Early prognosis of metastasis risk in inflammatory breast cancer by texture analysis of tumour microscopic images

Abstract

Inflammatory breast cancer (IBC) is a rare and aggressive type of locally advanced breast cancer. The purpose of this study was to determine the value of microscopic tumour histomorphology texture for prognosis of local and systemic recurrence at the time of initial IBC diagnosis. This retrospective study included a group of 52 patients selected on the basis of non-metastatic IBC diagnosis, stage IIIB. Gray-Level-Co-Occurrence-Matrix (GLCM) texture analysis was performed on digital images of primary tumour tissue sections stained with haematoxylin/eosin. Obtained values were categorized by use of both data- and outcome-based methods. All five acquired GLCM texture features significantly associated with metastasis outcome. By accuracies of 69-81% and AUCs of 0.71-0.81, prognostic performance of GLCM parameters exceeded that of standard major IBC clinical prognosticators such as tumour grade and response to induction chemotherapy. Furthermore, a composite score consisting of tumour grade, contrast and correlation as independent features resulted in further enhancement of prognostic performance by accuracy of 89%, discrimination efficiency by AUC of 0.93 and an outstanding hazard ratio of 71.6 (95%CI, 41.7-148.4). Internal validation was successfully performed by bootstrap and split-sample cross-validation, suggesting that the model is generalizable. This study indicates for the first time the potential use of primary breast tumour histology texture as a highly accurate, simple and cost-effective prognostic indicator of metastasis risk in IBC. Clinical relevance of the obtained results rests on the role of prognosis in decisions on induction chemotherapy and the resulting impact on quality of life and survival.

Effects of different preprocessing algorithms on the prognostic value of breast tumour microscopic images

“…Descriptive statistics and standard clinicopathological prognostic factors are not shown as these were already described for both patient groups used in this study (Kolarevic et al ., , Vujasinovic et al ., ).…”

“…Descriptive statistics and standard clinicopathological prognostic factors are not shown as these were already described for both patient groups used in this study (Kolarevic et al, 2015, Vujasinovic et al, 2015b. Image preprocessing by use of six algorithms was followed by image analysis using five different GLCM, two monofractal and ten multifractal algorithms.…”

“…This rationale is based on the hypothesis that such images contain distributed prognostic information which remain unused by either visual or computational evaluations targeting specific structures. There have been several previous reports utilising this approach for prognostic purposes (Tambasco & Magliocco, 2008;Tambasco et al, 2010;Vujasinovic et al, 2015a;Kolarevic et al, 2015;Rajkovic et al, 2016a,b;Mezheyeuski et al, 2016). Such unselective approach to histopathology image analysis has a potential for routine clinical application as it often outperforms the major molecular prognostic tools such as multiplex ELISA (Jia et al, 2014) and gene expression microarrays (Buyse et al, 2006).…”

Predicting the pathological response to chemoradiotherapy of non-mucinous rectal cancer using pretreatment texture features based on intravoxel incoherent motion diffusion-weighted imaging

“…LARC locally advanced rectal cancer, pCR pathological complete response, T2WI T2-weighted imaging, VOI volume of interest, IVIM-DWI intravoxel incoherent motion diffusion-weighted imaging, nCRT neoadjuvant chemoradiotherapy, TRG tumor regression grade generated from dynamic contrast-enhanced MRI images report can discriminate between the cured and relapsed cervical cancer patients who underwent chemoradiotherapy, whereas the first-order features cannot [38]. These observations imply the advantage of GLCM analysis over the firstorder statistics in discriminating the pathological response of LARC, which may be because the first-order texture features do not provide spatial information of images and therefore they cannot provide sufficient information on tumor heterogeneity [17,33,34,39].…”

“…These findings revealed that baseline GLCM features derived from IVIM-DWI could act as imaging predictors of pCR in patients with LARC. Among these GLCM features, Correlation and SumAverg are not directly related to heterogeneity, whereas Contrast and DifVarnc are measures of tissue heterogeneity [39], that is to say, lower Contrast or DifVarnc means less heterogeneity [35,36,39,40]. In the present study, all the values of Contrast and Dif-Varnc derived from the ADC, D, and D* maps in the pCR group were lower than those in the non-pCR group, implying that higher heterogeneity on IVIM-DWI maps may be associated with poorer therapeutic effect.…”