Furtado, Larissa V.

Abstract Diffuse leptomeningeal glioneuronal tumors (DLGNT) are rare central nervous system tumors. Our understanding of the clinical course, prognostic factors, and optimal treatment strategies remains limited. Histologic, molecular, and clinical data were collected and analyzed for 25 patients with DLGNT reviewed at St. Jude Children’s Research Hospital. The median age at diagnosis was 6 years (11 months-20 years). Nineteen (72%) patients were male. The disease was localized at diagnosis in 12 (48.0%) patients, with most tumors localized in the spinal cord. BRAF fusion was present in 22 (95.6%) of 23 patients who had material for testing. Of 18 patients who had tissue for DNA methylation profiling, 3 (16.7%) matched to DLGNT MC-1, 3 (16.7%) to DLGNT MC-2, and 12 (66.7%) did not match to a DLGNT subclass. Median follow-up was 63 months. Twenty-four (96.0%) patients received adjuvant therapy after initial diagnosis or at progression. Thirty-five cytotoxic chemotherapy and 24 single-agent targeted therapy regimens were used in 22 and 13 patients, respectively. With targeted therapy, 3 (12%) patients had partial responses. Eleven (44.0%) received radiotherapy. Median progression-free survival (PFS) after a patient’s first chemotherapy, targeted therapy, or radiotherapy regimen was 44 months (1-77), 11 months (4-39), and 16.5 months (9-23), respectively. The 5-year PFS was 18.7%±9.2, while the 5-year overall survival (OS) was 82.4%±9.2. The 5-year OS was lower in patients older than 9 years at diagnosis (60.0%±21.9 vs 91.7%±8.0, p = 0.02). Ki67 >4% was associated with worse PFS (0%±0 vs 24.6%±11.9 at 5 years, p = 0.03). Initial disease dissemination (localized vs. disseminated) and 1q gain status do not influence PFS or OS. PFS for DLGNT remains poor across all treatment modalities, including targeted therapies, despite the high frequency of BRAF fusions. Due to its rarity, establishing optimal therapeutic strategies for DLGNT requires additional treatment and outcome data and prospective multi-institutional trials.

Context.— Genomic reports are primarily organized in a narrative and unstructured format with variations in content and format. Regulatory requirements and professional guidelines for genetic test reporting exist but provide little guidance for effective communication of information. Objective.— To assess clinical genomic reporting practices across 5 disciplines within molecular diagnostics, including germline, somatic solid tumors, somatic hematologic malignancies, pharmacogenomics, and prenatal cell-free DNA screening. Design.— Reporting practices were assessed by using a structured review of clinical genomic reports from multiple laboratories in 5 molecular disciplines spanning different practice settings. Report content was reviewed by the presence/absence of from 27 to 44 elements, including 23 elements required by the College of American Pathologists and/or the Clinical Laboratory Improvement Amendments of 1988 (CLIA). If present, the element’s location on the report was recorded. Results.— A total of 69 genomics reports from 31 laboratories were reviewed. Overall, the reports were compliant with regulatory requirements but showed variability in both format and content. Six of 7 required reporting elements (per CLIA, 42 CFR [Code of Federal Regulations] 493.1291) were included in 90% of the reports. However, these elements were often located in different report sections. Only patient demographics were always found in a specific report section (header). Conclusions.— These results show that reports are overall compliant with regulatory requirements, despite some reporting elements being less consistently reported. The lack of consistent presentation of the data elements presents an opportunity to improve the communication of molecular testing results to clinicians and patients.

Context.— Machine learning applications in the pathology clinical domain are emerging rapidly. As decision support systems continue to mature, laboratories will increasingly need guidance to evaluate their performance in clinical practice. Currently there are no formal guidelines to assist pathology laboratories in verification and/or validation of such systems. These recommendations are being proposed for the evaluation of machine learning systems in the clinical practice of pathology. Objective.— To propose recommendations for performance evaluation of in vitro diagnostic tests on patient samples that incorporate machine learning as part of the preanalytical, analytical, or postanalytical phases of the laboratory workflow. Topics described include considerations for machine learning model evaluation including risk assessment, predeployment requirements, data sourcing and curation, verification and validation, change control management, human-computer interaction, practitioner training, and competency evaluation. Data Sources.— An expert panel performed a review of the literature, Clinical and Laboratory Standards Institute guidance, and laboratory and government regulatory frameworks. Conclusions.— Review of the literature and existing documents enabled the development of proposed recommendations. This white paper pertains to performance evaluation of machine learning systems intended to be implemented for clinical patient testing. Further studies with real-world clinical data are encouraged to support these proposed recommendations. Performance evaluation of machine learning models is critical to verification and/or validation of in vitro diagnostic tests using machine learning intended for clinical practice.