Preprocessing Tools Applied to Improve the Assessment of Aldrin Effects on Prostate Cancer Cells Using Raman Spectroscopy

• Published in: Applied spectroscopy Vol. 72; no. 3; p. 489
• Main Authors: Olmos, Víctor, Bedia, Carmen, Tauler, Romà, Juan, Anna de
• Format: Journal Article
• Language: English
• Published: United States 01.03.2018
• Subjects: Aldrin - adverse effects; Cell Line, Tumor; Discriminant Analysis; Disease Progression; Humans; Least-Squares Analysis; Male; Pesticides - adverse effects; Prostate - drug effects; Prostate - metabolism; Prostate - pathology; Prostatic Neoplasms - chemically induced; Prostatic Neoplasms - chemistry; Prostatic Neoplasms - metabolism; Prostatic Neoplasms - pathology; Spectrum Analysis, Raman - methods; environmental; omics; Raman spectroscopy; Data preprocessing; partial least squares discriminant analysis; PLSDA
• ISSN: 1943-3530
• DOI: 10.1177/0003702817746947

The study of pollutant effects on living organisms provides information about the possible biological and environmental response to a contaminant. Progression of prostate cancer may be related to exposure to pesticides or other chemical substances. In this work, the effect of the pesticide aldrin on human prostate cancer cells (DU145) is studied using Raman spectroscopy and chemometric techniques. Prostate cancer cell line DU145 has been exposed acutely the pesticide aldrin. Individual Raman spectra coming from control and treated cell populations have been acquired. Partial least squares discriminant analysis (PLSDA) has been used to assess differences among treated and control samples and to identify spectral biomarkers associated with pollutant stress. Some preprocessing methodologies have been tested in order to improve the capability of discrimination between fingerprints. Partial least squares discriminant analysis results suggest that the best normalization-scaling preprocessing combination is provided by Euclidean normalization (EN)-SIMPLISMA-based scaling (SBS). SIMPLISMA-based scaling has been proposed as a scaling method focused on the classification objective, which enhances variables with high relative variation among samples. The most relevant spectral variables related to aldrin effect on DU145 seem to be mainly related to lipids, proteins, and variations in nucleic acids.