Our study underscores the crucial role of the adjuvant phase of the FLOT regimen in a real-world setting. The survival rates of our patients were surprisingly low. The 5-year survival rate was even lower than that of the perioperative chemotherapy arm of the now historic MAGIC trial [2]. We performed a dedicated analysis in order to identify and rule out factors that could have led to such unfavorable survival rates:

First, we found that the neoadjuvant phase in our cohort was implemented with competitive efficiency and cannot be used to explain the lack of survival benefit: The pathological complete response (pCR) rate in our study was 11.8%, which is comparable to the pCR of the FLOT control arms of several contemporary chemotherapy trials, including the MATTERHORN trial [8], or the DANTE Phase II trial [9]. Our real-world analysis showed that 84.4% of all patients had completed at least four cycles of neoadjuvant chemotherapy, which is only slightly lower than the 90% rate observed in the neoadjuvant phase of the original FLOT4 trial [3] or the 95% rate observed in the neoadjuvant phase of the FLOT control arm of the DANTE trial [10]. We further show that surgical procedures cannot be used to explain the unfavorable survival rates. The 30-day postoperative survival rates were even better than the 30-day survival rates reported in the FLOT4 trial [3], the median number of lymph nodes resected was comparable to that reported in the recently published SPACE-FLOT study [11] and the R0 resection rates were even slightly higher (90.3% versus 88.6%) than those reported in the SPACE-FLOT study [11].

Our data underscore that the implementation of adjuvant chemotherapy in daily clinical practice represents an obvious obstacle in the real world. Should all patients receive adjuvant chemotherapy whenever possible, and if not, which subgroup should be targeted?

We hypothesized that the omission of adjuvant chemotherapy would affect patient survival. Surprisingly, however, at first glance there was no significant difference in survival between patients with and without adjuvant chemotherapy, which was started in only 42.9% and completed in only 27.2% of our patients.

Although a correlation between TRG and patient prognosis in the perioperative setting has been frequently described, its prognostic value is still debated due to different scoring systems with different cut-off values complicated by inter- and intra-observer variability [5]. In the past, we have confirmed superior survival rates for patients with < 10% tumor regression using digital image analysis, a cut-off used in the Becker TRG scoring system [5]. A central question of our current study was how the continuation or omission of adjuvant chemotherapy would affect real-world outcomes depending on the degree of tumor regression after neoadjuvant chemotherapy. Real-world studies from the time before or during the transition to the FLOT era paint an ambiguous picture in this regard: In their study from the pre-FLOT era, Deng et al. described that complete and non-responders to neoadjuvant chemotherapy did not benefit from adjuvant chemotherapy, based on a comparison between T stage at diagnosis and yT stage of the resection specimen after neoadjuvant chemotherapy [12]. Saunders et al. described in their study from the pre-FLOT era that adjuvant therapy would only benefit those patients, who showed a histopathological response after neoadjuvant chemotherapy and based their findings on the old Mandard TRG scoring system [13]. In contrast, the study by Glatz et al., which analyzed a mixed population receiving the now historic chemotherapy regimens ECF and EOX, but also the FLOT regimen, described that patients with poor histopathologic regression according to Becker et al. (TRG 3) benefited from adjuvant chemotherapy [14]. To address this highly relevant therapeutic issue in a contemporary context, real-world studies involving only patients receiving the current chemotherapy regimen with FLOT are needed.

Our real-world study included only patients who had received the current FLOT regimen. We analyzed whether the effect of adjuvant chemotherapy on survival depended on tumor regression and restricted our analysis to those patients who had received at least all four neoadjuvant chemotherapy cycles. We found that adjuvant chemotherapy did not improve survival in TRG1 patients. For TRG2 patients, there was a strong trend toward longer survival with any number of adjuvant chemotherapy cycles, but this trend did not reach statistical significance. However, TRG3 patients lived significantly longer with any number of adjuvant chemotherapy cycles, regardless of dose reductions or the omission of chemotherapeutic drugs.

Our findings are partially supported by the results of the recently published retrospective SPACE-FLOT study [11], which confirms a key finding of our study: lack of benefit of adjuvant chemotherapy in patients with complete tumor regression. Furthermore, they described a significant benefit of adjuvant chemotherapy in patients with a partial histopathologic response, which supports the strong trend for longer survival seen in our collective for TRG2 patients. However, in the SPACE-FLOT study population, patients with minimal or no tumor regression (TRG3) did not benefit from adjuvant chemotherapy, which is in contrast to our findings.

German medical centers did not participate in the SPACE-FLOT study, but we do not believe that geographic differences were responsible for the different results regarding the subgroup of patients with minimal or no tumor regression. Furthermore, we do not believe that the different results are due to the concepts of TRG evaluation. In our study, TRG evaluation was centrally performed by two independent board-certified surgical pathologists to reduce interobserver variability, whereas in the SPACE-FLOT study TRG was not centrally assessed [11]. Equivalent to the international Delphi consensus [15] used in the SPACE-FLOT study, in our study the entire tumor was always embedded and analyzed. While we consistently used the Becker scoring system, the SPACE-FLOT cohort used seven different TRG systems. A practical approach was needed to unify these systems for data analysis [11]. Nevertheless, we believe that with respect to the definition of minimal/no tumor regression the TRG scoring between our study and the SPACE-FLOT study is comparable, as they defined minimal/no tumor regression as the presence of > 50% residual tumor in the resection specimen, which is equivalent to TRG3 according to Becker et al.

We believe that the relevant discrepancy regarding the effect of adjuvant chemotherapy on survival of patients with minimal/no tumor regression may be explained by different approaches to the analysis of adjuvant chemotherapy efficacy and by different R0 resection rates.

In our study, only patients who had received at least all four neoadjuvant chemotherapy cycles, and thus had received optimal neoadjuvant treatment, were included in the subsequent survival analysis related to TRG and adjuvant chemotherapy administration. In the SPACE-FLOT patient subgroup with minimal or no histopathologic response, 87.9% of those with adjuvant treatment and only 66.8% of those without adjuvant treatment had received all four cycles of neoadjuvant chemotherapy. Although the rationale of neoadjuvant chemotherapy with FLOT is the reduction of tumor size for surgery, it is conceivable that it could also influence metastatic spread, relapse and survival. Incomplete administration of neoadjuvant chemotherapy, whether due to internal or external factors, may greatly affect the histopathologic response and the conclusions drawn from it.

The R0 resection rate in our study cohort was high at 90.3% overall and 85.9% in the subgroup with minimal or no tumor regression (100% in the subgroup that received adjuvant chemotherapy and 90.0% in the subgroup that did not receive adjuvant chemotherapy with FLOT). However, in the SPACE-FLOT trial, the R0 resection rate in the subgroup with minimal or no tumor regression was only 82.7% for patients who received adjuvant chemotherapy and 76.5% for patients who did not receive adjuvant treatment. Therefore, in the SPACE-FLOT trial, adjuvant chemotherapy had to compensate for considerably more patients in whom complete surgical removal of the cancer site was not possible than in our study cohort, in which considerably better surgical results were achieved. In addition, because in the SPACE-FLOT trial the R0 resection rates were comparatively low in both subgroups with or without adjuvant chemotherapy, no statistical difference was observed in the univariate analysis and we assume that the resection rate was therefore not considered in the subsequent Cox regression analysis and propensity score matching analysis of the trial. We postulate that in the case of patients with minimal or no tumor regression after neoadjuvant chemotherapy, adjuvant chemotherapy cannot compensate for incomplete surgical resection of the cancer site because the poorly responding primary tumor is still present in its original setting.

The focus of adjuvant chemotherapy is to eliminate occult distant metastases, which have a different tumor microenvironment than the primary tumor site. In the past we have shown that adjuvant chemotherapy in the perioperative setting addresses a completely different tumor [5, 16, 17]. This differs from the current concept of adjuvant chemotherapy in the FLOT era and the conclusions drawn by the SPACE-FLOT trial. We believe that a minimal or poor histopathologic response at the primary site is not predictive of chemotherapy efficacy at the metastatic site, as a completely different tumor environment is being targeted. The data presented here demonstrate that the TRG3 subset of patients benefited from adjuvant treatment and we postulate that this effect was seen, in contrast to the SPACE-FLOT trial, because the R0 resection rate in our cohort was high and the effect of adjuvant chemotherapy with FLOT in our cohort reflects its effect on the elimination of peripheral micrometastases rather than on residual tumor cells at the primary tumor site.

The subgroup of patients with partial or no response to neoadjuvant chemotherapy with FLOT have an overall poor prognosis and should be the focus of future therapeutic efforts. Our study therefore provided a dedicated biomarker analysis that could serve as a basis for the design of future randomized controlled trials. We see a need for personalization of the adjuvant regimen based on the biomarker profile of the resected tissue specimen—not the initial biopsies. Patients with partial or no histopathologic response may benefit from the addition of targeted therapies to the chemotherapy backbone in the adjuvant setting. Recently, the results of the VESTIGE trial were published, which evaluated adjuvant immunotherapy in patients with resected GC or AEG following preoperative chemotherapy and high risk for recurrence (ypN + and/or R1 status) [18]. The chemotherapy-free approach had not been successful in this high-risk subset of patients [18], thereby underscoring the role of the chemotherapy backbone. Future studies are needed to investigate the potential benefits of adding targeted therapy to the chemotherapy backbone in the adjuvant setting.

The limitation of our study is its retrospective nature and of course this is only a cohort of patients who underwent tumor surgery. As comorbidities naturally increase with age, an assessment of comorbidities would have been desirable and should be included in future real-world studies. The ECOG performance status was not regularly documented in our retrospective cohort, which limits our results in this regard. In addition, we believe that the relatively small size of the TRG2 subgroup resulted in the observation of only a strong trend and not a significant survival benefit for TRG2 patients receiving adjuvant chemotherapy as seen in the SPACE-FLOT trial.