Methylation classifiers: brain tumors, sarcomas and what's next

Deep-Learning Model for Tumor-Type Prediction Using Targeted Clinical Genomic Sequencing Data

Tumor type guides clinical treatment decisions in cancer, but histology-based diagnosis remains challenging. Genomic alterations are highly diagnostic of tumor type, and tumor-type classifiers trained on genomic features have been explored, but the most accurate methods are not clinically feasible, relying on features derived from whole-genome sequencing (WGS), or predicting across limited cancer types. We use genomic features from a data set of 39,787 solid tumors sequenced using a clinically targeted cancer gene panel to develop Genome-Derived-Diagnosis Ensemble (GDD-ENS): a hyperparameter ensemble for classifying tumor type using deep neural networks. GDD-ENS achieves 93% accuracy for high-confidence predictions across 38 cancer types, rivaling the performance of WGS-based methods. GDD-ENS can also guide diagnoses of rare type and cancers of unknown primary and incorporate patient-specific clinical information for improved predictions. Overall, integrating GDD-ENS into prospective clinical sequencing workflows could provide clinically relevant tumor-type predictions to guide treatment decisions in real time.

SIGNIFICANCE

We describe a highly accurate tumor-type prediction model, designed specifically for clinical implementation. Our model relies only on widely used cancer gene panel sequencing data, predicts across 38 distinct cancer types, and supports integration of patient-specific nongenomic information for enhanced decision support in challenging diagnostic situations. See related commentary by Garg, p. 906. This article is featured in Selected Articles from This Issue, p. 897.

Molecular pathology as basis for timely cancer diagnosis and therapy

Precision and personalized therapeutics have witnessed significant advancements in technology, revolutionizing the capabilities of laboratories to generate vast amounts of genetic data. Coupled with computational resources for analysis and interpretation, and integrated with various other types of data, including genomic data, electronic medical health (EMH) data, and clinical knowledge, these advancements support optimized health decisions. Among these technologies, next-generation sequencing (NGS) stands out as a transformative tool in the field of cancer treatment, playing a crucial role in precision oncology. NGS-based workflows are employed across a range of applications, including gene panels, exome sequencing, and whole-genome sequencing, supporting comprehensive analysis of the entire cancer genome, including mutations, copy number variations, gene expression profiles, and epigenetic modifications. By utilizing the power of NGS, these workflows contribute to enhancing our understanding of disease mechanisms, diagnosis confirmation, identifying therapeutic targets, and guiding personalized treatment decisions. This manuscript explores the diverse applications of NGS in cancer treatment, highlighting its significance in guiding diagnosis and treatment decisions, identifying therapeutic targets, monitoring disease progression, and improving patient outcomes.

Meningeal Solitary Fibrous Tumor: A Single-Center Retrospective Cohort Study

Background

Meningeal solitary fibrous tumors (SFTs) are rare, malignant, mesenchymal tumors of the central nervous system. While surgical gross total resection is widely accepted as a positive prognostic factor for local control (LC), the role of postoperative radiotherapy (PORT) remains controversial. We sought to report our institutional experience with a particular focus on outcomes after PORT.

Materials and Methods

In this single-center, retrospective cohort study, 20 patients with the primary diagnosis of histopathologically confirmed meningeal SFT were analyzed. Data on patient characteristics, imaging, treatment modalities, histopathology, and oncological outcomes were collected. LC and overall survival (OS) were assessed using the Kaplan-Meier estimator.

Results

The median follow-up time was 95.8 months. After surgery only, 9 out of 11 patients (81.8%) developed a local recurrence while, after surgery and PORT, 3 out of 9 patients (33.33%) showed local failure. The 5- and 10-year LC rates were 50.5% and 40.4% in the surgery-only group and 80% at both time points in the surgery with the PORT group. In the surgery-only group, 4 out of 11 patients (36.4%) died, and 4 out of 9 patients (44.4%) died in the surgery and PORT group. OS rates after 5 and 10 years were 88.9% and 66.7% in the surgery-only group and 88.9% and 76.2% in the surgery with PORT group.

Conclusions

Our findings suggest that PORT may improve LC in patients with meningeal SFT. The low incidence of meningeal SFT impedes prospective studies and requires further international collaborative efforts to exploit retrospective datasets and molecular analysis to improve patient outcomes.

Assessment of The Utility of The Sarcoma DNA Methylation Classifier In Surgical Pathology

Diagnostic classification of soft tissue tumors is based on histology, immunohistochemistry, genetic findings, and radiologic and clinical correlations. Recently, a sarcoma DNA methylation classifier was developed, covering 62 soft tissue and bone tumor entities. The classifier is based on large-scale analysis of methylation sites across the genome. It includes DNA copy number analysis and determines O 6 methylguanine DNA methyl-transferase methylation status. In this study, we evaluated 619 well-studied soft tissue and bone tumors with the sarcoma classifier. Problem cases and typical examples of different entities were included. The classifier had high sensitivity and specificity for fusion sarcomas: Ewing, synovial, CIC -rearranged, and BCOR -rearranged. It also performed well for leiomyosarcoma, malignant peripheral nerve sheath tumors (MPNST), and malignant vascular tumors. There was low sensitivity for diagnoses of desmoid fibromatosis, neurofibroma, and schwannoma. Low specificity of matches was observed for angiomatoid fibrous histiocytoma, inflammatory myofibroblastic tumor, Langerhans histiocytosis, schwannoma, undifferentiated sarcoma, and well-differentiated/dedifferentiated liposarcoma. Diagnosis of lipomatous tumors was greatly assisted by the detection of MDM2 amplification and RB1 loss in the copy plot. The classifier helped to establish diagnoses for KIT-negative gastrointestinal stromal tumors, MPNSTs with unusual immunophenotypes, and undifferentiated melanomas. O 6 methylguanine DNA methyl-transferase methylation was infrequent and most common in melanomas (35%), MPNSTs (11%), and undifferentiated sarcomas (11%). The Sarcoma Methylation Classifier will likely evolve with the addition of new entities and refinement of the present methylation classes. The classifier may also help to define new entities and give new insight into the interrelationships of sarcomas.

The Pathologic Diagnosis of Pediatric Soft Tissue Tumors in the Era of Molecular Medicine: The Sarcoma Pediatric Pathology Research Interest Group Perspective

CONTEXT.—

Pediatric soft tissue tumors are one of the areas of pediatric pathology that frequently generate consult requests. Evolving classification systems, ancillary testing methods, new treatment options, research enrollment opportunities, and tissue archival processes create additional complexity in handling these unique specimens. Pathologists are at the heart of this critical decision-making, balancing responsibilities to consider expediency, accessibility, and cost-effectiveness of ancillary testing during pathologic examination and reporting.

OBJECTIVE.—

To provide a practical approach to handling pediatric soft tissue tumor specimens, including volume considerations, immunohistochemical staining panel recommendations, genetic and molecular testing approaches, and other processes that impact the quality and efficiency of tumor tissue triage.

DATA SOURCES.—

The World Health Organization Classification of Soft Tissue and Bone Tumors, 5th edition, other recent literature investigating tissue handling, and the collective clinical experience of the group are used in this manuscript.

CONCLUSIONS.—

Pediatric soft tissue tumors can be difficult to diagnose, and evaluation can be improved by adopting a thoughtful, algorithmic approach to maximize available tissue and minimize time to diagnosis.

[Supratentorial neuroepithelial tumor with PLAGL1 gene fusion - a new type of morphologically variable pediatric brain neoplasm defined by a distinct DNA methylation class. A case report and literature review]

BACKGROUND

Methylation analysis has become a powerful diagnostic tool in modern neurooncology. This technique is valuable to diagnose new brain tumor types.

OBJECTIVE

To describe the MRI and histological pattern of neuroepithelial tumor with PLAGL1 gene fusion.

MATERIAL AND METHODS

We present a 6-year-old patient with small right frontal intraaxial tumor causing drug resistant epilepsy. Despite indolent preoperative clinical course and MRI features suggesting glioneuronal tumor, histological evaluation revealed characteristics of high-grade glioma, ependymoma and neuroblastoma.

RESULTS

Methylation analysis of tumor DNA confirmed a new type of a recently discovered neoplasm - neuroepithelial tumor with PLAGL1 fusion (NET PLAGL1). PCR confirmed fusion of PLAGL1 and EWSR1 genes. No seizures were observed throughout the follow-up period. There was no tumor relapse a year after surgery.

CONCLUSION

Methylation analysis in neurooncology is essential for unclear tumor morphology or divergence between histological and clinical data. In our case, this technique confirmed benign nature of tumor, and we preferred follow-up without unnecessary adjuvant treatment.

Osteosarcomas With Few Chromosomal Alterations or Adult Onset Are Genetically Heterogeneous

Osteosarcoma is the most common primary bone malignancy, often detected in children and adolescents and commonly associated with TP53 alterations along with a high number of chromosomal rearrangements. However, osteosarcoma can affect patients of any age, and some tumors display less genetic complexity. Besides TP53 variants, data on key driving mutations are lacking for many osteosarcomas, particularly those affecting adults. To detect osteosarcoma-specific alterations, we screened transcriptomic and genomic sequencing and copy number data from 150 bone tumors originally diagnosed as osteosarcomas. To increase the precision in gene fusion detection, we developed a bioinformatic tool denoted as NAFuse, which extracts gene fusions that are verified at both the genomic and transcriptomic levels. Apart from the already reported genetic subgroups of osteosarcoma with TP53 structural variants, or MDM2 and/or CDK4 amplification, we did not identify any recurrent genetic driver that signifies the remaining cases. Among the plethora of mutations identified, we found genetic alterations characteristic of, or similar to, those of other bone and soft tissue tumors in 8 cases. These mutations were found in tumors with relatively few other genetic alterations or in adults. Due to the lack of clinical context and available tissue, we can question the diagnosis only on a genetic basis. However, our findings support the notion that osteosarcomas with few chromosomal alterations or adult onset seem genetically distinct from conventional osteosarcomas of children and adolescents.

Current status of DNA methylation profiling in neuro-oncology as a diagnostic support tool: A review

Over the last 2 decades, high throughput genome-wide molecular profiling has revealed characteristic genetic and epigenetic alterations associated with different types of central nervous system (CNS) tumors. DNA methylation profiling has emerged as an important molecular platform for CNS tumor classification with improved diagnostic accuracy and patient risk stratification in comparison to the standard of care histopathological analysis and any single molecular tests. The emergence of DNA methylation arrays have also played a crucial role in refining existing types and the discovery of new tumor types or subtypes. The adoption of methylation data into neuro-oncology has been greatly aided by the development of a freely accessible machine learning-based classifier. In this review, we discuss methylation workflow, address the utility of DNA methylation profiling in CNS tumors in a routine diagnostic setting, and provide an overview of the methylation-based tumor types and new types or subtypes identified with this platform.

Sellar Mass in 2 Patients With Acute-Onset Headache and Visual Symptoms: Not Your Usual Pituitary Adenoma

Background/Objective

Clinical diagnosis of rare aggressive sellar malignancies requires a high index of suspicion. The objective was to report 2 patients with primary sellar atypical teratoid (AT)/rhabdoid tumor (RT) who presented with acute-onset headache and visual symptoms.

Case Report

Patient 1 was a 45-year-old woman who presented with 3 weeks of headache and 1 week of eye pain and diplopia. Magnetic resonance imaging (MRI) identified a 2.2-cm sellar mass. Pituitary hormone testing showed elevated prolactin and suppressed luteinizing hormone, follicle-stimulating hormone, and estradiol levels. Patient 2 was a 32-year-old woman who presented with 1 month of headache and 1 week of diplopia. MRI showed a 2.1-cm sellar mass. Hormonal test results were reportedly unremarkable. Both patients did not have a significant medical history. They each underwent transsphenoidal resection. Surgical histology and molecular studies were consistent with primary sellar AT/RT. After surgery, patient 1 developed bilateral blindness and was lost to follow-up. Patient 2 developed hypopituitarism; her visual symptoms improved temporarily but recurred 2 weeks later. Pituitary MRI showed sellar recurrence. She underwent further debulking, but the tumor recurred promptly again. Despite radiation therapy, she died 4 months after the original presentation.

Discussion

AT/RT appears to be the most aggressive sellar malignancy.

Conclusion

Based on the 2 cases presented and the literature, I conclude that rapidly progressive headache with subsequent visual impairment in women with large sellar masses is almost pathognomonic of sellar AT/RT.

[Peripheral neuroblastic tumors in childhood]

Peripheral neuroblastic tumors represent the fourth-largest group of malignant tumors in childhood. The majority of these tumors are neuroblastomas, which can be classified into undifferentiated, poorly differentiated, and differentiating subtypes. In addition, peripheral neuroblastic tumors include ganglioneuroblastoma, a composite tumor composed of Schwannian cell stroma and neuroblasts as well as benign ganglioneuroma. In this overview, histopathological diagnostic criteria and grading systems, as well as common molecular alterations that are of prognostic and therapeutic significance, are discussed.

NCT/DKFZ MASTER handbook of interpreting whole-genome, transcriptome, and methylome data for precision oncology

Analysis of selected cancer genes has become an important tool in precision oncology but cannot fully capture the molecular features and, most importantly, vulnerabilities of individual tumors. Observational and interventional studies have shown that decision-making based on comprehensive molecular characterization adds significant clinical value. However, the complexity and heterogeneity of the resulting data are major challenges for disciplines involved in interpretation and recommendations for individualized care, and limited information exists on how to approach multilayered tumor profiles in clinical routine. We report our experience with the practical use of data from whole-genome or exome and RNA sequencing and DNA methylation profiling within the MASTER (Molecularly Aided Stratification for Tumor Eradication Research) program of the National Center for Tumor Diseases (NCT) Heidelberg and Dresden and the German Cancer Research Center (DKFZ). We cover all relevant steps of an end-to-end precision oncology workflow, from sample collection, molecular analysis, and variant prioritization to assigning treatment recommendations and discussion in the molecular tumor board. To provide insight into our approach to multidimensional tumor profiles and guidance on interpreting their biological impact and diagnostic and therapeutic implications, we present case studies from the NCT/DKFZ molecular tumor board that illustrate our daily practice. This manual is intended to be useful for physicians, biologists, and bioinformaticians involved in the clinical interpretation of genome-wide molecular information.

DNA methylation-based diagnosis confirmation in a pediatric patient with low-grade glioma: a case report

Central nervous system (CNS) tumors in children comprise a highly heterogenous and complex group of diseases. Historically, diagnosis and confirmation of these tumors were routinely based on histological examination. However, recently obtained data demonstrate that such a diagnostic approach is not completely accurate and could lead to misdiagnosis. Also, in recent times, the quantity and quality of molecular diagnostic methods have greatly improved, which influences the current classification methods and treatment approach for pediatric CNS tumors. Nowadays, molecular methods, such as DNA methylation profiling, are an integral part of diagnosing brain and spinal tumors in children. In this paper, we present the case of an infant with a posterior fossa tumor who demonstrated a non-specific morphology and whose diagnosis was verified only after DNA methylation.

EpiGe: A machine-learning strategy for rapid classification of medulloblastoma using PCR-based methyl-genotyping

Molecular classification of medulloblastoma is critical for the treatment of this brain tumor. Array-based DNA methylation profiling has emerged as a powerful approach for brain tumor classification. However, this technology is currently not widely available. We present a machine-learning decision support system (DSS) that enables the classification of the principal molecular groups-WNT, SHH, and non-WNT/non-SHH-directly from quantitative PCR (qPCR) data. We propose a framework where the developed DSS appears as a user-friendly web-application-EpiGe-App-that enables automated interpretation of qPCR methylation data and subsequent molecular group prediction. The basis of our classification strategy is a previously validated six-cytosine signature with subgroup-specific methylation profiles. This reduced set of markers enabled us to develop a methyl-genotyping assay capable of determining the methylation status of cytosines using qPCR instruments. This study provides a comprehensive approach for rapid classification of clinically relevant medulloblastoma groups, using readily accessible equipment and an easy-to-use web-application.t.

Deep Learning Model for Tumor Type Prediction using Targeted Clinical Genomic Sequencing Data

Tumor type guides clinical treatment decisions in cancer, but histology-based diagnosis remains challenging. Genomic alterations are highly diagnostic of tumor type, and tumor type classifiers trained on genomic features have been explored, but the most accurate methods are not clinically feasible, relying on features derived from whole genome sequencing (WGS), or predicting across limited cancer types. We use genomic features from a dataset of 39,787 solid tumors sequenced using a clinical targeted cancer gene panel to develop Genome-Derived-Diagnosis Ensemble (GDD-ENS): a hyperparameter ensemble for classifying tumor type using deep neural networks. GDD-ENS achieves 93% accuracy for high-confidence predictions across 38 cancer types, rivalling performance of WGS-based methods. GDD-ENS can also guide diagnoses on rare type and cancers of unknown primary, and incorporate patient-specific clinical information for improved predictions. Overall, integrating GDD-ENS into prospective clinical sequencing workflows has enabled clinically-relevant tumor type predictions to guide treatment decisions in real time.

Rapid Classification of Sarcomas Using Methylation Fingerprint: A Pilot Study

Sarcoma classification is challenging and can lead to treatment delays. Previous studies used DNA aberrations and machine-learning classifiers based on methylation profiles for diagnosis. We aimed to classify sarcomas by analyzing methylation signatures obtained from low-coverage whole-genome sequencing, which also identifies copy-number alterations. DNA was extracted from 23 suspected sarcoma samples and sequenced on an Oxford Nanopore sequencer. The methylation-based classifier, applied in the nanoDx pipeline, was customized using a reference set based on processed Illumina-based methylation data. Classification analysis utilized the Random Forest algorithm and t-distributed stochastic neighbor embedding, while copy-number alterations were detected using a designated R package. Out of the 23 samples encompassing a restricted range of sarcoma types, 20 were successfully sequenced, but two did not contain tumor tissue, according to the pathologist. Among the 18 tumor samples, 14 were classified as reported in the pathology results. Four classifications were discordant with the pathological report, with one compatible and three showing discrepancies. Improving tissue handling, DNA extraction methods, and detecting point mutations and translocations could enhance accuracy. We envision that rapid, accurate, point-of-care sarcoma classification using nanopore sequencing could be achieved through additional validation in a diverse tumor cohort and the integration of methylation-based classification and other DNA aberrations.

Sarcoma classification by DNA methylation profiling in clinical everyday life: the Charité experience

BACKGROUND

Sarcomas are a heterogeneous group of rare malignant tumors with more than 100 subtypes. Accurate diagnosis remains challenging due to a lack of characteristic molecular or histomorphological hallmarks. A DNA methylation-based tumor profiling classifier for sarcomas (known as sarcoma classifier) from the German Cancer Research Center (Deutsches Krebsforschungszentrum) is now employed in selected cases to guide tumor classification and treatment decisions at our institution. Data on the usage of the classifier in daily clinical routine are lacking.

METHODS

In this single-center experience, we describe the clinical course of five sarcoma cases undergoing thorough pathological and reference pathological examination as well as DNA methylation-based profiling and their impact on subsequent treatment decisions. We collected data on the clinical course, DNA methylation analysis, histopathology, radiological imaging, and next-generation sequencing.

RESULTS

Five clinical cases involving DNA methylation-based profiling in 2021 at our institution were included. All patients' DNA methylation profiles were successfully matched to a methylation profile cluster of the sarcoma classifier's dataset. In three patients, the classifier reassured diagnosis or aided in finding the correct diagnosis in light of contradictory data and differential diagnoses. In two patients with intracranial tumors, the classifier changed the diagnosis to a novel diagnostic tumor group.

CONCLUSIONS

The sarcoma classifier is a valuable diagnostic tool that should be used after comprehensive clinical and histopathological evaluation. It may help to reassure the histopathological diagnosis or indicate the need for thorough reassessment in cases where it contradicts previous findings. However, certain limitations (non-classifiable cases, misclassifications, unclear degree of sample purity for analysis and others) currently preclude wide clinical application. The current sarcoma classifier is therefore not yet ready for a broad clinical routine. With further refinements, this promising tool may be implemented in daily clinical practice in selected cases.