Risk Stratification of Postoperative Cardiopulmonary Toxicity after Trimodality Therapy for Esophageal Cancer

Abstract

<h3>Purpose/Objective(s)</h3> Postoperative toxicity after trimodality therapy for esophageal cancer profoundly impacts patient quality of life and may impact overall survival (OS). We interrogated a large, multi-institutional database of patients to refine a surrogate to report cardiopulmonary total toxicity burden (CPTTB) and study whether toxicity attributed to chemoradiation (CRT) predicted for this endpoint. <h3>Materials/Methods</h3> Our group assessed a multi-institutional database of patients diagnosed with biopsy-proven esophageal cancer and treated with neoadjuvant CRT, followed by surgical resection. CPTTB was defined using the pre-existing analysis of total toxicity burden (Lin et al. JCO 2020). A subset of these variables (myocardial infarction, non-ischemic cardiac toxicity, pleural effusion, pneumonia, and acute respiratory distress syndrome) was utilized. (Table). CPTTB was validated for its impact on OS, length of postoperative hospital stay (LOS), and death or readmission within 60 days of surgery (DR60). To predict for CPTTB, recursive partitioning analysis was used. Variables utilized in the highest fidelity RPA models were incorporated into the final score. <h3>Results</h3> From 3 institutions, 571 patients were included. Most patients had stage 2A (28%) or 3 (61%) adenocarcinoma (93%). Patients were treated with 3D (37%), IMRT (44%), and proton therapy (19%). Median CPTTB was 0 (interquartile range [IQR]: 0-30), with 61 patients having major CPTTB (defined as CPTTB ≥ 70). Increasing CPTTB as a continuous variable and major CPTTB were predictive of decreased OS (p<0.001 and p=0.005), LOS (p<0.001) and DR60 (p<0.001). Next, an RPA-based model was developed predictive of major CPTTB. This risk score involved one point for: age ≥ 65, grade ≥ 2 nausea or esophagitis attributed to CRT, and grade ≥ 3 hematologic toxicity attributed to CRT. This model was predictive of CPTTB (area under the curve [AUC]: 0.66) and DR60 (p=0.017). A score of 1 showed an odds ratio (OR) of 5.15 (95% CI: 1.54-17.26, p=0.008), and a score of 2-3 showed OR of 9.78 (95% CI: 2.91-32.82, p<0.001) for major CPTTB. The CPTTB risk score was also predictive of continuous CPTTB (p<0.001). Patients treated with 3D radiotherapy had increased major CPTTB (18.5% vs. 6.1%, p<0.001), but the model demonstrated higher predictive power for major CPTTB (AUC=0.70) among patients treated with IMRT or proton therapy. <h3>Conclusion</h3> CPTTB predicts for OS, LOS, and DR60. Patients with age ≥ 65 years and CRT toxicity are at highest risk for major CPTTB and may benefit from further pre-operative management or even omission of surgery. 3D radiotherapy is associated with increased toxicity.