Introduction

Over the past decades, treatment of esophagogastric adenocarcinoma (EGA) has developed from mere tumor resection to sophisticated multimodal treatment strategies, in order to overcome limitations of surgery alone by improving local resectability as well as systemic tumor control [1,2,3,4,5,6,7,8,9]. In Western countries, where more than 70% of junctional and gastric adenocarcinomas are diagnosed in advanced stages [10], perioperative chemotherapy has become standard treatment for locally advanced EGA [11, 12] after 2006 based on the results of the MAGIC trial [1]. Hence, perioperative triplet epirubicin-, platinum-, and fluorouracil-based chemotherapy regimens (EPF) became state of the art in treating locally advanced EGA. Results of the MAGIC trial later were confirmed by the French FNCLCC/FFCD trial using a platinum/fluoropyrimidine doublet therapy [2]. Despite these advances, outcome of EGA patients remained unsatisfactory. A new combination consisting of fluorouracil, leucovorin, oxaliplatin, and docetaxel (FLOT) was first evaluated in metastatic EGA and proved to be highly active [13,14,15]. Thereupon, a phase II/III randomized controlled trial (RCT) was conducted comparing perioperative FLOT to perioperative anthracycline-based triplet chemotherapy with epirubicin, cisplatinum, 5-fluorouracil/capecitabine (ECF/ECX) for treatment of locally advanced EGA. The FLOT4 trial showed a significantly higher complete pathologic response (pCR) rate of 16% after FLOT compared to 6% after ECF/ECX [16] and revealed increased OS in the FLOT group compared to the ECF/ECX group (50 vs. 35 months median) [17]. These results internationally defined FLOT as the new standard perioperative chemotherapy protocol for treatment of EGA. However, data from real-life application of perioperative FLOT chemotherapy in clinical practice is scarce. This is the first study to report real-life experience on perioperative FLOT compared to anthracycline-based triplet chemotherapy (EPF). The aim of this single-center retrospective study was to investigate, whether the results of the FLOT4 trial can be reproduced in a heterogeneous patient population with comparable results in a real-life environment.

Methods

Study design and patient population

This study included patients with primary EGA who underwent elective surgery in curative intent at the University Hospital of Heidelberg, Department of Surgery between August 2010 and April 2018. Inclusion criteria were locally advanced primary tumor (cT3/4) and/or nodal positive disease (cN +) and/or distant metastasis (cM1) according to pretherapeutic clinical staging. Patients with distant metastasis (cM1) presented with oligo-metastatic disease and underwent surgery in curative intent as an individual treatment decision as previously described [28]. The metastatic lesions either had complete response to preoperative chemotherapy or were resected additionally to the primary tumor (supplement table). The metastatic lesions were judged to be resectable by experienced surgeons during an interdisciplinary oncological tumor board. All patients received preoperative chemotherapy either according to FLOT (FLOT group) or ECF/ECX/EOF/EOX protocol (EPF group). Clinicopathological and follow-up data of 356 patients (Fig. 1) were collected in a prospective database and analyzed retrospectively.

Fig. 1
figure 1

Patient selection and study collective

Full size image

Informed consent was obtained from all patients. The conduct of this retrospective study was approved by the institutional ethics committee.

Comorbidity

The American Society of Anaesthesiologists (ASA) Physical Status Classification System was applied in order to assess medical comorbidities and the perioperative risks of patients [18,19,20]. Pretreatment comorbidities of patients were assessed by experienced anesthesiologists and surgeons. The following conditions, which were clinically judged pertinent to perioperative risk assessment, were considered severe: decompensated renal insufficiency, decompensated cardiac insufficiency, liver cirrhosis, status post (s/p) myocardial infarction, s/p valve replacement, s/p stroke, s/p carotid stenosis, severe coronary heart disease, complicated diabetes mellitus, chronic pancreatitis, chronic obstructive pulmonary disease (COPD), or lung emphysema.

Pretreatment staging

Initial staging comprised upper endoscopy including biopsies with or without endoscopic ultrasound and computerized tomography (CT) of the chest and upper abdomen for all patients. Clinical tumor stage and localization were assessed according to the 8th edition of the UICC (Union for International Cancer Control) staging system. As for the cN category, it was only differentiated between cN0 and cN + according to lymph node diameter, shape, and contrast enhancement. The cM category was evaluated by CT or biopsies.

Chemotherapy

Chemotherapy was administered in form of FLOT (n = 263), ECF/ECX (n = 37), or EOF/EOX protocol (n = 46) according to FLOT4, MAGIC, and REAL-2 trials as described previously. Surgery was scheduled 4–6 weeks after the last application of preoperative chemotherapy. Restaging was done before surgery in form of CT of the chest and abdomen and endoscopy if necessary. Adjuvant treatment was administered in form of perioperative treatment if patients had undergone neoadjuvant treatment [21]. Postoperative chemotherapy started 4–8 weeks after surgery. Patients received treatment on an outpatient basis either by the National Center for Tumor Diseases Heidelberg (NCT) or other treating oncologists.

Surgery

The surgical approach for tumor resection was chosen depending on tumor localization. Standard surgical procedures were a right abdominothoracic en-bloc esophagectomy with a 2-field lymphadenectomy (Ivor-Lewis procedure) for adenocarcinoma of the esophagogastric junction (AEG) I, a transhiatal extended gastrectomy (THG) with an extended D2 lymphadenectomy for AEG III and or proximal gastric cancer (GC), and AEG II were treated either like AEG I or III depending on tumor extension into the esophagus [22]. In the case of GC, a total gastrectomy with D2 lymphadenectomy was performed for tumors in the middle or distal third of the stomach, and a subtotal gastrectomy for distal GC, if an adequate proximal resection margin was possible. All perioperative complications were recorded and graded according to Clavien-Dindo classification [23]. In patients with cM1 and intraoperative confirmation of metastasis, the surgical procedure was extended by resection of the metastatic lesions.

Histopathology and postoperative staging

Resection specimens were histologically examined in the Department of Pathology of the University Hospital. Histopathologic work up included extent of primary tumor, regional lymph node spread, presence of distant metastases ((y)pTNM categories according to the UICC classification, 8th edition), R category, tumor differentiation, and growth pattern according to Laurén. The lymph node ratio was calculated by dividing the number of LNs involved by the total number of LNs removed [24]. Tumor regression was evaluated according to Becker [25]. Less than 10% residual vital primary tumor (Becker Ia + Ib) was classified as histopathological response.

Follow-up

Patients were followed on an outpatient basis by our Medical Oncology Department according to a standardized protocol or other treating physicians as previously published [26]. Last follow-up was September 2, 2020. Complete follow-up information was available for all 439 patients. The median follow-up time for surviving patients (n = 219) was 60.2 months.

Propensity score matching

In order to reduce effects of selection bias and confounding factors in survival comparison, propensity score matching (PSM) was performed to create comparable groups. Propensity scores were estimated using a multivariable logistic regression model with the treatment groups (FLOT vs. EPF) as the dependent variables and potential confounders (age, sex, BMI, ASA, severe, cardiovascular, and pulmonary comorbidities, tumor localization, grading, Laurén classification, signet ring cells, cT, cN, and cM category) as covariates. One-to-one matching without replacement was performed using a 0.1 caliper width [27]. This allowed us to generate 81 (FLOT vs. EPF) score-matched pairs, which were used in subsequent analyses as indicated.

Statistical analysis

Overall survival (OS) was calculated from time of diagnosis till death or last follow-up. Survival rates were estimated using the Kaplan–Meier method. Differences in survival amongst groups were calculated using the log-rank test. To compare categorical variables, we used χ2 and Fisher exact test, for comparison of continuous variables the Mann–Whitney-U test was used. All tests were two-sided and a p-value < 0.05 was considered as statistically significant. Analyses were performed using SPSS version 20.0 (SPSS Inc., Chicago, Illinois, USA).

Results

Pretreatment patient and tumor characteristics

The median age at time of diagnosis was 61 (range 25–82) years. 208 patients (58.4%) were diagnosed with junctional (AEG) and 148 (41.6%) with gastric cancer (GC). FLOT chemotherapy was administered to 263 (73.9%) and epirubicin-based triplet chemotherapy (EPF) to 93 (26.1%) patients. Detailed information on patient demographics, tumor characteristics, and treatment is summarized in Table 1.

Table 1 a) Pretreatment, b) treatment-associated, and c) postoperative patient and tumor characteristics of all patients
Full size table

In comparison to the EPF group, more patients were clinically staged cN + (90.5% vs. 82.8%, p = 0.018) and cM1 (18.2 vs. 6.5%, p = 0.004) in the FLOT group (Table 1a). To create comparable groups, a propensity score matching (PSM) was performed. Comparison of pretreatment characteristics after PSM between 81 FLOT vs. 81 EPF patients revealed no significant differences (Table 2a).

Table 2 a) Pretreatment, b) treatment-associated, and c) postoperative patient and tumor characteristics of matched cohorts
Full size table

Treatment

All planned cycles of preoperative chemotherapy were completed by 94.3% of patients in the FLOT group and 90.3% in the EPF group (p = 0.228). Preoperatively a median of 4 (2–13) cycles of FLOT and 3 (1–4) cycles of EPF were given. An overview on chemotherapy and surgical treatment is given in Table 1b and Table 2b.

Perioperative morbidity

Overall surgical complication rate was 35.1%. Of all surgical complications 25.8% were grade 3 or higher according to Clavien-Dindo. Anastomotic leakage occurred in 12.9% of patients. 30-day mortality was 1.2% and in-hospital mortality was 2.5%. While FLOT and EPF groups differed little regarding surgical complications, there was a trend towards more medical and especially more pulmonary complications in EPF compared to FLOT patients (Table 1b and Table 2b).

Histopathology

Distribution of individual (y)pT categories was distinct between FLOT and EPF groups (Table 1c and Table 2c). Yet the proportion of locally advanced primary tumors ((y)pT3/4) was similar in the FLOT and EPF groups before (68.1% vs. 65.6%, p = 0.700) and after PSM (63.0% vs. 64.2%, p = 1.000).

A median of 28 lymph nodes (LNs) was removed and a median of 1 lymph node was positive, resulting in a median lymph node ratio (LNR) of 0.05. Distribution of individual (y)pN categories did not differ significantly (Table 1c). After PSM, the median number of positive and removed lymph nodes, and the distribution of LNR and ypN categories was similar (Table 2c).

The FLOT group revealed a trend towards higher R0 resection rates than the EPF group and after PSM this difference proved to be significant (R0 92.6 vs. 79.0%, p = 0.023) (Table 1c and Table 2c).

There was a trend towards a higher histopathologic response rate after FLOT compared to EPF before (31.9 vs. 26.9%, p = 0.297) and after PSM (34.6 vs. 25.9%, p = 0.169). However, the pCR rate (Becker Ia) was higher in the EPF group (6.1 vs. 11.8%, p = 0.059, after PSM 4.9 vs. 12.3%, p = 0.024), whereas after FLOT a markedly greater proportion of patients revealed regression grade Becker Ib (26.2 vs. 15.1%, p = 0.059, after PSM 29.6 vs. 13.6%, p = 0.024) (Table 1c and Table 2c).

Survival

Upon last follow-up, 183 of 356 patients (51.4%) had died (FLOT n = 130, EPF n = 53). Median OS from time of diagnosis for all patients was 49.7 (95% CI 28.3–71.1) months. Estimated 3- and 5-year OS rate was 55.8% and 46.6%.

Median OS was 52.1 (95% CI 27.4–76.8) months for FLOT and 46.4 (95% CI 24.3–68.5) months for EPF (Fig. 2a). Estimated 3- and 5-year OS rate was 57.8% and 48.5% respectively in the FLOT group, 53.9% and 44.7% in the EPF group. There was no significant difference in OS between FLOT and EPF groups (p = 0.577).

Fig. 2
figure 2

Overall survival according to treatment strategy (perioperative FLOT- vs. epirubicin-based chemotherapy (EPF)) in a all patients, b matched cohorts, c patients without distant metastases (cM0)

Full size image

After PSM, there was a trend towards improved OS after FLOT compared to EPF (median OS not reached/46.4 months, 3-year OS 59.0/54.8%, 5-year OS 54.5/44.3%, p = 0.234) (Fig. 2b).

There was also no significant difference in OS in the subgroup analysis of patients without distant metastasis (cM0) with 87.2 months median (95% CI 43.9–130.5) after FLOT vs. 50.2 months median (95% CI 13.7–86.8) after EPF chemotherapy (p = 0.189) (Fig. 2c).

Subgroup analyses

The relative treatment effect of FLOT on OS was compared to EPF (Fig. 3) across subgroups according to pretreatment and postoperative characteristics. A significant OS benefit in the FLOT group compared to the EPF group could be detected in patients without cardiovascular comorbidities (p = 0.007) and locally limited primary tumors in clinical (cT1/2; p = 0.006) and pathological (ypT0/1/2; p = 0.027) staging. No significant OS difference could be detected in the remaining subgroups.

Fig. 3
figure 3

Treatment effect of FLOT compared to EPF on overall survival according to pretreatment and postoperative patient and tumor characteristics. The forest plot shows hazard ratios for death (oblongs) and 95% confidence intervals (I bars). *Data was not available for all patients; values in bold print indicate a significance-level of p ≤ 0.5

Full size image

Patients with oligo-metastatic disease

In the study, we included 54 patients who presented with distant metastases in the clinical staging at time of diagnosis. A total of 48 of those patients received FLOT and 6 EPF. After the chemotherapy treatment, 31 (64.4%) patients in the FLOT group and 4 (66.7%) patients in the EPF group still presented with oligo-metastatic disease in preoperative restaging investigations. Furthermore, some of the metastatic lesions had complete response to preoperative chemotherapy and could not be detected during the surgery or could not be confirmed by the pathological examination neither during the surgery by frozen section nor the final pathological examination. An overview of the exact localization of the metastasis is given in the supplement table.

Discussion

The randomized controlled FLOT4 trial demonstrated improved OS in patients with EGA treated with perioperative FLOT chemotherapy compared to ECF/ECX. Median OS was increased by 15 months (50 vs. 35 months) and the estimated 5-year OS rate by 9% (45% vs. 36%) [17]. In our retrospective single-center study, we analyzed, real-life data on survival after perioperative FLOT chemotherapy compared to perioperative epirubin, platinum, and fluorouracil-based triplet chemotherapy 10 years after introduction of FLOT protocol at our institution. Our data show a median OS of 52 months (5-year OS rate 49%) after perioperative FLOT and surgery which is comparable to a median OS of 50 months (5-year OS rate 45%) in the FLOT4 trial [17]. However, in our series, patients who failed to proceed to surgery or did not undergo resection in curative intent were excluded from our study collective, whereas data from the FLOT-4 trial resulted from an intent-to-treat analysis with 5% of all patients who were not resected, mostly due to disease progression. On the other hand, we included selected patients with metastatic disease (18.2% with cM1 and 15% with pM1 in the FLOT group), who nevertheless underwent resection in curative intent as individual treatment decision as previously described [28]. Our results demonstrate a slightly worse survival than an analysis of Glatz et al. with 228 EGA patients all treated with perioperative FLOT followed by curative resection reporting a median OS of 61 months (5-year OS rate 51%) and a median recurrence free survival (RFS) of 42 months (5-year RFS rate 46%) [29]. However, this study included only patients after surgical resection without metastasis. In survival analyses of the matched FLOT cohorts with less advanced cT and cN categories, and especially a lesser proportion of metastatic disease, median OS was excellent (not reached/87 months, 5-year OS rate 55%/62% respectively).

Overall survival of the EPF group in our study was longer (46 months median, 5-year OS rate 45%) than in the ECF/ECX arm of the FLOT4 trial (35 months median, 5-year OS rate 36%) and the ECF arm of the MAGIC trail (25 months median, 5-year OS rate 36%) [1, 17]. This could be explained by the fact that in contrast to the FLOT4 or MAGIC trial we included only patients, who underwent surgery in curative intent.

In our collective real-life application of FLOT, chemotherapy failed to provide a significant survival benefit compared to EPF. This fact is most likely explained by the patient selection obviously occurring in clinical reality, since FLOT is expected to be the more aggressive treatment associated with more efficacy but also potentially causing more side effects. Patients receiving perioperative FLOT revealed more nodal positive and metastatic disease in clinical staging compared to EPF patients. The presence of nodal and distant metastases represents two of the most important and well-known negative prognosticators in EGA [36,37,38]. The inclusion of M1 patients might be controversial. However, we did not exclude patients with resectable metastatic disease, in order to reflect clinical reality from a surgical point of view. We rather performed subgroup analyses, which could not reveal a significant survival advantage for FLOT compared to EPF in non-metastatic EGA, neither.

After compensating for heterogeneity between groups by performance of PSM, no significant differences could be detected and preoperative patient and tumor characteristics were mostly well balanced between groups. As a result, survival analyses of the matched cohorts revealed a trend towards improved OS after FLOT compared to EPF. Possibly, these differences in OS failed to reach statistical significance, because patient numbers were limited after PSM.

No significant difference was found between FLOT and EPF groups with regard to pathological staging. Hence, in contrast to the FLOT4 study results, our analyses do not clearly indicate that perioperative FLOT is able to achieve better downstaging than EPF regimens [16, 17]. However, the proportion of histopathological responders (Becker grade Ia + b) was 35% after FLOT and 26% after EPF. Phase II results of the FLOT4 trial yielded comparable histopathological response rates (Becker Ia + b 37% after FLOT vs. 23% after ECF/ECX, p = 0.02) [16]. Remarkably, the pCR rate (Becker Ia) in our collective was only 5% in the FLOT compared to 12% in the EPF group contradictory to the results of FLOT4 trial with 16% pCR after FLOT compared to 6% after ECF/ECX [16] and quite low compared to other studies reporting a pCR rate of 14–17% after FLOT chemotherapy and even up to 31% in case of intestinal and/or AEG tumors [29, 39].

In accordance with the results of the FLOT4 trial, perioperative complication and mortality rates as well as duration of ICU and hospital stay were similar in both groups [17]. This is in line with the results of multiple studies including the benchmark RCTs, which have shown no increase in perioperative morbidity and mortality [1,2,3, 17, 40].

Limitations of the study are the retrospective, non-randomized, and single center design. There was strong selection bias resulting in marked heterogeneity between the FLOT and EPF groups rendering comparison of prognosis between the groups impossible. However, this patient selection reflects clinical reality. Even though propensity score matching was performed, remnant heterogeneity between groups cannot be excluded. Besides, despite a considerable sample size of the entire study population, it might still be too small for adequate subgroup analyses. Low patient numbers in the EPF group resulted in limited sample size of matched cohorts.

Ultimately, the issue of the ideal treatment strategy for EGA remains unresolved. Perioperative FLOT chemotherapy has become standard treatment for patients with locally advanced EGA after results of the FLOT4 trial [17]. Junctional cancers are alternatively treated with preoperative chemoradiation according to the CROSS protocol [3] depending on regional standards. Whether patients with AEG should receive perioperative FLOT or neoadjuvant chemoradiation according to CROSS protocol is currently investigated by the phase III ESOPEC trial [41]. To what extent patients could benefit from the addition of radiation to perioperative chemotherapy in the treatment of resectable EGA is currently evaluated in two RCTs, the RACE [42] and TOPGEAR trials [43, 44]. Further trials are necessary to clarify which treatment strategy is most suitable for particular patient subgroups or even individual patients with EGA. In the recently published CheckMate 577, a global RCT involving patients with resected esophageal or junctional cancer after neoadjuvant chemoradiotherapy, disease-free survival was significantly improved after a 1-year course of adjuvant nivolumab therapy compared to placebo (22.4 vs. 11.0 months median, HR 0.69) [45]. Even though OS data are not mature, adjuvant nivolumab will probably become a new standard of care. The adjuvant use of checkpoint inhibitors in patients undergoing perioperative chemotherapy (KEYNOTE-585, NCT03221426) [46] or definitive chemoradiotherapy (KEYNOTE-975, NCT04210115) [47] is subject of ongoing investigations. Implementation of targeted treatment strategies, such as biomarker-based therapy or immunotherapy with monoclonal antibodies such as trastuzumab, ramucirumab, or checkpoint inhibitors into the perioperative setting, might help to further improve the survival of EGA patients.

Conclusion

Ten years after the introduction of perioperative FLOT chemotherapy for the treatment of EGA at our institution real-life data revealed an OS of patients undergoing perioperative FLOT followed by surgical resection comparable to clinical trials. However, in our collective, we could not demonstrate a significant survival benefit after perioperative FLOT in comparison to perioperative ECX/ECF/EOX/EOF. Therefore, the current monopoly of FLOT as perioperative chemotherapy regimen of choice for resectable EGA must be critically questioned. Yet analyses of matched cohorts support the findings of the FLOT4 trial. Therefore, currently, FLOT represents the preferred perioperative chemotherapy regimen in the standard treatment of EGA.