Biomarker – vom Sein und Wesen

Abstract

Zusammenfassung:Biomarker, ein inzwischen geradezu inflationär verwendeter Begriff! Während in den vergangenen Jahrzehnten die Bezeichnung „Marker“ vor allem mit Tumorerkrankungen und deren klinisch-chemischer Diagnostik verbunden war, hat die „-omics“-Welle der letzten Dekade eine Unmenge an neuen „Markern“ für alles und jedes in die medizinische Literatur gespült. Insbesondere der unkritische Umgang mit jenen Markern und die Unerfahrenheit derjenigen, die durch neue Techniken in vormals rein naturwissenschaftlichen Fachgebieten in die Lage versetzt wurden, diese neuen Marker zu messen, haben nicht nur zu einer großen Verunsicherung in bezug auf die Wertigkeit von Biomarkern an sich, sondern auch zu einer großen Enttäuschung in den wie Strohfeuer aufflammenden „-omics“-Disziplinen geführt. Kaum einer der oft hervorragend publizierten Biomarker hat den Weg in die Klinik gefunden, hochzitierten Wissenschaftlern konnten elementarste Fehler in (Prä-)analytik und Interpretation nachgewiesen werden und selbst die bisher als anerkannt angesehenen „klassischen“ Tumormarker sind vielfach in Misskredit geraten. Zu Unrecht! Denn mit alten wie neuen Markern hat die Labormedizin hervorragende Werkzeuge in der Hand, die, richtig angewandt, nicht nur das Potential haben, die Labordiagnostik zu revolutionieren, sondern auch das Gesicht des Faches verändern werden, weg von einer in vitro quantifizierenden Hilfsdisziplin hin zu einer integrativen und interpretativen Wissenschaft.

Accuracy and Calibration of Computational Approaches for Inpatient Mortality Predictive Modeling

“…Although lab requests usually should follow uniform diagnostic paths, their patterns are frequently highly variable–between disciplines, between clinics, and between physicians. Open “menu” request systems without digital expert systems or administrative restrictions facilitate the selection of favorite sets of “biomarkers”, that are neither mirrored by the actual guidelines nor by computational evidence [ 29 ], and even if published or hospital-based recommendations exist, they are scarcely followed [ 30 ]. The high degree of collinearity present in many routinely measured lab results not only points at tests potentially not additionally informative, but also blurs the contribution of a single parameter to a certain prediction or differentiation.…”

Sparse Proteomics Analysis – a compressed sensing-based approach for feature selection and classification of high-dimensional proteomics mass spectrometry data

“…Pancreatic cancer is not only a common and increasingly frequent [44], but also still a fatal disease, with a survival rate of 3-5% five years after diagnosis [45]. The conventional tumor marker, Carbohydrate Antigen 19-9 (CA19-9), as a blood group antigen not present in a significant proportion of the patients [46], shows insufficient diagnostic sensitivity and specificity (AUC 0.71), even in combination with the second-line tumor marker Carcinoembryonic Antigen (CEA, combined AUC 0.75) [47]. The need for better markers for screening and differential diagnosis is evident, as panceratic carcinoma would be principally curable if detected and identified very early in the course of the disease.…”

“…the so-called Warburg effect) [ 46 ], but also alters proteolytic activity [ 47 ]. Therefore, it might be naïve to expect a single marker capable to indicate presence, progression and exact type of the malignancy at once [ 48 ] 9– it might even be overly reductionistic to attribute these capabilities to a single model, even if it consists of several entities measured by different “- omics” technologies [ 43 ]. As Raftery states “basing inferences on a single “best” model as if the single selected model were true ignores model uncertainty, which can result in underestimating uncertainty about quantities of interest” [ 49 ], and the larger the “-omics” data-sets grow, the larger is the ‘probability, that there is not one “single best” predictive marker model, but instead several with comparable selectivity [ 48 ].…”

“…Biomarkers for clinical diagnostics comprise a wide field of applications (e.g. population-wide screening, early diagnostics, characterization, treatment guidance, efficacy and toxicity monitoring, prognosis, susceptibility estimation and many more) [ 43 ], each with special requirements for sensitivity and specificity, that are only partially condensed in the AUC as an overall selectivity measure [ 48 ]. Especially for screening purposes, sensitivity is extremely important [ 45 ], and clinically applied tests e.g.…”

“…for newborn screening frequently surpass the 0.99 hallmark [ 53 ]. Compared with the conventional, “not-for-screening” marker CA19-9, the SPA-based model shows considerable improvement, however there is still a big gap to screening suitability, which in the next years might be bridged by improved sensitivity of new instrumentation, refined algorithms (as the SPA), and combination with other “markers” from the “big data” field, enabling a more holistic view – not only of the disease, but also of the affected patient [ 43 ].…”