Radiomics is a promising technique for discovering image based biomarkers of therapy response in cancer. Reproducibility of radiomics features is a known issue that is addressed by the image ...biomarker standardisation initiative (IBSI), but it remains challenging to interpret previously published radiomics signatures. This study investigates the reproducibility of radiomics features calculated with two widely used radiomics software packages (IBEX, MaZda) in comparison to an IBSI compliant software package (PyRadiomics). Intensity histogram, shape and textural features were extracted from 334 diffusion weighted magnetic resonance images of 59 head and neck cancer (HNC) patients from the PREDICT-HN observational radiotherapy study. Based on name and linear correlation, PyRadiomics shares 83 features with IBEX and 49 features with MaZda, a sub-set of well correlated features are considered reproducible (IBEX: 15 features, MaZda: 18 features). We explore the impact of including non-reproducible radiomics features in a HNC radiotherapy response model. It is possible to classify equivalent patient groups using radiomic features from either software, but only when restricting the model to reliable features using a correlation threshold method. This is relevant for clinical biomarker validation trials as it provides a framework to assess the reproducibility of reported radiomic signatures from existing trials.
In radiation oncology, predicting patient risk stratification allows specialization of therapy intensification as well as selecting between systemic and regional treatments, all of which helps to ...improve patient outcome and quality of life. Deep learning offers an advantage over traditional radiomics for medical image processing by learning salient features from training data originating from multiple datasets. However, while their large capacity allows to combine high-level medical imaging data for outcome prediction, they lack generalization to be used across institutions. In this work, a pseudo-volumetric convolutional neural network with a deep preprocessor module and self-attention (PreSANet) is proposed for the prediction of distant metastasis, locoregional recurrence, and overall survival occurrence probabilities within the 10 year follow-up time frame for head and neck cancer patients with squamous cell carcinoma. The model is capable of processing multi-modal inputs of variable scan length, as well as integrating patient data in the prediction model. These proposed architectural features and additional modalities all serve to extract additional information from the available data when availability to additional samples is limited. This model was trained on the public Cancer Imaging Archive Head-Neck-PET-CT dataset consisting of 298 patients undergoing curative radio/chemo-radiotherapy and acquired from 4 different institutions. The model was further validated on an internal retrospective dataset with 371 patients acquired from one of the institutions in the training dataset. An extensive set of ablation experiments were performed to test the utility of the proposed model characteristics, achieving an AUROC of Formula: see text, Formula: see text and Formula: see text for DM, LR and OS respectively on the public TCIA Head-Neck-PET-CT dataset. External validation was performed on a retrospective dataset with 371 patients, achieving Formula: see text AUROC in all outcomes. To test for model generalization across sites, a validation scheme consisting of single site-holdout and cross-validation combining both datasets was used. The mean accuracy across 4 institutions obtained was Formula: see text, Formula: see text and Formula: see text for DM, LR and OS respectively. The proposed model demonstrates an effective method for tumor outcome prediction for multi-site, multi-modal combining both volumetric data and structured patient clinical data.
A minority of patients with metastatic head and neck squamous cell carcinoma (HNSCC) present with oligometastatic disease. Oligometastasis not only reflects a disease state, but might also present an ...opportunity for cure in the metastatic setting. Radical ablation of all oligometastatic sites may confer prolonged survival and possibly achieve cure in some patients. However, substantial debate remains about whether patients with oligometastatic disease could benefit from curative intent therapy or whether aggressive treatments expose some patients to futile toxicity. Optimal selection of patients, carefully balancing the currently known prognostic factors against the risks of toxicity is critical. Emerging evidence suggests that patients with a limited burden of disease, viral-related pharyngeal cancer, metachronous metastasis and lung-only metastasis may benefit most from this approach. Efforts are underway to identify biomarkers that can detect oligometastasis and better select patients who would derive the maximum benefit from an aggressive radical approach. The combination of radiotherapy and immunotherapy promises to enhance the anti-tumoral immune response and help overcome resistance. However, optimization of management algorithms, including patient selection, radiation dose and sequencing, will be critical in upcoming clinical trials. This review summarizes recent knowledge about the characteristics and investigational efforts regarding oligometastasis in HNSCC.
To determine which parameters allow for CyberKnife fiducial-less tumor tracking in stereotactic body radiation therapy (SBRT) for early-stage non-small cell lung cancer.
A total of 133 lung SBRT ...patients were preselected for direct soft-tissue tracking based on manufacturer recommendations (peripherally located tumors ≥1.5 cm with a dense appearance) and staff experience. Patients underwent a tumor visualization test to verify adequate detection by the tracking system (orthogonal radiographs). An analysis of potential predictors of successful tumor tracking was conducted looking at: tumor stage, size, histology, tumor projection on the vertebral column or mediastinum, distance to the diaphragm, lung-to-soft tissue ratio, and patient body mass index.
Tumor visualization was satisfactory for 88 patients (66%) and unsatisfactory for 45 patients (34%). Median time to treatment start was 6 days in the success group (range, 2-18 days) and 15 days (range, 3-63 days) in the failure group. A stage T2 (P=.04), larger tumor size (volume of 15.3 cm(3) vs 6.5 cm(3) in success and failure group, respectively) (P<.0001), and higher tumor density (0.86 g/cm(3) vs 0.79 g/cm(3)) were predictive of adequate detection. There was a 63% decrease in failure risk with every 1-cm increase in maximum tumor dimension (relative risk for failure = 0.37, CI=0.23-0.60, P=.001). A diameter of 3.6 cm predicted a success probability of 80%. Histology, lung-to-soft tissue ratio, distance to diaphragm, patient's body mass index, and tumor projection on vertebral column and mediastinum were not found to be predictive of success.
Tumor size, volume, and density were the most predictive factors of a successful XSight Lung tumor tracking. Tumors >3.5 cm have ≥80% chance of being adequately visualized and therefore should all be considered for direct tumor tracking.
This study aimed to evaluate the prognostic value of pre-treatment NLR in patients with oropharyngeal cancer.
Patients who completed definitive radiotherapy (RT) for oropharyngeal cancer and had ...blood counts taken pre-RT from 2002 to 2013 were included. NLR was calculated as total neutrophil/lymphocytes. Survival rates were estimated using the Kaplan-Meier method. Univariable and multivariable analyses were conducted with linear and Cox regression methods. NLR was analysed posteriori and dichotomised on the discovered median.
Eight hundred and forty-eight patients were analysed. The median pre-RT NLR was 3. Patients with NLR of <3 had improved overall survival (OS) than those with NLR ≥ 3 (5-year OS 85 vs 74%, p < 0.0001). OS differences remained significant when stratified according to HPV status (HPV-positive p = 0.011; HPV-negative p = 0.003). Freedom from any recurrence (FFR), locoregional control (LRC) and freedom of distant recurrence (FDR) were better in those with NLR < 3. The negative impact of elevated pre-RT NLR on OS (HR = 1.64, p = 0.001), FFR (HR = 1.6, p = 0.006) and LRC (HR = 1.8, p = 0.005) remained significant on multivariable analysis.
Pre-RT NLR is an independent prognostic factor in patients with oropharyngeal cancer regardless of HPV status. Patients with lower NLR had more favourable OS and disease control.
The efficacy of immunotherapy targeting the PD-1/PD-L1 pathway has previously been demonstrated in metastatic head and neck squamous cell carcinoma (HNSCC). Stereotactic Body Radiotherapy (SBRT) aims ...at ablating metastatic lesions and may play a synergistic role with immunotherapy. The purpose of this study is to assess the safety and efficacy of triple treatment combination (TTC) consisting of the administration of durvalumab and tremelimumab in combination with SBRT in metastatic HNSCC.
This is a phase I/II single arm study that will include 35 patients with 2-10 extracranial metastatic lesions. Patients will receive durvalumab (1500 mg IV every 4 weeks (Q4W)) and tremelimumab (75 mg IV Q4W for a total of 4 doses) until progression, unacceptable toxicity or patient withdrawal. SBRT to 2-5 metastases will be administered between cycles 2 and 3 of immunotherapy. The safety of the treatment combination will be evaluated through assessment of TTC-related toxicities, defined as grade 3-5 toxicities based on Common Terminology Criteria for Adverse Events (v 4.03), occurring within 6 weeks from SBRT start, and that are definitely, probably or possibly related to the combination of all treatments. We hypothesize that dual targeting of PD-L1 and CTLA-4 pathways combined with SBRT will lead to < 35% grade 3-5 acute toxicities related to TTC. Progression free survival (PFS) will be the primary endpoint of the phase II portion of this study and will be assessed with radiological exams every 8 weeks using the RECIST version 1.1 criteria.
The combination of synergistic dual checkpoints inhibition along with ablative radiation may significantly potentiate the local and systemic disease control. This study constitutes the first clinical trial combining effects of SBRT with dual checkpoint blockade with durvalumab and tremelimumab in the treatment of metastatic HNSCC. If positive, this study would lead to a phase III trial testing this treatment combination against standard of care in metastatic HNSCC.
NCT03283605 . Registration date: September 14, 2017; version 1.
Purpose
The purpose of this study was to develop and validate accurate methods for determining iodine content and virtual noncontrast maps of physical parameters, such as electron density, in the ...context of radiotherapy.
Methods
A simulation environment is developed to compare three methods allowing extracting iodine content and virtual noncontrast composition: (a) two‐material decomposition, (b) three‐material decomposition with the conservation of volume constraint, and (c) eigentissue decomposition. The simulation allows comparing the performance of the methods using iodine‐based contrast agent contents in tissues from a reference dataset with variable density and elemental composition. The comparison is performed in two ways: (a) with a priori knowledge on the composition of the targeted tissue, and (b) without a priori knowledge on the base tissue. The three methods are tested with patient images scanned with dual‐energy CT and iodine‐based contrast agent. An experimental calibration adapted to the presence of iodine is performed by imaging tissue equivalent materials and diluted contrast agent solutions with known atomic composition.
Results
Results show that in the case of known a priori on the composition of the targeted tissue, the two‐material decomposition is robust to variable densities and atomic compositions without biasing the results. In the absence of a priori knowledge on the target tissue composition, the eigentissue decomposition method yields minimal bias and higher robustness to variations. Results from the experimental calibration and the images of two patients show that the extracted quantities are accurate and the bias is negligible for both methods with respect to clinical applications in their respective scope of use. For the patient imaged with a contrast agent, virtual noncontrast electron densities are found in good agreement with values extracted from the scan without contrast agent.
Conclusion
This study identifies two accurate methods to quantify iodine‐based contrast agents and virtual noncontrast composition images with dual‐energy CT. One is the two‐material decomposition with a priori knowledge of the constituent components focused on organ‐specific applications, such as kidney or lung function assessment. The other method is the eigentissue decomposition and is useful for general radiotherapy applications, such as treatment planning where accurate dose calculations are needed in the absence of contrast agent.
Human papillomavirus (HPV)-associated oropharyngeal squamous cell carcinoma (OPSCC) has increased in incidence in recent decades. With higher cure rates in younger populations, long-term survivors ...may live with acute- and long-term toxicity, leading to increased interest in de-escalation treatment strategies for HPV-related OPSCC. Herein, we have examined the current landscape of clinical trials in this context.
A review of active clinical trials related to de-escalation of HPV-associated OPSCC treatment was performed using the clinicaltrials.gov database from inception to January 2022. A search using the key words "oropharyngeal cancer" and "HPV" was completed. Three investigators independently reviewed each trial, with any discrepancies settled by a fourth. Data collected from each study included study phase, study design, primary, and secondary endpoints, and de-escalation treatment strategies. A final 24 articles were selected for full text review.
Many trials (n=19, 79%) were non-randomized, and most studies employed a phase II design (n=14, 58%). Only 13% (n=3) were randomized trials, and 8% (n=2) included a phase III component. The most frequent primary endpoint was progression-free survival (PFS) (n=9, 37.5%). With regards to the identified de-escalation strategies, all the studies (n=24) had at least one component assessing changes in RT dose/fractionation and/or a reduction in RT volumes. A smaller percentage of trials assessed surgical interventions (n=9, 37.5%) and/or changes in systemic therapy (n=8, 33.3%).
A small number of randomized trials are underway, and a transition to more randomized phase III trials in the future will be important to change clinical practice.
•Head and neck cancer radiotherapy lacks predictive tools due to its heterogeneity.•Deep learning models can successfully integrate clinical data and radiomics.•They promise to predict different ...outcomes to guide individualized decision.•Understanding, validating, expanding them is critical for clinical implementation.
Head and neck radiotherapy induces important toxicity, and its efficacy and tolerance vary widely across patients. Advancements in radiotherapy delivery techniques, along with the increased quality and frequency of image guidance, offer a unique opportunity to individualize radiotherapy based on imaging biomarkers, with the aim of improving radiation efficacy while reducing its toxicity. Various artificial intelligence models integrating clinical data and radiomics have shown encouraging results for toxicity and cancer control outcomes prediction in head and neck cancer radiotherapy. Clinical implementation of these models could lead to individualized risk-based therapeutic decision making, but the reliability of the current studies is limited. Understanding, validating and expanding these models to larger multi-institutional data sets and testing them in the context of clinical trials is needed to ensure safe clinical implementation. This review summarizes the current state of the art of machine learning models for prediction of head and neck cancer radiotherapy outcomes.