External quality assessment of tumour marker analysis: state of the art and consequences for estimating diagnostic sensitivity and specificity

This review shows the current analytical quality for the following analytes used as tumour markers in the external quality assessment (EQA)-programmes of Instand e.V., a national EQA-organiser in Germany: Corticotropin (ACTH), growth hormone (GH, hGH), prolactin (PRL), chorionic gonadotropin (CG, hCG), calcitonin (CT, hCT), thyroglobulin (Tg), carcinoembryonic antigen (CEA), CA-Antigens 125, 72-4, 15-3 and 19-9, alpha foetoprotein (AFP) and prostate-specific antigen (PSA). The results from the participants show a large variation in the precision of the methods used as well as in the comparability of results between methods for the same analyte. In general, the hormones used as tumour markers show better performance than the "CA-markers", which are often inadequately standardised and defined. In the case of one CA-marker (CA 72-4/TAG 72-4), the differences between the lowest kit median concentration and highest kit median concentration for one sample pair were 440% and 580%. The corresponding figures for ACTH were 123% and 156% and for CEA 180% and 184%. The classical tumour markers such as carcinoembryonic antigen (CEA) and alpha foetoprotein (AFP) performed markedly better than the CA-markers and PSA with regards to both inter- and intra-method comparability. The inter-laboratory precision for a given kit and marker was acceptable in many cases. The results show that only results from the same kit/method for each tumour marker can be used for cumulative or time-dependent comparison of results - for example pre-operative and post-operative follow up. In the case of prostate specific antigen (PSA), the kits used for free and total PSA must come from the same producer, if the generally accepted ratios are to have any diagnostic value. The need for kit- and laboratory-specific reference ranges and cut-off values for setting diagnostic specificity and sensitivity is highlighted from the EQA-results. The situation for inter-method comparability for the CA-Markers has not improved over the past decade. With the exception of calcitonin for detecting medullary thyroid carcinoma, chorionic gonadotropin in germ-cell tumours in men and thyroglobulin after total thyroidectomy, none of the remaining analytes appear to be suitable for screening purposes.