Stereotactic body radiotherapy for early-stage non-small cell lung cancer: a multicentre study by the Oncologic Group for the Study of Lung Cancer (Spanish Radiation Oncology Society)

Stratification of tumour cell radiation response and metabolic signatures visualization with Raman spectroscopy and explainable convolutional neural network

Reprogramming of cellular metabolism is a driving factor of tumour progression and radiation therapy resistance. Identifying biochemical signatures associated with tumour radioresistance may assist with the development of targeted treatment strategies to improve clinical outcomes. Raman spectroscopy (RS) can monitor post-irradiation biomolecular changes and signatures of radiation response in tumour cells in a label-free manner. Convolutional Neural Networks (CNN) perform feature extraction directly from data in an end-to-end learning manner, with high classification performance. Furthermore, recently developed CNN explainability techniques help visualize the critical discriminative features captured by the model. In this work, a CNN is developed to characterize tumour response to radiotherapy based on its degree of radioresistance. The model was trained to classify Raman spectra of three human tumour cell lines as radiosensitive (LNCaP) or radioresistant (MCF7, H460) over a range of treatment doses and data collection time points. Additionally, a method based on Gradient-Weighted Class Activation Mapping (Grad-CAM) was used to determine response-specific salient Raman peaks influencing the CNN predictions. The CNN effectively classified the cell spectra, with accuracy, sensitivity, specificity, and F1 score exceeding 99.8%. Grad-CAM heatmaps of H460 and MCF7 cell spectra (radioresistant) exhibited high contributions from Raman bands tentatively assigned to glycogen, amino acids, and nucleic acids. Conversely, heatmaps of LNCaP cells (radiosensitive) revealed activations at lipid and phospholipid bands. Finally, Grad-CAM variable importance scores were derived for glycogen, asparagine, and phosphatidylcholine, and we show that their trends over cell line, dose, and acquisition time agreed with previously established models. Thus, the CNN can accurately detect biomolecular differences in the Raman spectra of tumour cells of varying radiosensitivity without requiring manual feature extraction. Finally, Grad-CAM may help identify metabolic signatures associated with the observed categories, offering the potential for automated clinical tumour radiation response characterization.

PET/CT for Target Delineation of Lung Cancer Before Radiation Therapy

In clinical routine of patients suffering from lung cancer, radiotherapy/radiation oncology represents one of the therapeutic hallmarks in the multimodal treatment besides or in combination with other local treatments such as surgery, but also systemic treatments such as chemotherapy, tyrosine kinase, and immune check-point inhibitors. Conventional morphological imagings such as CT or MR are commonly used for staging, response assessment, but also for radiotherapy planning. However, advanced imaging techniques such as PET do continuously get increasing access to clinical routine overcoming limitations of standard imaging techniques by visualizing and quantifying molecular processes such as glucose metabolism, which is also of relevance for radiotherapy planning. This review article summarizes the current place of radiotherapy within the treatment regimens of patients with lung cancer and elucidates current concepts of standard morphological imaging for staging and radiotherapy planning. Moreover, the place of PET-based radiotherapy planning in a clinical context is presented and current methodological/technical advances that do comprise a potential role for radiotherapy planning in lung cancer patients are discussed.

Establishment of a Prediction Model for Overall Survival after Stereotactic Body Radiation Therapy for Primary Non-Small Cell Lung Cancer Using Radiomics Analysis

Simple Summary

Lung cancer remains the leading cause of cancer-related mortality worldwide. Although early-stage non-small cell lung cancer (NSCLC) is likely to be controlled with stereotactic body radiation therapy (SBRT), approximately 18% of patients lead to recurrence. The aim of this study was to identify prognostic factors and establish a predictive model for survival outcomes of patients with non-metastatic NSCLC treated with SBRT. Several radiomic features were selected as predictive factors and two prediction models were established from the pre-treatment computed tomography images of 250 patients in the training cohort. One radiomic factor remained a significant prognostic factor of overall survival (OS) (p = 0.044), and one predicting model could estimate OS time (mean: 37.8 months) similar to the real OS time (33.7 months). In this study, we identified one radiomic factor and one prediction model that can be widely used.

Abstract

Stereotactic body radiation therapy (SBRT) for early-stage non-small cell lung cancer (NSCLC) leads to recurrence in approximately 18% of patients. We aimed to extract the radiomic features, with which we predicted clinical outcomes and to establish predictive models. Patients with primary non-metastatic NSCLC who were treated with SBRT between 2002 and 2022 were retrospectively reviewed. The 358 primary tumors were randomly divided into a training cohort of 250 tumors and a validation cohort of 108 tumors. Clinical features and 744 radiomic features derived from primary tumor delineation on pre-treatment computed tomography were examined as prognostic factors of survival outcomes by univariate and multivariate analyses in the training cohort. Predictive models of survival outcomes were established from the results of the multivariate analysis in the training cohort. The selected radiomic features and prediction models were tested in a validation cohort. We found that one radiomic feature showed a significant difference in overall survival (OS) in the validation cohort (p = 0.044) and one predicting model could estimate OS time (mean: 37.8 months) similar to the real OS time (33.7 months). In this study, we identified one radiomic factor and one prediction model that can be widely used.