Predicting Relapse in Patients With Medulloblastoma by Integrating Evidence From Clinical and Genomic Feature

Despite significant progress in the molecular understanding of medulloblastoma, stratification of risk in patients remains a challenge. Focus has shifted from clinical parameters to molecular markers, such as expression of specific genes and selected genomic abnormalities, to improve accuracy of treatment outcome prediction. Here, we show how integration of high-level clinical and genomic features or risk factors, including disease subtype, can yield more comprehensive, accurate, and biologically interpretable prediction models for relapse versus no-relapse classification. We also introduce a novel Bayesian nomogram indicating the amount of evidence that each feature contributes on a patient-by-patient basis. The training set consisted of 96 samples for which relapse status at 30 months post-treatment was known. Matched normal blood samples were collected through the COG Tumor Bank (protocol ACNS02B3) and from Children's Hospital Boston under institutional review board approval. The test set included 78 samples: 47 samples from our original study6 not used for training, 16 samples from Kool et al,33 and 15 samples from the COG Tumor Bank. All training and test samples correspond to patients at least 3 years old treated with conventional chemotherapy, surgical resection, and craniospinal irradiation. In the test set, 15 samples (Children's Oncology Group Tumor Bank) were processed as described above. Of those 15, a total of six have DNA copy number data. Another set of 47 samples was taken from the data set from Pomeroy et al. (Affymetrix Hu6800; Affymetrix). None of those samples had DNA copy number data. The remaining 16 samples, all with DNA copy number data, came from Kool et al. (Affymetrix U133; Affymetrix). A Bayesian cumulative log-odds model of outcome was developed from a training cohort of 96 children treated for medulloblastoma, starting with the evidence provided by clinical features of metastasis and histology (model A) and incrementally adding the evidence from gene-expression-derived features representing disease subtype-independent (model B) and disease subtype-dependent (model C) pathways, and finally high-level copy-number genomic abnormalities (model D). The models were validated on an independent test cohort (n = 78). Information on Kool et al. samples (GSM) used: In total 62 medulloblastoma samples were selected for Affymetrix gene expression profiling. The set included 60 primary tumors and 2 local relapses. All samples were snap frozen in the institutional pathology departments immediately upon arrival. All samples were reviewed by experienced neuropathologists and examined for tumor content. Samples were excluded from analysis when less than 70% of the sample contained tumor cells. RNA and DNA was isolated using Trizol (Invitrogen, Carlsbad, CA). RNA purification was performed using the RNeasy mini kit (Qiagen, Germantown, USA). DNA and RNA quantity and quality was determined by spectrophotometry (Nanodrop, Wilmington, USA) and microfluidics-based electrophoresis (Agilent 2100 Bioanalyzer, Agilent, Palo Alto, USA). Four µg total RNA was used for hybridization of Affymetrix HG-U133 plus 2.0 Genechips according to the manufacturer's instructions (Affymetrix Inc. Santa Barbara, USA). Quality of the arrays was ensured by inspection of the beta-actin and GAPDH 5′-3′ ratio's as well as the percentage of present calls generated by the MAS5.0 algorithm (Affymetrix Inc. Santa Barbara, USA). For clustering and gene expression analyses data were normalized using the gcRMA algorithm in the Bioconductor package (http://www.bioconductor.org) of the R statistical environment (http://www.r-project.org/). Present call information from the MAS5.0 algorithm was added to the gcRMA gene expression measures. Expression data have been deposited in NCBI's Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE10237 Information on Pomeroy et al. samples (Brain_MD) used: Patients included 60 children with medulloblastomas, 10 young adults with malignant gliomas (WHO grades III and IV), 5 children with AT/RTs, 5 with renal/extrarenal rhabdoid tumours, and 8 children with supratentorial PNETs (see Supplementary Information I). Medulloblastoma patients were treated with craniospinal irradiation to 2,400–3,600 centiGray (cGy) with a tumour dose of 5,300–7,200 cGy. All patients with medulloblastoma were treated with chemotherapy consisting of cisplatin and vincristine, plus combinations of carboplatin, etoposide, cyclophosphamide or lumustine (1-(2-chloroethyl)-3-cyclohexyl-1-nitrosourea, CCNU) (details in Supplementary Information II). Samples were snap frozen in liquid nitrogen and stored at -80 °C. Studies were done with approval of the Committee for Clinical Investigation of Boston Children's Hospital. The data were organized into three sets: data set A (42 samples containing 10 medulloblastomas, 10 malignant gliomas, 10 AT/RTs, 8 PNETs and 4 normal cerebella); data set B (34 samples containing 9 desmoplastic medulloblastomas and 25 classic medulloblastomas); and data set C (60 samples containing 39 medulloblastoma survivors and 21 treatment failures). The clinical attributes of each of the patients in the study are available in Supplementary Information II. Tissues were homogenized in guanidinium isothiocyanate and RNA was isolated by centrifugation over a CsCl gradient. RNA integrity was assessed either by northern blotting or by gel electrophoresis. Ten–twelve micrograms total RNA was used to generate biotinlylated antisense RNAs, which were hybridized overnight to HuGeneFL arrays containing 5,920 known genes and 897 expressed sequence tags, as described previously6. Arrays were scanned on Affymetrix scanners and the expression value for each gene was calculated using GENECHIP software (Affymetrix, Santa Clara, California). Minor differences in microarray intensity were corrected using a linear scaling method as detailed in Supplementary Information I. Scans were rejected if the scaling factor exceeded 3, fewer than 1,000 genes received ‘present’ calls, or microarray artefacts were visible. Please note that processed data is provided in the *.gct files as indicated in each sample description field.