Quantification of DNA content using 3D Optical Projection Tomographic Microscopy for Early Cancer Diagnosis

Abstract

Aneuploidy, an accepted biomarker for malignancy, is typically assessed by flow (FCM) and image cytometry (ICM). We used optical projection tomographic microscopy (OPTM) for assessing cellular DNA content using different absorption and fluorescence stains. OPTM combines some attributes of both FCM and ICM and provides quantitative cytometry with 3D visualization of morphology. It generates submicron, isometric three-dimensional (3D) images of nuclear features, performing multimodal imaging with exact co-registration. We further present a novel automated method for 3D nucleus segmentation. First, nuclei of chicken erythrocyte, trout erythrocyte and triploid trout erythrocyte standards were used to calibrate the OPTM for stoichiometric DNA ploidy assessment. Thionin, as an alternative to Feulgen stain was optimized for stoichiometry by varying the staining time. DNA indices were extracted from 50 images of each standard and their ratios were compared to FCM. A similar comparison was done with hematoxylin, Feulgen and sytox green labeling. The mean square error (MSE) of the three ratios was highest with hematoxylin and least with Feulgen. Forty-five minutes (45mins) of thionin staining time had least MSE among different thionin stains. Sytox green MSE was greater than both thionin and Feulgen but less than hematoxylin. For all three ratios, variance of hematoxylin was significantly greater than 45min thionin, which was greater than Feulgen (p<0.01). The variance of sytox green was significantly greater than Feulgen but not less than 45min thionin (p>0.01). From qualitative assessment of seven cancer cell lines, morphology was preserved only for hematoxylin and Feulgen stain. Thus, Feulgen stain, like in 2D analysis, is stoichiometric with least variance while also preserving nuclear morphology in 3D. We further tested feulgen stain on seven different cancer cell lines for DNA index. We compared our results with FCM and obtained an agreement on five cancer cell lines. The possible reason for discrepancies in two cancer cell lines could be due to inaccurate three-dimensional segmentation. In conclusion we present a novel technique for quantitative measurement of DNA content, while preserving the nuclear morphology in 3D. The addition of this quantitative biomarker could further strengthen existing classifiers and improve early diagnosis of cancer using 3D microscopy.