Age-specific discrimination of blood plasma samples of healthy and ovarian cancer prone mice using laser-induced breakdown spectroscopy

Abstract

Epithelial ovarian cancer (EOC) mortality rates are strongly correlated with the stage at which it is diagnosed. Detection of EOC prior to its dissemination from the site of origin is known to significantly improve the patient outcome. However, there are currently no effective methods for early detection of the most common and lethal subtype of EOC. We sought to determine whether laser-induced breakdown spectroscopy (LIBS) and classification techniques such as linear discriminant analysis (LDA) and random forest (RF) could classify and differentiate blood plasma specimens from transgenic mice with ovarian carcinoma and wild type control mice. Herein we report results using this approach to distinguish blood plasma samples obtained from serially bled (at 8, 12, and 16weeks) tumor-bearing TgMISIIR-TAg transgenic and wild type cancer-free littermate control mice. We have calculated the age-specific accuracy of classification using 18,000 laser-induced breakdown spectra of the blood plasma samples from tumor-bearing mice and wild type controls. When the analysis is performed in the spectral range 250nm to 680nm using LDA, these are 76.7 (±2.6)%, 71.2 (±1.3)%, and 73.1 (±1.4)%, for the 8, 12 and 16weeks. When the RF classifier is used, we obtain values of 78.5 (±2.3)%, 76.9 (±2.1)% and 75.4 (±2.0)% in the spectral range of 250nm to 680nm, and 81.0 (±1.8)%, 80.4 (±2.1)% and 79.6 (±3.5)% in 220nm to 850nm. In addition, we report, the positive and negative predictive values of the classification of the two classes of blood plasma samples. The approach used in this study is rapid, requires only 5μL of blood plasma, and is based on the use of unsupervised and widely accepted multivariate analysis algorithms. These findings suggest that LIBS and multivariate analysis may be a novel approach for detecting EOC.