BACKGROUND
The authors built a model for lung cancer diagnosis previously based on the blood biomarkers progastrin-releasing peptide (ProGRP), carcinoembryonic antigen (CEA), squamous cell carcinoma antigen (SCC), and cytokeratin 19 fragment (CYFRA21-1). In the current study, they examined whether modification of the model to include relevant clinical information, risk factors, and low-dose chest computed tomography screening would improve the performance of the biomarker panel in large cohorts of Chinese adults.
METHODS
The current study was a large-scale multicenter study (ClinicalTrials.gov identifier NCT01928836) performed in a Chinese population. A total of 715 participants were enrolled from 5 regional centers in Beijing, Henan, Nanjing, Shanghai, and Chongqing between October 2012 and February 2014. Serum biomarkers ProGRP, CEA, SCC, and CYFRA21-1 were analyzed on the ARCHITECT i2000SR. Relevant clinical information was collected and used to develop a patient risk model and a nodule risk model.
RESULTS
The resulting patient risk model had an area under the receiver operating characteristic (ROC) curve of 0.7037 in the training data set and 0.7190 in the validation data set. The resulting nodule risk model had an area under the ROC curve of 0.9151 in the training data set and 0.5836 in the validation data set. Moreover, the nodule risk model had a relatively higher area under the ROC curve (0.9151 vs 0.8360; P = 0.001) compared with the American College of Chest Physician model in patients with lung nodules.
CONCLUSIONS
Both the patient risk model and the nodule risk model, developed for the early diagnosis of lung cancer, demonstrated excellent discrimination, allowing for the stratification of patients with different levels of lung cancer risk. These new models are applicable in high-risk Chinese populations. Cancer 2017. © 2017 American Cancer Society.
http://ift.tt/2xmNx1g
Δεν υπάρχουν σχόλια:
Δημοσίευση σχολίου