基于18F-FDG PET/CT代谢参数的风险预测模型对胃癌淋巴结分期的预测价值及效能评价

Predictive value and efficacy assessment of a risk prediction model based on the metabolic parameters of 18F-FDG PET/CT for lymph node staging of gastric cancer

    摘要
  • 摘要:
    目的  探讨基于18F-氟脱氧葡萄糖（FDG） PET/CT代谢参数的风险预测模型在预测胃癌淋巴结分期中的价值并评价其效能。
    方法  回顾性分析2011年10月至2022年9月在福建省肿瘤医院行胃癌根治术的167例患者的临床资料及18F-FDG PET/CT影像资料，其中男性129例、女性38例，年龄62（56，69）岁。采用多元Logistic回归筛选出影响胃癌淋巴结分期的独立影响因素。采用受试者工作特征（ROC）曲线分析评估多个参数及其联合参数对胃癌淋巴结分期的价值。构建风险预测模型和列线图，确定最终诊断效能，采用Hosmer-Lemeshow（H-L）检验评价模型的校准度。
    结果  多元Logistic回归分析结果表明，性别、肿瘤浸润深度、肿瘤长径、糖酵解总量（TLG）是影响胃癌N0分期的独立影响因素，其比值比（OR）（95%CI）分别为0.217（0.060~0.784）、5.907（1.984~17.589）、0.622（0.427~0.905）和1.010（1.004~1.017），均P<0.05。癌胚抗原（CEA）、肿瘤长径分别是影响胃癌N3a、N3b分期的独立影响因素，其OR（95%CI）分别为1.018（1.004~1.033）、1.258（1.074~1.473），均P<0.05。ROC曲线分析结果显示，性别、肿瘤浸润深度、肿瘤长径、TLG预测胃癌N0分期的曲线下面积（AUC）分别为0.565、0.706、0.725、0.652。CEA、肿瘤长径预测胃癌N3a、N3b分期的AUC分别为0.648、0.710。ROC曲线评价风险预测模型预测胃癌N0分期的AUC为0.796（95%CI：0.716~0.876，P<0.001），灵敏度为56.4%，特异度为89.8%，阳性预测值为62.9%，阴性预测值为87.1%，准确率为82.0%。H-L检验结果显示，该模型具有较好的校准度（χ2=9.067，P=0.337）。
    结论  基于18F-FDG PET/CT代谢参数的风险模型在预测胃癌N0分期中有较大应用价值。

     

    Abstract:
    Objective To explore the predictive value and assess the effectiveness of a risk prediction model based on 18F-fluorodeoxyglucose (FDG) PET/CT metabolic parameters for lymph node staging in gastric cancer (GC).
    Methods The clinical data and 18F-FDG PET/CT imaging data of 167 patients who underwent radical GC surgery at Fujian Cancer Hospital from October 2011 to September 2022 were retrospectively analyzed. The cohort included 129 males and 38 females, with a median age of 62(56, 69) years. Multiple Logistic regression analysis was used to identify independent factors that affect the lymph node staging of GC. The predictive value of individual metabolic parameters and their combinations was evaluated using receiver operating characteristic (ROC) curve analysis. A risk prediction model and nomogram were constructed, and their diagnostic performance was assessed. The calibration of the model was evaluated using the Hosmer-Lemeshow (H-L) test.
    Results Multiple Logistic regression analysis show that gender, tumor infiltration depth, tumor length diameter, and total lesion glycolysis (TLG) are independent factors affecting the N0 stage of GC. The odds ratios (OR) (95%CI) are 0.217(0.060–0.784), 5.907(1.984–17.589), 0.622(0.427–0.905), and 1.010(1.004–1.017), all P<0.05. Carcinoembryonic antigen (CEA) and tumor length diameter are independent factors affecting the N3a and N3b stages of GC, respectively. Their OR (95%CI) are 1.018(1.004–1.033) and 1.258(1.074–1.473), both P<0.05. ROC curve analysis show that the area under curve (AUC) for predicting the N0 stage of GC using gender, tumor infiltration depth, tumor length diameter, and TLG are 0.565, 0.706, 0.725, and 0.652, respectively. The AUCs for CEA and tumor length diameter in predicting the N3a and N3b stages of GC are 0.648 and 0.710, respectively. The risk prediction model for predicting the N0 stage has an AUC of 0.796(95%CI: 0.716–0.876, P<0.001), with a sensitivity of 56.4%, specificity of 89.8%, positive predictive value of 62.9%, negative predictive value of 87.1%, and accuracy of 82.0%. The H-L test results indicates good calibration of the model (χ2=9.067, P=0.337).
    Conclusion The risk prediction model based on 18F-FDG PET/CT metabolic parameters demonstrates good performance in predicting lymph node N0 stage in GC.

     

    • HTML全文
    • 参考文献(26)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回