Laurent Dercle

Promises and challenges for the implementation of computational medical imaging (radiomics) in oncology

Medical image processing and analysis (also known as Radiomics) is a rapidly growing discipline that maps digital medical images into quantitative data, with the end goal of generating imaging biomarkers as decision support tools for clinical practice. The use of imaging data from routine clinical work-up has tremendous potential in improving cancer care by heightening understanding of tumor biology and aiding in the implementation of precision medicine. As a noninvasive method of assessing the tumor and its microenvironment in their entirety, radiomics allows the evaluation and monitoring of tumor characteristics such as temporal and spatial heterogeneity. One can observe a rapid increase in the number of computational medical imaging publications-milestones that have highlighted the utility of imaging biomarkers in oncology. Nevertheless, the use of radiomics as clinical biomarkers still necessitates amelioration and standardization in order to achieve routine clinical adoption. This Review addresses the critical issues to ensure the proper development of radiomics as a biomarker and facilitate its implementation in clinical practice.

18F-FDG PET and CT-scan Detect New Imaging Patterns of Response and Progression in Patients with Hodgkin Lymphoma Treated by Anti-PD1 Immune Checkpoint Inhibitor

The response evaluation criteria in patients with Hodgkin lymphoma (HL) were designed for the assessment of chemotherapy and targeted molecular agents. We investigated the accuracy of 3-mo 18F-FDG PET/CT for the identification of HL patients responding to immune-checkpoint blockade by anti–programmed death 1 antibodies (anti-PD1). We also reported the frequency of new immune patterns of response and progression. Methods: Retrospectively, we recruited consecutive HL patients treated by anti-PD1 (pembrolizumab or nivolumab) at Gustave Roussy from 2013 to 2015. 18F-FDG PET/CT and contrast-enhanced CT scans were acquired every 3 mo. We recorded the best overall response according to the International Harmonization Project Cheson 2014 criteria and LYmphoma Response to Immunomodulatory therapy Criteria (LYRIC) (2016 revised criteria). Patients achieving an objective response at any time during the anti-PD1 treatment were classified as responders. Results: Sixteen relapsed or refractory classic HL patients were included. The median age was 39 y (age range, 19–69 y). The median previous lines of therapy was 6 (range, 3–13). The mean follow-up was 22.6 mo. Nine of 16 patients (56%) achieved an objective response. Two deaths occurred due to progressive disease at 7 mo. 18F-FDG PET/CT detected all responders at 3 mo and reclassified best overall response in 5 patients compared with CT alone. A decrease in tumor metabolism and volume (SUVmean, metabolic tumor volume) and increase in healthy splenic metabolism at 3 mo were observed in responders (area under the curve > 0.85, P < 0.04). Five of 16 patients (31%) displayed new imaging patterns related to anti-PD1; we observed 2 transient progressions consistent with indeterminate response according to the LYRIC (2016) (IR2b at 14 mo and IR3 at 18 mo) and 3 patients with new lesions associated with immune-related adverse events. Conclusion: Three-month 18F-FDG PET/CT scans detected HL patients responding to anti-PD1. New patterns were encountered in 31% of patients, emphasizing the need for further evaluation in larger series and close collaboration between imaging and oncology specialists on a per-patient basis.

Prediction of cervical cancer recurrence using textural features extracted from 18 F-FDG PET images acquired with different scanners

Objectives To identify an imaging signature predicting local recurrence for locally advanced cervical cancer (LACC) treated by chemoradiation and brachytherapy from baseline 18F-FDG PET images, and to evaluate the possibility of gathering images from two different PET scanners in a radiomic study. Methods 118 patients were included retrospectively. Two groups (G1, G2) were defined according to the PET scanner used for image acquisition. Eleven radiomic features were extracted from delineated cervical tumors to evaluate: (i) the predictive value of features for local recurrence of LACC, (ii) their reproducibility as a function of the scanner within a hepatic reference volume, (iii) the impact of voxel size on feature values. Results Eight features were statistically significant predictors of local recurrence in G1 (p < 0.05). The multivariate signature trained in G2 was validated in G1 (AUC=0.76, p<0.001) and identified local recurrence more accurately than SUVmax (p=0.022). Four features were significantly different between G1 and G2 in the liver. Spatial resampling was not sufficient to explain the stratification effect. Conclusion This study showed that radiomic features could predict local recurrence of LACC better than SUVmax. Further investigation is needed before applying a model designed using data from one PET scanner to another.

Limits of radiomic-based entropy as a surrogate of tumor heterogeneity: ROI-area, acquisition protocol and tissue site exert substantial influence

Entropy is a promising quantitative imaging biomarker for characterizing cancer imaging phenotype. Entropy has been associated with tumor gene expression, tumor metabolism, tumor stage, patient prognosis, and treatment response. Our hypothesis states that tumor-specific biomarkers such as entropy should be correlated between synchronous metastases. Therefore, a significant proportion of the variance of entropy should be attributed to the malignant process. We analyzed 112 patients with matched/paired synchronous metastases (SM#1 and SM#2) prospectively enrolled in the MOSCATO-01 clinical trial. Imaging features were extracted from Regions Of Interest (ROI) delineated on CT-scan using TexRAD software. We showed that synchronous metastasis entropy was correlated across 5 Spatial Scale Filters: Spearman’s Rho ranged between 0.41 and 0.59 (P = 0.0001, Bonferroni correction). Multivariate linear analysis revealed that entropy in SM#1 is significantly associated with (i) primary tumor type; (ii) entropy in SM#2 (same malignant process); (iii) ROI area size; (iv) metastasis site; and (v) entropy in the psoas muscle (reference tissue). Entropy was a logarithmic function of ROI area in normal control tissues (aorta, psoas) and in mathematical models (P < 0.01). We concluded that entropy is a tumor-specific metric only if confounding factors are corrected.

The Role of 18F-FDG PET/CT and PET/MRI in Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) remains a difficult disease to treat and continues to portend a poor prognosis, as most patients are unresectable at diagnosis. 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) combined with CT (PET/CT) has been a cornerstone in oncological imaging of different cancers; however, the role of PET/CT in PDAC is continually evolving and currently not well established. Studies have shown the potential of PET/CT in guiding the management of patients with PDAC, with possible added benefit over anatomic imaging with CT or MRI in certain scenarios. PET/CT may be useful in diagnosis, initial staging, treatment response assessment, differentiation of recurrent tumor from post-treatment fibrosis, and radiotherapy planning. Additionally, PET/CT may be a cost-effective modality due to upstaging of patients originally deemed as surgical candidates. Recently, the advent of simultaneous PET/MRI represents an exciting advancement in hybrid functional imaging with potential applications in the imaging of PDAC. The advantages of PET/MRI include simultaneous acquisition to improve registration of fusion images, lower radiation dose, superior soft tissue contrast, and availability of multiparametric imaging. Studies are underway to evaluate the utility of PET/MRI in PDAC, including in initial staging and treatment response assessment and to determine the subgroup of patients that will benefit from PET/MRI. Further studies are warranted in both PET/CR and PET/MRI to better understand the role of these modalities in PDAC.

Vol-PACT: A Foundation for the NIH Public-Private Partnership That Supports Sharing of Clinical Trial Data for the Development of Improved Imaging Biomarkers in Oncology

PURPOSE To develop a public-private partnership to study the feasibility of a new approach in collecting and analyzing clinically annotated imaging data from landmark phase III trials in advanced solid tumors. PATIENTS AND METHODS The collection of clinical trials fulfilled the following inclusion criteria: completed randomized trials of > 300 patients, highly measurable solid tumors (non-small-cell lung cancer, colorectal cancer, renal cell cancer, and melanoma), and required sponsor and institutional review board sign-offs. The new approach in analyzing computed tomography scans was to transfer to an academic image analysis laboratory, draw contours semi-automatically by using in-house-developed algorithms integrated into the open source imaging platform Weasis, and perform serial volumetric measurement. RESULTS The median duration of contracting with five sponsors was 12 months. Ten trials in 7,085 patients that covered 12 treatment regimens across 20 trial arms were collected. To date, four trials in 3,954 patients were analyzed. Source imaging data were transferred to the academic core from 97% of trial patients (n = 3,837). Tumor imaging measurements were extracted from 82% of transferred computed tomography scans (n = 3,162). Causes of extraction failure were nonmeasurable disease (n = 392), single imaging time point (n = 224), and secondary captured images (n = 59). Overall, clinically annotated imaging data were extracted in 79% of patients (n = 3,055), and the primary trial end point analysis in each trial remained representative of each original trial end point. CONCLUSION The sharing and analysis of source imaging data from large randomized trials is feasible and offer a rich and reusable, but largely untapped, resource for future research on novel trial-level response and progression imaging metrics.

Diagnostic and prognostic value of 18F-FDG PET, CT, and MRI in perineural spread of head and neck malignancies

AbstractObjectivesWe assessed whether quantitative imaging biomarkers derived from fluorodeoxyglucose-positron emission tomography (18F-FDG PET) could be extracted from perineural spread (PNS) in head and neck malignancies (HNM) to improve patient risk stratification.MethodsA case–control exploratory study (1:2 ratio) enrolled 81 patients with FDG-avid HNM. The case-group comprised 28 patients with documented PNS (reference: expert consensus), including 14 squamous cell carcinomas (SCC). Imaging biomarkers were extracted from the PNS on 18F-FDG PET, CT-scan, and MRI. The control-group enrolled 53 SCCs. The Cox proportional-hazards regression model explored the association with overall survival by univariate and multivariate analyses.ResultsThe rate of PNS detection by 18F-FDG PET was 100% in the case-group. Quantitative imaging biomarkers were not associated with the presence of sensory (p>0.20) or motor (p>0.10) symptoms. In SCC patients (case: 14; control: 53), PNS was associated with a hazard ratio of death of 5.5 (95%CI: 1.4:20.9) by multivariate analysis. Increased cranial nerve SUVmax was significantly associated with poorer overall survival by univariate analysis (p=0.001).ConclusionsOur pilot study showed the feasibility of extracting 18F-FDG PET biomarkers from PNS in FDG-avid HNM. Our results encourage the development of new PET/CT- or PET/MRI-guided management strategies in further prospective studies.Key Points• 18F-FDG PET/CT detects PNS in FDG-avid HNM. • PNS metabolism is more heterogeneous than healthy tissue. • PNS diagnosis is crucial: most patients were asymptomatic, N0 and M0. • PNS diagnosis is associated with poorer overall survival in SCC. • PET/CT- or PET/MRI-guided management strategies should be evaluated.

Abstract A051: Prediction of clinical outcomes of cancer patients treated with anti-PD-1/PD-L1 using a radiomics-based imaging score of immune infiltrate

Background: The discovery of biomarkers identifying responders to immunotherapy is a major challenge. Tumor and peritumoral immune infiltration has been shown to be associated with response to anti-PD-1/PD-L1. The aim of this study was to develop a radiomics-based imaging tool of tumor immune infiltrate and to assess whether such a tool could predict clinical outcomes of patients treated with anti-PD1/PDL1. Methods: A predictive radiomics-based model of tumor-infiltrating CD8+ T cells was trained using data from the head and neck cohort of The Cancer Imaging Archive (HNSC-TCIA). Two cohorts from our institute were used for validation. Contrast-enhanced CTs of 57 patients from the HNSC-TCIA were manually segmented (tumor and surrounding tissue) and 76 radiomics features extracted. A radiomics-based score was build using radiomics features to predict tumor-infiltrating CD8+ T-cells9 abundance, which was estimated using RNA-sequencing data from The Cancer Genome Atlas, and the Microenvironment Cell Populations-counter signature. As a first validation, this signature was applied to an independent cohort of 100 patients for whom the pathologic tumor immune infiltrate was postulated as either favorable (lymphoma, melanoma, lung, bladder, renal, MSI+ cancers, and adenopathy; 70 patients) or unfavorable (adenoid cystic carcinoma, low-grade neuroendocrine tumors, uterine leiomyoma; 30 patients). The signature was then applied on baseline-CTs of a second external cohort of 139 patients prospectively enrolled in anti PD-1/PD-L1 phase 1 trials. The median of the radiomics-based CD8+ score was used to separate patients into two groups (high and low score). Survival was estimated using Cox-proportional hazards model. Results: We developed a radiomics-based CD8+ signature using the six radiomics features that had highest performance on random forest. In the first external cohort, the radiomics-based CD8 T-cells score was associated with the postulated tumor immune infiltrate (Wilcoxon test, P Conclusions: The radiomics-based signature of CD8+ T cells appears as a promising tool to estimate tumor immune infiltrate and to infer the outcome of patients treated with anti-PD-1/PD-L1. Citation Format: Roger Sun, Elaine Johanna Limkin, Laurent Dercle, Sylvain Reuze, Stephane Champiat, David Brandao, Loic Verlingue, Samy Ammari, Sandrine Aspeslagh, Antoine Hollebecque, Christophe Massard, Aurelien Marabelle, Jean-Yves Scoazec, Charlotte Robert, Jean-Charles Soria, Eric Deutsch, Charles Ferte. Prediction of clinical outcomes of cancer patients treated with anti-PD-1/PD-L1 using a radiomics-based imaging score of immune infiltrate [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A051.

Interobserver variability in tumor contouring affects the use of radiomics to predict mutational status

Abstract. Radiomic features characterize tumor imaging phenotype. Nonsmall cell lung cancer (NSCLC) tumors are known for their complexity in shape and wide range in density. We explored the effects of variable tumor contouring on the prediction of epidermal growth factor receptor (EGFR) mutation status by radiomics in NSCLC patients treated with a targeted therapy (Gefitinib). Forty-six early stage NSCLC patients (EGFR mutant:wildtype = 20:26) were included. Three experienced radiologists independently delineated the tumors using a semiautomated segmentation software on a noncontrast-enhanced baseline and three-week post-therapy CT scan images that were reconstructed using 1.25-mm slice thickness and lung kernel. Eighty-nine radiomic features were computed on both scans and their changes (radiomic delta-features) were calculated. The highest area under the curves (AUCs) were 0.87, 0.85, and 0.80 for the three radiologists and the number of significant features (AUC>0.8) was 3, 5, and 0, respectively. The AUCs of a single feature significantly varied among radiologists (e.g., 0.88, 0.75, and 0.73 for run-length primitive length uniformity). We conclude that a three-week change in tumor imaging phenotype allows identifying the EGFR mutational status of NSCLC. However, interobserver variability in tumor contouring translates into a significant variability in radiomic metrics accuracy.