CEA, CYFRA 21-1, NSE, and ProGRP in the diagnosis of lung cancer: a multivariate approach / CEA, CYFRA 21-1, NSE und ProGRP in der Diagnostik des Lungenkarzinoms: eine multivariate Analyse

Abstract

Abstract We retrospectively studied the single and combined diagnostic value of carcinoembryonic antigen (CEA), cytokeratin fragment 19 (CYFRA 21-1), neuron specific enolase (NSE) and pro-gastrin-releasing peptide (ProGRP), which were routinely analysed in patients with lung tumours of unknown origin at the time of admission to hospital. Inclusion criteria were the determination of CEA (AxSYM/Abbott), CYFRA 21-1 (ElecSys/Roche) and NSE (Kryptor/Brahms). We examined 1747 patients, where 1325 suffered from lung cancer (LC; small cell lung cancer, SCLC: n=194; non-small cell lung cancer, NSCLC: n=1015; others: n=116), 318 from benign lung diseases and 104 from lung metastases due to another primary malignancy. As ProGRP (ELISA ALSI/IBL) became available only recently, there are less data points of this marker. In total, 99.8% of LC patients released at least one of the four biomarkers (defined as values exceeding the median of healthy controls), and for the discrimination between benign disease (BD) and malignant lung disease each marker reached 100% tumour specificity at high levels (CEA: 20 ng/mL; CYFRA 21–1: 40 ng/mL; NSE: 45 ng/mL; ProGRP: 250 pg/mL). At a specificity of >99%, ProGRP reached the highest diagnostic efficacy for SCLC with 57% true positive results, CEA had the highest capacity (17%) to detect malignant lung tumours in general and adenocarcinomas of the lung with 29%. CYFRA 21-1 was dominant for squamous cell carcinomas (12%). Combining the four markers leads with the prerequisite of high specificity (>99%) to 50% true positives for malignant lung tumours, 44% for NSCLC, 36% for squamous cell carcinomas, 53% for adenocarcinomas, and 78% for SCLC, respectively. In cases of lung tumours of unknown origin, the combined use of CEA, CYFRA 21-1, NSE and ProGRP is useful for the differentiation between benign and primary or secondary malignant disease and suggests the assignment to histological subtypes.