Abstract
<title>Abstract</title> <p>Background Accurate prediction of disease-free survival (DFS) in patients with esophageal squamous cell carcinoma (ESCC) after neoadjuvant therapy (NAT) remains challenging. Although the recently validated Neoadjuvant Esophageal (NAE) score can capture local anatomical downstaging with remarkable precision, it lacks the ability to reflect the dynamic evolution of the host systemic immune-inflammatory microenvironment. This study aimed to develop a comprehensive, multidimensional prognostic model that bridges the gap between systemic immune dynamics and robust local pathological features. Methods This multicenter retrospective study enrolled 328 patients with ESCC who received NAT followed by esophagectomy. We systematically evaluated and compared the prognostic performance of dynamic changes in systemic inflammatory indicators (ΔSII, ΔNLR, and ΔPLR). The most robust dynamic biomarker was identified through multivariable Cox proportional hazards regression and integrated with conventional pathological features to construct a nomogram for predicting DFS. Model performance was comprehensively assessed using the concordance index (C-index), time-dependent receiver operating characteristic (ROC) curves, and decision curve analysis (DCA). Results During the follow-up period, 63 recurrence events were observed. Multivariable Cox regression confirmed that a high ΔPLR (≥ 74.55), positive lymphovascular invasion (LVI), and a higher NAE score were independent risk factors for DFS. The resulting three-variable nomogram demonstrated favorable predictive performance (overall C-index = 0.807; 1-, 2-, and 3-year AUCs of 0.813, 0.819, and 0.828, respectively). Furthermore, the model exhibited incremental prognostic value beyond traditional major pathological response (MPR), effectively identifying occult high-risk individuals even within the MPR subgroup (HR = 7.23, P = 0.0004). Regarding postoperative management guidance, low-risk patients assessed by the model maintained very low recurrence rates (< 7%) with no significant survival differences regardless of whether they received observation alone or adjuvant therapy of varying intensities, suggesting that foregoing postoperative adjuvant therapy may represent a potential de-escalation strategy for model-identified low-risk patients; however, this requires validation in prospective studies. Interaction analyses revealed that LVI and ΔPLR were stable risk factors across treatment regimens, whereas the prognostic value of the NAE score was significantly amplified in the immuno-combination therapy subgroup (P for interaction = 0.028). Conclusions This study constructed a novel multidimensional nomogram that successfully integrates dynamic systemic immune-inflammatory evolution (ΔPLR) with local pathological residuals (LVI and NAE score), providing a practical individualized risk stratification tool for patients with ESCC after neoadjuvant therapy. The model compensates for the limitations of traditional MPR assessment and offers preliminary evidence and stratification basis for exploring "postoperative adjuvant therapy-free" de-escalation strategies in low-risk patients, while also revealing the treatment modality-dependent nature of prognostic factors, thereby providing a reference for precision surveillance and comprehensive management of different intervention groups in the era of immunotherapy.</p>