The identification of survival predictors is crucial for early intervention to improve outcome in acute myeloid leukemia (AML). This study aim to identify chest computed tomography (CT)-derived features to predict prognosis for acute myeloid leukemia (AML). 952 patients with pathologically-confirmed AML were retrospectively enrolled between 2010 and 2020. CT-derived features (including body composition and subcutaneous fat features), were obtained from the initial chest CT images and were used to build models to predict the prognosis. A CT-derived MSF nomogram was constructed using multivariate Cox regression incorporating CT-based features. The performance of the prediction models was assessed with discrimination, calibration, decision curves and improvements. Three CT-derived features, including myosarcopenia, spleen_CTV, and SF_CTV (MSF) were identified as the independent predictors for prognosis in AML (P < 0.01). A CT-MSF nomogram showed a performance with AUCs of 0.717, 0.794, 0.796 and 0.792 for predicting the 1-, 2-, 3-, and 5-year overall survival (OS) probabilities in the validation cohort, which were significantly higher than the ELN risk model. Moreover, a new MSN stratification system (MSF nomogram plus ELN risk model) could stratify patients into new high, intermediate and low risk group. Patients with high MSN risk may benefit from intensive treatment (P = 0.0011). In summary, the chest CT-MSF nomogram, integrating myosarcopenia, spleen_CTV, and SF_CTV features, could be used to predict prognosis of AML.

中文翻译：

---

基于胸部 CT 的列线图预测急性髓系白血病的生存率

生存预测因子的识别对于早期干预以改善急性髓系白血病 (AML) 的预后至关重要。本研究旨在识别胸部计算机断层扫描 (CT) 衍生的特征，以预测急性髓系白血病 (AML) 的预后。回顾性纳入2010年至2020年间952例经病理证实的AML患者。从最初的胸部CT图像中获得CT衍生特征（包括身体成分和皮下脂肪特征），并用于建立模型来预测预后。使用多元 Cox 回归并结合基于 CT 的特征构建了 CT 衍生的 MSF 列线图。通过区分、校准、决策曲线和改进来评估预测模型的性能。三个 CT 衍生特征，包括肌少症、脾_CTV 和 SF_CTV (MSF) 被确定为 AML 预后的独立预测因子 (P < 0.01)。 CT-MSF 列线图显示，预测验证队列中 1 年、2 年、3 年和 5 年总生存 (OS) 概率的 AUC 分别为 0.717、0.794、0.796 和 0.792，显着高于ELN 风险模型。此外，新的MSN分层系统（MSF列线图加ELN风险模型）可以将患者分为新的高、中、低风险组。 MSN 高风险患者可能会受益于强化治疗 (P = 0.0011)。总之，整合肌少症、脾_CTV 和 SF_CTV 特征的胸部 CT-MSF 列线图可用于预测 AML 的预后。