Dear Editor,

We read with great interest the paper by Chen et al. reporting the 2-year update analysis of the ATTRACTION-2 trial [1]. At 2 years of follow-up, nivolumab was confirmed to confer an overall-survival (OS) benefit over placebo in heavily pretreated advanced gastric or gastroesophageal junction (G/GEJ) cancer patients (10.6% vs 3.2%) and this was regardless of the presence of potential predictors such as PD-L1 positivity. Radiologic response, however, seemed to be associated with prolonged survival in the nivolumab arm. Other clinical features were not associated with treatment efficacy as shown in a forest plot provided in the supplementary material. In particular, no interaction with the presence of liver or peritoneal metastasis was demonstrated. However, presence of lymph-node metastasis was not included in the subgroup analysis. Lymph-node metastases have been found to be particularly responsive to immune checkpoint inhibitors (ICIs) [2, 3] and also the three complete responses reported in the ATTRACTION-2 were in patients with predominantly lymph node-based metastatic disease.

We recently reviewed 57 patients with mGC treated in second-line with ICIs [4]. Herein we report that patients with lymph node-only metastatic disease had a significantly longer median overall survival as compared to other metastatic patterns (mOS 20.6 vs 4.8 months, HR 0.28, 95% CI 0.12–0.63, p 0.001, see Figure S1). Progression free survival (PFS) and objective response rate (ORR) were similarly significantly improved in the lymph node-only subgroup (mPFS 11 vs. 2 months, HR 0.43, p 0.01 and ORR 33% vs 11%, p 0.05, respectively). Presence of lymph node-only metastasis was confirmed to be an independent prognostic factor in a multivariate cox-regression analysis including gender, primary tumor location, age, Lauren’s subtype, PD-L1 expression, resection of the primary, number of metastatic sites and ECOG performance status (Fig. S1).

Lymph node-only metastatic disease might represent a biologically and phylogenetically distinct entity with peculiar prognosis and response to treatment [5]. Moreover, it can be hypothesized that the increased density and infiltration of lymphocytes in the lymph-node tissue might favor an enhanced ICI-induced immune response against cancer cells metastasizing to the lymph nodes.

We suggest the inclusion of specific metastatic patterns in future subgroup analyses of ICIs in G/GEJ cancer, and in particular we suggest the assessment of the effect of exclusive or predominant nodal metastatic disease in terms of interaction with immunotherapy efficacy.