Abstract

Conventional karyotype and chromosomal comparative genomic hybridization (CGH) studies of diffuse large B-cell lymphoma (DLBCL) have revealed few chromosomal abnormalities associated with outcome. In order to evaluate the association between genomic copy number changes in DLBCL and clinical outcome at a higher resolution, we assayed a panel of 64 newly-diagnosed DLBCL specimens by array-CGH. For each specimen, the respective patient had a known response to anthracycline-based therapy, and the median follow-up was 5-years. Forward and reverse hybridizations were performed for all specimens to avoid dye-bias, using BAC/PAC arrays with 1-4 Mb resolution coverage of the genome (Spectral Genomics). After customized normalization, the circular binary segmentation (CBS) algorithm was used to identify regional copy number changes along each chromosome in each specimen. Of the 64 specimens, 59 (92%) displayed regional copy number changes. The frequency of gain/loss of each clone for all 64 specimens was then calculated based on the CBS results, and, additionally, based on singleton clone changes if they were outside of CBS-defined regions. Clones displaying change ≥ in 10% of specimens for at least two contiguous clones were further considered as recurrent sites for clinical correlations. 27 sites of recurrent gain and 25 sites of recurrent loss were identified. Association between loss or gain of sites and International Prognostic Index (IPI) were evaluated using Fisher’s exact test, and associations between loss or gain of sites and time to treatment failure (TTF) and overall survival (OS) were tested using the log-rank test. Table 1 lists the sites as independent markers that were significant at p < 0.05. Of these sites, five (with asterisk) were found to significantly predict outcome after stratifying by IPI using the stratified log-rank test. It is noted that 6 of the sites were 5 Mb or less in size, facilitating the identification of target genes. Losses of chromosome 7 (8–13 Mb) and chromosome 13 (56.7–63 Mb) were found to contribute to IPI prediction of outcome, but were not significantly associated with clinical features as independent markers. In summary, array-CGH has lead to the identification of several genetic markers with prognostic significance in DLBCL, which in some cases are of a size amenable for target gene identification.

Significant Genetic Markers Associated with Clinical Outcome

Chrom. Start (Mb) End (MB) Change # of cases IPI TTF OS 
78.2 79.1 Loss 19   Better 
2.4 4.1 Loss   Worse* 
0.2 72.3 Gain 14  Better Better 
101.3 136 Gain 18 Low Better  
154.7 188.7 Gain 18 Low Better Better 
62.2 170.5 Loss 32 Low Better Better 
18.8 19.2 Loss 11  Better*  
8.3 13 Loss 13  Better  
83.9 96.3 Gain   Better 
122.1 132.8 Gain 12  Better Better* 
12 133.3 Gain 31 High   
16 0.1 4.7 Gain 13   Better 
16 33.8 35.6 Loss  Worse* Worse* 
19 43.1 63.7 Gain 13  Better Better 
Chrom. Start (Mb) End (MB) Change # of cases IPI TTF OS 
78.2 79.1 Loss 19   Better 
2.4 4.1 Loss   Worse* 
0.2 72.3 Gain 14  Better Better 
101.3 136 Gain 18 Low Better  
154.7 188.7 Gain 18 Low Better Better 
62.2 170.5 Loss 32 Low Better Better 
18.8 19.2 Loss 11  Better*  
8.3 13 Loss 13  Better  
83.9 96.3 Gain   Better 
122.1 132.8 Gain 12  Better Better* 
12 133.3 Gain 31 High   
16 0.1 4.7 Gain 13   Better 
16 33.8 35.6 Loss  Worse* Worse* 
19 43.1 63.7 Gain 13  Better Better 

Author notes

Corresponding author