食管鳞癌患者同步放化疗期间发生严重淋巴细胞减少的影响因素分析及预测模型建立

Analysis of influencing factors and establishment of prediction model for severe lymphocytopenia in patients with ESCC during concurrent chemoradiotherapy

目的 分析食管鳞癌(ESCC)患者同步放化疗(CCRT)期间发生严重绝对淋巴计数(ALC)减少的影响因素,建立ESCC患者CCRT期间发生严重ALC减少的预测模型并进行效能验证。方法 收集接受CCRT的286例ESCC患者的临床基线资料、放化疗方案,放疗剂量学参数等资料。根据治疗前ALC及治疗期间每周ALC,分析CCRT期间ALC动力学。根据患者CCRT期间是否发生严重ALC减少分为严重ALC减少组与非严重ALC减少组,比较两组收集的上述资料,并对CCRT期间发生严重ALC减少患者进行多因素Logistic回归分析筛选ESCC患者CCRT期间发生严重ALC减少的独立影响因素,基于影响因素构建严重ALC减少的风险预测模型(列线图)。采用受试者工作特征(ROC)、一致性指数(C-index)、校准曲线、决策分析(DCA)曲线评估列线图模型的区分度、一致性、校准度及临床适用性。结果 ESCC患者在CCRT期间ALC呈现逐渐下降趋势(P<0.05)。两组BMI、治疗前ALC、平均椎体剂量(MVD)、循环免疫细胞估计剂量(EDIC)比较差异有统计学意义(P均<0.05)。BMI(OR=0.829,95%CI:0.760~0.900)、治疗前ALC(OR=0.412,95%CI:0.253~0.648)、MVD(OR=1.079,95%CI:1.011~1.155)和EDIC(OR=1.711,95%CI:1.422~2.292)是ESCC患者CCRT期间发生严重ALC减少的独立影响因素(P<0.05)。基于上述影响因素构建列线图模型。列线图模型预测CCRT期间发生严重ALC减少的ROC下面积为0.799(95%CI:0.748~0.850);C-index为0.789;预测ALC减少风险与实际发生的风险较一致;预测发生严重ALC减少的可能在0.25~0.90概率范围内具有较好的临床适用性。结论 ESCC患者在CCRT期间发生严重ALC减少的独立影响因素为BMI、治疗前ALC、MVD、EDIC,基于上述影响因素建立的预测ESCC患者在CCRT期间发生严重ALC减少的列线图模型具有良好的区分度、一致性、校准度及临床实用性。

Objective To evaluate the influencing factors for severe absolute lymphatic count(ALC) reduction during concurrent chemoradiotherapy(CCRT) in patients with esophageal squamous cell carcinoma(ESCC),to develop a predictive model for severe ALC reduction during CCRT in patients with ESCC,and to validate its efficacy.Methods Clinical baseline data,radiotherapy regimen,and radiotherapy dosimetry parameters were collected from 286 ESCC patients who underwent CCRT.Trends in ALC during CCRT were analyzed based on pre-treatment ALC and weekly ALC during treatment.Patients were divided into the severe ALC reduction group and non-severe ALC reduction group according to whether severe ALC reduction occurred during CCRT.We compared the above information between the two groups,and screened the independent risk factors for severe ALC reduction during CCRT in ESCC patients by multifactorial Logistic regression analysis,constructed the risk prediction model based on risk factors,and plotted a nomogram(nomogram model).Receiver operating characteristics(ROC) curve,consistency index(C-index),calibration curve,and decision analysis(DCA) curve were used to assess the discrimination,consistency,calibration,and clinical applicability of the nomogram model.Results ESCC patients showed a gradual decrease in ALC during CCRT(P<0.05).Statistically significant differences were found in BMI,pre-treatment ALC,mean vertebral dose(MVD),and estimated dose of circulating immune cells(EDIC) between the two groups(all P<0.05).BMI(OR=0.829,95%CI:0.760-0.900),pretreatment ALC(OR=0.412,95% CI:0.253-0.648),MVD(OR=1.079,95% CI:1.011-1.155) and EDIC(OR=1.711,95%CI:1.422-2.292) were independent influencing factors for severe ALC reduction during CCRT in ESCC patients(all P<0.05).A nomogram model was constructed based on the above risk factors.The area under the ROC of the nomogram model in predicting the occurrence of severe ALC reduction during CCRT was 0.799(95% CI:0.748-0.850),the C-index was 0.789,and the predicted risk of ALC reduction was consistent with the risk of the actual occurrence;the prediction of severe ALC reduction had good clinical applicability in the probability range of 0.25 to 0.90.Conclusions The independent influencing factors for severe ALC reduction during CCRT in ESCC patients are BMI,pretreatment ALC,MVD,and EDIC,and the nomogram model for predicting severe ALC reduction during CCRT in ESCC patients based on the above mentioned influencing factors has good differentiation,consistency,calibration,and clinical applicability.