Radiomic feature clusters and prognostic signatures specific for Lung and Head & Neck cancer

Abstract

Radiomics provides a comprehensive quantification of tumor phenotypes by extracting and mining large number of quantitative image features. To reduce the redundancy and compare the prognostic characteristics of radiomic features across cancer types, we investigated cancer-specific radiomic feature clusters in four independent Lung and Head &Neck (H) cancer cohorts (in total 878 patients). Radiomic features were extracted from the pre-treatment computed tomography (CT) images. Consensus clustering resulted in eleven and thirteen stable radiomic feature clusters for Lung and H cancer, respectively. These clusters were validated in independent external validation cohorts using rand statistic (Lung RS = 0.92, p < 0.001, H RS = 0.92, p < 0.001). Our analysis indicated both common as well as cancer-specific clustering and clinical associations of radiomic features. Strongest associations with clinical parameters: Prognosis Lung CI = 0.60 ± 0.01, Prognosis H CI = 0.68 ± 0.01; Lung histology AUC = 0.56 ± 0.03, Lung stage AUC = 0.61 ± 0.01, H HPV AUC = 0.58 ± 0.03, H stage AUC = 0.77 ± 0.02. Full utilization of these cancer-specific characteristics of image features may further improve radiomic biomarkers, providing a non-invasive way of quantifying and monitoring tumor phenotypic characteristics in clinical practice.

Figure 1

(a) Radiomic analysis overview: For both Lung and H & N cancer datasets, we extracted radiomic features from pre-treatment CT images. Cluster analysis was performed on the feature data. (b) Datasets overview: Four independent radiomic cohorts of Lung and Head & Neck cancer were included in the analysis. Lung1 and HN1 were used for training; Lung2 and HN2 were used for validation.

Figure 2. Heatmap showing the prognostic performance of radiomic features in Lung2 and HN2 cohorts.

Prognostic performance was evaluated using the concordance index (CI). Note that a large number of features are prognostic in both cancer types. However, also a large number of features are cancer-type specific, e.g. prognostic only in Lung or only in H & N cancer.

Figure 3. Heatmaps for radiomic features of Lung and H & N training cohorts ordered with respect to the obtained Lung and H & N clusters.

(a–b) Cluster consensus maps of Lung cancer (11 clusters) and H & N cancer (13 clusters) cohorts. (c–d) Radiomic feature expressions of Lung and H & N radiomic clusters. (e–f) Clinical relevance (CI & AUC) of radiomic clusters of Lung and H & N cancer.

Figure 4. Heatmap depicting cluster overlap and clinical relevance (CI & AUC).

Center matrix in green & white color represents the overlap (Jaccard index) between the clusters of Lung (rows) and H & N (columns) radiomic cohorts. Top and left side panels in blue & white color depicts the average CI & AUC of the corresponding Lung and H & N radiomic clusters.