A new method for lung cancer prognosis via centrosome image feature analysis
Dansheng Song1, Tatyana A Zhukov, Olga Markov
1Molecular Epidemiology Program, Department of Cancer Epidemiology and Genetics, H. Lee Moffitt Cancer Center & Research Institute, Tampa, USA.
Objective:
To predict survival of resected stage I non-small cell lung cancer (NSCLC) patients through quantitative analysis and classification of centrosome features.
Study Design:
Disordered centrosome amplification leads to the loss of regulated chromosome segregation, aneuploidy and chromosome instability and may be a biomarker of cancer prognosis. Resected, stage I NSCLC tissues from survivor and fatal cases were immunostained with gamma-tubulin and scanned by confocal microscopy. Regions of interest were selected to include 1 cell and at least 1 centrosome. Four hundred forty-six regions were imaged, including 903 centrosomes whose features were extracted and measured. After segmentation, 12 centrosome features were measured. After optimization, 6 non-redundant features were selected for statistical analysis and classification.
Results:
Two statistical methods showed that for each feature, centrosomes from survivors differed significantly from centrosomes of fatalities. Centrosomes were classified into survival or fatal outcomes by centrosome features using linear discriminant analysis, support vector machines (SVMs) and further optimized using SVMs with bagging. Ten-fold cross-validation was applied. Classification accuracies were 74%, 79% and 85%, respectively.
Conclusion:
Centrosome features can be a prognostic biomarker for resected stage I NSCLC and may indicate patients who would benefit from additional adjuvant therapy.
