基于18F-FDG PET代谢参数及临床参数的列线图生存预测模型预测弥漫大B细胞淋巴瘤患者预后的价值

The prognostic value of nomogram survival prediction model based on 18F-FDG PET metabolic parameters and clinical parameters of patients with diffuse large B-cell lymphoma

    摘要
  • 摘要:
    目的 构建基于18F-氟脱氧葡萄糖（FDG）PET代谢参数及临床参数的列线图生存预测模型，并验证其对弥漫大B细胞淋巴瘤（DLBCL）患者预后的预测价值。
    方法 回顾性分析2011年3月至2019年11月于南京大学医学院附属鼓楼医院以及南京医科大学第一附属医院经组织病理学检查确诊的383例未经治疗的DLBCL患者的18F-FDG PET/CT影像学资料和临床资料，其中男性204例、女性179例，年龄19~93（47.3±14.9）岁。按照7∶3的比例采用随机数字表法将患者分为训练组（n=268例）和验证组（n=115例）。勾画并计算患者总肿瘤代谢体积（TMTV）和病灶糖酵解总量（TLG）。采用Kaplan Meier生存分析、单因素和多因素Cox比例风险回归模型对患者无进展生存（PFS）期及总生存（OS）期进行预后分析并构建生存预测模型。通过训练组和验证组的校准曲线、一致性指数（C-index）以及临床决策曲线分析（DCA）对预测模型进行评估。
    结果 训练组单因素分析结果显示，年龄、乳酸脱氢酶（LDH）水平、美国东部肿瘤协作组行为状态（ECOG PS）评分、Ann Abor分期、大包块、TMTV及TLG为预测PFS期的危险影响因素（HR=1.670~3.277，均P<0.05）；年龄、LDH水平、B症状、ECOG PS评分、Ann Abor分期、大包块、TMTV及TLG为预测OS期的危险影响因素（HR=1.661~4.193，均P<0.05）。训练组多因素分析结果表明，年龄、LDH水平、Ann Abor分期及TLG是预测DLBCL患者PFS期和OS期的独立影响因素（HR=1.589~3.367，均P<0.05）。校准曲线显示，该预测模型具有较好的预测一致性；C-index评估结果显示，训练组和验证组预测模型具有较高的准确性（PFS期：0.724对0.762； OS期：0.749对0.753）。临床DCA结果表明，预测模型可以给患者带来更多的临床获益。
    结论 基于18F-FDG PET代谢参数TLG及临床参数（年龄、LDH水平、Ann Abor分期）生存预测模型能够很好地对DLBCL患者进行预后评估，为精准个性化治疗提供可能。

     

    Abstract:
    Objective To construct prognostic nomogram models of 18F-fluorodeoxyglucose (FDG) PET-based metabolic and clinical parameters and validate their importance to survival prediction of patients with diffuse large B-cell lymphoma (DLBCL).
    Methods 18F-FDG PET image and clinical characteristics of 383 patients with DLBCL who received no treatments and underwent histopathology in the Affiliated Drum Tower Hospital, Medical School of Nanjing University, and the First Affiliated Hospital of Nanjing Medical University from March 2011 to November 2019 were retrospectively analyzed. The patients included 204 males and 179 females, aged 19–93(47.3±14.9) years old. The patients were randomly allocated as the training group (n=268) and validation group (n=115) at a 7∶3 ratio. The total metabolic tumor volume (TMTV) and total lesion glycolysis (TLG) were computed. Kaplan-Meier survival analysis, univariate and multivariate Cox proportional hazard regression models were used to evaluate progression-free survival (PFS) and overall survival (OS). The models were construct, and performance was assessed and validated with regard to calibration, discrimination, and clinical usefulness by calibration curve, concordance index (C-index), and decision curve analysis (DCA).
    Results Univariate analysis indicated that age, lactate dehydrogenase (LDH) level, Eastern Cooperative Oncology Group performance status (ECOG PS) score, Ann Abor stage, bulky, TMTV, and TLG were factors for predicting PFS in the training group (HR=1.670–3.277, all P<0.05). Age, LDH level, B symptoms, ECOG PS score, Ann Abor stage, bulky, TMTV, and TLG were factors for predicting OS in the training group (HR=1.661–4.193, all P<0.05). Multivariate Cox regression analyses showed that age, LDH level, Ann Arbor stage, and TLG were independent predictors of PFS and OS of DLBCL patients in the training group (HR=1.589–3.367, all P<0.05). Calibration curves showed that the models had good consistency for survival. The C-index showed that the models exhibited significant prognostic superiority in training and validation group (PFS: 0.724 vs. 0.762; OS: 0.749 vs. 0.753). Clicinal DCA showed that the prediction model could bring more clinical usefulness to patients.
    Conclusion 18F-FDG PET metabolic (TLG) and clinical (age, LDH level, and Ann Abor stage) parameters can successfully predict patient prognosis, which may promote precision medicine.

     

    • HTML全文
    • 参考文献(18)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回