Department of Gastroenterology and Hepatology, Tianjin, Medical University General Hospital, Tianjin, China.
Tianjin Institute of Digestive Disease, Tianjin Medical University General Hospital, Tianjin, China.
JPEN J Parenter Enteral Nutr. 2021 Feb;45(2):403-413. doi: 10.1002/jpen.1841. Epub 2020 May 2.
Alterations in body compositions are related to poor outcomes and the presence of complications in cirrhosis. However, no predictive tools combining all these anthropometric parameters are applicable in the clinical setting. We aimed to clarify the potential utility of body compositions and develop a nomogram incorporating any independent factor for prognosticating long-term mortality in cirrhosis.
A total of 414 patients were randomized into primary (n = 274) and validation (n = 140) cohorts. X-tile was performed to identify optimal cut points for stratifying participants. Multivariate Cox regression was performed, and nomogram incorporating body compositions were generated. The utility of developed models was evaluated by Harrell concordance index (C-index), calibration curve, and decision curve analysis (DCA).
Stratifying by X-tilederived cut points, low skeletal muscle index (myopenia), high intramuscular adipose tissue content (myosteatosis), and the ratio of high visceral to subcutaneous adipose tissue area (adiposity) was independently associated with 3-year mortality. A sex-stratified nomogram incorporating anthropometric indices and clinical factors resulted in moderate discriminative accuracy, with a C-index of 0.787 (95% CI, 0.736-0.838) and 0.789 (95% CI, 0.727-0.851) in males and females, respectively. The calibration curve showed predictive survival corresponding optimally with the actual outcomes. Our models were feasible in the clinical settings based on DCA. Similar results were observed in the validation cohort. Additionally, participants could be classified into 3 distinct risk groups by the nomogram.
Our proposed nomogram embedding body compositions rendered an individualized predictive tool for long-term mortality in cirrhosis.
身体成分的改变与肝硬化不良预后和并发症的发生有关。然而，目前尚无将所有这些人体测量参数结合起来的预测工具可应用于临床。我们旨在阐明身体成分的潜在效用，并开发一个包含任何独立因素的列线图，以预测肝硬化患者的长期死亡率。
共 414 例患者被随机分为主要队列（n = 274）和验证队列（n = 140）。X-tile 用于确定分层参与者的最佳切点。进行多变量 Cox 回归，生成包含身体成分的列线图。通过 Harrell 一致性指数（C 指数）、校准曲线和决策曲线分析（DCA）评估开发模型的实用性。
通过 X-tile 衍生的切点分层，低骨骼肌指数（肌减少症）、高肌内脂肪组织含量（肌脂病）和高内脏与皮下脂肪组织面积比（肥胖）与 3 年死亡率独立相关。一个包含人体测量指数和临床因素的性别分层列线图产生了中等的判别准确性，男性和女性的 C 指数分别为 0.787（95%CI，0.736-0.838）和 0.789（95%CI，0.727-0.851）。校准曲线显示预测生存率与实际结果最佳对应。基于 DCA，我们的模型在临床环境中是可行的。在验证队列中也观察到了类似的结果。此外，通过列线图可以将参与者分为 3 个不同的风险组。
我们提出的嵌入身体成分的列线图为肝硬化的长期死亡率提供了一个个体化的预测工具。
