Combined CT radiomics of primary tumor and metastatic lymph nodes improves prediction of loco-regional control in head and neck cancer

Abstract

Loco-regional control (LRC) is a major clinical endpoint after definitive radiochemotherapy (RCT) of head and neck cancer (HNC). Radiomics has been shown a promising biomarker in cancer research, however closer related to primary tumor control than composite endpoints. Radiomics studies often focus on the analysis of primary tumor (PT). We hypothesize that the combination of PT and lymph nodes (LN) radiomics better predicts LRC in HNC treated with RCT. Radiomics analysis was performed in CT images of 128 patients using Z-Rad implementation (training n = 77, validation n = 51). 285 features were extracted from PT and involved LN. Features were preselected with the maximum relevance minimum redundancy method and the multivariate Cox model was trained using least absolute shrinkage and selection operator. The mixed model was based on the combination of PT and LN radiomics, whereas the PT model included only the PT features. The mixed model showed significantly higher performance than the PT model (p < 0.01), c-index of 0.67 and 0.63, respectively; and better risk group stratification. The clinical nodal status was not a significant predictor in the combination with PT radiomics. This study shows that the LRC can be better predicted by expansion of radiomics analysis with LN features.

Baseline MRI-Radiomics Can Predict Overall Survival in Non-Endemic EBV-Related Nasopharyngeal Carcinoma Patients

“…Even though usually radiomic models obtained from multiparametric MRI are the ones that perform better [ 13 ], this is not the first time that monomodal radiomic models result to be the best [ 16 ]. The results are also partially in contrast with [ 17 ], in which the addition of lymph node features significantly improved the results, but this may be explained taking into account the differences in terms of type of pathology (NPC vs. generic head and neck cancer), the definition of the lymphnodal ROI (all the nodal masses vs. the largest lymph node), the different imaging technique (computed tomography vs. MRI).…”

“…Only N+ patients were considered for the analysis since it has been shown in previous studies related to head and neck cancer that the addition of radiomic features from the lymphnodal ROI provides a better prognostic ability than radiomics from the main tumor alone [ 17 , 35 ]. Metastatic lymph node involvement was defined according to Ho et al [ 36 ]: criteria included, namely, central necrosis, extra capsular spread, shortest diameter of cervical or medial retropharyngeal lymph nodes >1 cm and >5 mm for lateral retropharyngeal lymph node(s).…”

Comparison of patient stratification by computed tomography radiomics and hypoxia positron emission tomography in head-and-neck cancer radiotherapy

“…However, in our study the AUC value determined for the validation cohort was still 0.59 but the significance of the CT radiomics model to stratify patient risk groups could not be confirmed in this cohort in contrast to other published studies [37][38][39]. In one study by Bogowicz et al [36] a CT radiomics model based on the primary tumour volume could not be validated in contrast to a model, which was applied to primary tumour and lymph node volumes. According to these findings, the fact that in our study CT radiomics was assessed for the primary tumour only whereas loco-regional failure was used as a prediction variable might be a further limitation.…”

“…Several recent studies published CT radiomics models for predicting local control or overall survival in HNC patients following CRT [36] , [37] , [38] , [39] . Similar to our findings, those studies identified features related to CT value homogeneity as most relevant for outcome prognosis.…”

Applicability of radiomics in interstitial lung disease associated with systemic sclerosis: proof of concept

“…Radiomics has attracted increased attention in recent years, and several studies show that radiomics can be of benefit in terms of prognosis and diagnosis of multiple diseases, [21][22][23]. These studies have shown that radiomics features show great potential to serve as surrogate imaging markers for tissue biopsies [40] and reliably predict outcome [41][42][43][44] and drug response [45,46]. Currently, there are different approaches for the evaluation of HRCT, namely (1) visual analysis, (2) semiquantitative analysis, and (3) quantitative analysis or automated approaches using artificial intelligence.…”