A Multicenter Nomogram Integrating Dynamic ΔPLR and Pathological Features for High- Accuracy Prognostic Prediction in Esophageal Cancer Following Neoadjuvant Therapy

Technical Summary: A Multicenter Nomogram Integrating Dynamic ΔPLR and Pathological Features for High-Accuracy Prognostic Prediction in Esophageal Cancer Following Neoadjuvant Therapy

Problem Statement
Accurately predicting disease-free survival (DFS) in patients with esophageal squamous cell carcinoma (ESCC) following neoadjuvant therapy (NAT) remains a significant clinical challenge. Current prognostic standards rely heavily on "static" snapshots, such as traditional pathological staging (ypTNM) and Major Pathological Response (MPR). While the recently validated Neoadjuvant Esophageal (NAE) score improves upon these by quantifying local anatomical downstaging, it fails to capture the dynamic evolution of the host's systemic immune-inflammatory microenvironment. Conversely, while systemic inflammatory markers (e.g., NLR, SII, PLR) are known prognostic indicators, most studies utilize single-time-point baseline measurements, missing the critical longitudinal response to therapy. There is an urgent need for a multidimensional model that bridges systemic immune dynamics with local pathological residuals to enable precise risk stratification and guide postoperative management, particularly in the era of immunotherapy.

Methodology
This multicenter retrospective study analyzed 328 patients with ESCC who underwent NAT followed by curative esophagectomy across three centers in Fujian Province between January 2020 and December 2025.

• Data Collection: The study integrated clinicopathological data with paired peripheral blood samples collected at two key time points: baseline (within 7 days pre-NAT) and preoperatively (within 7 days pre-surgery).
• Biomarker Evaluation: The authors calculated dynamic changes ($\Delta$) in systemic inflammatory indicators: $\Delta$NLR, $\Delta$SII, and $\Delta$PLR (defined as preoperative value minus baseline value).
• Model Construction:
  1. Biomarker Selection: Univariable and multivariable Cox proportional hazards regressions were used to compare the prognostic performance of the dynamic indicators. $\Delta$PLR was selected as the optimal immune biomarker based on the highest concordance index (C-index) and lowest Akaike information criterion (AIC).
  2. Variable Integration: The selected $\Delta$PLR was integrated with core pathological features: Lymphovascular Invasion (LVI) and the NAE score.
  3. Statistical Rigor: The study employed maximally selected rank statistics to determine optimal cutoff values. Multicollinearity was assessed (VIF < 5), and the proportional hazards assumption was verified. To address overfitting, the model underwent rigorous internal validation using 1000 bootstrap resamples. A sensitivity analysis incorporating the "year of enrollment" was performed to control for time-dependent confounding due to the shift from conventional therapy to immuno-combination therapy during the study period.
• Validation: Model performance was evaluated using the C-index, time-dependent ROC curves, calibration plots, and Decision Curve Analysis (DCA). Subgroup and interaction analyses were conducted across treatment regimens and MPR status.

Key Results

• Independent Risk Factors: Multivariable Cox regression identified three independent risk factors for DFS: positive LVI (HR = 3.23), higher NAE score (HR = 1.02), and high $\Delta$PLR (cutoff $\ge$ 74.55, HR = 2.04).
• Model Performance: The resulting three-variable nomogram demonstrated robust predictive accuracy with an overall optimism-corrected C-index of 0.801. The time-dependent AUCs for 1-, 2-, and 3-year DFS were 0.813, 0.819, and 0.828, respectively.
• Superiority over MPR: The model provided incremental prognostic value over traditional MPR assessment. Crucially, it identified occult high-risk individuals within the MPR subgroup (HR = 7.23 for high-risk vs. low-risk within MPR patients), capturing 81% of recurrence events in this group. Conversely, it identified a low-risk subset within the non-MPR population who had excellent outcomes.
• Treatment Regimen Interaction: Interaction analysis revealed that LVI and $\Delta$PLR were stable risk factors across both conventional NAT and immuno-combination therapy subgroups. However, the prognostic value of the NAE score was significantly amplified in the immuno-combination subgroup (P for interaction = 0.028), showing predictive significance only in that group.
• De-escalation Potential: In the low-risk group identified by the nomogram, recurrence rates were very low (< 7%) regardless of whether patients received observation alone or adjuvant therapy. Propensity score matching (PSM) between low-risk patients receiving observation versus those receiving adjuvant therapy showed no significant difference in DFS (log-rank P = 0.98), suggesting that foregoing postoperative adjuvant therapy may be a viable de-escalation strategy for this specific cohort.

Significance and Claims
The authors claim this study presents the first parsimonious nomogram that successfully integrates longitudinal systemic immune-inflammatory evolution ($\Delta$PLR) with local pathological residuals (LVI and NAE score). The significance of this work lies in several areas:

1. Dynamic Integration: It moves beyond static pathological assessment by incorporating the dynamic trajectory of the host immune response, offering a more comprehensive view of tumor biology post-NAT.
2. Precision Stratification: The model compensates for the limitations of MPR, which can be obscured by immune cell infiltration or fibrosis in the immunotherapy era, by providing a refined risk gradient that identifies both hidden high-risk patients and low-risk candidates for treatment de-escalation.
3. Clinical Guidance: It offers a practical tool for individualized postoperative management, potentially supporting the omission of adjuvant therapy in low-risk patients to avoid unnecessary toxicity, while flagging high-risk patients for intensified surveillance or intervention.
4. Treatment Modality Dependence: The study highlights that prognostic factors are not universally static; specifically, the NAE score's utility appears dependent on the use of immunotherapy, providing a reference for precision management in different intervention groups.

The authors conclude that while the model provides a high-precision tool for risk stratification, the findings regarding de-escalation strategies require validation in prospective studies, and the model's long-term calibration may be influenced by the retrospective nature of the data and evolving treatment paradigms.