Despite advances in the multimodal treatment of locally advanced esophagogastric cancer, more than half of patients will relapse during follow-up and die due to metastatic disease.

As a consequence, the proportion of patients presenting with metastatic disease having already received perioperative systemic treatment for previous localized disease remains high. The development of tumor clones resistant to previously applied drugs as well a residual toxicity from previous chemotherapy, complicate the selection of an appropriate treatment regimen for recurrent disease.

The present study aimed to analyze prognostic factors in patients with a relapse after multimodally treated esophagogastric cancer.

The majority of patients in our cohort had received FLOT-based perioperative treatment; for this subgroup, no data from recent phase III trials exploring second-line treatment can be applied (Wilke et al. 2014; Shitara et al. 2018), as this group of patients has already been exposed to both a platinum and a taxane compound.

In a univariate analysis OS relapse was clearly correlated to the time span between the end of primary therapy and the relapse occurring: patients with an early relapse, directly or within 6 months after completion of surgery with/without adjuvant chemotherapy, have a significantly worse OS relapse of 6.3 months, comparable to that seen with second-line chemotherapy in Caucasian patients (Wilke et al. 2014; Ford et al. 2014; Thuss-Patience et al. 2011).

Using a different definition of early recurrence (modified early relapse, defined as a relapse within 6 months after surgery), a similar separation in term of OS relapse and OS first-line-tx was observed. However, using this category, the prognostic value in terms of PFS first-line was worse, with no significant difference being observed any more. This is most likely due to the fact that patients with an early relapse after adjuvant chemotherapy have a worse outcome assuming platinum- and optional taxane-refractory tumors, even if the relapse occurs more than 6 months after surgery.

Among relapsed patients, those with an MSI tumor had a very favorable outcome, although not all of them had received checkpoint inhibition. These data are in line with several reports suggesting a beneficial outcome of stage IV MSI tumors (Haag et al. 2019; Giampieri et al. 2017; Polom et al. 2018; Busch et al. 2021). The higher immunological control caused by a higher mutational burden associated with the presentation of neoantigens might contribute to the indolent clinical course observed in this molecular subgroup.

Irinotecan-based regimens were chosen in the majority of patients as a consequence of proposed resistance against platinum- or taxane-containing therapies (especially in early-relapsed patients). Irinotecan-based regimens have shown efficacy both as palliative first-line therapy (Guimbaud et al. 2014) as well as in platinum-pretreated patients (Thuss-Patience et al. 2011; Hironaka et al. 2013). In addition, persisting sensory polyneuropathy is a common problem among patients having received FLOT-based chemotherapy perioperatively, so irinotecan-based regimens were also often selected among patients with a later relapse. The ongoing phase III Ramiris trial will help to define the benefit of the irinotecan-based Folfiri regimen and ramucirumab in taxane-pretreated patients (NCT03081143).

Recently the combination of PD-1 inhibition and platinum-based chemotherapy has shown a superior OS in patients with Her-2 negative adenocarcinoma and a PD-L1 combined prognostic score (CPS) of more than 5 or 10 respectively (Janjigian et al. 2021; Sun et al. 2021). Although it is unknown whether the addition of a PD-1 inhibitor to a platinum-based chemotherapy can overcome a resistance against platinum compounds in patients relapsing after perioperative platinum-based treatment, activity of PD-1 inhibitors both as monotherapy and in combination with chemotherapy across different lines of treatment merits further investigation in relapsed patients with a PD-L1 CPS of more than 5 or 10 respectively (Janjigian et al. 2021; Sun et al. 2021; Wainberg et al. 2021; Lei et al. 2021). Beyond PD-L1 expression and Her-2 expression, further molecularly stratified approaches are highly needed to improve outcome after failure of perioperative chemotherapy, given the modest efficacy of any cytotoxic drug in patients having failed platinum- and taxane-based treatment, such as FLOT.

A subset of our patients had a unilocular relapse that was suitable for local therapy in forms of surgical resection or irradiation after interdisciplinary consultation. Clearly, those patients would not have been identified without a structured follow-up, as clinical symptoms cannot be expected in an oligo-relapsed situation. The OS observed within this small subgroup clearly favored this approach with long-time control that could lead to sustained remission or even cure in single patients. So far, a structured follow-up has not shown to increase survival in esophagogastric cancer in a randomized setting (Bjerring et al. 2019). The plateau observed among patients having received a local therapy after relapse strongly suggest that the benefit in terms of OS is not caused by a lead-time bias, as a systemic treatment alone would probably not have been associated with the same sustained efficacy. Despite the promising overall survival, only an adequately powered randomized trial exploring a structured follow-up and interdisciplinary discussion of localized relapses in comparison to symptom-based follow-up could answer the question regarding the benefit of early identification and local treatment of relapses and thereby overcome the limitations of this retrospective analysis.

When analyzing the prognostic factors in multivariate analysis, MSI and a local intervention are confirmed as prognostic markers, however, the role of the relapse-free interval as a prognostic factor is less clear, as the correlation did not reach statistical significance. Apart from the limited number of patients in each subgroup, another possible explanation could be that, given the possible resistance to chemotherapeutical compounds, the relapse-free interval seems to be more relevant in patients receiving systemic treatment compared to those receiving local therapy, e.g. irradiation of an isolated lymph node metastasis. In addition, there is an overlap between patients with a late relapse and patients having received a local therapy in forms of radiation or surgery, as local interventions were predominantly performed in patients with a late relapse.

In summary, our study shows that patients with a relapse after multimodal therapy for esophagogastric adenocarcinoma have a heterogeneous prognosis based on clinical and molecular biomarkers. The beneficial outcome observed with local interventions in patients with an oligometastatic relapse merits further investigation in a randomized trial.