吉林大学学报(医学版) ›› 2024, Vol. 50 ›› Issue (4): 1098-1108.doi: 10.13481/j.1671-587X.20240424

• 临床研究 • 上一篇    下一篇

多发性骨髓瘤患者胸部CT第4胸椎平面人体成分与预后的关联性分析

白雪,王晨晨,石张镇,毕林涛()   

  1. 吉林大学中日联谊医院肿瘤血液科,吉林 长春 130033
  • 收稿日期:2023-05-13 出版日期:2024-07-28 发布日期:2024-08-01
  • 通讯作者: 毕林涛 E-mail:bilt@jlu.edu.cn
  • 作者简介:白 雪(1997-),女,山东省菏泽市人,在读硕士研究生,主要从事内科学临床方面的研究。
  • 基金资助:
    吉林省科技厅自然科学基金项目(YDZJ202201ZYTS117)

Analysis on correlation between body components at T4 thoracic vertebra plane on chest CT in patients with multiple myeloma and prognosis

Xue BAI,Chenchen WANG,Zhangzhen SHI,Lintao BI()   

  1. Department of Tumor Hematology,China-Japan Union Hospital,Jilin University,Changchun 130033,China
  • Received:2023-05-13 Online:2024-07-28 Published:2024-08-01
  • Contact: Lintao BI E-mail:bilt@jlu.edu.cn

摘要:

目的 利用深度学习模型自动分割初诊多发性骨髓瘤(MM)患者胸部CT第4胸椎平面4种人体成分,探讨4种人体成分与MM患者预后的相关性。 方法 对2017年1月-2021年12月于本院确诊的MM患者的临床资料进行回顾性分析,收集患者的年龄、性别、体质量、身高和体质量指数(BMI)等临床信息,收集患者血清中乳酸脱氢酶(LDH)、血钙(Ca)、血肌酐(Scr)、白蛋白(Alb)、血红蛋白(Hb)、β2-微球蛋白(β2-MG)和血清游离轻链水平等实验室数据,利用深度学习模型分别将79例规律进行疗效评价MM患者的胸部CT影像结果分割为胸大肌、胸小肌、皮下脂肪和纵隔脂肪4种人体成分,采用Image J软件分别测量第4胸椎平面4种人体成分面积,分析其与MM患者预后的相关性,并进行Kaplan-Meier生存分析。 结果 单因素分析,皮下脂肪面积、血Ca水平、Scr水平和国际分期系统(ISS)分期与MM患者总生存期(OS)有关(HR=2.260,95%CI:1.116~4.578,P=0.024;HR=2.088,95%CI:1.007~4.327,P=0.048; HR=2.209,95%CI:1.105~4.414,P=0.025;HR=1.730,95%CI:1.040~2.879,P=0.035)。多因素分析,4种人体成分中皮下脂肪面积是影响MM患者预后的独立危险因素(95%CI:1.228~5.782,P=0.013)。Log-Rank检验,在所有患者中,与皮下脂肪面积高值组比较,皮下脂肪面积低值组MM患者OS缩短(P=0.018);在不同性别患者中,皮下脂肪面积高值组与皮下脂肪面积低值组MM患者OS比较差异无统计学意义(P>0.05);在未行造血干细胞移植的患者中,与皮下脂肪面积高值组比较,皮下脂肪面积低值组患者OS缩短(P=0.037)。 结论 第4胸椎平面4种人体成分中,皮下脂肪组织面积与MM患者OS有关,是MM患者预后的独立危险因素,而纵隔脂肪组织、胸大肌和胸小肌面积对MM患者的预后无预测作用。

关键词: 多发性骨髓瘤, 计算机断层扫描, 人体成分, 深度学习模型

Abstract:

Objective To automatically segment four body components at the T4 thoracic veertebra plane on chest CT in the newly diagnosed multiple myeloma (MM) patients by deep learning model, and to discuss the correlation between the four body components and the prognosis of the MM patients. Methods The retrospective analysis was conducted on the clinical data of the MM patients diagnosed in our hospital from January 2017 to December 2021. The clinical informations such as age, gender, weight, height, and body mass index (BMI) of the patients were collected. The laboratory data of the patients were collected, including serum levels of lactate dehydrogenase (LDH), calcium (Ca), creatinine (Scr), albumin (Alb), hemoglobin (Hb), β2-microglobulin (β2-MG), and serum free light chains. The chest CT images of 79 regularly evaluated MM patients detected by deep learning model were divide into four body components: pectoralis major, pectoralis minor, subcutaneous fat, and mediastinal fat. Image J software was used to detect the areas of the four body components at the T4 thoracic vertebra plane, and their correlation with the prognosis of the MM patients was analyzed by Kaplan-Meier survival analysis. Results The univariate analysis results showed that the area of subcutaneous fat, serum Ca levels, Scr levels, and International Staging System (ISS) stage were related to the overall survival (OS) of the MM patients (HR=2.260, 95% CI: 1.116-4.578, P=0.024; HR=2.088, 95% CI: 1.007-4.327, P=0.048; HR=2.209, 95% CI: 1.105-4.414, P=0.025; HR=1.730, 95% CI: 1.040-2.879, P=0.035). The multivariate analysis results showed that the area of subcutaneous fat among the four body components was an independent risk factor affecting the prognosis of the MM patients (95% CI: 1.228-5.782, P=0.013). The Log-Rank test results showed that compared with high subcutaneous fat area group, the OS of the patients in low subcutaneous fat area group was decreased(P=0.018). There was no significant difference in OS of the patients with different genders between high subcutaneous fat area group and low subcutaneous fat area group (P>0.05). In the patients without hematopoietic stem cell transplantation, compared with high subcutaneous fat area group, the OS of the patients in low subcutaneous fat area group was decreased (P=0.037). Conclusion Among the four body components at the T4 thoracic vertebra plane, the area of subcutaneous fat is related to the OS of the MM patients and it is an independent risk factor for the prognosis of the MM patients, while the areas of mediastinal fat, pectoralis major, and pectoralis minor have no predictive value for the prognosis of the MM patients.

Key words: Multiple myeloma, Computed tomography, Body composition, Deep learning model

中图分类号: 

  • R733.3