RESUMEN
BACKGROUND: Serum antibodies to the Merkel oncoprotein (AMERK) are detectable in approximately 50% of patients with Merkel cell carcinoma (MCC) and can be used to monitor for recurrence. The objective of this study was to characterize AMERK levels in patients receiving curative-intent radiation therapy (RT) for MCC and identify associations between AMERK and recurrence. METHODS: This was a retrospective study of patients with MCC who had baseline AMERK measurements before they received curative-intent RT from 2010 to 2020. Event-free survival (EFS) was calculated using the Kaplan-Meier method and Cox regression. The cumulative incidence of MCC-related recurrence (CIMR) was analyzed with death as a competing risk and the Gray test. RESULTS: The authors identified 88 patients who had baseline AMERK measurements, including 52 (59%) with detectable levels. AMERK positivity was associated with younger median age (67.8 vs. 72.0 years; p = .02) and tumor site (p = 0.02), with lower rates for those who had disease in the head/neck region (17.3% vs. 44.4%). EFS (71.3% vs. 60.4%; p = .30) and CIMR (24.4% vs. 39.6%; p = .23) were more favorable in AMERK-positive patients. Two patients had recurrences in the RT field, and both were AMERK-negative at baseline. The median time to AMERK nadir after RT was 11.2 months; and, in a 6-month post-RT landmark analysis, the proportion of patients who were AMERK-positive who became negative or who had levels that decreased by ≥50% were not associated with EFS (87.1% vs. 85.0%; p = .90) or CIMR (12.9% vs. 15.0%; p = .62). CONCLUSIONS: Positive AMERK baseline levels were correlated with younger age at MCC diagnosis and nonhead and neck tumor location, possibly related to the distribution of viral etiology. A specific post-RT AMERK decline correlating with EFS could not be identified.
RESUMEN
Purpose To develop, externally test, and evaluate clinical acceptability of a deep learning pediatric brain tumor segmentation model using stepwise transfer learning. Materials and Methods In this retrospective study, the authors leveraged two T2-weighted MRI datasets (May 2001 through December 2015) from a national brain tumor consortium (n = 184; median age, 7 years [range, 1-23 years]; 94 male patients) and a pediatric cancer center (n = 100; median age, 8 years [range, 1-19 years]; 47 male patients) to develop and evaluate deep learning neural networks for pediatric low-grade glioma segmentation using a stepwise transfer learning approach to maximize performance in a limited data scenario. The best model was externally tested on an independent test set and subjected to randomized blinded evaluation by three clinicians, wherein they assessed clinical acceptability of expert- and artificial intelligence (AI)-generated segmentations via 10-point Likert scales and Turing tests. Results The best AI model used in-domain stepwise transfer learning (median Dice score coefficient, 0.88 [IQR, 0.72-0.91] vs 0.812 [IQR, 0.56-0.89] for baseline model; P = .049). With external testing, the AI model yielded excellent accuracy using reference standards from three clinical experts (median Dice similarity coefficients: expert 1, 0.83 [IQR, 0.75-0.90]; expert 2, 0.81 [IQR, 0.70-0.89]; expert 3, 0.81 [IQR, 0.68-0.88]; mean accuracy, 0.82). For clinical benchmarking (n = 100 scans), experts rated AI-based segmentations higher on average compared with other experts (median Likert score, 9 [IQR, 7-9] vs 7 [IQR 7-9]) and rated more AI segmentations as clinically acceptable (80.2% vs 65.4%). Experts correctly predicted the origin of AI segmentations in an average of 26.0% of cases. Conclusion Stepwise transfer learning enabled expert-level automated pediatric brain tumor autosegmentation and volumetric measurement with a high level of clinical acceptability. Keywords: Stepwise Transfer Learning, Pediatric Brain Tumors, MRI Segmentation, Deep Learning Supplemental material is available for this article. © RSNA, 2024.
Asunto(s)
Neoplasias Encefálicas , Aprendizaje Profundo , Imagen por Resonancia Magnética , Humanos , Niño , Neoplasias Encefálicas/diagnóstico por imagen , Neoplasias Encefálicas/patología , Imagen por Resonancia Magnética/métodos , Masculino , Adolescente , Preescolar , Estudios Retrospectivos , Femenino , Lactante , Adulto Joven , Glioma/diagnóstico por imagen , Glioma/patología , Interpretación de Imagen Asistida por Computador/métodosRESUMEN
Pediatric glioma recurrence can cause morbidity and mortality; however, recurrence pattern and severity are heterogeneous and challenging to predict with established clinical and genomic markers. Resultingly, almost all children undergo frequent, long-term, magnetic resonance (MR) brain surveillance regardless of individual recurrence risk. Deep learning analysis of longitudinal MR may be an effective approach for improving individualized recurrence prediction in gliomas and other cancers but has thus far been infeasible with current frameworks. Here, we propose a self-supervised, deep learning approach to longitudinal medical imaging analysis, temporal learning, that models the spatiotemporal information from a patient's current and prior brain MRs to predict future recurrence. We apply temporal learning to pediatric glioma surveillance imaging for 715 patients (3,994 scans) from four distinct clinical settings. We find that longitudinal imaging analysis with temporal learning improves recurrence prediction performance by up to 41% compared to traditional approaches, with improvements in performance in both low- and high-grade glioma. We find that recurrence prediction accuracy increases incrementally with the number of historical scans available per patient. Temporal deep learning may enable point-of-care decision-support for pediatric brain tumors and be adaptable more broadly to patients with other cancers and chronic diseases undergoing surveillance imaging.
RESUMEN
Intrinsic, expansile pontine tumors typically occur in the pediatric population. These tumors characteristically present as diffuse intrinsic pontine glioma (DIPG), which is now considered as diffuse midline glioma (DMG), H3K27-mutated of the pons. DIPG has limited treatment options and a poor prognosis, and the value of tissue diagnosis from an invasive biopsy remains controversial. This study presents the case of a 19-year-old female with clinical and imaging hallmarks of DIPG, who underwent a biopsy of a tumor in the region of the right middle cerebellar peduncle. Her lesional cells were negative for H3K27M alterations and had low-grade histologic features. Next-generation sequencing revealed a frameshift mutation in the NF1 gene as the likely driver mutation. These features suggest a diagnosis of a low-grade glioma associated with NF1 loss of function, with far-reaching consequences regarding both treatment strategy and prognosis. This case provides support for the utility of diagnostic tissue biopsy in cases of suspected DIPG.
RESUMEN
BACKGROUND: Prognosis for patients with high-risk neuroblastoma (HR-NBL) is guarded despite aggressive therapy, and few studies have characterized outcomes after radiotherapy in relation to radiation treatment fields. METHODS: Multi-institutional retrospective cohort of 293 patients with HR-NBL who received autologous stem cell transplant (ASCT) and EBRT between 1997-2021. LRR was defined as recurrence at the primary site or within one nodal echelon beyond disease present at diagnosis. Follow-up was defined from the end of EBRT. Event-free survival (EFS) and OS were analyzed by Kaplan-Meier method. Cumulative incidence of locoregional progression (CILP) was analyzed using competing risks of distant-only relapse and death with Gray's test. RESULTS: Median follow-up was 7.0 years (range: 0.01-22.4). Five-year CILP, EFS, and OS were 11.9 %, 65.2 %, and 77.5 %, respectively. Of the 31 patients with LRR and imaging review, 15 (48.4 %) had in-field recurrences (>12 Gy), 6 (19.4 %) had marginal failures (≤12 Gy), and 10 (32.3 %) had both in-field and marginal recurrences. No patients receiving total body irradiation (12 Gy) experienced marginal-only failures (p = 0.069). On multivariable analyses, MYCN amplification had higher risk of LRR (HR: 2.42, 95 % CI: 1.06-5.50, p = 0.035) and post-consolidation isotretinoin and anti-GD2 antibody therapy (HR: 0.42, 95 % CI: 0.19-0.94, p = 0.035) had lower risk of LRR. CONCLUSIONS: Despite EBRT, LRR remains a contributor to treatment failure in HR-NBL with approximately half of LRRs including a component of marginal failure. Future prospective studies are needed to explore whether radiation fields and doses should be defined based on molecular features such as MYCN amplification, and/or response to chemotherapy.
Asunto(s)
Recurrencia Local de Neoplasia , Neuroblastoma , Humanos , Neuroblastoma/radioterapia , Neuroblastoma/mortalidad , Estudios Retrospectivos , Masculino , Femenino , Preescolar , Lactante , Niño , Dosificación Radioterapéutica , AdolescenteRESUMEN
Purpose To develop and externally test a scan-to-prediction deep learning pipeline for noninvasive, MRI-based BRAF mutational status classification for pediatric low-grade glioma. Materials and Methods This retrospective study included two pediatric low-grade glioma datasets with linked genomic and diagnostic T2-weighted MRI data of patients: Dana-Farber/Boston Children's Hospital (development dataset, n = 214 [113 (52.8%) male; 104 (48.6%) BRAF wild type, 60 (28.0%) BRAF fusion, and 50 (23.4%) BRAF V600E]) and the Children's Brain Tumor Network (external testing, n = 112 [55 (49.1%) male; 35 (31.2%) BRAF wild type, 60 (53.6%) BRAF fusion, and 17 (15.2%) BRAF V600E]). A deep learning pipeline was developed to classify BRAF mutational status (BRAF wild type vs BRAF fusion vs BRAF V600E) via a two-stage process: (a) three-dimensional tumor segmentation and extraction of axial tumor images and (b) section-wise, deep learning-based classification of mutational status. Knowledge-transfer and self-supervised approaches were investigated to prevent model overfitting, with a primary end point of the area under the receiver operating characteristic curve (AUC). To enhance model interpretability, a novel metric, center of mass distance, was developed to quantify the model attention around the tumor. Results A combination of transfer learning from a pretrained medical imaging-specific network and self-supervised label cross-training (TransferX) coupled with consensus logic yielded the highest classification performance with an AUC of 0.82 (95% CI: 0.72, 0.91), 0.87 (95% CI: 0.61, 0.97), and 0.85 (95% CI: 0.66, 0.95) for BRAF wild type, BRAF fusion, and BRAF V600E, respectively, on internal testing. On external testing, the pipeline yielded an AUC of 0.72 (95% CI: 0.64, 0.86), 0.78 (95% CI: 0.61, 0.89), and 0.72 (95% CI: 0.64, 0.88) for BRAF wild type, BRAF fusion, and BRAF V600E, respectively. Conclusion Transfer learning and self-supervised cross-training improved classification performance and generalizability for noninvasive pediatric low-grade glioma mutational status prediction in a limited data scenario. Keywords: Pediatrics, MRI, CNS, Brain/Brain Stem, Oncology, Feature Detection, Diagnosis, Supervised Learning, Transfer Learning, Convolutional Neural Network (CNN) Supplemental material is available for this article. © RSNA, 2024.