Abstract：
Plasma and tissue biopsy have both used for targeting actionable driver gene mutations in lung cancer, whose concordance is imperfect. A reliable method to predict the concordance is urgently needed to ease clinical application.
A total of 1012 plasma samples, including 519 with paired-tissue biopsy samples, derived from lung adenocarcinoma patients were retrospectively enrolled. We assessed the associations of several clinicopathological characteristics and serum tumor markers with the concordance between plasma and tissue biopsies.
When carcinoembryonic antigen (CEA) levels were higher than thresholds of 15.01 ng/ml and 51.15 ng/ml, the positive predictive value of concordance reached 90% and 95%, respectively. When CEA levels were lower than thresholds of 5.19 ng/ml and 3.26 ng/mL, the negative predictive value of concordance reached 45% and 50%. The performance of CYFRA21-1 in predicting concordance was similar but inferior to CEA (AUC: 0.727 vs. 0.741, p = 0.633). The performance of CEA combined with CYFRA21-1 in predicting the concordance was similar to that of the combination of independent factors derived from the LASSO regression model (AUC: 0.796 vs. 0.818, p = 0.067). CEA (r = 0.47, p < 0.01) and CYFRA21-1 levels (r = 0.45, p < 0.05) were significantly correlated with the maximum variant allele frequency, respectively.
CEA combined with CYFRA21-1 could effectively predict the concordance between plasma and tissue biopsies, which could be used for evaluating the priority of plasma and tissue biopsies for gene testing to timely guide clinical applications in advanced lung adenocarcinoma patients.
A total of 1012 plasma samples, including 519 with paired-tissue biopsy samples, derived from lung adenocarcinoma patients were retrospectively enrolled. We assessed the associations of several clinicopathological characteristics and serum tumor markers with the concordance between plasma and tissue biopsies.
When carcinoembryonic antigen (CEA) levels were higher than thresholds of 15.01 ng/ml and 51.15 ng/ml, the positive predictive value of concordance reached 90% and 95%, respectively. When CEA levels were lower than thresholds of 5.19 ng/ml and 3.26 ng/mL, the negative predictive value of concordance reached 45% and 50%. The performance of CYFRA21-1 in predicting concordance was similar but inferior to CEA (AUC: 0.727 vs. 0.741, p = 0.633). The performance of CEA combined with CYFRA21-1 in predicting the concordance was similar to that of the combination of independent factors derived from the LASSO regression model (AUC: 0.796 vs. 0.818, p = 0.067). CEA (r = 0.47, p < 0.01) and CYFRA21-1 levels (r = 0.45, p < 0.05) were significantly correlated with the maximum variant allele frequency, respectively.
CEA combined with CYFRA21-1 could effectively predict the concordance between plasma and tissue biopsies, which could be used for evaluating the priority of plasma and tissue biopsies for gene testing to timely guide clinical applications in advanced lung adenocarcinoma patients.
摘要：
背景：血浆和组织活检均用于靶向肺癌中可行的驱动基因突变，其一致性是不完美的。迫切需要一种可靠的方法来预测一致性，以简化临床应用。
方法：总共1012个血浆样本，包括519个配对组织活检样本，来自肺腺癌的患者被回顾性纳入.我们评估了几种临床病理特征和血清肿瘤标志物与血浆和组织活检之间一致性的关联。
结果：当癌胚抗原（CEA）水平高于15.01ng/ml和51.15ng/ml的阈值时，一致性的阳性预测值达到90%和95%，分别。当CEA水平低于阈值5.19ng/ml和3.26ng/mL时，一致性的阴性预测值分别达到45%和50%。CYFRA21-1在预测一致性方面的表现相似，但不如CEA(AUC：0.727vs.0.741,p=0.633)。CEA联合CYFRA21-1在预测一致性方面的表现与从LASSO回归模型得出的独立因素组合相似（AUC：0.796vs.0.818,p=0.067)。CEA（r=0.47，p<0.01）和CYFRA21-1水平（r=0.45，p<0.05）与最大变异等位基因频率显着相关，分别。
结论：CEA联合CYFRA21-1可有效预测血浆和组织活检的一致性。可用于评估血浆和组织活检基因检测的优先级，以及时指导晚期肺腺癌患者的临床应用。
方法：总共1012个血浆样本，包括519个配对组织活检样本，来自肺腺癌的患者被回顾性纳入.我们评估了几种临床病理特征和血清肿瘤标志物与血浆和组织活检之间一致性的关联。
结果：当癌胚抗原（CEA）水平高于15.01ng/ml和51.15ng/ml的阈值时，一致性的阳性预测值达到90%和95%，分别。当CEA水平低于阈值5.19ng/ml和3.26ng/mL时，一致性的阴性预测值分别达到45%和50%。CYFRA21-1在预测一致性方面的表现相似，但不如CEA(AUC：0.727vs.0.741,p=0.633)。CEA联合CYFRA21-1在预测一致性方面的表现与从LASSO回归模型得出的独立因素组合相似（AUC：0.796vs.0.818,p=0.067)。CEA（r=0.47，p<0.01）和CYFRA21-1水平（r=0.45，p<0.05）与最大变异等位基因频率显着相关，分别。
结论：CEA联合CYFRA21-1可有效预测血浆和组织活检的一致性。可用于评估血浆和组织活检基因检测的优先级，以及时指导晚期肺腺癌患者的临床应用。
