Aniek J.G. Even

Radiomics: the bridge between medical imaging and personalized medicine

Radiomics, the high-throughput mining of quantitative image features from standard-of-care medical imaging that enables data to be extracted and applied within clinical-decision support systems to improve diagnostic, prognostic, and predictive accuracy, is gaining importance in cancer research. Radiomic analysis exploits sophisticated image analysis tools and the rapid development and validation of medical imaging data that uses image-based signatures for precision diagnosis and treatment, providing a powerful tool in modern medicine. Herein, we describe the process of radiomics, its pitfalls, challenges, opportunities, and its capacity to improve clinical decision making, emphasizing the utility for patients with cancer. Currently, the field of radiomics lacks standardized evaluation of both the scientific integrity and the clinical relevance of the numerous published radiomics investigations resulting from the rapid growth of this area. Rigorous evaluation criteria and reporting guidelines need to be established in order for radiomics to mature as a discipline. Herein, we provide guidance for investigations to meet this urgent need in the field of radiomics.

In Vivo Quantification of Hypoxic and Metabolic Status of NSCLC Tumors Using [18F]HX4 and [18F]FDG-PET/CT Imaging

Purpose: Increased tumor metabolism and hypoxia are related to poor prognosis in solid tumors, including non–small cell lung cancer (NSCLC). PET imaging is a noninvasive technique that is frequently used to visualize and quantify tumor metabolism and hypoxia. The aim of this study was to perform an extensive comparison of tumor metabolism using 2[18F]fluoro-2-deoxy-d-glucose (FDG)-PET and hypoxia using HX4-PET imaging. Experimental Design: FDG- and HX4-PET/CT images of 25 patients with NSCLC were coregistered. At a global tumor level, HX4 and FDG parameters were extracted from the gross tumor volume (GTV). The HX4 high-fraction (HX4-HF) and HX4 high-volume (HX4-HV) were defined using a tumor-to-blood ratio > 1.4. For FDG high-fraction (FDG-HF) and FDG high-volume (FDG-HV), a standardized uptake value (SUV) > 50% of SUVmax was used. We evaluated the spatial correlation between HX4 and FDG uptake within the tumor, to quantify the (mis)match between volumes with a high FDG and high HX4 uptake. Results: At a tumor level, significant correlations were observed between FDG and HX4 parameters. For the primary GTV, the HX4-HF was three times smaller compared with the FDG-HF. In 53% of the primary lesions, less than 1 cm3 of the HX4-HV was outside the FDG–HV; for 37%, this volume was 1.9 to 12 cm3. Remarkably, a distinct uptake pattern was observed in 11%, with large hypoxic volumes localized outside the FDG-HV. Conclusion: Hypoxic tumor volumes are smaller than metabolic active volumes. Approximately half of the lesions showed a good spatial correlation between the PET tracers. In the other cases, a (partial) mismatch was observed. The addition of HX4-PET imaging has the potential to individualize patient treatment. Clin Cancer Res; 20(24); 6389–97. ©2014 AACR.

Modern clinical research: How rapid learning health care and cohort multiple randomised clinical trials complement traditional evidence based medicine.

ABSTRACT Background. Trials are vital in informing routine clinical care; however, current designs have major deficiencies. An overview of the various challenges that face modern clinical research and the methods that can be exploited to solve these challenges, in the context of personalised cancer treatment in the 21st century is provided. Aim. The purpose of this manuscript, without intending to be comprehensive, is to spark thought whilst presenting and discussing two important and complementary alternatives to traditional evidence-based medicine, specifically rapid learning health care and cohort multiple randomised controlled trial design. Rapid learning health care is an approach that proposes to extract and apply knowledge from routine clinical care data rather than exclusively depending on clinical trial evidence, (please watch the animation: http://youtu.be/ZDJFOxpwqEA). The cohort multiple randomised controlled trial design is a pragmatic method which has been proposed to help overcome the weaknesses of conventional randomised trials, taking advantage of the standardised follow-up approaches more and more used in routine patient care. This approach is particularly useful when the new intervention is a priori attractive for the patient (i.e. proton therapy, patient decision aids or expensive medications), when the outcomes are easily collected, and when there is no need of a placebo arm. Discussion. Truly personalised cancer treatment is the goal in modern radiotherapy. However, personalised cancer treatment is also an immense challenge. The vast variety of both cancer patients and treatment options makes it extremely difficult to determine which decisions are optimal for the individual patient. Nevertheless, rapid learning health care and cohort multiple randomised controlled trial design are two approaches (among others) that can help meet this challenge.

PET-based dose painting in non-small cell lung cancer: Comparing uniform dose escalation with boosting hypoxic and metabolically active sub-volumes

BACKGROUND AND PURPOSE We compared two imaging biomarkers for dose-escalation in patients with advanced non-small cell lung cancer (NSCLC). Treatment plans boosting metabolically active sub-volumes defined by FDG-PET or hypoxic sub-volumes defined by HX4-PET were compared with boosting the entire tumour. MATERIALS AND METHODS Ten NSCLC patients underwent FDG- and HX4-PET/CT scans prior to radiotherapy. Three isotoxic dose-escalation plans were compared per patient: plan A, boosting the primary tumour (PTVprim); plan B, boosting sub-volume with FDG >50% SUVmax (PTVFDG); plan C, boosting hypoxic volume with HX4 tumour-to-background >1.4 (PTVHX4). RESULTS Average boost volumes were 507 ± 466 cm(3) for PTVprim, 173 ± 127 cm(3) for PTVFDG and 114 ± 73 cm(3) for PTVHX4. The smaller PTVHX4 overlapped on average 87 ± 16% with PTVFDG. Prescribed dose was escalated to 87 ± 10 Gy for PTVprim, 107 ± 20 Gy for PTVFDG, and 117 ± 15 Gy for PTVHX4, with comparable doses to the relevant organs-at-risk (OAR). Treatment plans are available online (https://www.cancerdata.org/10.1016/j.radonc.2015.07.013). CONCLUSIONS Dose escalation based on metabolic sub-volumes, hypoxic sub-volumes and the entire tumour is feasible. Highest dose was achieved for hypoxia plans, without increasing dose to OAR. For most patients, boosting the metabolic sub-volume also resulted in boosting the hypoxic volume, although to a lower dose, but not vice versa.

Decision support systems for personalized and participative radiation oncology.

Abstract A paradigm shift from current population based medicine to personalized and participative medicine is underway. This transition is being supported by the development of clinical decision support systems based on prediction models of treatment outcome. In radiation oncology, these models ‘learn’ using advanced and innovative information technologies (ideally in a distributed fashion — please watch the animation: http://youtu.be/ZDJFOxpwqEA) from all available/appropriate medical data (clinical, treatment, imaging, biological/genetic, etc.) to achieve the highest possible accuracy with respect to prediction of tumor response and normal tissue toxicity. In this position paper, we deliver an overview of the factors that are associated with outcome in radiation oncology and discuss the methodology behind the development of accurate prediction models, which is a multi‐faceted process. Subsequent to initial development/validation and clinical introduction, decision support systems should be constantly re‐evaluated (through quality assurance procedures) in different patient datasets in order to refine and re‐optimize the models, ensuring the continuous utility of the models. In the reasonably near future, decision support systems will be fully integrated within the clinic, with data and knowledge being shared in a standardized, dynamic, and potentially global manner enabling truly personalized and participative medicine. Graphical abstract Figure. No caption available.

PET imaging of zirconium-89 labelled cetuximab: A phase I trial in patients with head and neck and lung cancer

BACKGROUND AND PURPOSE PET imaging of cetuximab uptake may help selecting cancer patients with the highest chance of benefit. The aim of this phase I trial was to determine the safety of the tracer 89Zr-cetuximab and to assess tumour uptake. METHODS Two dose schedules were used; two consecutive doses of 60MBq 89Zr-cetuximab or a single dose of 120MBq, both preceded by 400mg/m2 of unlabelled cetuximab. Toxicity (CTCAE 3.0) was scored twice weekly. PET-CT scans were acquired on days 4, 5 and 6 (step 1) or 5, 6, 7 (step 2). Because tumour uptake could not be assessed satisfactorily, a third step was added including EGFR overexpressing tumours. RESULTS Nine patients were included (6 NSCLC; 3 HNC). No additional toxicity was associated with administration of 89Zr-cetuximab compared to standard cetuximab. A tumour to blood ratio (TBR)>1 was observed in all but one patient, with a maximum of 4.56. TBR was not different between dose schedules. There was a trend for higher TBR at intervals>5days after injection. CONCLUSIONS Both presented 89Zr-cetuximab administration schedules are safe. The recommended dose for future trials is 60MBq, with a minimum time interval for scanning of 6days.

Data from: Quantitative assessment of Zirconium-89 labeled cetuximab using PETCT imaging in patients with advanced head and neck cancer - a theragnostic approach

Biomarkers predicting treatment response to the monoclonal antibody cetuximab in locally advanced head and neck squamous cell carcinomas (LAHNSCC) are lacking. We hypothesize that tumor accessibility is an important factor in treatment success of the EGFR targeting drug. We quantified uptake of cetuximab labeled with Zirconium-89 (89Zr) using PET/CT imaging. Seventeen patients with stage III-IV LAHNSCC received a loading dose unlabeled cetuximab, followed by 10 mg 54.5±9.6 MBq 89Zr-cetuximab. PET/CT images were acquired either 3 and 6 or 4 and 7 days post-injection. 89Zr-cetuximab uptake was quantified using standardized uptake value (SUV) and tumor-to-background ratio (TBR), and correlated to EGFR immunohistochemistry. TBR was compared between scan days to determine optimal timing. Uptake of 89Zr-cetuximab varied between patients (day 6-7: SUVpeak range 2.5-6.2). TBR increased significantly (49±28%, p < 0.01) between first (1.1±0.3) and second scan (1.7±0.6). Between groups with a low and high EGFR expression a significant difference in SUVmean (2.1 versus 3.0) and SUVpeak (3.2 versus 4.7) was found, however, not in TBR. Data is available at www.cancerdata.org (DOI: 10.17195/candat.2016.11.1). In conclusion, 89Zr-cetuximab PET imaging shows large inter-patient variety in LAHNSCC and provides additional information over FDG-PET and EGFR expression. Validation of the predictive value is recommended with scans acquired 6-7 days post-injection.

Defining the hypoxic target volume based on positron emission tomography for image guided radiotherapy – the influence of the choice of the reference region and conversion function

Abstract Background: Hypoxia imaged by positron emission tomography (PET) is a potential target for optimization in radiotherapy. However, the implementation of this approach with respect to the conversion of intensities in the images into oxygenation and radiosensitivity maps is not straightforward. This study investigated the feasibility of applying two conversion approaches previously derived for 18F-labeled fluoromisonidazole (18F-FMISO)-PET images for the hypoxia tracer 18F-flortanidazole (18F-HX4). Material and methods: Ten non-small-cell lung cancer patients imaged with 18F-HX4 before the start of radiotherapy were considered in this study. PET image uptake was normalized to a well-oxygenated reference region and subsequently linear and non-linear conversions were used to determine tissue oxygenations maps. These were subsequently used to delineate hypoxic volumes based partial oxygen pressure (pO2) thresholds. The results were compared to hypoxic volumes segmented using a tissue-to-background ratio of 1.4 for 18F-HX4 uptake. Results: While the linear conversion function was not found to result in realistic oxygenation maps, the non-linear function resulted in reasonably sized sub-volumes in good agreement with uptake-based segmented volumes for a limited range of pO2 thresholds. However, the pO2 values corresponding to this range were significantly higher than what is normally considered as hypoxia. The similarity in size, shape, and relative location between uptake-based sub-volumes and volumes based on the conversion to pO2 suggests that the relationship between uptake and pO2 is similar for 18F-FMISO and 18F-HX4, but that the model parameters need to be adjusted for the latter. Conclusions: A non-linear conversion function between uptake and oxygen partial pressure for 18F-FMISO-PET could be applied to 18F-HX4 images to delineate hypoxic sub-volumes of similar size, shape, and relative location as based directly on the uptake. In order to apply the model for e.g., dose-painting, new parameters need to be derived for the accurate calculation of dose-modifying factors for this tracer.

Predicting tumor hypoxia in non-small cell lung cancer by combining CT, FDG PET and dynamic contrast-enhanced CT

Abstract Background: Most solid tumors contain inadequately oxygenated (i.e., hypoxic) regions, which tend to be more aggressive and treatment resistant. Hypoxia PET allows visualization of hypoxia and may enable treatment adaptation. However, hypoxia PET imaging is expensive, time-consuming and not widely available. We aimed to predict hypoxia levels in non-small cell lung cancer (NSCLC) using more easily available imaging modalities: FDG-PET/CT and dynamic contrast-enhanced CT (DCE-CT). Material and methods: For 34 NSCLC patients, included in two clinical trials, hypoxia HX4-PET/CT, planning FDG-PET/CT and DCE-CT scans were acquired before radiotherapy. Scans were non-rigidly registered to the planning CT. Tumor blood flow (BF) and blood volume (BV) were calculated by kinetic analysis of DCE-CT images. Within the gross tumor volume, independent clusters, i.e., supervoxels, were created based on FDG-PET/CT. For each supervoxel, tumor-to-background ratios (TBR) were calculated (median SUV/aorta SUVmean) for HX4-PET/CT and supervoxel features (median, SD, entropy) for the other modalities. Two random forest models (cross-validated: 10 folds, five repeats) were trained to predict the hypoxia TBR; one based on CT, FDG, BF and BV, and one with only CT and FDG features. Patients were split in a training (trial NCT01024829) and independent test set (trial NCT01210378). For each patient, predicted, and observed hypoxic volumes (HV) (TBR > 1.2) were compared. Results: Fifteen patients (3291 supervoxels) were used for training and 19 patients (1502 supervoxels) for testing. The model with all features (RMSE training: 0.19 ± 0.01, test: 0.27) outperformed the model with only CT and FDG-PET features (RMSE training: 0.20 ± 0.01, test: 0.29). All tumors of the test set were correctly classified as normoxic or hypoxic (HV > 1 cm3) by the best performing model. Conclusions: We created a data-driven methodology to predict hypoxia levels and hypoxia spatial patterns using CT, FDG-PET and DCE-CT features in NSCLC. The model correctly classifies all tumors, and could therefore, aid tumor hypoxia classification and patient stratification.

Clustering of multi-parametric functional imaging to identify high-risk subvolumes in non-small cell lung cancer

BACKGROUND AND PURPOSE We aimed to identify tumour subregions with characteristic phenotypes based on pre-treatment multi-parametric functional imaging and correlate these subregions to treatment outcome. The subregions were created using imaging of metabolic activity (FDG-PET/CT), hypoxia (HX4-PET/CT) and tumour vasculature (DCE-CT). MATERIALS AND METHODS 36 non-small cell lung cancer (NSCLC) patients underwent functional imaging prior to radical radiotherapy. Kinetic analysis was performed on DCE-CT scans to acquire blood flow (BF) and volume (BV) maps. HX4-PET/CT and DCE-CT scans were non-rigidly co-registered to the planning FDG-PET/CT. Two clustering steps were performed on multi-parametric images: first to segment each tumour into homogeneous subregions (i.e. supervoxels) and second to group the supervoxels of all tumours into phenotypic clusters. Patients were split based on the absolute or relative volume of supervoxels in each cluster; overall survival was compared using a log-rank test. RESULTS Unsupervised clustering of supervoxels yielded four independent clusters. One cluster (high hypoxia, high FDG, intermediate BF/BV) related to a high-risk tumour type: patients assigned to this cluster had significantly worse survival compared to patients not in this cluster (p = 0.035). CONCLUSIONS We designed a subregional analysis for multi-parametric imaging in NSCLC, and showed the potential of subregion classification as a biomarker for prognosis. This methodology allows for a comprehensive data-driven analysis of multi-parametric functional images.