distal resection margins for complete removal of the gastric mucosa is necessary, because residual gastric mucosa can degenerate and result in a gastric cancer.24
Biologic Characteristics and Molecular Biology
Prognostic Factors
The most valuable prognostic indicators relate to tumor extent. With either blood-borne metastasis or peritoneal seeding, cure is rare to nonexistent. Survival decreases with progressive direct tumor extension both within and beyond the wall of the stomach.25,26,27,28,29 Lymph node involvement, per se, is not as important as the number and location of nodes.26,30 Minimal lymph node involvement adjacent to the primary lesion is the most favorable.31 The solitary finding of either involved lymph nodes or extension beyond the gastric wall is usually not as ominous as the presence of both.26,27,28,29,30
Tumor grade and gross and microscopic pathologic appearance of the primary malignancy appear to provide some prognostic information, but none of these factors is an independent prognostic variable relative to tumor stage. Prognosis is generally worse with higher-grade and diffuse-type carcinomas, which usually present with a higher stage of disease. Borrmann’s type I and II carcinomas have relatively favorable 5-year OS, whereas patients with type IV tumors (linitis plastica) have a very poor prognosis.31,32
Prognostic factors evaluated in a British study group trial33 included tumor size, macroscopic aspect, number of sites involved, wall invasion, involvement of surgical resection limits, nodal stage, and histologic grade. Multivariate analyses established as prognostic features the depth of wall invasion, nodal involvement, and involvement of margins of surgical resection.
Some investigators have suggested that tumors of the gastric cardia may have different epidemiologic factors than cancers of the distal stomach34,35 and may exhibit different tumor biology.36 Prognosis appears worse for cardia lesions,37,38 and flow cytometry reveals a greater incidence of aneuploidy when compared with tumors of the antrum and body.39
Molecular Biology
The molecular biology of gastric cancer reflects the heterogeneity of its causes and its histologic subtypes. Identification of the genetic and phenotypic variables existing among gastric cancers may lead to more directed therapeutic approaches and a more accurate prediction of clinical outcome. Changes that may affect the behavior of gastric tumor cells involve four major types of alterations. Loss of tumor-suppressor gene function, especially inactivation of the TP53 gene, certainly plays a critical role. The TP53 gene is located on the short arm of chromosome 17 and plays a key role in tumor suppression and cell-cycle regulation.40 This gene puts a brake on DNA replication and triggers programmed cell death in response to DNA damage.41 Loss of TP53 function is associated with the development of malignancy, impacts the effectiveness of chemotherapy and irradiation,42,43 and predisposes cells to genetic instability. The latter is particularly important because TP53 mutations typically occur early in carcinogenesis.
A second major aberration affecting gastric epithelial cells is the impact of alterations in mismatch repair genes. Two such genes, MSH2 and MLH1, on chromosomes 2 and 3, respectively, account for replication errors throughout the genome. Mutations in these genes are implicated in cancer family syndromes and hereditary nonpolyposis colorectal cancer, a syndrome associated with an increased tendency for the development of colorectal and gastric tumors.44 Mutations in these genes generate genetic instability and have the potential to lead to further alterations in oncogenes.
Two proto-oncogenes, MET and FGFR2, are associated with scirrhous carcinoma of the stomach. The former encodes hepatocyte growth factor, which is a potent endogenous promoter of gastric epithelial cell growth.45 Its overexpression correlates with tumor progression and metastasis.47 The latter encodes a tyrosine kinase receptor family.46 In scirrhous carcinoma, MET and FGFR2 amplification may occur independently. There is a tendency for FGFR2 to be activated in women with gastric cancer younger than 40 years of age and for MET to be amplified in men older than 50 years of age.47,48
Flow cytometry provides valuable prognostic information for gastric cancer and may be an independent prognostic factor.39 As noted previously, aneuploidy is associated with unfavorable tumor location such as the cardia39,49 but is also associated with lymph node metastasis49–51 and direct tumor extension.28 Unfavorable DNA flow cytometry characteristics seem to relate to an unfavorable prognosis.39,49,50 In one series in which multivariate analysis of DNA ploidy was analyzed with other known prognostic factors such as stage, age, and sex, DNA ploidy carried statistically significant independent prognostic information.51
The presence of several peptides including estrogen receptor,52 epidermal growth factor receptor,53 and the ERBB2 protein54 appears to affect prognosis adversely. The expression of epidermal growth factor receptor and high levels of epidermal growth factor correlate with a higher incidence of primary tumor infiltration, poor histologic differentiation, and linitis plastica. The pathophysiologic relationship between these peptide receptors and poor patient prognosis is not clear.
ERBB2 overexpression has been shown to occur in approximately 20% of patients with gastric cancer and is associated with a poorer prognosis. Perhaps more importantly, the use of the anti-ERBB2 antibody trastuzumab has been demonstrated to improve survival by approximately 2.5 months in patients with metastatic disease.55
Modern molecular biology observations confirm the heterogeneity of human gastric cancer. Genetic alterations, detected and potentially associated with worse prognosis, included CD44 expression, telomerase reactivation, TP53 gene inactivation, dysfunction of repair genes such as MSH2 and MLH1, overexpression of proto-oncogenes such as ERBB2, BCL2, MET, and FGFR2, estrogenic receptor expression, and presence of viral genome.56 Gastric cancers with class II major histocompatibility complex antigen expression (HLA-DR) have a better prognosis, but the loss of expression is not an independent prognostic factor.57
Pathology And Pathways Of Spread
Pathology
Adenocarcinoma (ACA) histology accounts for 90% to 95% of all gastric malignancies, and the terms gastric cancer and stomach cancer usually refer to such tumors. Other histologies include lymphoma (usually unfavorable histology), leiomyosarcoma, carcinoid, adenoacanthoma, and squamous cell carcinoma.58
Gastric carcinomas have been categorized with regard to both microscopic and gross pathologic features. The Lauren classification system includes an intestinal type with improved prognosis (that predominates in regions with high prevalence of gastric cancer) and a diffuse histologic type with poor prognosis (that occurs more commonly in countries with low prevalence).59 Diffuse carcinomas occur more often in young patients and develop throughout the stomach but especially in the cardia. Intestinal-type lesions are frequently ulcerative and occur in the distal stomach more commonly than the diffuse type. Grossly, gastric cancers can be categorized according to Borrmann’s five types: I, polypoid or fungating; II, ulcerating lesions surrounded by elevated borders; III, ulceration with invasion of the gastric wall; IV, diffusely infiltrating (linitis plastica); and V, not classifiable.60 The Japanese Research Society for Gastric Cancer has a classification system that divides lesions into protruded (I), superficial (II) (with elevated [IIa], flat [IIb], and depressed [IIc] subtypes), and excavated (III) types.58
Pathways of Spread
Lymphatics
Because of the numerous pathways of lymphatic drainage in the stomach, it is difficult to perform a complete nodal dissection (see web-only Fig. 45-3). Although initial drainage is usually to lymph nodes along the lesser and greater curvatures (perigastric or N1 nodes using the Japanese Research Society for Gastric Cancer designation), primary nodal drainage includes nodes along all three branches of the celiac (left gastric, common hepatic, splenic) and the celiac itself (N2).58 Node groups that are more distal include hepatoduodenal, peripancreatic, root of mesentery (N3), periaortic, and middle colic (N4). When proximal gastric lesions extend into the distal esophagus, that nodal system is at risk.
Clinical Manifestations, Patient Evaluation, And Staging
Patient Evaluation
The diagnosis of gastric cancer is usually confirmed by upper gastrointestinal endoscopy or radiographs (Table 45-1). Double-contrast radiographs may reveal small lesions limited to the superficial (inner) layers of the gastric wall. Endoscopy is now the preferred diagnostic study, because it allows direct tumor visualization, cytology, and histologic biopsy that yield the diagnosis in 90% or more of patients with exophytic lesions. Ulcerated cancers and linitis plastica lesions may be harder to diagnose endoscopically, but multiple biopsies and washings enhance the probability of accurate diagnosis. Endoscopic ultrasonography (EUS) has a high degree of accuracy in determining depth of tumor invasion before resection (i.e., whether the lesion extends beyond the muscularis propria) but is less accurate in detecting regional nodal metastasis.61,62 Needle biopsies of suspicious nodes can be performed at the time of endoscopy with EUS. Although EUS may not affect therapy decisions with most current management approaches, its use may lead to selective management strategies based on tumor stage.
Diagnostic Procedure | Diagnosis and Staging Capability | Recommend Routine Use |
---|---|---|
Primary Tumor/Regional Nodes* | ||
Single-contrast UGI | Useful in detecting and defining primary lesions in stomach | Optional; consider along with double-contrast UGI |
Gastroscopy | Very accurate modality to detect and define primary lesion, ≈90% confirmation rate | Yes; use to confirm lesion detected in UGI series and to screen high-risk patients |
Ultrasound-endoscopy | Most accurate method of determining extension within and beyond gastric wall | Yes; biopsy node(s), if feasible |
Double-contrast UGI | Useful in detecting early gastric cancers | Consider along with single-contrast UGI |
CT: abdomen + chest | Most valuable modality to determine degree of extragastric extension and distant metastases | Yes; include CT chest for GE junction cancers |
Metastatic Tumors | ||
Chest films | Good for detecting metastases | Yes |
Laparoscopy | May allow visualization of small serosal implants or liver metastases | Optional; recommended if planning preoperative chemotherapy or chemoradiation |
PET | Excellent for detecting unsuspected metastases | Optional; recommended if planning preoperative chemotherapy or chemoradiation |
UGI, upper gastrointestinal study.
* Laboratory studies: complete blood cell count, creatinine level, liver function studies (alkaline phosphate, bilirubin, aspartate transaminase, lactate dehydrogenase), albumin level.
The value of laparoscopy in the staging of gastric cancer is still being defined. In 71 patients with CT criteria of resectable disease, 69 completed a laparoscopic evaluation.63 Laparoscopy confirmed disease in 16 of the 17 patients with peritoneal metastasis (avoiding 12 laparotomies) and in 1 of 3 patients with small liver metastases and negative CT scan. The combination of CT and laparoscopy staging information yielded a resectability rate of 93% for patients defined as potential candidates for curative gastric cancer surgery.
Treatment planning in gastric cancer patients could be improved by accurate preoperative classifications of depth of invasion (T category) and lymph node involvement (N category). A prospective study in 108 consecutive patients evaluated EUS, CT, and intraoperative surgical assessment for T and N classification.64 Staging of the T category was correct with CT in 43% of the cases, in 86% with endoscopic ultrasonography, and in 56% with intraoperative assessment. Staging of N1 and N2 lymph nodes was correct with CT in 51% of the cases, with endoscopic ultrasonography in 74%, and with intraoperative assessment in 54%. Advanced gastric tumors tended to be more accurately staged with CT; CT, in general, overstaged the T category and understaged the N category. Endoscopic ultrasonography showed high and similar accuracy for all the T categories, although it also tended to understage N categories. Finally, intraoperative assessment was equally accurate for all N categories but tended to overstage early T stages and to understage N categories.
Staging
Table 45-2 shows the current American Joint Commission on Cancer (AJCC) TNM (tumor, lymph nodes, metastasis) staging system.65 The TNM system is the standard system for reporting outcomes in gastric and gastroesophageal junction cancer.
Primary Therapy
Surgical Method
Although only one small prospective randomized trial has been performed with regard to extent of the gastric resection,66 extensive experience exists with various surgical procedures, and appropriate generalizations can be made. The preferred treatment for lesions arising in the body or antrum of the stomach is a radical distal subtotal resection. This removes approximately 80% of the stomach along with the first portion of the duodenum, the gastrohepatic and gastrocolic omenta, and the nodal tissue adjacent to the three branches of the celiac axis. Extensive or proximal cancers will require a total gastrectomy to achieve an adequate proximal gastric margin. However, total gastrectomy is not necessary when subtotal gastrectomy will provide a 5-cm clearance of the gross tumor.67,68 The propensity for gastric carcinoma to spread via submucosal and subserosal lymphatics dictates the need for a 5-cm surgical resection margin of normal stomach beyond the visible lesion. It may be necessary to extend the resection to include some (or additional) esophagus or duodenum if frozen section pathologic evaluation of the surgical margins fails to confirm the adequacy of proximal and distal resection margins. If total gastrectomy is necessary, a splenectomy is sometimes performed, particularly in gastric cancers of the proximal third of the stomach and tumors of the body near the greater curvature. Cancers in these locations are more likely to metastasize to lymph nodes in the splenic hilum that cannot be completely excised without a splenectomy. Although the value of routine splenectomy has not been addressed in prospective randomized trials, retrospective Japanese and American data do not provide evidence of a survival benefit with the routine use of this procedure.69,70
The optimal extent of lymph node dissection for gastric cancer remains controversial. Several studies have shown that the presence and extent of lymph node metastasis correlate with the depth of primary tumor invasion.71–73 Although several nonrandomized clinical trials have suggested that extended lymphadenectomies may improve survival,70,74–78 other nonrandomized79,80 and randomized trials81–83,84–87 have not demonstrated an advantage. A large multicenter phase III study in the Netherlands accrued 996 evaluable patients and provides additional objective data on the lack of value of extended lymphadenectomy in gastric carcinoma.82 Among the 711 patients having curative resections, both morbidity and mortality were significantly higher with the more extensive nodal dissection.83,84,85 A randomized study of 400 patients conducted by the Medical Research Council (MRC) in the United Kingdom has also demonstrated higher morbidity and mortality in the extended lymphadenectomy cohort.86,87 No benefit in disease-free survival (DFS) or OS was found in either the Dutch or British trials (Table 45-3). Any apparent survival benefit seen with the extended node dissection performed in Japan may be caused by the phenomenon of a shift in stage rather than superior surgery. Although most patients with five or more lymph node metastases or node metastasis not adjacent to the primary tumor have a very poor outcome, extended node dissection seems reasonable when it can be performed by experienced surgeons without significant additional surgical morbidity or mortality. A recent systematic review and meta-analysis88,89 demonstrate the safety and feasibility of D2 resection with or without para-aortic nodal dissection but fail to show a benefit in OS.
Endoscopic laser surgery has been used in selected patients with early gastric cancer.90 Small lesions (≤3 cm) that are pedunculated, do not invade the submucosa, and are well differentiated infrequently have lymph node metastasis (<5%). Nearly 75% of these select tumors can be completely removed endoscopically. Although early gastric cancer may have a long natural history before progression, standard surgical resection rather than endoscopic removal is still preferable.91 Irradiation plus chemotherapy should be considered as adjuvant therapy when the tumor is endoscopically treated.
Quality of life after gastrectomy is being analyzed.92,93 In a University of Heidelberg report of 104 patients,94 12 months after gastrectomy, postoperative symptoms of heartburn or dumping in the group undergoing total gastrectomy with pouch reconstruction did not differ significantly from the group having only distal gastric resection. In a recent prospective trial of 120 patients requiring total gastrectomy plus pouch who were randomized to two different types of reconstruction (jejunal interposition versus Roux-en-Y),95 there were no differences in food passage, body weight, or quality of life parameters. However, for all patients there are significant quality of life issues related to decreased absorption of fats, iron, and calcium, as well as the loss of intrinsic factor, and thus vitamin B12 deficiency. These losses need to be addressed with dietary supplements (iron, calcium), monitoring for osteoporosis, and possible monthly vitamin B12 injections.
Survival Results of Surgery Alone
Overall survival with surgery alone remains poor, despite improved perioperative care that has resulted in a substantial decline in postoperative mortality.96 A large review from Europe97 reported excellent 5-year survival for early gastric cancer patients (83%) but a marked diminution in survival for more invasive cancers (31% for tumors of the antrum, 24% for the midstomach, and 16% for the cardia). Survival in excess of 90% has been achieved throughout the world with resection of lesions confined to the mucosa or submucosa.98–102 In reports with lower 5-year survivals for the early lesions, most of the mortality is from noncancer causes.99,103 For gastric cancers with deeper invasion or nodal involvement, survival decreases proportional to the degree of invasion or nodal involvement. When N1 or N2 nodes are involved, Western reports continue to show 5-year OS of 10% to 30%,104,105 whereas Japanese authors report 5-year OS of 25% to 60% (vs. <10% with N3 or N4 involvement)70,106–109 (see web-only Table 45-1). Although pathologic staging differences, different tumor biology, and more radical surgical extirpation have been proposed as explanations, the cause of the difference in U.S. and Japanese results with N1 or N2 disease remains uncertain. Even in the Far East, half of all patients with more invasive gastric cancer do not survive their disease, a fact that underlies the need for new nonsurgical therapies. Results of phase III trials, to be discussed subsequently, demonstrate improvements in survival with the addition of adjuvant chemoradiation to surgical resection.
Relapse Patterns after “Curative Resection”
Local relapse or failure in the tumor bed and regional lymph nodes or distant failures via hematogenous or peritoneal routes are all common mechanisms of relapse after “curative resection” in clinical, reoperative, and autopsy series.* For lesions of the GE junction, both the liver and lungs are common sites of distant metastases. With gastric lesions that do not extend to the esophagus, the initial site of distant metastases is usually the liver, and many relapses could be prevented if an effective “abdominal” or liver therapy could be combined with treatment of the primary tumor and regional lymph nodes.
Locoregional failures occur commonly within the region of the gastric bed and nearby lymph nodes (Table 45-4 and Fig. 45-1). Tumor relapse in anastomoses, the gastric remnant, or the duodenal stump is also frequent. In a University of Minnesota reoperative analysis,28,29 locoregional failure occurred as the only evidence of relapse in 29% of the 86 patients with relapse (23% of the 105 evaluable patients at risk) and as any component of failure in 88%. More extensive operative procedures including routine splenectomy, omentectomy, and radical lymph node dissection neither improved survival116 nor decreased the incidence of locoregional failure28,29 in the reoperative analysis. Subsequent relapse within the scope of the initial node dissection occurred in a high percentage of the patients even when radical node dissections were performed (removal of N1 and N2 with or without N3 nodes)9 (see web-only Table 45-2). These data indicate the difficulty of performing a complete lymph node and lymphatic resection in this anatomic location and provide a partial explanation for the lack of survival benefit with a D2 versus D1 node dissection in phase III surgical trials discussed previously. In a more recent clinical analysis of patterns of relapse from Memorial Sloan-Kettering Cancer Center, 50% of patients with relapse had a locoregional component.118
Patterns of relapse by stage were analyzed in detail in a series of 130 patients who underwent resection with curative intent at the Massachusetts General Hospital.112 Locoregional relapse occurred as any component of failure in 49 patients (38%) and as the sole failure in 21 (16% of 130 patients at risk and 24% of the 88 with disease progression). The incidence of locoregional relapse by stage was in excess of 35% for modified Astler-Coller stages B2 (T3N0), B3 (T4N0), C2 (T3N1-3), and C3 (T4N1-3). The sites at higher risk for locoregional relapse included the gastric bed (27 of 130 patients, 21%) and the anastomosis or gastric remnant (33 of 130 patients, 25%). The true incidence of gastric bed, regional lymph node, and peritoneal failures may be higher because this was neither a reoperative nor an autopsy series (see comparative findings in Table 45-4). Additional information on patterns of relapse by stage exist in both the University of Minnesota reoperation analysis28,29 and the University of Washington autopsy analysis.115 Although patterns of relapse data are more accurate in such analyses, patient selection is biased.
Adjuvant Chemotherapy
The role of adjuvant chemotherapy in gastric cancer has been extensively studied over the past decades, being the first attempt to improve the prognosis of resected gastric cancer.* Results of randomized clinical trials of adjuvant single-agent chemotherapy after curative resection in gastric cancer generally have not shown any survival benefit when compared with surgical resection alone. For an expanded discussion of this topic, the reader is referred to the Expert Consult website.
Two phase III clinical trials evaluated the impact of combination chemotherapy with 5-FU, doxorubicin (Adriamycin), and mitomycin C (the FAM regimen) after curative surgery in resected high-risk gastric cancer.119,120 Coombs and associates119 included 315 patients, and no difference was observed in either OS (FAM 45% vs. control 35%) or relapse rates (FAM 56% vs. control 56%). In a Southwest Oncology Group study,120 no advantage to adjuvant treatment with the FAM regimen was reported either.
A Spanish trial121 compared high-dose mitomycin C after surgical resection (33 patients) versus surgery alone (37 patients). This small trial showed a significant 5-year OS advantage (p = .001) in the group receiving chemotherapy (76% vs. 30%), and a low rate of relapse in the treated arm (7/33 patients or 21%) versus the control arm (23/37 patients or 62%). This trial was extended to 134 patients with median follow-up of 8.75 years and still showed a survival advantage for postoperative mitomycin (p = .025). A subsequent randomized Spanish trial122 observed a significantly better 5-year OS for mitomycin plus tegafur (67%) versus mitomycin alone (44%) in resected high-risk gastric cancer, suggesting a strong benefit in patients with node-negative and early-stage disease.
Neri and colleagues123 treated 68 surgically resected, node-positive patients with the addition of adjuvant epirubicin, 5-FU, and leucovorin (the EFL regimen) and compared them with 69 surgery-alone control patients. The authors noted an improvement in 5-year OS from 13% to 30% for those receiving adjuvant chemotherapy.
Adjuvant chemotherapy with FAM2, seven cycles, was compared with surgery alone in a European Organization for Research and Treatment of Cancer (EORTC) randomized phase II trial that included 314 patients with stage II and III gastric adenocarcinoma.124 Five-year OS was 70% for stage II and 32% for stage III disease with no significant differences between groups. Toxicity was high with the FAM2 regimen, but time to disease progression was significantly longer in this group. DFS increased to a borderline significant value. Similarly, an adjuvant FAM randomized trial done by the Southwest Oncology Group, failed to show improved survival results.125
The EORTC and the International Collaborative Cancer Group (ICCG) conducted independent randomized phase III studies of either adjuvant FAMTX or FEMTX (5-FU, epirubicin, and MTX) compared with surgery alone.126 Both studies were closed when they failed to reach accrual after being open for 7 years. A pooled analysis of the two studies demonstrated that they both caused substantial toxicity but had no significant effect on OS, which was the primary endpoint of each study.
Perioperative Chemotherapy
In the British MAGIC trial, 503 patients with resectable adenocarcinoma of the stomach (372 patients), GE junction (58 patients), or lower esophagus (73 patients) were randomized to receive either surgery alone or surgery plus three cycles of preoperative epirubicin, cisplatin, and infusional 5-FU (ECF regimen) and three cycles of postoperative ECF therapy.127 The use of perioperative chemotherapy was associated with significant improvement in survival compared with surgery alone (median survival 24 vs. 20 months, 5-year OS 36% vs. 23%, p = .009). Overall, 103 patients completed all six cycles of chemotherapy (Table 45-5).
Meta-Analysis of Adjuvant Trials
Because many of the trials conducted to date have been underpowered to adequately assess potential differences between the control and treatment arms in regard to OS, seven meta-analyses have been performed since 1990 to better assess the use of adjuvant chemotherapy.128,129,130–133,134 Three recent meta-analyses showed a significant survival advantage to the use of chemotherapy after surgical resection of gastric cancer. Seventeen randomized clinical trials, involving 3118 patients, were included in the meta-analysis by Panzini and associates.128 The pooled odds ratio was 0.72 (95% confidence interval [CI], 0.62 to 0.84). The meta-analysis by Janunger and coauthors129 included 21 randomized clinical trials, involving 3962 patients, a slightly larger and somewhat different group of studies. The pooled odds ratio from this study was 0.84 (95% CI, 0.74 to 0.96). In a subanalysis, Western and Asian studies were assessed separately. The meta-analysis of the Western studies did not show a benefit to adjuvant therapy (pooled odds ratio of 0.96; 95% CI, 0.83 to 1.12), whereas the Asian studies did show evidence of benefit (pooled odds ratio of 0.58; 95% CI, 0.44 to 0.76). In 2010, a meta-analysis by Paoletti and colleagues134 identified 17 trials with individual patient data availability (3838 patients) and a median follow-up exceeding 7 years. There were 1000 deaths in the 1924 patients assigned to chemotherapy and 1067 deaths in 1857 patients assigned to surgery only. Adjuvant chemotherapy achieved a significant benefit in 5-year OS (49.6% vs. 55.3%, p = .001).
Adjuvant Irradiation
Irradiation has been only minimally evaluated as the sole adjuvant treatment after complete surgical resection in randomized phase III trials. Adjuvant EBRT reduced locoregional failures when compared with the surgery-alone control arm in the British adjuvant trial noted later, but no survival benefits were found.141,148 Although phase III trials from Japan149,150 and China151 suggest some survival benefit for IORT versus a surgery-alone control arm, the advantage was found only in subset analyses.
Postoperative External Beam Irradiation
The British Stomach Cancer Group (BSCG) completed a prospectively randomized trial of surgery alone versus postoperative FAM or EBRT (45 Gy in 25 fractions ± 5-Gy boost).141,148 A total of 436 patients were randomized and observed for a minimum of 12 months; the study arms were well balanced with regard to prognostic factors. No survival differences by treatment arm were seen (median, 15 months) (see Table 45-5). However, locoregional failure was documented in only 15 of 153 patients (10%) in the adjuvant irradiation arm versus 39 of 145 (27%) in the surgery-alone arm and 26 of 138 (19%) in the FAM group. Interpretation of the results is complicated by the inclusion of 93 patients with resection but gross residual disease and of 78 patients (18%) with gross total resection but microscopically positive resection margins. Neither group of patients would be candidates for current gastric surgical adjuvant trials in the United States. In addition, nearly one third of the patients randomized to receive adjuvant treatment did not receive the assigned therapy. Of 153 patients randomized to the irradiation study arm, only 104 (68%) received a dose of 40.5 Gy or more. Only 62% of the patients received six or more cycles of chemotherapy. The results of this study are similar to results seen in the adjuvant treatment of rectal cancer, in which adjuvant preoperative and postoperative irradiation improve local control but did not increase patient survival unless combined with chemotherapy.
Preoperative External Beam Irradiation
Three prospective randomized Russian trials have evaluated preoperative irradiation in potentially resectable gastric cancer.152–154 The first trial randomly assigned 293 patients to receive either surgery alone, surgery after preoperative EBRT (20 Gy in four fractions), or surgery after the same EBRT plus daily hyperthermia. The survival rates at 3 and 5 years were improved in both irradiation study arms compared with surgery alone, and the improvement with combined EBRT and hyperthermia was statistically significant at both 3 and 5 years.157 The second trial compared preoperative EBRT (20 Gy) versus surgery alone in 279 patients. Three- and 5-year survival rates were increased, and no increase in operative morbidity was observed.158 The third trial compared surgery alone versus preoperative EBRT (32 Gy with concomitant inhalation of oxygen) plus surgery. A survival advantage was observed with preoperative treatment, and the resection rate was increased by 17%.159 There are some methodologic uncertainties with all three of these trials, and their applicability to Western gastric carcinoma is not clear.
A double-blind, randomized trial from Beijing, conducted from 1978 to 1989, compared a surgery-alone control arm (n = 199) with preoperative EBRT plus surgery (n = 171) for patients with adenocarcinoma of the gastric cardia.155 Irradiation was given with 8-MV photons or cobalt with anterior-posterior/posterior-anterior fields to a dose of 40 Gy in 20 fractions of 2 Gy over 4 weeks. Surgery was performed 2 to 4 weeks after completion of irradiation. The addition of preoperative EBRT produced downstaging of disease and improvements in radical resection rates (radical resection rates of 80% vs. 62% with preoperative EBRT vs. surgery alone).
Survival and locoregional disease control were improved in the patients assigned to preoperative EBRT versus surgery alone. The 5- and 10-year OS were 30% versus 20% and 20% versus 13%, respectively (Fig. 45-2; p = .009 Kaplan-Meier log-rank) (see Table 45-5). The divergence in survival curves began in the first year of follow-up and persisted through 9 years. Locoregional disease control was also improved with combined modality treatment with local relapse rates of 39% versus 52% (p <.025) and regional node relapse rates of 39% versus 54% (p <.005). The rates of distant metastases were the same at 24% versus 25%. The improvements in survival and disease control (locoregional) were accomplished with no increase in treatment-related morbidity or mortality (operative mortality 0.6% vs. 2.5% with or without preoperative EBRT; intrathoracic leak rates were 1.8% and 4.2%, respectively).
Adjuvant Irradiation Plus Chemotherapy
Postoperative EBRT Plus Chemotherapy
Phase II Trials
Phase II single-institution gastric cancer trials that showed promise for postoperative adjuvant chemoradiation included series from Massachusetts General Hospital (MGH),156 Israel (Hadassah Medical Center),157 Thomas Jefferson University Hospital,158,159 the University of Pennsylvania,160 and the Mayo Clinic161 (see web-only Table 45-3). Gunderson and associates156 reported a median survival of 24 months and a 4-year survival of 43% in 14 patients from MGH who had complete resection of tumors with extension beyond the wall, nodal involvement, or both. Patients received postoperative EBRT (45 to 52 Gy, 1.8 Gy/day) plus concurrent 5-FU–based chemotherapy. Subsequent locoregional relapse was documented in 2 of 14 patients (14%), in contrast to a 42% incidence in similar high-risk patients treated with surgery alone.112 Information on the other phase II trials is provided in detail on the Expert Consult website.
In a gastric series from Thomas Jefferson University Hospital, 120 patients had surgical resection but were at high risk for relapse because of extension beyond the gastric wall, nodal metastases, or positive margins of resection.158 Seventy patients had surgery alone, and 50 received adjuvant therapy. Apparent improvements in local control as well as median and 5-year survival were noted with additional therapy. For patients with negative resection margins, 2-year local control with surgery alone was 55% versus 93% with adjuvant irradiation with or without chemotherapy (p = .03). For patients with T3 and T4 tumors and lymph node involvement, median survival was 9 months versus 13 months (surgery alone or plus adjuvant treatment) and 5-year OS was 4% versus 22% (p = .03). In a separate analysis at the same institution of 55 patients with cancers of the GE junction, local relapse with surgery alone was 74% versus 36% in patients with adjuvant EBRT ± chemotherapy (p = .0014).159 Survival trends for node-positive patients appeared better in patients who received adjuvant treatment (5-year OS 15% versus 0%, p >.01). In a University of Pennsylvania analysis, treatment intensification appeared to improve both disease control and survival. The incidence of local relapse with surgery alone was 75% (31 of 40) versus 24% with adjuvant irradiation (4 of 17) and 15% with adjuvant irradiation plus chemotherapy (4 of 27).160 Five-year survival trends favored adjuvant chemoradiation over surgery alone at 55% versus 31%. In a retrospective Mayo Clinic analysis, 63 patients received postoperative EBRT ± 5-FU after resection of carcinoma of the stomach or GE junction.161 Twenty-five of the 63 patients had complete resection with no residual disease but had high-risk factors for disease relapse (extension beyond gastric wall, 92% of the patients; involved nodes, 92%; both high-risk factors, 84%). Concurrent 5-FU ± leucovorin was given with EBRT in 84% of the 25 adjuvantly treated patients, but maintenance chemotherapy was given in only 20%. Locoregional control was achieved in 20 of the 25 (80%), with median survival of 19 months. Four-year OS was 31% in spite of the very poor prognostic factors in these 25 patients.
Phase III Trials: Mayo Clinic, U.S. GI Intergroup 0116
A prospective randomized trial was conducted at the Mayo Clinic162 and included 62 patients with a poor prognosis after complete resection of gastric cancers (Table 45-5). Patients were randomized to either surgery alone or surgery followed by EBRT plus concomitant 5-FU (37.5 Gy in 24 fractions over 4 to 5 weeks; 5-FU, 15 mg/kg/day 1 to 3 by IV bolus). A nonstratified, prerandomization scheme was used with a 2 : 3 ratio favoring treatment. Informed consent was requested only of the 39 patients randomized to treatment. Ten of the 39 patients refused further therapy and were observed. When analyzed by intent to treat, the adjuvant arm had statistically significant improvement in both relapse-free (RFS) and OS (5-year OS 23% versus 4%; p <.05). When patient outcome was compared by actual treatment received (29 adjuvant treatment, 33 surgery alone), 5-year OS still favored the adjuvant group (20% vs. 12%), but the differences were not statistically significant. The 10 patients who refused assignment to adjuvant treatment had more favorable prognostic findings than the other two groups of patients. When the two groups with equally poor prognostic factors were compared, 5-year OS was 20% versus 4%, with an advantage to those receiving adjuvant treatment. When analyzed by treatment delivered, locoregional relapse was decreased with adjuvant treatment (54% incidence with surgery alone vs. 39% with irradiation plus 5-FU).
Because of conflicting results in prior small phase III studies, a U.S. GI Intergroup trial (INT 0116) was initiated to evaluate postoperative combined 5-FU–based chemotherapy and irradiation to the gastric bed and regional nodes versus surgery only in resected but high-risk gastric cancer patients.163 Eligible patients had completely resected ACA of the stomach or GE junction with complete penetration of the muscularis propria (T2-4N0) or involved nodes (T1-4N1-3). After an en bloc resection, 556 patients were randomized to either surgery alone or postoperative combined modality therapy. This consisted of one 5-day cycle of 5-FU and leucovorin followed by concurrent chemoradiation (45 Gy in 25 fractions plus concurrent 5-FU and leucovorin, 4 days/week 1, 3 days/week 5) followed by two additional 5-day cycles of 5-FU and leucovorin given at 1-month intervals. Nodal metastases were present in 85% of patients. With median follow-up of 5 years, RFS at 3 years is 48% for adjuvant treatment and 31% for observation (p = .001); 3-year OS is 50% for treatment and 41% for observation (p = .005) (see Table 45-5). The median survival in the surgery-only group was 27 months, as compared with 36 months in the chemoradiation group. The median duration of RFS was 30 months in the chemoradiation group and 19 months in the surgery-only group (Fig. 45-3).
The results of this large randomized phase III U.S. GI Intergroup trial demonstrate a clear survival advantage to the use of postoperative chemoradiation in high-risk patients who have undergone resection of their disease.163 Furthermore, the results strongly support the integration of postoperative chemoradiation into the routine care of patients with curatively resected high-risk carcinoma of the stomach and GE junction. This approach is now viewed by many as the standard of care in the United States.
Quality control of irradiation field design in the INT 0116 trial was conducted during the cycle of chemotherapy given before the start of concurrent chemoradiation.164 The upfront quality control provided the mechanism to correct most of the major or minor deviations (35% incidence) in field design before the start of treatment and resulted in only a 6.5% final major deviation rate. Utilization of upfront quality control may have been a key factor in achieving a positive survival advantage for adjuvant chemoradiation.
Postoperative Chemoradiation Plus D2 Resection
The potential role of postoperative chemoradiation in D2 resected patients was evaluated in a series of 990 patients treated in Seoul, South Korea165 (Table 45-6 and see web-only Table 45-3). Surgery alone was used in 446 patients, and postoperative chemoradiation was given to 544 patients (disease and patient characteristics and method of chemoradiation paralleled the INT 0116 trial). Both disease control and survival were improved in patients who received trimodality treatment (median survival 95.3 vs. 62.6 months, 5-year OS 57% vs. 51%, p = .02; 5-year RFS 54.5% vs. 47.9%, p = .016; locoregional relapse 14.9% vs. 21.7%, p = .005). Five-year RFS and OS were consistently better with trimodality treatment for each stage grouping.