Correlation of gene expression with magnetic resonance imaging features of retinoblastoma: a multi-center radiogenomics validation study.

OBJECTIVES: To validate associations between MRI features and gene expression profiles in retinoblastoma, thereby evaluating the repeatability of radiogenomics in retinoblastoma. METHODS: In this retrospective multicenter cohort study, retinoblastoma patients with gene expression data and MRI were included. MRI features (scored blinded for clinical data) and matched genome-wide gene expression data were used to perform radiogenomic analysis. Expression data from each center were first separately processed and analyzed. The end product normalized expression values from different sites were subsequently merged by their Z-score to permit cross-sites validation analysis. The MRI features were non-parametrically correlated with expression of photoreceptorness (radiogenomic analysis), a gene expression signature informing on disease progression. Outcomes were compared to outcomes in a previous described cohort. RESULTS: Thirty-six retinoblastoma patients were included, 15 were female (42%), and mean age was 24 (SD 18) months. Similar to the prior evaluation, this validation study showed that low photoreceptorness gene expression was associated with advanced stage imaging features. Validated imaging features associated with low photoreceptorness were multifocality, a tumor encompassing the entire retina or entire globe, and a diffuse growth pattern (all p < 0.05). There were a number of radiogenomic associations that were also not validated. CONCLUSIONS: A part of the radiogenomic associations could not be validated, underlining the importance of validation studies. Nevertheless, cross-center validation of imaging features associated with photoreceptorness gene expression highlighted the capability radiogenomics to non-invasively inform on molecular subtypes in retinoblastoma. CLINICAL RELEVANCE STATEMENT: Radiogenomics may serve as a surrogate for molecular subtyping based on histopathology material in an era of eye-sparing retinoblastoma treatment strategies. KEY POINTS: • Since retinoblastoma is increasingly treated using eye-sparing methods, MRI features informing on molecular subtypes that do not rely on histopathology material are important. • A part of the associations between retinoblastoma MRI features and gene expression profiles (radiogenomics) were validated. • Radiogenomics could be a non-invasive technique providing information on the molecular make-up of retinoblastoma.

Optic nerve thickening on high-spatial-resolution MRI predicts early-stage postlaminar optic nerve invasion in retinoblastoma.

OBJECTIVES: To assess the diagnostic accuracy of nerve thickening on MRI to predict early-stage postlaminar optic nerve invasion (PLONI) in retinoblastoma. Furthermore, this study aimed to incorporate measurements into a multiparametric model for radiological determination of PLONI. METHODS: In this retrospective multicenter case-control study, high-spatial-resolution 3D T2-weighted MR images were used to measure the distal optic nerve. Histopathology was the reference standard for PLONI. Two neuroradiologists independently measured the optic nerve width, height, and surface at 0, 3, and 5 mm from the most distal part of the optic nerve. Subsequently, PLONI was scored on contrast-enhanced T1-weighted and 3D T2-weighted images, blinded for clinical data. Optic nerve measurements with the highest diagnostic accuracy for PLONI were incorporated into a prediction model for radiological determination of PLONI. RESULTS: One hundred twenty-four retinoblastoma patients (median age, 22 months [range, 0-113], 58 female) were included, resulting in 25 retinoblastoma eyes with histopathologically proven PLONI and 206 without PLONI. ROC analysis of axial optic nerve width measured at 0 mm yielded the best area under the curve of 0.88 (95% confidence interval: 0.79, 0.96; p < 0.001). The optimal width cutoff was ≥ 2.215 mm, with a sensitivity of 84% (95% CI: 64, 95%) and specificity of 83% (95% CI: 75, 89%) for detecting PLONI. Combining width measurements with the suspicion of PLONI on MRI sequences resulted in a prediction model with an improved sensitivity and specificity of respectively up to 88% and 92%. CONCLUSION: Postlaminar optic nerve thickening can predict early-stage postlaminar optic nerve invasion in retinoblastoma. CLINICAL RELEVANCE STATEMENT: This study provides an additional tool for clinicians to help determine postlaminar optic nerve invasion, which is a risk factor for developing metastatic disease in retinoblastoma patients. KEY POINTS: • The diagnostic accuracy of contrast-enhanced MRI for detecting postlaminar optic nerve invasion is limited in retinoblastoma patients. • Optic nerve thickening can predict postlaminar optic nerve invasion. • A prediction model combining MRI features has a high sensitivity and specificity for detecting postlaminar optic nerve invasion.

MRI Features for Identifying MYCN-amplified RB1 Wild-type Retinoblastoma.

Background MYCN-amplified RB1 wild-type (MYCNARB1+/+) retinoblastoma is a rare but clinically important subtype of retinoblastoma due to its aggressive character and relative resistance to typical therapeutic approaches. Because biopsy is not indicated in retinoblastoma, specific MRI features might be valuable to identify children with this genetic subtype. Purpose To define the MRI phenotype of MYCNARB1+/+ retinoblastoma and evaluate the ability of qualitative MRI features to help identify this specific genetic subtype. Materials and Methods In this retrospective, multicenter, case-control study, MRI scans in children with MYCNARB1+/+ retinoblastoma and age-matched children with RB1-/- subtype retinoblastoma were included (case-control ratio, 1:4; scans acquired from June 2001 to February 2021; scans collected from May 2018 to October 2021). Patients with histopathologically confirmed unilateral retinoblastoma, genetic testing (RB1/MYCN status), and MRI scans were included. Associations between radiologist-scored imaging features and diagnosis were assessed with the Fisher exact test or Fisher-Freeman-Halton test, and Bonferroni-corrected P values were calculated. Results A total of 110 patients from 10 retinoblastoma referral centers were included: 22 children with MYCNARB1+/+ retinoblastoma and 88 control children with RB1-/- retinoblastoma. Children in the MYCNARB1+/+ group had a median age of 7.0 months (IQR, 5.0-9.0 months) (13 boys), while children in the RB1-/- group had a median age of 9.0 months (IQR, 4.6-13.4 months) (46 boys). MYCNARB1+/+ retinoblastomas were typically peripherally located (in 10 of 17 children; specificity, 97%; P < .001) and exhibited plaque or pleomorphic shape (in 20 of 22 children; specificity, 51%; P = .011) with irregular margins (in 16 of 22 children; specificity, 70%; P = .008) and extensive retina folding with vitreous enclosure (specificity, 94%; P < .001). MYCNARB1+/+ retinoblastomas showed peritumoral hemorrhage (in 17 of 21 children; specificity, 88%; P < .001), subretinal hemorrhage with a fluid-fluid level (in eight of 22 children; specificity, 95%; P = .005), and strong anterior chamber enhancement (in 13 of 21 children; specificity, 80%; P = .008). Conclusion MYCNARB1+/+ retinoblastomas show distinct MRI features that could enable early identification of these tumors. This may improve patient selection for tailored treatment in the future. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Rollins in this issue.

Automatic segmentation and quantification of the optic nerve on MRI using a 3D U-Net.

Purpose: Pathological conditions associated with the optic nerve (ON) can cause structural changes in the nerve. Quantifying these changes could provide further understanding of disease mechanisms. We aim to develop a framework that automatically segments the ON separately from its surrounding cerebrospinal fluid (CSF) on magnetic resonance imaging (MRI) and quantifies the diameter and cross-sectional area along the entire length of the nerve. Approach: Multicenter data were obtained from retinoblastoma referral centers, providing a heterogeneous dataset of 40 high-resolution 3D T2-weighted MRI scans with manual ground truth delineations of both ONs. A 3D U-Net was used for ON segmentation, and performance was assessed in a tenfold cross-validation (n=32) and on a separate test-set (n=8) by measuring spatial, volumetric, and distance agreement with manual ground truths. Segmentations were used to quantify diameter and cross-sectional area along the length of the ON, using centerline extraction of tubular 3D surface models. Absolute agreement between automated and manual measurements was assessed by the intraclass correlation coefficient (ICC). Results: The segmentation network achieved high performance, with a mean Dice similarity coefficient score of 0.84, median Hausdorff distance of 0.64 mm, and ICC of 0.95 on the test-set. The quantification method obtained acceptable correspondence to manual reference measurements with mean ICC values of 0.76 for the diameter and 0.71 for the cross-sectional area. Compared with other methods, our method precisely identifies the ON from surrounding CSF and accurately estimates its diameter along the nerve's centerline. Conclusions: Our automated framework provides an objective method for ON assessment in vivo.

Magnetic resonance imaging based radiomics prediction of Human Papillomavirus infection status and overall survival in oropharyngeal squamous cell carcinoma.

OBJECTIVES: Human papillomavirus- (HPV) positive oropharyngeal squamous cell carcinoma (OPSCC) differs biologically and clinically from HPV-negative OPSCC and has a better prognosis. This study aims to analyze the value of magnetic resonance imaging (MRI)-based radiomics in predicting HPV status in OPSCC and aims to develop a prognostic model in OPSCC including HPV status and MRI-based radiomics. MATERIALS AND METHODS: Manual delineation of 249 primary OPSCCs (91 HPV-positive and 159 HPV-negative) on pretreatment native T1-weighted MRIs was performed and used to extract 498 radiomic features per delineation. A logistic regression (LR) and random forest (RF) model were developed using univariate feature selection. Additionally, factor analysis was performed, and the derived factors were combined with clinical data in a predictive model to assess the performance on predicting HPV status. Additionally, factors were combined with clinical parameters in a multivariable survival regression analysis. RESULTS: Both feature-based LR and RF models performed with an AUC of 0.79 in prediction of HPV status. Fourteen of the twenty most significant features were similar in both models, mainly concerning tumor sphericity, intensity variation, compactness, and tumor diameter. The model combining clinical data and radiomic factors (AUC = 0.89) outperformed the radiomics-only model in predicting OPSCC HPV status. Overall survival prediction was most accurate using the combination of clinical parameters and radiomic factors (C-index = 0.72). CONCLUSION: Predictive models based on MR-radiomic features were able to predict HPV status with sufficient performance, supporting the role of MRI-based radiomics as potential imaging biomarker. Survival prediction improved by combining clinical features with MRI-based radiomics.

Clinical impact of MRI-detected additional lesions in breast cancer patients with neoadjuvant systemic therapy at the Netherlands cancer institute.

BACKGROUND: In breast cancer patients treated with neoadjuvant systemic therapy (NST), MRI is used pre- and post-NST for response monitoring. The relevance of additional MRI-detected lesions in these patients is unclear. Therefore, we aimed to assess the impact of pre-NST MRI-detected additional lesions on surgical treatment and outcome. METHODS: We retrospectively selected all early-stage breast cancer patients with MRI pre-NST at our institute from January 2010-2015. MRI-detected lesions were defined as separated from the index tumor and occult at conventional mammography and ultrasound. Outcomes were change in surgical treatment and five-year recurrence-free and overall survival. RESULTS: Overall, MRI detected additional lesions in 206 (31%) of 656 patients: in 160 patients in the ipsilateral breast and in 78 contralateral breasts, including 32 bilateral cases. Ipsilateral lesions were mostly categorized BI-RADS 5 (54 %) and contralateral lesions BI-RADS 3 (64%). Targeted ultrasound was performed in 115 (56%) patients: in 70 ipsilateral and in 64 contralateral cases. Biopsy was obtained in 44 (28% of 160) ipsilateral and 50 (64% of 78) contralateral breasts, containing tumor foci in 20 (13% of 160) and 11 (14% of 78) cases, respectively. Surgical treatment changed in 54 (26% of 206) patients: 19 (9%) had mastectomy, 24 (12%) had wider local excision and 11 (5%) underwent contralateral surgery. Five-year recurrence-free and overall survival did not differ for patients with local excision or mastectomy. CONCLUSION: Pre-NST MRI-detected additional lesions in 31% of patients, resulting in more extensive surgery in 26% of these patients, including 5% contralateral surgeries.

Magnetic Resonance Imaging Can Reliably Differentiate Optic Nerve Inflammation from Tumor Invasion in Retinoblastoma with Orbital Cellulitis.

PURPOSE: To investigate the prevalence and magnetic resonance imaging (MRI) phenotype of retinoblastoma-associated orbital cellulitis. Additionally, this study aimed to identify postlaminar optic nerve enhancement (PLONE) patterns differentiating between inflammation and tumor invasion. DESIGN: A monocenter cohort study assessed the prevalence of orbital cellulitis features on MRI in retinoblastoma patients. A multicenter case-control study compared MRI features of the retinoblastoma-associated orbital cellulitis cases with retinoblastoma controls. PARTICIPANTS: A consecutive retinoblastoma patient cohort of 236 patients (311 eyes) was retrospectively investigated. Subsequently, 30 retinoblastoma cases with orbital cellulitis were compared with 30 matched retinoblastoma controls without cellulitis. METHODS: In the cohort study, retinoblastoma MRI scans were scored on presence of inflammatory features. In the case-control study, MRI scans were scored on intraocular features and PLONE patterns. Postlaminar enhancement patterns were compared with histopathologic assessment of postlaminar tumor invasion. Interreader agreement was assessed, and exact tests with Bonferroni correction were adopted for statistical comparisons. MAIN OUTCOME MEASURES: Prevalence of retinoblastoma-associated orbital cellulitis on MRI was calculated. Frequency of intraocular MRI features was compared between cases and controls. Sensitivity and specificity of postlaminar optic nerve patterns for detection of postlaminar tumor invasion were assessed. RESULTS: The MRI prevalence of retinoblastoma-associated orbital cellulitis was 6.8% (16/236). Retinoblastoma with orbital cellulitis showed significantly more tumor necrosis, uveal abnormalities (inflammation, hemorrhage, and necrosis), lens luxation (all P < 0.001), and a larger eye size (P = 0.012). The inflammatory pattern of optic nerve enhancement (strong enhancement similar to adjacent choroid) was solely found in orbital cellulitis cases, of which none (0/16) showed tumor invasion on histopathology. Invasive pattern enhancement was found in both cases and controls, of which 50% (5/10) showed tumor invasion on histopathology. Considering these different enhancement patterns suggestive for either inflammation or tumor invasion increased specificity for detection of postlaminar tumor invasion in orbital cellulitis cases from 32% (95% confidence interval [CI], 16-52) to 89% (95% CI, 72-98). CONCLUSIONS: Retinoblastoma cases presenting with orbital cellulitis show MRI findings of a larger eye size, extensive tumor necrosis, uveal abnormalities, and lens luxation. Magnetic resonance imaging contrast-enhancement patterns within the postlaminar optic nerve can differentiate between tumor invasion and inflammatory changes.

Multi-view convolutional neural networks for automated ocular structure and tumor segmentation in retinoblastoma.

In retinoblastoma, accurate segmentation of ocular structure and tumor tissue is important when working towards personalized treatment. This retrospective study serves to evaluate the performance of multi-view convolutional neural networks (MV-CNNs) for automated eye and tumor segmentation on MRI in retinoblastoma patients. Forty retinoblastoma and 20 healthy-eyes from 30 patients were included in a train/test (N = 29 retinoblastoma-, 17 healthy-eyes) and independent validation (N = 11 retinoblastoma-, 3 healthy-eyes) set. Imaging was done using 3.0 T Fast Imaging Employing Steady-state Acquisition (FIESTA), T2-weighted and contrast-enhanced T1-weighted sequences. Sclera, vitreous humour, lens, retinal detachment and tumor were manually delineated on FIESTA images to serve as a reference standard. Volumetric and spatial performance were assessed by calculating intra-class correlation (ICC) and dice similarity coefficient (DSC). Additionally, the effects of multi-scale, sequences and data augmentation were explored. Optimal performance was obtained by using a three-level pyramid MV-CNN with FIESTA, T2 and T1c sequences and data augmentation. Eye and tumor volumetric ICC were 0.997 and 0.996, respectively. Median [Interquartile range] DSC for eye, sclera, vitreous, lens, retinal detachment and tumor were 0.965 [0.950-0.975], 0.847 [0.782-0.893], 0.975 [0.930-0.986], 0.909 [0.847-0.951], 0.828 [0.458-0.962] and 0.914 [0.852-0.958], respectively. MV-CNN can be used to obtain accurate ocular structure and tumor segmentations in retinoblastoma.

MR Imaging Features to Differentiate Retinoblastoma from Coats' Disease and Persistent Fetal Vasculature.

Retinoblastoma mimickers, or pseudoretinoblastoma, are conditions that show similarities with the pediatric cancer retinoblastoma. However, false-positive retinoblastoma diagnosis can cause mistreatment, while false-negative diagnosis can cause life-threatening treatment delay. The purpose of this study is to identify the MR imaging features that best differentiate between retinoblastoma and the most common pseudoretinoblastoma diagnoses: Coats' disease and persistent fetal vasculature (PFV). Here, six expert radiologists performed retrospective assessments (blinded for diagnosis) of MR images of patients with a final diagnosis based on histopathology or clinical follow-up. Associations between 20 predefined imaging features and diagnosis were assessed with exact tests corrected for multiple hypothesis testing. Sixty-six patients were included, of which 33 (50%) were retinoblastoma and 33 (50%) pseudoretinoblastoma patients. A larger eye size, vitreous seeding, and sharp-V-shaped retinal detachment were almost exclusively found in retinoblastoma (p < 0.001-0.022, specificity 93-97%). Features that were almost exclusively found in pseudoretinoblastoma included smaller eye size, ciliary/lens deformations, optic nerve atrophy, a central stalk between optic disc and lens, Y-shaped retinal detachment, and absence of calcifications (p < 0.001-0.022, specificity 91-100%). Additionally, three newly identified imaging features were exclusively present in pseudoretinoblastoma: intraretinal macrocysts (p < 0.001, 38% [9/24] in Coats' disease and 20% [2/10] in PFV), contrast enhancement outside the solid lesion (p < 0.001, 30% [7/23] in Coats' disease and 57% [4/7] in PFV), and enhancing subfoveal nodules (38% [9/24] in Coats' disease). An assessment strategy was proposed for MR imaging differentiation between retinoblastoma and pseudoretinoblastoma, including three newly identified differentiating MR imaging features.