Radiotherapy for tumors of the stomach and gastroesophageal junction – a review of its role in multimodal therapy

Whether GEJ adenocarcinomas should be classified as gastric or esophageal carcinomas or as an own entity is an issue that has been discussed for years[12‐14]. The classification proposed by Siewert et al.[13] distinguished three types of GEJ carcinoma, Type I being considered as adenocarcinoma of the distal esophagus, infiltrating the GEJ from above, Type II being the true carcinoma of the cardia arising immediately at the GEJ, and Type III, the subcardial gastric carcinoma that infiltrates the GEJ from below. Although this classification was based purely on anatomical/topographic parameters, it was accepted by most surgeons[15, 16]. In the 7^th American Joint Committee on Cancer (AJCC) staging manual[17, 18] GEJ cancer is now classified as esophageal cancer even within the first 5 cm of the stomach if it invades the junction, meaning that in addition to Siewert I, the former Siewert II and even Siewert III lesions are now both classified as esophageal carcinoma[17]. This new classification which is currently not accepted by all authors[16] makes an appropriate interpretation of results from therapeutic trials difficult, as GEJ carcinomas that were formerly classified as gastric carcinoma are now staged as esophageal carcinoma.

Taken together, there is an ongoing discussion about classification of GEJ carcinomas as gastric or esophageal cancer. They were included in clinical trials on both cancers. Therefore detailed knowledge of the respective population in each trial is necessary when results of esophageal or gastric cancer trials are applied to GEJ adenocarcinoma patients.

With regard to the high incidence of gastric and the increasing incidence of GEJ carcinoma, it comes as a surprise that therapeutic approaches differ widely in terms of surgical approach especially concerning extended vs. limited lymph node dissection[19‐22], and (neo-)adjuvant therapy. Basically, according to the classification by the Japanese Gastric Cancer Association (JGCA), most perigastric lymph nodes are defined as group 1, whereas the distant perigastric nodes, the nodes along the common hepatic artery, splenic artery, celiac axis and proper hepatic artery are defined as group 2[23]. In the discussion about lymph node resection, most surgeons of the Western world favored a limited en bloc resection of group 1 nodes with the stomach (D1) because of increased perioperative mortality associated with the additional resection of group 2 nodes (D2)[19, 21].

In contrast to the trials in the West, in Eastern trials, extended resection (D2) approaches for gastric carcinoma were applied with good results since the 1960s[23]. Current data from Western populations are indicating an emerging consensus that D2 resection with sparing of pancreas and spleen (also called D1+) is not associated with increased perioperative mortality[20, 22]. Although a recent update of the Dutch D1D2 trial[20] indicated a significantly decreased gastric cancer-related death rate in the D2 group as compared to D1, it remains to be proven whether D2/D1+ vs. D1 resection in Western populations significantly increases overall survival (OS)[24].

When it comes to lymphadenectomy in esophageal carcinomas of the lower third and in GEJ adenocarcinomas, two-field lymphadenectomy with dissection of lymph nodes in the lower half of the chest and in the upper abdominal compartment has been widely accepted as a compromise between the benefits of extended surgery and postoperative morbidity[25].

In gastric- and GEJ carcinoma, therefore, the best surgical approach is still discussed, especially regarding the extension of lymph node dissection. Although surgical techniques have been improved greatly during the last years, there is still a high rate of failure in local control[26] as well as a high rate of distant metastases, stressing the importance of systemic and local therapeutic approaches in addition to surgery.

Discussions about therapeutic approaches in addition to surgery in gastric and GEJ carcinoma patients have emerged since the first trial in a randomized setting, US intergroup study (INT)-0116[27] demonstrated a significant improvement in OS and disease free survival (DFS) in patients with advanced gastric and junctional (20%) cancer with adjuvant radiochemotherapy (CRT) vs. surgery alone. This study was heavily criticized due to the high rate of D0 dissections (see Table1 for details); however, benefits on OS of the experimental arm of the INT-0116 treatment protocol were soon confirmed in a non-randomized trial on a population of D2-resected Eastern patients by Kim et al.[28], and extended to other CRT protocols by further randomized trials as detailed below. Recently, an update of the INT-0116 trial showed that results remained stable in the long term[29]. The authors additionally provided information on the risk of secondary malignancies in the treatment arm. Although they observed a higher incidence of second tumors in the CRT arm, Smalley et al. consider this being biased by: 1. improved survival in the CRT arm, and: 2. completeness of reporting for second primaries in the treatment arm. Therefore, it seems reasonable to conclude that benefits of CRT outweighed the risk of second malignancies in this setting[29, 30].

In contrast to the INT-0116 study, three randomized trials on Western patients, randomized to adjuvant CT with different therapeutic regimens (but without radiation) vs. surgery alone failed to significantly increase survival in stomach cancer[31‐33]. Despite the disappointing results on adjuvant CT, Cunningham et al. demonstrated the superiority of perioperative CT vs. surgery alone. The authors enrolled a mixed population of gastric (74%), GEJ (15%) and esophageal (11%) cancer patients in the Medical Research Council Adjuvant Gastric Cancer Infusional Chemotherapy (MAGIC) study[34].

Patients were randomized to surgery alone vs. perioperative CT with ECF, a combination of fluorouracil (FU), cisplatin and epirubicin, administered as described in Table1. The MAGIC trial was the first to convincingly demonstrate a survival benefit using CT in Western gastric carcinoma patients. Nevertheless, it has been criticized for similar reasons as the INT-0116 trial: the majority of patients received potentially insufficient lymphadenectomy (see Table1). Therefore some authors argued that CRT or perioperative CT only compensated for inadequate surgery[35]. In addition, many authors have criticized the MAGIC trial for the persistence of methodological biases related to poor staging accuracy[36], heterogeneity of tumors included, as well as the lack of quality control in the surgical approaches[35, 37]. Based on the results of the MAGIC study, perioperative CT has been widely adopted and included in guidelines[38] in Europe while CRT as performed in INT-0116 trial represents the therapy standard in most centers in the US[39].

A detailed comparison of the INT-0116 study and the MAGIC trial is provided in Table1; this comparison shows that OS in INT-0116 was about 42% after five years in the CRT arm while 36.3% of patients were alive after three years in the MAGIC study. The survival advantage of INT-0116 vs. MAGIC was coherent in terms of DFS, and was despite the fact that: 1. patients had more advanced carcinomas in INT-0116 trial (see Table1), and: 2. patients in the MAGIC trial received D2 resection in >40% of cases vs. 10% in INT-0116. On the other hand, by study design, patients included in INT-0116 trial had already undergone curative resection at randomization. Patients in the MAGIC trial were randomized before any treatment. Cunningham et al. reported that surgery “considered curative by the operating surgeon” was performed in 69.3% of patients in the CT arm, and 66.4% in control arm[34]. To our best knowledge, the exact rate of R0, R1 and R2 resections has never been reported by the authors, nonetheless the rate of 69.3% was recognized by some authors as R0 resection rate in the MAGIC trial[29, 40, 41]. According to a statement taken from personal communication published by Smalley et al. in their update on the INT-0116 trial, D. Cunningham stated that R1 resection was especially common in the GEJ carcinomas[29].

The differences in study design between both trials must be emphasized as they hamper any reasonable comparison and explain the worse OS in the control arm of the MAGIC trial (as compared to INT-0116) despite the higher rate of D2 resections performed in the study population.

Therefore a direct comparison between MAGIC and INT-0116 is not possible. This explains the many different interpretations in the literature and the different therapeutic approaches after publication of the MAGIC trial in the US and in Europe.

While the superiority of CRT / CT vs. surgery alone in Western trials might be due to inadequate surgery, this is certainly not the case in studies on Eastern populations. As already mentioned, Kim et al.[28] confirmed the survival advantage of adjuvant CRT in a non-randomized population of D2-dissected patients.

In addition, the advantage of adjuvant therapeutic regimens vs. surgery alone was confirmed by Eastern randomized trials. First, Sakuramoto and coworkers reported on S-1 (TS-1), an orally active combination of tegafur, gimeracil, and oteracil. The authors reported in the Adjuvant Chemotherapy Trial of TS-1 for Gastric Cancer (ACTS-GC) on 1059 randomly assigned patients with gastric carcinoma who underwent D2 surgery with or without adjuvant S-1[42]. In accordance with the other trials they found a significant increase in OS and DFS[43] (see Table1 for details). Nakajima et al. reported on 190 early stage gastric carcinoma patients (T2N1-2) randomized to adjuvant uracil-tegafur or surgery alone[44]. All patients received extended lymph node resection and both OS and DFS were significantly better in the CT group (HR 0.46, 13% difference in survival at 4 years).

The latest Eastern results were provided by the capecitabine and oxaliplatin adjuvant study in stomach cancer (CLASSIC), 1035 patients in South Korea, China, and Taiwan were randomized to adjuvant CT vs. surgery only[45]. The difference in 3-year OS was minor compared to other studies (83% vs. 78%, p=0.0493), nevertheless there was a significant and considerable difference in DFS; details are provided in Table1.

In the last years, therapeutic approaches on GEJ carcinoma patients were essentially based on results derived from the INT-0116[27] and the MAGIC[34] trial. Standard of care in the US was mainly adjuvant CRT as described in the intergroup protocol while European oncologists and surgeons preferred the perioperative CT as described in the MAGIC protocol[46]. However, given the shortcomings of both, the low proportion of GEJ carcinoma patients included, and the current discussion about a closer relation of GEJ to esophageal carcinoma[17], it is now time to reconsider therapeutic strategies for GEJ tumors. This applies even more as evidence shows that adjuvant therapeutic strategies (CT and CRT) in esophageal carcinoma patients have led to disappointing results[47‐50].

In contrast to the results in gastric carcinoma, neoadjuvant regimens seem to surpass adjuvant strategies in esophageal carcinoma patients[51]. Recently, more evidence for neoadjuvant treatments was provided by the OEO2 trial[52] which showed a significant survival benefit (5-year OS 23% vs. 17.1%; p=0.003) with neoadjuvant CT and surgery vs. surgery alone, nevertheless a similar trial in the US (RTOG trial 8911, US Intergroup-113) could not demonstrate a survival advantage[53].

Notwithstanding these contradictory results in single trials, meta-analyses[51, 54, 55] clearly indicate survival benefits of neoadjuvant CRT and CT regimens as compared to surgery alone in locally advanced esophageal and GEJ carcinoma. When interpreting these findings, it must be considered that the mentioned analyses include a majority of esophageal SCC[47‐51, 54, 55], therefore there is still an ongoing discussion which results and therapeutic approaches may be extrapolated on GEJ adenocarcinoma[41]. A trial reported by Walsh et al. in 1996 strongly suggested a superiority of neoadjuvant CRT and surgery vs. surgery alone in patients with (only) adenocarcinoma of the GEJ, the esophagus or the cardia. The authors randomized 113 patients with adenocarcinoma of the GEJ, the esophagus, or the cardia to neoadjuvant CRT (40Gy + FU + Cisplatin) and surgery vs. surgery alone[56]. Multimodal therapy resulted in significantly improved OS vs. surgery alone (median survival in CRT and surgery arm was 16 months, as compared with 11 months in the surgery only arm; p=0.01).

In essence, therapeutic approaches in GEJ carcinoma patients have recently been dominated by the protocols of the two landmark trials (INT-0116 and MAGIC). Although there was evidence in favor of neoadjuvant regimens, similar to esophageal therapeutic strategies, patients with GEJ adenocarcinoma were treated either perioperatively or with an adjuvant approach. The basic question if radiation therapy in addition to surgery and chemotherapy should be applied in GEJ- and in gastric cancer therapy has therefore been raised again by recent results.

As mentioned above, meta-analyses showed neoadjuvant therapy to improve OS compared with surgery alone for esophageal cancer, including GEJ. A benefit for neoadjuvant CRT vs. CT alone has been suggested but was not clearly demonstrated[51].

In the discussion about CRT vs. CT as a neoadjuvant therapy, evidence in favor of CRT has emerged during the last years. A phase III trial conducted by Stahl et al. who compared a total of 126 patients receiving either neoadjuvant CT followed by surgery or neoadjuvant CRT followed by surgery[57]. Although the study was closed early and differences were not statistically significant, results showed a statistical trend to a survival advantage for CRT as compared to CT in GEJ adenocarcinoma (3-year OS 27.7% vs. 47.4%, for details see Table2). Another recently published study is the ACCORD07 phase III trial[58]; the authors reported on a significantly improved survival by CT without irradiation as compared to surgery alone, their regimen was associated with marked general toxicity (see Table3 for comparison) their survival rates also being worse than reported by other recently reported CRT trials in comparable populations (see Table2). Strong evidence in favor of neoadjuvant CRT was provided by van Hagen and coworkers. The authors recently reported on 368 patients with esophagus and junctional carcinoma in the CROSS trial[59]. Patients with T1N+ or T2-T3N0-1 tumors were included, the majority presenting with adenocarcinoma of the distal or junctional esophagus. Patients were randomly assigned to CRT followed by surgery, and to surgery alone. CRT consisted of carboplatin and paclitaxel as well as a total radiation dose of 41.4Gy given in 23 fractions of 1.8Gy each. The planning target volume (PTV) included the primary tumor with radial margins of 1.5 cm and proximal and distal margins of 4 cm. In addition, any enlarged lymph nodes were included. If the tumor extended into the stomach, a distal margin of 3 cm was chosen; further details are summarized in Table2. Patients underwent surgery after CRT as soon as possible (median time: 6.6 weeks), and patients in the surgery group were treated as soon as possible after randomization. Interestingly, the authors found no significant difference in surgical complications or postoperative mortality between CRT and surgery only groups. Surgeons achieved significantly more R0 resections in the CRT group as compared to the surgery only group (92% vs. 69%; p<0.001). After completion of a median follow-up time of 45.4 months for surviving patients, an intention to treat analysis showed a median OS of 49.4 months in the CRT arm vs. 24.0 months in the surgery only group (for details see Table2). These results are markedly better than those achieved by other therapeutic approaches. The authors reported that the survival difference was significant for all histological subgroups and the benefit of CRT was consistent across subgroups without any significant interaction identified. Toxicity of CRT was very low in terms of hematologic and non-hematologic side effects (see Table3 for details).

In summary, recent data in GEJ carcinoma indicate a survival benefit of neoadjuvant CRT and CT in curatively resectable patients. Due to a trend for better survival of CRT vs. CT as reported by Stahl and coworkers and because of the substantially longer survival reported in the CROSS trial, CRT as performed in the CROSS trial should be considered as a standard treatment of patients with potentially curatively resectable GEJ carcinoma in tumor stages >T1N0.

Recent trials in Eastern populations include the already mentioned CLASSIC trial which demonstrated in a large population in South Korea, China, and Taiwan, that capecitabine and oxaliplatin in an adjuvant setting after D2 gastrectomy is significantly associated with longer DFS when compared to surgery alone (see Table1). In another Eastern population, Lee et al. conducted the ARTIST trial to compare CT vs. CRT; the concept being similar to the above mentioned Stahl trial, which was conducted in GEJ, to demonstrate superiority of CRT and surgery vs. CT and surgery in gastric carcinoma[60]. Although DFS was not significantly increased by CRT vs. CT, the authors observed a statistical trend to better DFS with CRT (78.2% vs. 74.2% after 3 years). In a subgroup analysis of patients with positive pathologic nodes, the advantage on DFS for CRT vs. CT was significant; after three years, a difference of 5% more disease-free patients was achieved by adjuvant CRT as compared to CT (p=0.0471; details are summarized in Table1).

Although subgroup analyses should be interpreted with caution, the high number of patients provides in our opinion sufficient power in favor of adjuvant CRT in patients with positive lymph nodes. For the whole population in the ARTIST trial, it must be taken into account that survival curves crossed after about 24 months. This indicates that better local tumor control by CRT as compared to CT outweighs toxic effects only after a sufficient observation time. After three years there are still a considerable number of patients at risk for relapse; so effects on tumor control might lack significance in this early analysis as compared to long-term observations. The secondary endpoint (OS) was not (yet) analyzed by the authors due to a low number of events[60]. The acute toxicity profile of both arms was comparable with only slightly more grade 3 neutropenia in the CRT arm and a higher rate of patients with grade 2–3 hand-foot syndrome in the CT group (13.7% vs. 10.6%; see Table1 for further details). On the basis of these data, the ARTIST-2 trial was proposed that should compare CT vs. CRT in patients with node-positive gastric cancer[60].

Interestingly, Zhu et al. recently reported on a similar trial conducted in a Chinese population. As further detailed in Table1, the authors compared adjuvant CT alone to adjuvant CRT with intensity-modulated radiotherapy (IMRT) in a randomized setting consisting of 380 curatively D2-resected patients[61]. When compared to Lee et al. (ARTIST;[60]), Zhu and coworkers observed a significant difference in DFS between the therapeutic regimens not only in patients with positive nodes but in the whole population after an observation time of at least five years or until death of the patient.

Although a direct comparison between the populations in terms of TNM-stage is hampered by different staging patterns (AJCC 6^th vs. 7^th version), it is most likely that the stronger effect on DFS as compared to the ARTIST trial is due to the: 1. more advanced patients, especially in terms of lymph node involvement; this is also suggested by the increased rate of patients in the Chinese trial as compared to the ARTIST study, who died mainly because of local recurrence (Zhu et al.: 19.7% vs. ARTIST: 6.6%) or: 2. the longer observation time, as described above, the observation time in the Korean trial might be relatively short, considering the high number of patients still at risk for relapse.

Both the data provided by the ARTIST trial and by the Chinese study lead to the assumption that gastric carcinoma patients benefit the more from CRT as compared to CT, the more advanced their carcinomas are. This assumption is further supported by a small trial reported by Yu et al., the authors described in 68 randomly assigned advanced gastric carcinoma patients (T3-4 and/or N+) a significant superiority of adjuvant CRT with 45Gy irradiation applied by IMRT and FU/FA as compared to adjuvant FU/FA only (3-yr OS: 67.7% vs. 44.1%; p=0.037)[62]. Although the collective in this study was supposedly underpowered, results add to the hypothesis that more advanced gastric carcinoma patients benefit from CRT as compared to CT. Additionally the Chinese trials emphasize the need for further implementation of IMRT in radiotherapy of gastric carcinomas.

If these encouraging results can be reproduced in Western populations is undetermined. As mentioned above, there are considerable inconsistencies in terms of clinical results between trials conducted on Western patients and trials on Eastern patients. It is not clear if these differences were caused by differences in tumor biology or by other epidemiological factors.

In Western patients, there are currently no large randomized post-MAGIC results on CRT vs. CT in gastric carcinoma patients; retrospective data provided by Dikken et al.[63] on a European population of 91 gastric cancer patients undergoing CRT even suggested D2-dissected patients in this population (n=25) did not benefit from adjuvant CRT in terms of survival as compared to surgery only. Given this inconsistency with earlier results and the underpowered population size, the authors proposed a large randomized trial to compare CRT vs. perioperative CT in Western patients with gastric cancer. This currently ongoing study is the CRITICS trial (NCT00407186;[40]) which compares perioperative CT (epirubicin, cisplatin, capecitabine; ECC) and surgery, with neoadjuvant CT, followed by surgery and adjuvant CRT (45Gy in 25 fractions; concurrent capecitabine and cisplatin).

Taken together, superiority of CRT vs. CT in gastric carcinoma patients is suggested by Eastern trials, especially for advanced but curatively resectable patients. Nevertheless, the situation is not as clear as in GEJ carcinomas, because perioperative CT as performed in the MAGIC trial has until now not been compared to CRT in a randomized setup. The most important ongoing trial to address this topic is the CRITICS trial, which is estimated to complete accrual in December 2013.

Given the results of the Korean and the Chinese trials, and the difficult comparison of the INT-0116 and the MAGIC trial, CRT seems to be at least equal to CT in gastric cancer (excluding GEJ carcinoma). In Eastern populations, recent results strongly indicate a longer DFS by CRT as compared to CT in advanced patients (especially N+); if these data can be reproduced in Western patients will be clarified by the CRITICS trial. Another reason for the good results of the Chinese trial might be the utilization of IMRT which successively replaces 3D conformal radiotherapy (3DCRT) in gastric and GEJ carcinoma treatment.

As mentioned above, local recurrence contributes to a worse prognosis in gastric and GEJ carcinoma. External beam radiotherapy (EBRT) in an adjuvant setting for gastric carcinoma or in a neoadjuvant setup for esophageal and GEJ adenocarcinoma offers the possibility to prevent local relapse if sufficiently large PTVs covering tumor bed, gastric stump, anastomosis and D2-3 lymph node regions[64] are treated[65]. For proximal tumors, the distal part of the esophagus should be included; for corpus/distal tumors[66] diaphragm/proximal duodenum should also be covered.

Fulfillment of these criteria results in a geometrical complex and large upper abdominal PTV in direct vicinity of the liver, both kidneys and spinal cord which may lead to therapy interruption and possible late toxicity. While total dose applied (45–50.4Gy) is below spinal cord tolerance, irradiation of the kidney with ablative doses can cause late renal toxicity which is becoming more and more relevant for long-term survivors[67‐69].

With conventional 2D techniques and 3DCRT, definition of the PTV was a compromise between kidney sparing and PTV coverage. PTV size was reduced at the cost of the risk of local relapse, but nevertheless at least the proximal part of most frequently the left kidney had to be irradiated with ablative doses[70, 71].

The introduction of IMRT allows steep and concave dose gradients and a highly conformal dose distribution also for large PTVs with complex geometry[72]. The combination of IMRT with image-guided radiotherapy (IGRT) allows the daily precise application of these highly conformal doses. Dosimetric analyses have shown that IMRT can reduce the median kidney doses below the tolerance limits[73] despite treating larger PTVs which cover all recommended regions at risk for local relapse[73‐75]. IMRT dose distribution within the kidneys is also more favorable if compared to conventional techniques: the functionally most important renal cortex is spared[65, 74] (see Figure1e-h). Additionally even these lower doses are delivered at lower single fractions, which can also have an impact on sparing functional capacity[76]. However this effect has to be investigated by functional imaging and laboratory values in long-term survivor populations. First results with functional MRI have shown intact kidney morphology, diffusion capacity and sodium concentration in long-term survivors of gastric cancer treated by IMRT combined with modern chemotherapy doublets[77]. These results were consistent with the observation that patients retained normal creatinine (and creatinine clearance) levels and did not show any clinical signs of late renal toxicity (see Figure2[77]).

In GEJ and esophageal carcinoma treatment, Kole and coworkers reported IMRT to reduce heart- and coronary artery doses as compared to 3DCRT[78]. However, no long-term clinical data are available about outcome in terms of cardiac toxicity of both modalities. In the neoadjuvant setting, IMRT with concurrent CT has also been shown to be well tolerated, accomplished excellent target coverage and normal structure sparing, and led to appropriate response rates[79].

In the last years, modern IMRT techniques such as rotational IMRT (VMAT/IMAT, RapidArc) and helical tomotherapy have been introduced[80, 81]. Volumetric intensity-modulated arc therapy (VMAT) has the capability to further decrease treatment times (and with that, the impact of intrafractional patient and organ motion on the effectively absorbed dose compared to IMRT while still providing similar PTV coverage and organ at risk sparing. Dosimetric comparisons indicated VMAT to be a favorable treatment option for esophageal and junctional carcinoma[82‐86].

Further technical developments such as integrating breath hold in the image-guided workflow can probably further improve dose distribution parameters[87]. Steep dose gradients can be precisely applied by all of these techniques if combined with daily 3D soft-tissue based image-guidance (see Figure1a-d for GEJ carcinoma and Figure1e-h for gastric carcinoma;[88, 89]). For gastric cancer, kV or MV cone beam computed tomography (CBCT) or ultrasound based techniques can be both applied[90‐93].

Data on radiation therapy of gastric- and GEJ carcinoma with protons are relatively rare. However, as demonstrated by Welsh et al., intensity-modulated proton therapy may be able to further at least theoretically, reduce normal tissue exposure if compared to IMRT in definitive therapy for locally advanced distal esophageal tumors[94].

Clinical results of IMRT compared to 3DCRT regarding outcome and acute and late toxicity profile have been evaluated by some retrospective studies[95, 96]; however no large randomized studies exist for gastric cancer. In our own retrospective analysis, adjuvant CRT with IMRT and modern chemotherapy doublets led to better survival of patients with advanced gastric cancer compared to the adjuvant combination of 3DCRT and conventional chemotherapy[95, 97], also in the long-term follow-up[98].

A similar comparison of survival and local control after IMRT vs. 3DCRT for esophageal cancer has been completed by Lin et al. on a large retrospective cohort[99]. In this study, OS, local control, and non-cancer-related death were significantly better with IMRT when compared to 3DCRT[99].

These first results are encouraging, however, large prospective trials and long-term clinical and functional investigations regarding risk organ functions are necessary to evaluate the role of modern radiotherapy techniques for gastric and GEJ carcinoma.