This study showed that IBMs from contrast-enhanced CT images might yield to predict 3-year OS and 3-year PFS for EC patients. An IBM score was revealed to be an independent prognosis factor for ESCC patients. Patients were successfully stratified into low-risk and high-risk subgroups by the IBM score, with significant differences in OS and PFS. The IBMs nomograms showed better discrimination capability than the traditional clinical staging, indicating the clinical value of the IBM score for individualized OS and PFS estimation to some extent.
The newly IBM score, consisting of 13 optimal IBMs, demonstrated to be significant associated with ESCC patients for 3-year OS and 3-year PFS. Range, Q75, Q975 were obtained from the histogram of voxel intensities and represented the heterogeneity of voxel intensities within the ROI [27]. The geometric IBMs, Sphericity and Major Axis Length, quantified the spherical and size nature of tumor. These IBMs can promote the objective evaluation of subtle changes within tumors and provide clues on lesion invasiveness and growth-patterns [30, 31]. A higher value of texture IBMs, included Maximum Probability, Sum of Square Variance and Low Gray Level Run Emphasis, indicated the greater distribution variability of gray-level intensity values in the image [32, 33]. Small Zone Emphasis measures the distribution of small size zones and small dependencies, and Zone Percentage assessed the distribution of large zones of the same intensity, and not of small groups of pixels or segments in any given direction [26, 34]. These texture IBMs containing spatial information among voxels could strongly reflect intra-tumor heterogeneity which was highly relevant to poor prognosis [12]. In order to correlate the multiple IBMs with pathophysiological basis of tumor in an intuitive method, we constructed the multi-feature IBM score, which provided novel oncological biomarkers for obtaining phenotypic information, potentially assisting clinicians in making management strategies.
Current guidelines recommended definitive CCRT as a standard component for locally advanced ESCC therapies. However, some studies suggested that subgroups of patients could not show to be beneficial from present definitive CCRT strategies. Therefore, accurately distinguishing the risk subgroups of ESCC patients will help improve the current prognostic system and guide more personalized treatment. A few studies have focused on the correlate between radiomics analysis and treatment outcomes evaluation. Zhai at al. [30] found that heterogeneous IBMs on CT images were significantly correlated with OS and helped improve the performance of clinical factors for OS among head and neck cancer patients. Mule et al. [35] investigated contrast-enhanced CT outcomes that might help predict survival in patients with advanced hepatocellular carcinoma treated with sorafenib. In the present study, we indicated that ESCC patients with higher IBM scores had a greater likelihood of worse survival rate and failed to response to CCRT. High-risk ESCC patients identified in the present studies might lead to an appropriate group for more effective systemic approaches to improve survival outcomes [36, 37]. Thus, the IBM score was a prognostic tool for ESCC patients after definitive CCRT. Therefore, patients with higher IBM score would have larger probability of poor survival outcomes.
TNM staging systemis the most useful tool to stratify ESCC patients into different stages according to their tumor burden. However, its role in survival prediction among ESCC patients with the same clinical stage were insignificant. To develop an individualized easy-to-use tool for clinicians, we attempted to constructed nomograms based on IBM score for the prediction of the prognosis of individual patients. These IBMs nomograms could be used to predict the probability of 1-year, 2-year and 3-year OS and PFS for individual ESCC patient. The nomograms performed well with significant C-index and showed good discrimination and clinical utility both in the training and validation cohorts, such as helping counsel patients, individualize therapy, and arrange the follow-up for ESCC patients. The decision curve analysis indicated that the IBM score was superior to the clinical stage, within a major range of reasonable threshold probability. Notably, the limited performance and validation of the IBM score was that it was developed based on 3-year OS and 3-year-PFS, and it might result in the lack of some prognostic information. Encouragingly, the IBMs nomograms still had good predictive power for OS and PFS. One possible explanation was that most deaths or tumor progression in ESCC patients after definitive CCRT would occur within 3 years follow-up [2, 3]. In our study, during the follow-up time, 3-year OS rate and 3-year PFS rates were 60.9% and 51.0%, respectively, while OS and PFS rates were 55.6% and 48.3%, respectively. However, more appropriate approaches for assessment of prognosis model for ESCC patients was needed to establish in the future.
For EC patients, contrast-enhanced CT scan is the main performed imaging tool in conventional clinical practice [38]. It has been reported that IBMs extracted from contrast-enhanced CT images might be correlated with the spatial variability in microvessel density [39]. However, in standard CT images, IBMs might be associated with the variability in tissue densities due to spatially variable fibrosis, cell density, and necrosis [13]. Badic at el. suggested that IBMs extracted from standard CT and contrast-enhanced CT images could provide complementary prognostic information from both approaches [40]. In view of the wide availability of contrast-enhanced CT scans among patients undergoing definitive radiotherapy, our study provides an important basis for conducting large-scale and multicenter research. It is important to note that quality assurance of contrast-enhanced CT scans will have a critical impact on radiomics based on these images. Furthermore, verification is needed on whether IBMs extracted from contrast-enhanced CT images could provide prognostic information for esophageal adenocarcinoma patients.
The limitations of our retrospective design include several aspects that were insufficient for the model [41]. This was a retrospective and single-center study, including the relatively small sample size. This would be addressed more thoroughly in the future by using a greater sample size with multicenter validation cohorts to acquire high-level evidence for survival outcomes. Compared to IBM score, the clinical factors showed poor discrimination ability in predicting 3-year OS and 3-year PFS used in this study, but other potential prognostic biomarker should be incorporated into our IBMs nomograms. A combination of multiple biomarkers and IBMs may improve the capability of predicting 3-year OS among ESCC patients underwent definitive CCRT.