SCP of Oral Mucosa Cells

Cells of the oral mucosa presented an easy target for establishing SCP as a diagnostic test for disease due to the easy availability of these cells. In addition, the medical collaborator (N.L.) at TMC in Boston has experienced a steady flow of patients from far-eastern origin in which oral cancer is quite prevalent. In fact, TMC offers these population groups free screening for oral cancer once a year. 

For the initial studies of oral mucosa cells, normal cytological samples were harvested from laboratory volunteers, whereas clinical samples were collected at TMC, as described in Section 5.2.1.1, and immediately inserted into a vial filled with SurePath fixative. Data acquisition was carried out via the PapMap approach as discussed in Section 5.2.3.2 thus, for the first time, thousands of cellular spectra were available for statistical analysis. This large number of cells allowed several ancillary studies, for example, the immediate effects of smoking and over-the-counter medication on the spectra of oral cells, as well as the dependence of spectral characteristics on the site of cell collection. It was found that cells from the top of the tongue, the mouth floor, and the inside of the cheeks had slightly different spectral characteristics, which could be attributed to some different expression of proteins in these cells. Cells from the tongue, for example, had shghtly different 

The earlier results on the differences between cells from the cheeks, the mouth floor, and the tongue were further validated with an increase in the number of normal volunteers in this study. Of the 93 volunteers in the data set, 19 samples were deemed normal for cheek, 20 for mouth floor, and 17 for tongue, with the different numbers because of the cellularity of samples and the concomitant ability to prepare a sample from the contents of the swab. Normal samples, in this context, were defined as samples collected from volunteers who indicated no use of tobacco or medications, no recent or active cold sores, no history of mononucleosis or abnormal oral infections, and no HPV vaccination. The entire data set from these volunteers comprise 13 651 cell spectra. Based on the number of cells in the smallest data set, 3500 spectra were selected randomly from each data set for the PCA analysis. 

The increased overlap of the different classes may be due to the fact that new and unexpected variances were introduced in the present study. However, the dependence of spectral characteristics on anatomical regions should only be taken as a baseline step for the clinical studies to be discussed next and are aimed at reducing variance because of the region of sample collection for these more important studies. 

The main motivation of this preclinical trial was to develop a prescreening tool for oral disease. To this end, clinical samples, that is, samples with diagnosed oral disease, were compared with normal samples for which patient information (metadata) was available. As mentioned before, the cellular samples prepared from the clinical specimens contained mostly cells with normal morphology, and the SCP-based distinction between normal and clinical samples was based on small changes occurring in the abnormal cells before they were manifested as morphological changes. 

---