- In the past 5 to 10 years, completed clinical studies have helped to further define the optimal perioperative strategies for resectable GEJ/gastric cancer; however, most of these approaches have shown no benefit and have even refuted accepted treatments.
- In the perioperative chemotherapy setting, the additions of epirubicin to a fluoropyrimidine/platinum doublet or bevacizumab to chemotherapy are not beneficial.
- In the adjuvant setting for initially resected disease, the options are either chemotherapy alone or chemoradiation—in both cases, clinical equipoise exists as to the choice of a fluoropyrimidine with or without a platinum drug.
- Based on studies in esophageal/GEJ cancer and pending the results of studies such as TOPGEAR, ESOPEC, and Neo-AEGIS, there is a strong rationale to consider preoperative chemoradiation for gastric tumors that involve the GEJ.
In the past 15 years, completed phase III studies have clearly established that perioperative strategies improve outcomes in patients with locally advanced, resectable gastroesophageal junction (GEJ) and gastric cancers. Broadly speaking, the two approaches are pre- or perioperative chemotherapy and surgery or up-front surgery and either adjuvant chemotherapy or chemoradiation. In the United States and Western Europe, the strong preference is for perioperative chemotherapy, whereas East Asian oncologists prefer initial surgery and adjuvant treatment.
Recently, several phase III studies have attempted to refine these strategies in one of several ways, including incorporating additional chemotherapy drugs or targeted agents, varying the duration of chemotherapy, and adding chemoradiation to chemotherapy. With few exceptions, strategies that have attempted to augment current therapies have not yielded any benefit, although an indirect inference from some of these studies is that the addition of radiation to preoperative strategies may facilitate complete surgical resection for patients whose tumors involve the GEJ.
In Europe, perioperative chemotherapy is the predominant approach, based primarily on the phase III MAGIC trial performed in the United Kingdom.1 This trial randomly assigned 503 patients with gastric cancer (85% of whom had tumors in the GEJ/stomach) to three cycles (9 weeks) each of pre- and postoperative epirubicin, cisplatin, and 5-fluorouracil (5-FU; ECF) and surgery or surgery alone. Perioperative chemotherapy resulted in significant improvement in 5-year overall survival (OS; 36% vs. 23%, p = 0.009), establishing this regimen as a standard of care.
A similar degree of benefit was also noted in the contemporaneous French FFCD 9703 trial of 224 patients with esophagogastric adenocarcinoma (89% had GEJ/gastric tumors).2 Patients were randomly assigned to 18 weeks of perioperative 5-FU and cisplatin and surgery or surgery alone. Perioperative chemotherapy resulted in a significant improvement in 5-year disease-free survival (DFS; 34% vs. 19%, p = 0.003) and OS (38% vs. 24%, p = 0.02). Although comparisons between different clinical trials must be made cautiously, the survival benefit seen with 5-FU and cisplatin on this trial appears to be nearly identical to that seen with ECF in the MAGIC trial.
The benefit of the anthracycline—and the duration of preoperative therapy—has now been definitively disputed by the MRC OEO5 study, which has so far only been presented in abstract form.3 This study randomly assigned 897 patients with esophageal/GEJ adenocarcinomas to preoperative chemotherapy with either 6 weeks of 5-FU and cisplatin or 12 weeks of epirubicin, cisplatin, and capecitabine (ECX) chemotherapy. Although the pathologic response rate was improved in the ECX group compared with the 5-FU and cisplatin group (including a pathologic complete response [pCR] rate of 11% and 3%, respectively, in the patients who underwent surgery), there was no difference in median progression-free survival (1.78 years vs. 1.53 years, p = 0.058) or OS (2.15 years vs. 2.02 years, p = 0.86) between both groups. Based in part on this study, National Comprehensive Cancer Network guidelines now consider the use of anthracycline-containing three-drug regimens in this context to be a Category 2B recommendation (an appropriate intervention predicated on “lower-level” evidence); the use of a fluoropyrimidine plus a platinum doublet remains a Category 1 recommendation (a uniform consensus based on “high-level” evidence). Nevertheless, because this study did not enroll patients with mid and distal gastric cancer, some investigators argue that a benefit for an anthracycline in such tumors has not been definitively excluded. However, this is potentially a minor point because most gastric cancers in the West occur in the proximal stomach.
The OEO5 study has also raised the provocative suggestion that as little as 6 weeks of preoperative chemotherapy conveys the same survival benefit as 12 weeks of chemotherapy. Although this may be counterintuitive based on the MAGIC and FFCD studies, as well as other studies in gastric cancer that have administered 6 to 12 months of adjuvant chemotherapy,4,5 these results are not without precedent. Specifically, the Dutch CROSS trial found benefit for preoperative chemoradiation in patients with esophageal/GEJ cancer (approximately 25% of tumors involved the GEJ). Patients enrolled on the CROSS trial received only five weekly treatments of carboplatin and paclitaxel as the entirety of their systemic chemotherapy.6,7 The absolute improvement in OS seen in the CROSS study is also very much in the range of 10% to 15% seen in other positive phase III studies discussed here.
It is also important to note that although the MAGIC and FFCD studies aimed to deliver 18 weeks of perioperative chemotherapy, only 50% of patients who underwent preoperative chemotherapy and surgery were able to receive or complete adjuvant chemotherapy on both studies, further suggesting that most patients derived the survival benefit of chemotherapy from receiving much less than 18 weeks of treatment. The generally poor ability to deliver therapy after surgery was also seen in the recently presented Dutch CRITICS study (discussed later in this article) and would suggest, especially given the absence of a biologic rationale to split up the same systemic treatment with surgery, that future experimental strategies should focus on exclusively preoperative approaches.
Although adding epirubicin to the doublet is not beneficial, the German FLOT4 study is evaluating a triplet regimen of perioperative docetaxel, 5-FU, and oxaliplatin (FLOT) compared with ECF/ECX chemotherapy in 714 patients with GEJ/gastric cancer (NCT01216644).8 This approach is based ultimately on the proven, albeit modest, benefit of adding docetaxel to a fluoropyrimidine and platinum doublet for patients with metastatic gastric cancer.9 To date, preliminary data in 265 patients show a superior pCR rate for the FLOT regimen (15.6% vs. 5.8%, p = 0.015). Although the survival data are eagerly awaited, the results of OEO5 sound a note of caution, as the improvement in pCR rate in the ECX group did not translate into any survival benefit.
Finally, in another attempt to augment treatment, the UK ST03 study added bevacizumab, the antibody against VEGF, to perioperative ECX chemotherapy in 1,063 patients with esophagogastric adenocarcinoma.10 This strategy showed no benefit (3-year survival, 48.9% vs. 47.6% for ECX vs. bevacizumab/ECX), and it was also associated with a higher anastomotic leak rate in patients undergoing esophagogastrectomy (9% vs. 23%).
The results of these studies are summarized in Table 1.
In comparison to preoperative therapy, trials in East Asia of resectable gastric cancer have frequently focused on postoperative chemotherapy alone. To date, two large phase III trials have demonstrated a benefit for this approach. These data support the adjuvant use of a fluoropyrimidine as monotherapy or combined with a platinum agent.
The ACTS-GC study was performed in Japan. In this study of 1,059 patients with stage II-III gastric cancer who had undergone D2 resections, patients were randomly assigned to 1 year of adjuvant S-1 or observation.4 Five-year outcomes for this trial confirm that adjuvant S-1 is associated with significant improvements in 5-year relapse-free survival (65.4% vs. 53.1%; HR 0.65, 95% CI [0.54, 0.79]) and OS (71.7% vs. 61.1%; HR 0.67, 95% CI [0.54, 0.83]) compared with observation alone.11
The second of these is the CLASSIC trial, which was performed with 1,035 East Asian patients who had undergone a D2 resection of stage II-IIIB gastric cancer.5 Patients were randomly assigned to 6 months of adjuvant capecitabine and oxaliplatin (CapeOx) or observation. Updated survival data confirm improved 5-year OS for patients who received chemotherapy (78% vs. 69%; HR 0.66, p = 0.0015); 5-year DFS was also improved (68% vs. 53%; HR 0.58, p < 0.0001).12
Because both of these studies are associated with a similar 10% improvement in 5-year OS, it is not clear how much additional benefit comes from adding a platinum drug to a fluoropyrimidine. Some Asian investigators use the lymph node status to guide the choice of chemotherapy regimen. Long-term follow-up of the ACTS-GC study suggested a trend toward greater benefit for S-1 in patients with N0 (HR 0.317, 95% CI [0.127, 0.970]) compared with N1 (HR 0.608, 95% CI [0.440, 0.840]) or N2 (HR 0.839, 95% CI [0.612, 1.150]) tumors, although this was not statistically significant (p = 0.0882).11 On the other hand, a subset analysis of the CLASSIC study revealed that adjuvant CapeOx seemed to benefit only patients with N1 or N2 disease (HR 0.57, 95% CI [0.46, 0.72]) and not those with N0 tumors (HR 0.79, 95% CI [0.39, 1.60]).12 Again, this was not statistically significant (p = 0.35), and patients with N0 tumors made up only about 10% of the study population. It is, of course, essential to emphasize that the results of post hoc subset analyses that fail to meet statistical significance should remain hypothesis generating but should not be practice changing.
Since the completion of these seminal studies, several other trials have also attempted to augment treatment in the adjuvant setting by adding other chemotherapy drugs that have proven activity in the metastatic setting (Table 2), yet all have been uniformly negative.
The first of these to be published was the South Korean AMC0201 study, which enrolled 855 patients who had undergone D2 surgical resection for gastric cancer.13 Patients were treated with either two doses of mitomycin and doxifluridine (an oral 5-FU prodrug) daily for 3 months or one dose of mitomycin, followed by 6 monthly doses of cisplatin and doxifluridine daily for 12 months. There was no difference in the 5-year relapse-free survival (61.1% vs. 57.9%, p = 0.39) or OS (66.5% vs. 65%, p = 0.33) for the less- or more-intensified arms, respectively.
A lack of benefit for adding a taxane to a fluoropyrimidine was reported in the Japanese SAMIT study, which enrolled 1,433 evaluable patients with T4a or T4b gastric tumors who had undergone initial surgery.14 Patients were randomly assigned to either an oral fluoropyrimidine alone or paclitaxel preceding it (a second random assignment was between the oral 5-FU prodrugs S-1 and tegafur-uracil). There was no improvement in 3-year DFS for the group that also received a taxane (57.2% vs. 54.0%, p = 0.273), although there was a non-statistically significant trend toward benefit for the taxane arm in patients with more advanced disease (stage IIIB, by the Japanese classification).
Finally, the Italian ITACA study randomly selected 1,106 patients with gastric cancer who had undergone a D1 or more surgical resection to receive infusional 5-FU and leucovorin every 2 weeks for nine cycles or bolus and infusional 5-FU, leucovorin, and irinotecan (FOLFIRI) every 2 weeks for four cycles followed by cisplatin and docetaxel every 3 weeks for three cycles.15 No difference was noted in either the 5-year DFS (44.6% vs. 44.6%, p = 0.974) or OS rates (50.6% vs. 51.0%, p = 0.865) in the standard- and intensified-chemotherapy arms, respectively.
In the United States, the standard of care for GEJ/gastric cancers undergoing up-front resection is postoperative chemoradiation, based primarily on the results of the Intergroup 116 trial.16 This trial randomly assigned 556 patients to adjuvant chemotherapy and chemoradiation with bolus 5-FU and leucovorin or observation alone following surgery. Patients who received adjuvant chemoradiation had an improvement in 3-year relapse-free survival (48% vs. 31%, p < 0.001) and 3-year OS (51% vs. 40%, p = 0.005).
Based on the results of the Intergroup 116 trial, the Cancer and Leukemia Group B (CALGB) launched and completed the 80101 trial, which enrolled 546 patients with gastric cancer. The standard arm consisted of systemic bolus 5-FU and leucovorin preceding and following chemoradiation with infusional 5-FU, while the experimental arm intensified the systemic chemotherapy by replacing the bolus 5-FU and leucovorin with the ECF regimen. Results have been presented only in abstract form and reveal no improvement in 3-year DFS (47% vs. 46%) or OS (52% vs. 50%) with the addition of an anthracycline and platinum compound to 5-FU.17 These results are also virtually identical to the outcomes in the adjuvant chemoradiation arm of the Intergroup 116 trial.
The widespread adoption of the perioperative chemotherapy approach used in the MAGIC study has led to a decline in the use of adjuvant chemoradiation advocated by the Intergroup 116 study. In addition, despite its positive results, this trial is frequently criticized because of the relatively inadequate surgical resections that were performed; 54% of patients had a D0 resection, which is less than an optimal resection of regional lymph nodes. It has been argued that radiation in this setting potentially compensated for inadequate surgery because the greatest impact of adjuvant chemoradiation was a reduction in local recurrence of cancer. Such benefits may not be seen for radiotherapy if a more complete D1 or D2 surgical resection is undertaken.
Investigators attempted to answer the question of whether postoperative radiation provides a benefit for patients who have undergone a D2 gastrectomy in the South Korean ARTIST trial. This study randomly assigned 458 patients with stage IB-IV gastric cancer who had undergone D2 resections to either six cycles of adjuvant chemotherapy with capecitabine and cisplatin or to two cycles of capecitabine and cisplatin before and after chemoradiation with capecitabine.18
In the overall population, patients in the chemoradiation arm had a non-statistically significant trend toward improved 3-year DFS (78.2% vs. 74.2%, p = 0.09). In a subgroup analysis of 396 patients with lymph node–positive disease, there was a statistically significant improvement in 3-year DFS in the chemoradiation arm (77.5% vs. 72.3%, p = 0.04). In an updated analysis, the benefit seen only in the lymph node–positive group was confirmed but appeared to be restricted only to tumors with intestinal histology by Lauren classification (and not tumors with diffuse histology).19 The results of these adjuvant studies are summarized in Table 2.
Unfortunately, the results of the ARTIST trial do not provide definitive evidence for incorporating radiation into adjuvant therapy for patients with optimally resected disease that has lymph node involvement. The finding that radiation appears to benefit patients with lymph node–positive disease is somewhat counterintuitive because these patients are presumed to be at greater risk for developing distant metastases than patients with lymph node–negative disease and might, therefore, be expected to derive less benefit from an approach designed to improve locoregional control. Finally, even if one were to adopt a strategy of adjuvant chemoradiation for this population, it is entirely unclear that the systemic chemotherapy should consist of a fluoropyrimidine and platinum doublet because the negative CALGB 80101 study has shown no benefit to adding cisplatin (and an anthracycline) to a fluoropyrimidine (although, admittedly, a subset analysis for this study of patients with lymph node–negative vs. lymph node–positive tumors has yet to be presented).
Based on the results of the ARTIST study, a follow-up study (ARTIST-II; NCT01761461) is ongoing for patients with resected lymph node–positive disease. In addition to patients being randomly assigned to chemoradiation or chemotherapy alone, the systemic chemotherapy will consist of the 5-FU prodrug S-1 alone or with the platinum compound oxaliplatin. Hopefully, this study will, therefore, also address the benefit of adding a platinum drug to a fluoropyrimidine in the adjuvant setting, given the similar outcomes of the ACTS-GC and CLASSIC studies.
Finally, the Dutch CRITICS study has evaluated an alternative to the perioperative ECF approach by randomly assigning patients with GEJ/gastric cancer to perioperative ECX or epirubicin, oxaliplatin, and capecitabine (EOX) chemotherapy (the control arm) or preoperative ECX/EOX and adjuvant chemoradiation with capecitabine; all patients underwent D1 or D2 resections (Table 1). The results were recently presented and revealed no difference in 5-year progression-free survival (38.5% vs. 39.5%, p = 0.99 for chemotherapy alone vs. adjuvant chemoradiation) and OS (40.8% vs. 40.9%, p = 0.99) for either treatment arm.20 Although subgroup analyses are planned, the nearly superimposable Kaplan-Meier survival curves confirm, for now at least, that patients who have received preoperative chemotherapy should not receive adjuvant chemoradiation in a standard fashion.
Completed studies that have enrolled patients with gastric cancer (which include tumors that involve the GEJ) have not included preoperative chemoradiation, which is a standard of care in esophageal/GEJ cancer, based on several studies, including the CROSS study (Table 1).6 However, the results of the contemporary MAGIC, OEO5, and ST03 studies have indirectly raised the question of whether radiation is needed for tumors that involve the GEJ to facilitate an R0 surgical resection.
The R0 resection rates in the 1,400 patients treated in the contemporary MAGIC and OEO5 studies were consistently less than 70%; in the OEO5 study, the majority of patients were carefully assessed to be surgical candidates based on preoperative endoscopic ultrasound, PET scan, and laparoscopy.
The ST03 study of 1,063 patients actually reported an R0 resection rate of less than 60% in the intention-to-treat population. It also offered an informative comparison on the R0 resection rates based on the location of the primary tumor in patients who underwent surgery; it was 87% for gastric tumors but 60%, 71%, and 75%, respectively, for Siewert type I, II, and III tumors. Based on this classification, type I tumors involve the lower esophagus and GEJ, type II tumors are “true” GEJ tumors, and type III tumors arise in the gastric cardia and involve the GEJ.21 These results suggest that patients with greater tumor involvement of the GEJ and/or esophagus experience higher rates of incomplete resection than patients with pure gastric tumors when treated with preoperative chemotherapy.
Although none of these studies stratified outcomes based on R0 or R1 surgery, older studies suggest poor outcomes for patients with incomplete resection. Specifically, the USA Intergroup 113 study, a negative study that evaluated perioperative 5-FU and cisplatin compared with surgery alone for esophageal/GEJ cancer, noted dismal outcomes in the patients who underwent R1 resections (3-year survival 12%), comparable to patients who had palliative R2 resections or no surgery at all (4% each) and much worse than patients with R0 resections (39%).22 The only long-term survivors from the group of patients who had undergone R1 resections were patients who received adjuvant chemoradiation. Similarly, the UK OEO2 study of preoperative 5-FU and cisplatin for esophageal/GEJ cancer (which, unlike USA Intergroup 113, was a positive study with 6% improvement in 5-year OS) also reported inferior outcomes for patients who had undergone R1 versus R2 versus no resection versus R0 resections (3-year OS 18.0% vs. 8.6% vs. 1.4% vs. 42.4%, respectively).23
Therefore, the incorporation of preoperative chemoradiation for Siewert type I-III GEJ tumors might improve the R0 resection rate and offer a long-term survival benefit. Evidence for this comes from the CROSS study of esophageal/GEJ tumors, where chemoradiation improved the R0 resection rate to 92% in the patients who underwent surgery compared with 69% in the surgery-only group,6 a number strikingly similar to the R0 resection rate for GEJ tumors in the ST03 study.
The possible superiority of preoperative chemoradiation over preoperative chemotherapy has been suggested by the German POET study, in which patients with GEJ adenocarcinomas were randomly assigned to either 5-FU, leucovorin, and cisplatin followed by surgery or 5-FU, leucovorin, and cisplatin followed by chemoradiation with cisplatin and etoposide and then surgery.24 Only 119 eligible patients were randomly assigned before the trial was closed because of poor accrual, limiting the power of this study to detect a difference between the treatment groups. Nevertheless, there were strong trends toward an improvement in local control (76.5% vs. 59.0%, p = 0.06) and in 3-year OS (47.4% vs. 27.7%, p = 0.07) for the chemoradiation group.
Several ongoing studies are attempting to definitively answer the relative benefit of perioperative chemotherapy compared with preoperative chemoradiation with carboplatin and paclitaxel (CROSS). The perioperative chemotherapy regimen in the German ESOPEC study (NCT02509286) is FLOT chemotherapy, whereas the Irish Neo-AEGIS study (NCT01726452) administers ECF or ECX. The Australian TOPGEAR trial (NCT01924819) is adopting a complementary strategy of comparing MAGIC-style chemotherapy with preoperative ECF/ECX and chemoradiation prior to surgery and adjuvant ECF/ECX.
In the past 5 to 10 years, completed clinical studies have helped to further define the optimal perioperative strategies for resectable GEJ/gastric cancer. Unfortunately, most of these approaches have shown no benefit and have even refuted accepted treatments.
In the perioperative chemotherapy setting, it is now clear that the additions of epirubicin to a fluoropyrimidine and platinum doublet or bevacizumab to chemotherapy are not beneficial. It is also clearly established that adjuvant chemotherapy after surgery, following preoperative chemotherapy, is poorly tolerated and possible in only half of patients. The optimal duration of chemotherapy remains unclear but may not exceed 6 weeks of preoperative treatment. The value of adding a taxane to the doublet regimen will have to await presentation of the FLOT4 results.
In the adjuvant setting for initially resected disease, the options are either chemotherapy alone or chemoradiation. In both cases, clinical equipoise exists as to the choice of a fluoropyrimidine with or without a platinum drug, although some Asian oncologists propose choosing treatment based on lymph node status. In particular, systemic chemotherapy and chemoradiation may offer a benefit over chemotherapy alone for patients with lymph node–positive intestinal-type gastric cancer. The augmentation of systemic chemotherapy—with a taxane, irinotecan, and/or anthracycline—is not clearly beneficial. Similarly, there is no routine role for adjuvant chemoradiation following preoperative chemotherapy.
Finally, for the more than 2,000 patients with gastric tumors that involve the GEJ who were treated collectively in the MAGIC, OEO5, and ST03 studies, preoperative chemotherapy alone is associated with unacceptably low R0 resection rates. Based on studies in esophageal/GEJ cancer and pending the results of studies such as TOPGEAR, ESOPEC, and Neo-AEGIS, there is a strong rationale to consider preoperative chemoradiation for such tumors.
Although these results in totality are certainly disappointing, they do serve to direct and guide future strategies. For example, future clinical trial design should focus on preoperative approaches only based on improved tolerance and treatment delivery, even as it is defensible in current standard clinical practice to consider such an approach.
It is beyond the scope of this editorial, but there are numerous, innovative approaches that are being planned or are ongoing in this context, including the use of PET scans to direct therapy (e.g., the Alliance A021302 study; NCT02485834), the introduction of molecular subtyping of gastric cancer based on The Cancer Genome Atlas,25 and the addition of immune checkpoint inhibitors. As we forge ahead with cautious optimism, we must acknowledge the essential lessons from these supposedly negative studies and incorporate them into future clinical trial design so that we may avoid having to relearn these lessons.
About the Authors: Dr. Ku is a medical oncologist in the Gastrointestinal Oncology Service, Department of Medicine, at Memorial Sloan Kettering Cancer Center. Dr. Ilson is a medical oncologist in the Gastrointestinal Oncology Service, Department of Medicine, at Memorial Sloan Kettering Cancer Center.