Date Published: March 12, 2019
Publisher: Public Library of Science
Author(s): JL González-Solís, Anandhi Upendran.
Based in high sensitivity and specificity reported recently in detection of the cancer, the technique of Raman spectroscopy is proposed to discriminate between breast cancer, leukemia and cervical cancer using blood serum samples from patients officially diagnosed. In order to classify Raman spectra, clustering method known as Super Paramagnetic Clustering based on statistical physics concepts with a stochastic approach was implemented. Comparing firstly average Raman spectra of the three cancers, some peaks that allowed differentiating one cancer from other were identified, however, other peaks allowed concluding that there are biochemical similarities among them. According to these spectra, the band associated with amide I (1654 cm−1) and one of two shoulders assigned to amide III (1230-1282 cm−1) allowed discriminating leukemia from breast and cervical cancer, whereas band 714 cm−1 (polysaccharides) achieves to differentiate cervical cancer from leukemia and breast cancer, and bulged region, 1040 − 1100 cm−1 (phenylalanine, phospholipid) discriminated breast cancer from leukemia and cervical cancer. Subsequently, Super Paramagnetic Clustering method was applied to Raman spectra to study similarity relationships between cancers based on the biochemical composition of serum samples. Finally, as a cross check method, the standard method to classify Raman spectra of breast cancer, leukemia and cervical cancer, known as principal components analysis, was used showing excellent agreement with results of Super Paramagnetic Clustering method. Preliminary results demonstrated that Raman spectroscopy and Super Paramagnetic Clustering method can be used to discriminate between breast cancer, leukemia and cervical cancer samples using blood serum samples.
Although some of the most deadly cancers affect different parts of the body, among them there are similarities of different types according to several reported researches, ie, these studies had found relationships between certain forms of breast, lung, colon, cervical cancers and leukemia. The best knowledge of these similarities based on the same molecular origin could facilitate the comparison of therapeutic data bank between the cancers difficult to treat, suggesting that the treatments could be performed with the same chemotherapy drugs.
In this work, discrimination between breast cancer, leukemia, cervical cancer based on blood serum samples Raman spectroscopy and SPC method was studied. According to mean Raman spectra of breast cancer, leukemia and cervical cancer samples, peak associated to amide I (1654 cm−1) and one of two shoulders assigned to amide III (1230-1282 cm−1) allowed discriminating breast and cervical cancer from leukemia; the band 714 cm−1 (polysaccharides), the cervical cancer from leukemia and breast cancer and bulged region, 1040 − 1100 cm−1 (phenylalanine, phospholipid), breast cancer from the leukemia and cervical cancer. The tree structure obtained using SPC method also allowed concluding that although these degenerative diseases seem to be very different from each other, there are some aspects of biochemical type that allowed establishing relationships between them. This tree structure showed that clusters corresponding to leucemia and cervical cancer spectra belonged to a larger cluster, which together with another one corresponding to breast cancer, make up the largest cluster formed by all the spectra, ie, leukemia maintains greater similarities with cervical cancer but without the loss of relationships between the three cancers. Motivated by researches of genetic and biochemical type about cancer, reporting relationships between certain forms of leukemia, breast, lung, colon and cervical cancer, in this work were studied similarities between cancers using alternative techniques to those already reported, hoping that the best knowledge of these similarities based on the same molecular origin could facilitate the comparison of therapeutic data bank suggesting that certain treatments could be performed with same chemotherapy drugs. Finally, as a cross check, the standard method to classify Raman spectra of breast cancer, leukemia and cervical cancer, PCA, was used and results were in complete agreement with result obtained by applying SPC method. Thus, preliminary results demonstrated that Super Paramagnetic Clustering method allows properly classifying Raman spectra and discriminate between breast cancer, leukemia and cervical cancer patients based on chemical composition of blood serum samples indicated by the bands in a Raman spectrum.