Samuel L. Volchenboum

Advances in Risk Classification and Treatment Strategies for Neuroblastoma

Risk-based treatment approaches for neuroblastoma have been ongoing for decades. However, the criteria used to define risk in various institutional and cooperative groups were disparate, limiting the ability to compare clinical trial results. To mitigate this problem and enhance collaborative research, homogenous pretreatment patient cohorts have been defined by the International Neuroblastoma Risk Group classification system. During the past 30 years, increasingly intensive, multimodality approaches have been developed to treat patients who are classified as high risk, whereas patients with low- or intermediate-risk neuroblastoma have received reduced therapy. This treatment approach has resulted in improved outcome, although survival for high-risk patients remains poor, emphasizing the need for more effective treatments. Increased knowledge regarding the biology and genetic basis of neuroblastoma has led to the discovery of druggable targets and promising, new therapeutic approaches. Collaborative efforts of institutions and international cooperative groups have led to advances in our understanding of neuroblastoma biology, refinements in risk classification, and stratified treatment strategies, resulting in improved outcome. International collaboration will be even more critical when evaluating therapies designed to treat small cohorts of patients with rare actionable mutations.

Genomics of Ovarian Cancer Progression Reveals Diverse Metastatic Trajectories Including Intraepithelial Metastasis to the Fallopian Tube

Accumulating evidence has supported the fallopian tube rather than the ovary as the origin for high-grade serous ovarian cancer (HGSOC). To understand the relationship between putative precursor lesions and metastatic tumors, we performed whole-exome sequencing on specimens from eight HGSOC patient progression series consisting of serous tubal intraepithelial carcinomas (STIC), invasive fallopian tube lesions, invasive ovarian lesions, and omental metastases. Integration of copy number and somatic mutations revealed patient-specific patterns with similar mutational signatures and copy-number variation profiles across all anatomic sites, suggesting that genomic instability is an early event in HGSOC. Phylogenetic analyses supported STIC as precursor lesions in half of our patient cohort, but also identified STIC as metastases in 2 patients. Ex vivo assays revealed that HGSOC spheroids can implant in the fallopian tube epithelium and mimic STIC lesions. That STIC may represent metastases calls into question the assumption that STIC are always indicative of primary fallopian tube cancers. SIGNIFICANCE We find that the putative precursor lesions for HGSOC, STIC, possess most of the genomic aberrations present in advanced cancers. In addition, a proportion of STIC represent intraepithelial metastases to the fallopian tube rather than the origin of HGSOC. Cancer Discov; 6(12); 1342-51. ©2016 AACR.See related commentary by Swisher et al., p. 1309This article is highlighted in the In This Issue feature, p. 1293.

Bioinformatic Analysis and Post-Translational Modification Crosstalk Prediction of Lysine Acetylation

Recent proteomics studies suggest high abundance and a much wider role for lysine acetylation (K-Ac) in cellular functions. Nevertheless, cross influence between K-Ac and other post-translational modifications (PTMs) has not been carefully examined. Here, we used a variety of bioinformatics tools to analyze several available K-Ac datasets. Using gene ontology databases, we demonstrate that K-Ac sites are found in all cellular compartments. KEGG analysis indicates that the K-Ac sites are found on proteins responsible for a diverse and wide array of vital cellular functions. Domain structure prediction shows that K-Ac sites are found throughout a wide variety of protein domains, including those in heat shock proteins and those involved in cell cycle functions and DNA repair. Secondary structure prediction proves that K-Ac sites are preferentially found in ordered structures such as alpha helices and beta sheets. Finally, by mutating K-Ac sites in silico and predicting the effect on nearby phosphorylation sites, we demonstrate that the majority of lysine acetylation sites have the potential to impact protein phosphorylation, methylation, and ubiquitination status. Our work validates earlier smaller-scale studies on the acetylome and demonstrates the importance of PTM crosstalk for regulation of cellular function.

Comparison of Primary Neuroblastoma Tumors and Derivative Early-Passage Cell Lines Using Genome-Wide Single Nucleotide Polymorphism Array Analysis

Stromal contamination is one of the major confounding factors in the analysis of solid tumor samples by single nucleotide polymorphism (SNP) arrays. As we propose to use genome-wide SNP microarray analysis as a diagnostic platform for neuroblastoma, the sensitivity, specificity, and accuracy of these studies must be optimized. To investigate the effects of stromal contamination, we derived early-passage cell lines from nine primary tumors and compared their genomic signature with that of the primary tumors using 100K SNP arrays. The average concordance between tumor and cell line for raw loss of heterozygosity (LOH) calls was 96% (range, 91-99%) and for raw copy number alterations, 71% (range, 43-87%). In general, there were a larger number of LOH events identified in the cell lines compared with the matched tumor samples (mean increase, 3.2% +/- 1.9%). We have developed an algorithm that shows that the presence of stroma contributes to under-reporting of LOH and copy number loss. Notable findings in this sample set were uniparental disomy of chromosome arms 11p, 1q, 14q, and 15q and a novel area of amplification on chromosome band 11p15. Our analysis shows that LOH was identified significantly more often in derived cell lines compared with the original tumor samples. Although these may in part be due to clonal selection during adaptation to tissue culture, our study indicates that stromal contamination may be a major contributing factor in underestimation of LOH and copy number loss events.

Progress in Defining and Treating High-Risk Neuroblastoma: Lessons From the Bench and Bedside

Neuroblastoma (NB) is remarkable for its biologic heterogeneity and broad range of clinical behavior. Although survival rates are greater than 90% for patients with biologically favorable disease, outcomes for children with a high-risk clinical phenotype remains poor, with long-term survival still less than 40%. This issue of Journal of Clinical Oncology includes four studies that are focused on defining and treating high-risk NB. Modern molecular techniques have provided the tools to better define high-risk NB, and a modest improvement in outcome has been seen with dose intensification. However, each of the studies emphasizes the need for more effective therapeutic approaches. Efforts to identify variables that accurately predict outcome for patients with NB have been ongoing for more than 35 years. In the 1970s, the dramatic influence of stage and age on prognosis were reported. During this era, aggressive multimodality therapy was recommended for patients with skeletal metastases, whereas infants were treated more gently. In the 1980s, MYCN amplification emerged as a powerful marker of adverse prognosis in NB. The clinical significance of tumor histology, ploidy and other genetic aberrations, including 1p loss, 11q loss, and 17q gain were also demonstrated. Combinations of clinical and biologic prognostic variables are now routinely used for risk-group assignment and treatment stratification, although the criteria used to define risk vary greatly throughout the world. Because definitions of risk are not uniform, direct comparison of risk-based clinical trials conducted in different regions of the world has not been possible. To address this problem, an international task force recently established the International Neuroblastoma Risk Group (INRG) classification system, on the basis of the analysis of 13 prognostic variables in an 8,800 patient cohort. By defining homogenous pretreatment patient cohorts, the INRG classification system will greatly facilitate the development of international collaborative studies. To date, the strategy to improve outcome in high-risk patients has largely been focused on delivering increasingly intensive multimodality therapy. Matthay et al report the long-term results of a randomized Children’s Oncology Group (COG) trial comparing a more intensive arm of consolidation therapy (myeloablative therapy plus autologous bone marrow transplant [ABMT]) versus a less intensive arm (continued conventional-dose chemotherapy). All patients without evidence of disease after consolidation were then eligible for a second randomization to 13-cis-retinoic acid (cis-RA) versus no additional therapy. The initial results of the study, reported almost 10 years ago, demonstrated significantly better 3-year event-free survival (EFS) for the group randomly assigned to myeloablative therapy and ABMT and for patients randomly assigned to cis-RA. With additional followup, patients randomly assigned to the more intensive arm (myeloablative therapy and ABMT) continue to have significantly higher 5-year EFS. Similar results have been reported by European groups, although survival still remains poor. In the COG study, 30% SE 4% of the patients randomly assigned to the superior arm of therapy were event-free at 5 years. Although overall survival (OS) in the COG study was found to be significantly higher for each randomization at 5 years using a test of the log( log(.)) transformation, a significant advantage for OS was not observed in the European studies. Thus, the majority of the patients are not cured with intensive treatment strategies that include myeloablative therapy and stem-cell rescue. Furthermore, long-term follow-up studies have demonstrated that many survivors have serious late effects of therapy, including delays in growth and development, hearing loss, renal and cardiac dysfunction, learning problems, and treatment-related leukemia and second cancers. The study by Canete et al focuses on the outcome of infants younger than 12 months of age with high-risk disease treated on a International Society of Paediatric Oncology European Neuroblastoma (SIOPEN) clinical trial. In this cohort of 35 infants, 2-year EFS was 29% (SE 0.07) and OS was 30% (SE 0.08) after treatment with intensive multimodality therapy, including high-dose busulfan and melphalan with peripheral stem-cell support and cis-RA. Many of the tumors were resistant to chemotherapy, with 30% progressing or failing to respond to induction therapy. In the COG, infants with high-risk NB are treated on the same clinical trial as older patients with high-risk disease. Because of small numbers, international collaboration will likely be needed to determine if the clinical behavior of high-risk tumors diagnosed in infancy differs from high-risk tumors in older children. To ensure that patients with a lowor intermediate-risk clinical phenotype are spared toxic, dose-intensive, high-risk treatment regimens, accurate risk-group classification is critical. Historically, an age cutoff of 12 months has been used for risk stratification. However, recent analysis of a large series of 3,666 patients has demonstrated statistical evidence for increasing the age cutoff to 15 to 19 months. On the basis of these results, COG has modified eligibility criteria for its intermediate-risk clinical study to include toddlers, age 12 to 18 months, with favorable biology tumors. Similarly, an age cutoff of 18 months has been included in the INRG classification system. This new age cutoff will shift approximately 10% of patients previously classified as high risk to a lower-risk group. Newly designed JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 27 NUMBER 7 MARCH 1 2009

Scalable and accurate deep learning with electronic health records

Predictive modeling with electronic health record (EHR) data is anticipated to drive personalized medicine and improve healthcare quality. Constructing predictive statistical models typically requires extraction of curated predictor variables from normalized EHR data, a labor-intensive process that discards the vast majority of information in each patient’s record. We propose a representation of patients’ entire raw EHR records based on the Fast Healthcare Interoperability Resources (FHIR) format. We demonstrate that deep learning methods using this representation are capable of accurately predicting multiple medical events from multiple centers without site-specific data harmonization. We validated our approach using de-identified EHR data from two US academic medical centers with 216,221 adult patients hospitalized for at least 24 h. In the sequential format we propose, this volume of EHR data unrolled into a total of 46,864,534,945 data points, including clinical notes. Deep learning models achieved high accuracy for tasks such as predicting: in-hospital mortality (area under the receiver operator curve [AUROC] across sites 0.93–0.94), 30-day unplanned readmission (AUROC 0.75–0.76), prolonged length of stay (AUROC 0.85–0.86), and all of a patient’s final discharge diagnoses (frequency-weighted AUROC 0.90). These models outperformed traditional, clinically-used predictive models in all cases. We believe that this approach can be used to create accurate and scalable predictions for a variety of clinical scenarios. In a case study of a particular prediction, we demonstrate that neural networks can be used to identify relevant information from the patient’s chart.Artificial intelligence: Algorithm predicts clinical outcomes for hospital inpatientsArtificial intelligence outperforms traditional statistical models at predicting a range of clinical outcomes from a patient’s entire raw electronic health record (EHR). A team led by Alvin Rajkomar and Eyal Oren from Google in Mountain View, California, USA, developed a data processing pipeline for transforming EHR files into a standardized format. They then applied deep learning models to data from 216,221 adult patients hospitalized for at least 24 h each at two academic medical centers, and showed that their algorithm could accurately predict risk of mortality, hospital readmission, prolonged hospital stay and discharge diagnosis. In all cases, the method proved more accurate than previously published models. The authors provide a case study to serve as a proof-of-concept of how such an algorithm could be used in routine clinical practice in the future.

Second malignancies in patients with neuroblastoma: the effects of risk-based therapy.

To investigate the incidence of second malignant neoplasms (SMN) for patients with neuroblastoma, we analyzed patients from the SEER database according to three treatment eras (Era 1: 1973–1989, Era 2: 1990–1996, and Era 3: 1997–2006) corresponding to the introduction of multi‐agent chemotherapy, risk‐based treatment, and stem cell transplant.

CAPriCORN: Chicago Area Patient-Centered Outcomes Research Network

The Chicago Area Patient-Centered Outcomes Research Network (CAPriCORN) represents an unprecedented collaboration across diverse healthcare institutions including private, county, and state hospitals and health systems, a consortium of Federally Qualified Health Centers, and two Department of Veterans Affairs hospitals. CAPriCORN builds on the strengths of our institutions to develop a cross-cutting infrastructure for sustainable and patient-centered comparative effectiveness research in Chicago. Unique aspects include collaboration with the University HealthSystem Consortium to aggregate data across sites, a centralized communication center to integrate patient recruitment with the data infrastructure, and a centralized institutional review board to ensure a strong and efficient human subject protection program. With coordination by the Chicago Community Trust and the Illinois Medical District Commission, CAPriCORN will model how healthcare institutions can overcome barriers of data integration, marketplace competition, and care fragmentation to develop, test, and implement strategies to improve care for diverse populations and reduce health disparities.

Pharmacogenomics-Based Point-of-Care Clinical Decision Support Significantly Alters Drug Prescribing

Changes in behavior are necessary to apply genomic discoveries to practice. We prospectively studied medication changes made by providers representing eight different medicine specialty clinics whose patients had submitted to preemptive pharmacogenomic genotyping. An institutional clinical decision support (CDS) system provided pharmacogenomic results using traffic light alerts: green = genomically favorable, yellow = genomic caution, red = high risk. The influence of pharmacogenomic alerts on prescribing behaviors was the primary endpoint. In all, 2,279 outpatient encounters were analyzed. Independent of other potential prescribing mediators, medications with high pharmacogenomic risk were changed significantly more often than prescription drugs lacking pharmacogenomic information (odds ratio (OR) = 26.2 (9.0–75.3), P < 0.0001). Medications with cautionary pharmacogenomic information were also changed more frequently (OR = 2.4 (1.7–3.5), P < 0.0001). No pharmacogenomically high‐risk medications were prescribed during the entire study when physicians consulted the CDS tool. Pharmacogenomic information improved prescribing in patterns aimed at reducing patient risk, demonstrating that enhanced prescription decision‐making is achievable through clinical integration of genomic medicine.

Gene expression profiling of Ewing sarcoma tumours reveals the prognostic importance of tumour–stromal interactions: a report from the Children's Oncology Group

Relapse of Ewing sarcoma (ES) can occur months or years after initial remission, and salvage therapy for relapsed disease is usually ineffective. Thus, there is great need to develop biomarkers that can predict which patients are at risk for relapse so that therapy and post‐therapy evaluation can be adjusted accordingly. For this study, we performed whole genome expression profiling on two independent cohorts of clinically annotated ES tumours in an effort to identify and validate prognostic gene signatures. ES specimens were obtained from the Childrens Oncology Group and whole genome expression profiling performed using Affymetrix Human Exon 1.0 ST arrays. Lists of differentially expressed genes between survivors and non‐survivors were used to identify prognostic gene signatures. An independent cohort of tumours from the Euro‐Ewing cooperative group was similarly analysed as a validation cohort. Unsupervised clustering of gene expression data failed to segregate tumours based on outcome. Supervised analysis of survivors versus non‐survivors revealed a small number of differentially expressed genes and several statistically significant gene signatures. Gene‐specific enrichment analysis demonstrated that integrin and chemokine genes were associated with survival in tumours where stromal contamination was present. Tumours that did not harbour stromal contamination showed no association of any genes or pathways with clinical outcome. Our results reflect the challenges of performing RNA‐based assays on archived bone tumour specimens. In addition, they reveal a key role for tumour stroma in determining ES prognosis. Future biological and clinical investigations should focus on elucidating the contribution of tumour:micro‐environment interactions on ES progression and response to therapy.