Clinical Outcome in Patients with Carcinoma of the Esophagogastric Junction Treated with Neoadjuvant Radiochemotherapy or Perioperative Chemotherapy: A Two-Center Retrospective Analysis

Adenocarcinoma of the esophagogastric junction (AEG) is a rare solid tumor entity with a steadily increasing incidence in the Western parts of the world [1, 2]. Due to its location at the crossing from the esophagus to the stomach, it is related to both organs. This circumstance has led many authors to consider AEG as a distinct tumor entity with characteristic genetic alterations and risk factors. Multimodal treatment with either neoadjuvant radiochemotherapy (NRCT) or neoadjuvant/perioperative chemotherapy (NACT) is the standard treatment approach to adenocarcinoma of the upper gastrointestinal tract. Most clinical trials describing effects of multimodal treatment, however, include patients suffering from either gastric cancer, AEG, or squamous cell carcinoma of the esophagus [3‒6]. Accordingly, these trials show that neoadjuvant/perioperative treatment (NAT) with either NRCT or NACT is better than surgery alone. However, any superiority of either NRCT or NACT still is a matter of debate, as a direct comparison of these treatment concepts has not been performed yet.

Clinical trials considering AEG alone are rare. Two trials tried to investigate the optimal multimodal treatment concept for AEG. However, the POET trial was planned as a randomized phase III trial but was closed early due to insufficient accrual and missing statistical significance in the interim analysis [7], and the NeoRes trial was a randomized phase II trial including 181 patients of both AEG and squamous cell carcinoma of the esophagus [8, 9]. Both trials were not able to describe a difference according to the type of NAT. Due to that, the evidence for optimal treatment of AEG is sparse.

Multimodal treatment concepts are key to optimal clinical outcomes in patients with AEG. They require consensual interdisciplinary decisions and respective clinical experience. Rising overall survival (OS) as reported in various clinical trials over the last decades [3, 4, 7, 8] might indicate improvements in these respects, because treatment regimens have improved only marginally until the most recent FLOT4 trial [3].

In our retrospective analysis, we describe the clinical outcomes of a patient cohort that has characteristics similar to those of patient populations in modern clinical trials investigating multimodal treatment concepts in AEG. All patients were either treated by NRCT or NACT followed by resection. We especially focused on a possible difference in OS according to type of NAT, the relapse pattern, and if the extent of the OS benefit described in the clinical trials can be reached in this real-life setting.

We retrospectively identified 85 patients receiving either NRCT (53 patients) or NACT (33 patients) followed by resection for locally advanced AEG I–III at the Ordensklinikum Linz and the Kepler University Hospital Linz. The patients included for analysis had been treated between October 2010 and August 2019. The data were obtained from the prospectively maintained clinical data registry of the clinical cancer center at the Ordensklinikum Linz and the Kepler University. Required for inclusion into the analysis were (1) histologically proven AEG (Siewert AEG I to AEG III), (2) a locally advanced but resectable stage with indication for multimodal treatment (assessed by CT scan and endosonography), (3) an ECOG PS of 0–2, and (4) complete information about the treatment. The observation period started at the date of the first diagnosis, defined by the date of histology, and ended on December 31, 2019.

The choice of multimodal treatment concepts with either NRCT or NACT was up to the decision of the local multidisciplinary tumor boards based on local guidelines. Briefly, at the beginning of the observation period, EOX was the treatment standard, which was based on the data from the MAGIC trial. NRCT became standard for AEG I and II after the presentation of the CROSS trial. Currently, the treatment standard is the perioperative FLOT protocol for all patients, irrespective of the primary tumor location. This is based on the data from the FLOT4 trial.

Data were routinely collected by medical doctors and clinical documenters via tumor documentation software (e.g., Celsius37). Information collected included personal information on the patients, histology, stage at diagnosis and after resection, ECOG PS, Charlson Comorbidity Index (CCI), start and end of NAT, date of resection, follow-up, and vital state data. The CCI [10] at the date of diagnosis was calculated from chart reviews.

Disease-free survival (DFS) was defined as the time period from diagnosis to relapse (distant and/or local) or death. OS was defined as the time from first diagnosis of the disease by histology to death. Time-to-event data were analyzed by Kaplan-Meier product limit estimation. The reverse Kaplan-Meier method was used for calculation of the median follow-up time. The Kruskal-Wallis test and Fisher’s exact test were used for comparing independent samples of continuous and categorical data, respectively. The software used for analysis was RStudio version 1.2.5019.

We identified 85 patients meeting the inclusion criteria. The median follow-up time for OS was 56 months (IQR 25.00; 80.00) and for DFS 57 months (IQR 34.00; 82.00). The median follow-up was longer for patients receiving NACT (OS: 70.00 months [IQR 25.00; 86.00]; DFS: 73.00 months [IQR 25.00; 93.00]) than for patients treated with NRCT (OS: 48.00 months [IQR 30.00; 77.00]; DFS: 48.00 months [IQR 34.00; 65.00]). The median age was 63 years (IQR 55.00; 69.00), with 78.8% of the patients <70 years. As expected, there were considerably fewer women than men (10.6 vs. 89.4%). The CCI at diagnosis was retrospectively assessed for 84 patients (1 missing); 21.4% of the patients had a CCI of 0 or 1 points, 71.5% of 2–4 points, and 7.2% of 5–8 points. Only 1.2% of the patients had an ECOG PS of 2; all others had an ECOG PS of 0 or 1. There was no substantial difference between body weight at diagnosis (83 kg; range 42–112) and body weight at the time of resection (79.5 kg; range 41–115). There were only 8 patients (9.4%) with the primary tumor at Siewert position III. In the remaining 77 patients (90.6%), the primary location was almost equally distributed between AEG I and II (Table 1).

The majority of the patients (61.2%) received NRCT, whereas 38.8% received NACT. In the NRCT group, the CROSS protocol [4] was used for 80.8% of the patients, whereas the remaining 19.2% were treated with 5FU-based treatment regimens. EOX [5] was used for 64.7% of the patients in the NACT group, and the FLOT protocol [3] for the remaining 35.3%. In case of NACT, 43% of the patients also received adjuvant chemotherapy (data not shown). The median time from first diagnosis to resection was 16.43 weeks, and in a median time of <4 weeks, NAT was initiated after the diagnosis. The median time to resection after the end of NAT was 6.43 weeks (Table 2). These time intervals were not different between the NRCT and the NACT group.

In general, NRCT showed a higher local response rate with a pathological complete response (pCR, defined as ypT0, of 29.8%, which was significantly higher than in the NACT group, where the pCR rate was 12.1% (p = 0.01). Along with that, patients with nodal negativity (ypN0) were significantly more frequent in the NRCT group (59.6 vs. 33.3%; p < 0.001), although tumor size and nodal involvement at diagnosis were equally distributed between the groups (Table 1). Resection was R0 in all patients treated with NRCT and 90.6% in patients treated with NACT. Relapses, either local and/or distant, were statistically not significantly different between the two groups. Distant relapses trended to be more prevalent in the NRCT group (42.9 vs. 30.3%; p = 0.35; Table 3).

The median OS was 44 months (95% CI: 33–NA; Fig. 1) and the 3- and 5-year survival rates were 52.3% with NRCT and 42.9% with NACT. The median OS was not different between the patients treated with NRCT and those treated with NACT (HR 1.21; 95% CI: 0.63–2.35; Fig. 2). The patient characteristics were well balanced for this comparison (Table 3). Patients older than 65 years showed a worse CCI and a slightly worse ECOG PS than patients younger than 65 years (Table 4). However, there was no statistically significant difference in OS between these age groups (HR 1.14; 95% CI: 0.45–1.7; Fig. 3). The CCI did not show a relevant impact in our data set (data not shown).

Median DFS of the whole population was 30 months (95% CI: 13–34). The 3- and 5-year relapse-free survival rates were identical at 35.4%, indicating that the risk for relapse is highest before the third year after diagnosis. There was no difference in median DFS between patients treated with NRCT and those treated with NACT (Fig. 4).

NRCT or NACT followed by resection are current standard multimodal treatment options in the Western world for patients suffering from locoregional AEG. Evidence for that is coming from data showing only modest effects of adjuvant treatment on survival [11], even when escalating treatment by combining different cytotoxic agents [12]. This is in contrast to treatment guidelines in Asia, where adjuvant treatment has been proven to be efficacious [13, 14]. Using either option in Caucasians, NRCT or NACT, is based on data derived from several clinical phase III trials, showing superiority of multimodal treatment over surgery alone [3‒5]. Data regarding the direct comparison of NRCT and NACT are, however, scarce. The POET trial was set up as a randomized phase III trial to investigate a possible difference between NRCT and NACT in patients with AEG I and II [7]. However, the trial did not reach its targeted number of 354 patients and ceased due to poor accrual and due to statistically insignificant results after the randomization of 126 patients. On this basis, the POET trial failed to reach its primary endpoint (OS benefit at 3 years) and did not show any significant difference in OS between NRCT and NACT, although the authors suggested a benefit from NRCT due to higher local response rates, despite a higher rate of distant relapses [7, 15]. Similarly, the NeoRes trial, a randomized phase II trial, failed to show a difference in survival between NRCT and NACT, although the pathological response rates were better in the NRCT group [8, 9]. Notably, The NeoRes trial also included squamous cell carcinoma, like the CROSS trial. The lack of a reliable direct comparison between NRCT and NACT has led to different usage of treatment concepts, partly influenced by local availability of radiotherapy. In our data set, we investigated the treatment outcome of patients suffering from gastroesophageal junction cancer (AEG I–III) in a real-life patient cohort. In terms of patient demographics, especially age, ECOG PS, and TNM stage, we intended to identify patients that were highly likely represented in the respective clinical trials as well. The median age of 63 years and the age categories with 42.4% of the patients below 60 years, 36.5% between 60 and 70 years, and 21.2% older than 70 years were similar to those in the FLOT4 [3], MAGIC [5], OE05 [16], and CROSS [4] trials. The same is valid for the ECOG PS, with only 1 patient (1.2%) with an ECOG PS of 2 and all others with an ECOG PS below 2. Furthermore, the rate of AEG I compared to AEG II and III and the rate of clinical stage T4 carcinoma were similar to those in the FLOT4 and CROSS trials.

As expected, NRCT resulted in a higher pCR rate (29.8%) compared to NACT (12.1%). These results are in line with the abovementioned literature, underlining the significant impact of NRCT on local tumor control. This is also reflected in a lower local relapse rate in the NRCT group. Distant relapse rates, however, were lower in the NACT group. Both the local and the distant relapse rate were statistically nonsignificantly different but similar to those in the POET trial. As in the POET or NeoRes trials, these treatment effects did not translate into any OS or DFS benefit for the patients. The median OS was 44 months, which is between that reported in the MAGIC trial (36 months for the EOX arm) and those observed in the FLOT4 trial (50 vs. 35 months, FLOT vs. EOX). In the CROSS trial, the median OS was 49.4 months; according to the subgroup analysis, the effect for adenocarcinoma, however, was a little bit smaller. The OS data from modern clinical trials for upper gastrointestinal cancer put together show an OS between 40 and 50 months. The subgroup analyses in the MAGIC and FLOT4 trials show that the effect of chemotherapy is even more pronounced for AEG. The median OS was 21.1 and 30.8 months with NACT and NRCT, respectively, in the POET trial (not statistically significant), suggesting an increase in OS over time for AEG when compared with the CROSS and FLOT4 trials. The median OS of 44 months achieved in our patient cohort proposes that this clinical progress can be translated into routine practice.

This perspective might outweigh the main limitation of our analysis, which is its retrospective nature. With regard to NACT, the majority of our patients received EOX. The conclusions drawn in the abovementioned literature and our retrospective analysis suggest that EOX or a comparable 5FU-based regimen used as NACT might be similarly effective as CROSS-like NRCT. Additionally, the FLOT4 trial showed that perioperative FLOT is clinically more effective than EOX. Taking that into account, FLOT might be the more effective treatment for patients with upper gastrointestinal adenocarcinoma. Results from direct comparisons within phase III clinical trials are underway and will give more conclusive answers. The Australian TOPGEAR trial compares the perioperative modified MAGIC protocol with or without NRCT in both AEG and gastric cancer patients. In 2017, the TOPGEAR trial protocol was amended to also allow FLOT as treatment [17, 18]. Considering the superiority of FLOT over EOX/ECF, the results expected from the currently recruiting ESOPEC trial, comparing perioperative FLOT with the CROSS protocol, will be highly interesting [19]. Addition of NRCT (CROSS) to perioperative FLOT is compared with FLOT alone in the recently started RACE trial (NCT04028167). Data from these trials are expected within the next years.

In conclusion, we suggest that EOX-based perioperative treatment is similarly effective as neoadjuvant CROSS. Data regarding this particular issue are sparse, and due to that, our data add some evidence to this discussion despite its retrospective nature. We also show that outcome data regarding OS for patients suffering from AEG can be reproduced in clinical practice. Despite the fact that the FLOT4 trial was not a direct comparison between NACT and NRCT, we decided to use FLOT as a multimodal treatment standard.

The authors acknowledge the support of Florian Knotz, MA, for initial data documentation.

Our research complies with the guidelines for human studies and was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. This retrospective analysis was approved by the Ethics Committee of the Land Oberösterreich (Ethics Committee Land Oberösterreich, No. 1002/2020, February 12th, 2020). According to the vote of the ethics committee, no informed consent was necessary, because all the data processed for the analysis were retrieved from routine practice.

The authors have no conflicts of interest to declare.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Guarantor of integrity of the entire study: H.R.; study concepts and design: H.R. and S.M.-G.; literature and data research: H.R., H.V.W., B.D., and R.P.; data analysis: H.R., H.V.W., and S.M.-G.; interpretation of data: H.R., H.V.W., S.M.-G., R.P., B.D., C.V., C.S., and A.P.; manuscript preparation: H.R. and H.V.W.; manuscript editing: C.S., A.P., H.R., H.V.W., B.D., and C.V.; final approval for publishing: all authors.

H.V. Wundsam und B. Doleschal contributed equally to this paper.