Page 146 - 2018_12-Haematologica-web

L.V. Abruzzo et al.
del(17p), the site of TP53, and del(11q), the site of ATM and the miR34b/c cluster, are markers of poor prognosis.1,3 Cases with +12 have an overall survival (OS) that lies in the middle range between that of cases with del(13q) and those with del(11q) or del(17p).
Although +12 CLL comprises approximately 20% of cases, relatively little is known about its pathophysiology. These cases often demonstrate atypical morphological and immunophenotypic features.4 Patients with +12 CLL show a higher incidence of thrombocytopenia, Richter transformation, and other secondary cancers, their main cause of death.5 However, compared to other cytogeneti- cally-defined CLL subtypes, few studies have attempted to identify the critical protein-coding and microRNA genes that are relevant to its pathophysiology.6-9 For many cancer types, gene dosage correlates strongly with mRNA, microRNA, and protein expression.6,9 This suggests that at least a subset of the more than 1000 protein-coding genes (and more than 40 microRNA genes) on chromosome 12 in +12 CLL are likely to show a concordant increase in expression. Conceivably, some of these proteins modu- late expression of downstream targets, either on chromo- some 12 or other chromosomes, resulting in aberrant gene expression.
We aimed to identity protein-coding genes whose expression contributes to the unique pathophysiology of +12 CLL. We performed transcriptional profiling on CLL cases with +12 as the only cytogenetic abnormality, and compared them to cases with only del(13q), del(11q), or diploid cytogenetics. We demonstrate that CLL cases with +12 as the only cytogenetic abnormality express a unique set of activated pathways compared to other cyto- genetic subtypes, several of which may represent new therapeutic targets.
Methods
Sample collection and preparation
Between 2000 and 2008, we obtained peripheral blood (PB) from 250 treatment-naïve CLL patients. This study was approved by the Institutional Review Board and conducted according to principles expressed in the Declaration of Helsinki. We extracted nucleic acids from negatively-selected CLL cells (CD19+) and pre- pared cDNA, as described previously.10 Cases were divided into discovery and validation sets, as described in the Online Supplementary Methods.
Single nucleotide polymorphism (SNP) genotyping, IGHV and NOTCH1 mutation status, and ZAP70 protein expression
We assessed genomic copy number variations (CNV) by single nucleotide polymorphism (SNP) genotyping.10 The IGHV somatic mutation status was assessed as described previously, with minor modifications. Patient and germline sequences were aligned in VBASE II. Cases with <2% mutations compared to germline were designated “unmutated”; cases with ≥2% mutations were desig- nated “mutated”.11 NOTCH1 exon 34 mutation hotspots were assessed as described previously.12 We assessed ZAP70 protein expression by immunohistochemistry or flow cytometry.
Transcriptional profiling of protein-coding genes
For discovery, we performed transcriptional profiling using gene expression microarrays (Online Supplementary Methods).13 To iden- tify differentially expressed (DE) genes, we compared the +12
cohort to the others individually and jointly. We assessed DE genes between subtypes by performing probe-by-probe ANOVA (for multiple subtypes) or t-tests (for two subtypes). We fit a beta- uniform-mixture (BUM) model to the set of P-values in order to find the false discovery rate (FDR). Microarray data are available at http://silicovore.com/CLL/Trisomy12. For validation, we performed transcriptional profiling using an MF-QRT-PCR assay to a subset of DE genes identified using microarrays, along with 5 endoge- nous control genes.13 We performed gene-by-gene ANOVA and t-tests to validate DE genes discovered from the microarrays. A discovery was considered “validated” if the unadjusted P-value was <0.05.
Survivalanalysis
We performed time-to-event (survival) analysis using Cox pro- portional hazards models, and assessed significance using the log- rank (score) test. To assess multivariate models, we used a for- ward-backward stepwise algorithm to eliminate redundant factors and optimize the Akaike Information Criterion (AIC). We per- formed the computations using the survival package (v.2.40-1) in R v.3.3.0, and computed median follow-up times using the reverse Kaplan-Meier estimator.14
Pathway analysis
We performed pathway analysis using Ingenuity Pathway Analysis (IPA; Qiagen, Redwood City, CA, USA). The statistical significance of altered pathways was calculated using a one-sided Fisher exact test. The z-score indicates if a pathway is activated (positive) or down-regulated (negative). We also examined ratio values (the number of molecules with expression levels above or below the mean, divided by the total number of molecules in that pathway) for the canonical pathways.
Results
Patients' characteristics of discovery and validation sets
Based on SNP genotyping, we divided cases into cytoge- netic subsets defined by abnormalities that would be detected using a FISH probe panel to the common CLL cytogenetic abnormalities: del(11q), del(13q), del(17p), +12, and diploid. The proportion of cases in each subset is similar to that reported previously.7 Because patients were treatment-naïve, there was only one del(17p) case, which we excluded from subsequent analysis; a single case is insufficient to achieve statistical significance in a comparison of gene expression profiles to other cytoge- netic subtypes.
We focused our experiments on cases with only +12 compared to cases with only del(13q), del(11q), or diploid cytogenetics. We identified 147 CLL samples that met our inclusion criteria. One hundred and twenty-three con- tained a single abnormality or were diploid: 27 with +12 as the only abnormality (18%), 49 with del(13q) as the only abnormality (33%), and 47 diploid cases (32%). The discovery set was chosen to contain patients with only abnormalities who later went on to receive front-line ther- apy with FCR. Because there were relatively few cases with del(11q) as the only abnormality, to achieve statisti- cal significance, we used 10 cases with del(11q) as the only abnormality for the discovery set, but included 14 cases with del(11q) and del(13q) in the validation set. Patients' characteristics are summarized in Tables 1 and 2.
The discovery set contained no statistically significant
2070
haematologica | 2018; 103(12)