The relation and prognostic importance of proportion of tumor-stroma, tumor stem cells and entropy in colon adenocarcinoma

Aim: In this study, the relationship and prognostic value of the tumor stem cells and mathematical features including tumor/stroma proportion, entropy and complexity were evaluated. Method: Microscopic digital images were captured from tissue slides stained with hematoxylin-eosin from 248 patients which was treated with radical surgery and tumor, stroma and lumen were manually marked and recorded by using an image analysis programme which was developed by our study group. Mathematical calculation of features proportion, entropy and complexity and mean, maksimum, minimum and variance values for each feature were calculated for the whole analysed tumor area and on images representing invasive side seperately. Also, sub-images with different sizes were formed in main images and all calculations were done for these sub-images of 8x8 and 16x16 pixels. Immunohistochemical examination with CD133 and Lgr5 antibodies, which are reported to be stem cell markers were scored semi quantitatively. Statistical analysis was done with 80 parameters including clinic, histopathologic prognostic and all data acquired by calculated features. Results: Cox regression analysis revealed an increase of death risk of, 1.07 fold for age, 2,5 fold for pN stage, 2,2 fold for perineural invasion, 1,02 fold for tumor stroma proportion at the maximum group of 8x8 pixel images, 1,2 fold for tumor stroma proportion at the mean group of 16x16 pixel images and 1,9 fold for entropy at the invasive tumor border. Recurrence risk increased 2,4 fold for pT, 4,21 fold for pN and 1,002 times for variance of tumor stroma proportion at 16x16 pixel images. The risk of distant metastasis increased 3,357 times with pT stage, 9,479 times for pN stage. There was no relation with stem cell markers and prognosis. Statistically significant difference was identified for entropy feture values calculated in all tumor area and invasive side between CD133 subgroups and statistically significant difference was identified for entropy feature values and tumor/stroma proportion calculated in all tumor area between Lgr5 subgroups but these values did not change parallel to the scores. Conclusions: In contrast to reports of many studies, calculations for tumor/stroma proportion showed that statistical significance exists between tumor mass increase and death, recurrance and metastasis. Also, it was determined that increase in tumor/stroma proportion increased the risk of death and recurrance; increase in entropy fetaure value on invasive side increased the risk of death, as an independent prognostic factor in multivariant analysis. No statistical significance was found between stem cell markers and overall survival, recurrance and metastasis. In CD133 subgroups, there is a significant difference between the mean of entropy and complexity features, in Lgr5 subgroups there is a significant difference between the mean of entropy and tumor/stroma proportion fetures but these values did not change parallel to the scores. The original finding of this study is defining entropy and complexity features in tumors and defining entropy related to tumor/stroma on invasive side of colon tumor as an independant prognostic feature. The prognostically important data acquired in previous studies showing an increase in stroma not related to quantitative evaluation was not found to be significant by the quantitative method used in this study. Lgr5 and CD133, accepted as stem cell markers, were not found significant in this series as prognostic or computable features.