The diagnosis of malignant pleural effusion suffers from low sensitivity of its cytological evaluation. We therefore asked whether proteomic profiles obtained by matrix assisted laser desorption ionization-mass spectrometry (MALDI-MS) can discriminate malignant from non-malignant effusions.
MALDI-MS profiles were obtained from 49 cytologically proven malignant effusions and from 48 cytologically-negative exudative pleural effusions. We compared the two groups first in a training set of 27 malignant and 26 non-malignant exudative effusions, built a prediction model and tested its accuracy in a leave-one-out cross-validation class prediction analysis. We then tested the selected features in an independent test set of 22 malignant and 22 non-malignant effusions and estimated the misclassification rate.
In a training set of effusions, we identified a profile discriminating malignant effusions from others based on 897 MS features. Class prediction models based on the discriminatory protein peaks allowed correct classification of pleura from malignant and nonmalignant patients in 98% of the patients in the training set with a sensitivity of 96%, a specificity of 100%. This derviation set of discriminant features was validated in the remaining half of the samples. This model allowed correct classification in our blinded test set in 66% of the patients. We then compared the list of discriminant features obtained in pleural effusions to the one obtained in the serum distinguishing cases of lung cancer from matched controls. We identified 9 features that are likely to represent the same proteins or peptides between the discriminants in these two independent datasets.
We used MALDI-MS directly on 1μl of 1:10 diluted unfractionated pleural fluid to obtain protein expression profiles and distinguish individuals with cytologically proven pleural effusions from cytologically negative exudative effusions with 66% accuracy in the blinded set. We hope to further improve our prediction model in a larger cohort of patients with and without malignant effusions.
If these results are confirmed, such profiling may have important implications in the management of patients with malignant pleural effusions.
P. Yildiz, None.