Fuzzy logic-based tumor-marker profiles improved sensitivity in the diagnosis of lung cancer.

Background: The aim of this study was to improve the diagnostic efficiency of tumor markers in the diagnosis of lung cancer, by the mathematical evaluation of a tumor marker profile employing fuzzy logic modelling.

Methods: A panel of four tumor markers, i.e., carcinoembryonic antigen (CEA), cytokeratin 19 antibody (CYFRA 21-1), neuron-specific enolase (NSE), squamous cell carcinoma-related antigen (SCC) and, additionally, C-reactive protein (CRP), was measured in 175 newly diagnosed lung cancer patients with different histological types and stages. Results were compared with those in 120 control subjects, including 27 with chronic obstructive pulmonary diseases (COPD), 65 with pneumoconiosis, and 11 persons with acute inflammatory lung diseases. A classificator was developed using a fuzzy-logic rule-based system.

Results: Application of the fuzzy-logic rule-based system to the tumor marker values of CYFRA 21-1, NSE, and CRP yielded an increase in sensitivity of approximately 20%, i.e., 92%, compared with that of the best single marker, CYFRA 21-1(sensitivity, 72%). The corresponding specificity was 95%. The fuzzy classificator significantly improved the sensitivity of the tumor marker panel in stages I and IIIa for non-small-cell lung cancer, as well as in "limited disease" status for small-cell lung cancer. Also, the diagnosis of other stages of lung cancer was enhanced.

Conclusion: Fuzzy-logic analysis was proven to be more powerful than the measurement of single markers alone or combinations using multiple logistic regression analysis of all markers. Therefore, fuzzy logic offers a promising diagnostic tool to improve tumor marker efficiency.