Abstract
INTRODUCTION: Clinical decision making in Hodgkin lymphoma (HL) is primarily based on clinical variables in part because the scarcity of the malignant Hodgkin Reed Sternberg cells (HRS cells) hampers their molecular characterization. However, more recently investigation using laser capture microdissection has allowed a more detailed analysis of these cells. The objective of this study was to detect genomic alterations in HRS cells and correlate these changes with treatment outcome.
PATIENTS AND METHODS: We studied 53 patients with classical HL who were primarily treated at the BC Cancer Agency in Vancouver between 1984 and 2006. All received at least 4 cycles of polychemotherapy and stage-dependent radiotherapy if indicated. The cohort included 43 pretreatment samples and 10 biopsies taken at relapse. Treatment failure was defined as disease progression or relapse at any time (n=23), treatment success as absence of progression (n=30). Whole genome amplification (GenomePlex, Sigma) of pools from 500–1000 individually picked, microdissected HRS cells (Molecular Machines & Industries Cellcut with Nikon Eclipse TE2000-S microscope) was performed. 200 ng of amplified DNA was hybridized to 32k submegabase resolution BAC tiling arrays (SMRT) against sex-matched control-DNA. Scoring of array CGH data was performed by computational analysis using CNA-HMMer v0.1 (available at http://www.cs.ubc.ca/~sshah/acgh/) based on a Hidden Markov Model (HMM). Clustering of the 53 cases was performed using the K-medoids algorithm. Areas of amplification bias and known copy number polymorphisms were excluded.
RESULTS: On average whole genome amplification generated 500-fold amplification of genomic DNA. The most frequent copy number alterations (>20% of cases) included gains of 2p14–24.3, 9p12–24.3, 12p11.21–13.33, 16p11.2–13.3, 17p11.2–13.3, 17q11.1–25.1, 19p12–13.3, 19q12–13.43, 20q11.21–13.32, 21q22.11–22.2 and losses of 1p36.31–36.33, 6q11.1–27, 7q22.1–36.3, 8p23.1–23.3, 11q22.3–25, 13q33.3–34 and Xq11.2–28. We also identified several small changes (<5 Mbs) such as loss of 1p36.32, 5q31.1 or 6q23.3 and amplification of 1q32.1, 8q24.21, 17q21.31 or 20q13.2. When comparing the different outcome groups we more frequently identified gains of chromosomal regions 12p13.31–13.33 and 16p12.1–13.3 in treatment failures (39% vs. 13%, Fisher exact p=0.05 and 43% vs. 10%, p=0.009), and losses of chromosomal regions 16q12.1–12.2 and 17p13.1–13.2 (17% vs. 0%, p=0.061 and 27% vs. 4%, p=0.061) were more frequently observed in treatment successes. We did not find significant differences of these changes between pretreatment (n=13) and relapse biopsies (n=10) of patients failing treatment. Using unsupervised analysis we identified a sample cluster of eight cases characterized by simultaneous occurrence of gains of 2p, 16p, 17p, 19q and losses of 6q. Notably, treatment failed in six of these cases.
DISCUSSION: The combination of laser microdissection with subsequent WGA and high resolution array CGH provides a robust and sensitive platform for detecting chromosomal imbalances in microdissected HRS cells. We identified at high-resolution new and recurrent changes defining chromosomal regions that potentially harbor oncogenes and tumor suppressor genes crucial to the pathogenesis of HL. Furthermore, we found copy number alterations that are significantly associated with disease progression which, therefore, could serve as predictive factors for treatment outcome.
Evaluation of Phi29-based whole-genome amplification for microarray-based comparative genomic hybridisation
