Objective: The aim of this study was to propose and test a novel adverse pathology classification in AEG.
Background: Recent scientific advances show genomic and molecular concordance across all AEG types, suggesting a rationale for a biologic classification. We tested a 3-dimension adverse pathology classification across the entire junction and per Siewert anatomic subtype.
Methods: Of 1625 patients with AEG, 650 underwent radical surgery, 55% post-neoadjuvant therapy (NeoT). Adverse features defined a priori were poor differentiation (PD), lymphatic invasion (LI), vascular invasion (VI), and perineural invasion (PN), with 3 groupings: 0 (no adverse feature), 1 to 2, and 3 to 4. Multivariable logistic and Cox proportional hazards regression were applied.
Results: For adverse pathology, 31%, 46%, and 23% had 0, 1 to 2, and 3 to 4, respectively. Fifty percent of cases were AEG I, 25% AEG II, and 25% AEG III. Median survival was not reached, 49 and 17 months for 0, 1 to 2, and 3 to 4 adverse pathology, respectively (P < 0.001), and 76, 51, and 34 months for AEG I, II, and III, respectively (P < 0.001); AEG I was significantly (P< 0.001) associated with lower c (y)pT and c (y)pN stages, and LI, VI, PN, and PD (poor vs other). The pathology model was significant for survival along with (y)pT and (y)pN, and predicted response to chemotherapy and chemoradiation irrespective of anatomic subtype (P < 0.001).
Conclusion: A novel classification using standard pathology as proxy for poor biology is associated with survival and response to therapy. This effect is observed across the entire AEG spectrum, highlighting how biology should be aligned with anatomy in the modern paradigm of AEG management and design of clinical trials.