gms | German Medical Science

27. Deutscher Krebskongress

Deutsche Krebsgesellschaft e. V.

22. - 26.03.2006, Berlin

SELDI-TOF mass spectrometry enables detection of potential biomarkers for colorectal cancer in patient serum

Meeting Abstract

  • corresponding author presenting/speaker Jens K. Habermann - Universitätsklinikum Schleswig-Holstein, Lübeck, Deutschland
  • Uwe J. Roblick - Universitätsklinikum Schleswig-Holstein, Lübeck
  • Brian T. Luke - NCI Frederick, Frederick
  • DaRue Prieto - NCI Frederick, Frederick
  • Elisabeth Oevermann - Universitätsklinikum Schleswig-Holstein, Lübeck
  • Timothy D. Veenstra - NCI Frederick, Frederick
  • Michael Duchrow - Universitätsklinikum Schleswig-Holstein, Lübeck
  • Hans-Peter Bruch - Universitätsklinikum Schleswig-Holstein, Lübeck
  • Gert Auer - Karolinska Institutet, Stockholm
  • Thomas Ried - National Institutes of Health, Bethesda

27. Deutscher Krebskongress. Berlin, 22.-26.03.2006. Düsseldorf, Köln: German Medical Science; 2006. DocPO521

Die elektronische Version dieses Artikels ist vollständig und ist verfügbar unter:

Veröffentlicht: 20. März 2006

© 2006 Habermann et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen ( Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.



Introduction:Colorectal carcinomas are the third most common malignancies in the Western World. Despite screening programs about 70% of tumors are detected at advanced stages (UICC III/IV). A late diagnosis results in a significant reduction of average survival times. Therefore, alternative tools for early detection are needed.

Methods:We explored whether detection of malignant disease would be possible through identification of cancer specific protein signatures in serum samples. Sera from patients with colorectal malignancy (n=52) and control individuals (n=32) were analyzed using surface-enhanced laser desorption ionization time-of-flight (SELDI-TOF) mass spectrometry. Class comparison and custom developed algorithms identified a set of 13 serum features (m/z values) that distinguished the malignant samples and the healthy controls of the training set. These features were validated by class prediction using an independently collected, blinded validation set of 55 sera.The predictions utilized a combination of 16 classifiers based on K-Nearest Neighbors.This deterministic classification procedure has no adjustable parameters and is independent of the order of the training samples.We chose this conservative approach because of the controversy associated with previously published reports on SELDI analyses. Only samples that revealed identical results with all 16 models were considered for further classification into the malignant or control group.

Results: Applying this criterion, 48 of 55 samples were classifiable. Of those 48 samples, a total of 98% were classified correctly. None of the normal samples were assigned to the malignant group. The only misclassified sample represented an inoperable tumor whose characteristics were underrepresented for classifier training. Our SELDI based protein profiling with KNN class prediction based on 13 features permitted the discrimination of colorectal cancer-associated sera – including those with early stage disease – from healthy controls in an independent, blinded validation set with 96.7% sensitivity and 100% specificity.

Conclusion: Large scale prospective multicenter studies are now warranted to establish the clinical value of the identified discerning features for colorectal cancer screening.