Large but not small copy-number alterations correlate to high-risk genomic aberrations and survival in chronic lymphocytic leukemia: a high-resolution genomic screening of newly diagnosed patients

The known recurrent genomic aberrations, that is deletions of 11q, 13q, 17p and trisomy 12, are important prognostic markers, which reliably predict clinical outcome in chronic lymphocytic leukemia (CLL) patients.1 Approximately 50% of CLL patients carry deletions of 13q, which is correlated with an indolent disease course when detected as a sole aberration. In contrast, deletions of 11q and 17p, which cover the ATM and TP53 gene, respectively, are associated with poor prognosis. Furthermore, trisomy 12 is related to an intermediate prognosis, whereas deletion of 6q has been identified as a recurrent CLL progression marker.1, 2

Genomic microarrays are widely used for screening of copy-number alterations (CNAs) in cancers. Several studies on CLL have provided additional information on genome-wide alterations, such as gain of 2p and deletion of 22q.3, 4 Recently, an SNP-array (50K) study showed an association between genomic complexity (3 CNAs/sample) and time to first or second treatment as clinical end points.5 To date, copy-number neutral loss of heterozygosity has been identified on chromosome 11p, 13q, and 17p in CLL,3, 6 although the biological or clinical significance remains unknown.

In this study, we applied high-density SNP arrays for evaluation of genomic alterations and their clinical impact in 203 newly diagnosed CLL patients. All patients were selected from the Swedish part of a Scandinavian population-based case–control study called SCALE (Scandinavian Lymphoma Etiology).7 Peripheral blood samples were collected during 1999–2001, within a median of 3 months from diagnosis (range, 0–18 months) and follow-up was achieved in 2008 (median follow-up time 91 months). All samples showed a typical immunophenotype and >70% tumor cells. Clinical data obtained from medical records is presented in Table 1.

Table 1 Presentation of baseline data, laboratory results and outcome of 203 CLL patients

Genomic DNA was extracted and subjected to Affymetrix GeneChip Mapping 250K Nsp1 arrays as described earlier.8 All raw and normalized data from the SNP arrays can be accessed through Gene Expression Omnibus (, accession number: GSE16406). Copy-number analysis for detection of CNAs was performed in the BioDiscovery Nexus Copy Number 3.0 software (BioDiscovery Inc., El Segundo, CA, USA) using the built-in Rank Segmentation algorithm. We performed validation experiments with quantitative real-time PCR on genomic DNA to define the optimal setting for detection of CNAs in the copy-number analysis (primer sequences are available on request). A high number of verified alterations (23 of 27, 85%) was found for regions >200 kbp, when applying P=1 × 10−6 in the copy-number analysis, thus applied for the detection of CNAs in all samples. Moreover, different cutoffs in log2ratio for regions 200–500 kb and regions >500 kb were applied (log2ratio 0.2 and 0.15, respectively), as longer stretches of SNPs are less likely to be affected by technical fluctuations. Furthermore, to avoid inclusion of copy-number variations, which exist as genomic variants in the population, regions overlapping >50% with reported copy-number variations were removed from further analysis.

We here report that 182 samples (90%) of the 203 CLL samples investigated carried CNAs, whereas 21 samples (10%) presented a normal karyotype. In total, 455 aberrations were identified and losses were more commonly detected than gains, 70% vs 30%, respectively. The majority of samples (75%) carried between 1 and 3 CNAs. Moreover, the known recurrent aberrations were detected in 74% of patients and constituted approximately one-third of the total CNAs in this cohort of newly diagnosed CLL patients (Table 2). Deletion of 13q was the most common CNA detected in 54%, in close agreement to other reports.1, 3, 6 Homozygous del(13q) were identified in 11%, focused to 13q14, and were considerably smaller compared with the heterozygous del(13q) (average 1.0 vs 5.1 Mbp). Only 4% (1/23) of the homozygous 13q deletions compared with 39% of heterozygous deletions covered the RB1 (retinoblastoma gene) at 13q14.2. On the contrary, the majority of both hetero- and homozygous del(13q) (98 and 87%) covered miR-15 and -16 at 13q14.3. The second most common CNA was del(11q) (13%), followed by trisomy 12 (11%). The least common known recurrent aberration in this cohort was deletion of 17p (4.4%), in which eight samples carried large deletions of 17p and one sample (SCAN16 in Figure 1a) showed a narrow lesion of 980 kb at 17p13.1 covering the TP53 gene. However, del(11q), del(17p) and trisomy 12 were detected at somewhat lower frequencies than reported earlier,1 which reflects the population-based nature of the cohort. In contrast, no recurrent del(6q) were detected, and the absence of this deletion supports the finding that this aberration is a progression marker in CLL.2 Nineteen samples showed more than one of the known recurrent aberrations and were, therefore, classified according to the ‘hierarchal model’ for the investigation of overall survival, time to treatment, and genomic complexity.1

Table 2 Known recurrent alterations detected in 203 patients
Figure 1

Genomic map of chromosomes 1–23. Patients are indicated with SCAN number and red and green color indicates losses and gains, respectively. (a) Nine patients carried del(17p) and displayed additional genomic aberrations (genomic map). The patients with del(17p) have a higher genomic complexity with an average of 7.3 CNAs/sample and an average size of 15.6 Mbp. (b) Five patients displayed concomitant gain of chromosome 2p and del(11q).

Besides the known recurrent aberrations, we detected 140 CNAs with a size >1 Mbp. These aberrations were overrepresented by losses, 61 vs 39% gains, and had an average size of 20 Mbp. Most alterations >1 Mbp were non-overlapping, such as trisomy 7, monosomy 9, and loss of chromosome 3p. However, we identified several large recurrent CNAs, for example deletions of chromosome 4p and 8p in three samples each, and gains of chromosome 8q in five samples (overlapping regions, 26.4, 38.5, and 55 Mbp, respectively). Trisomy 18 and 19, detected in two and three samples, respectively, were only carried by patients with a concomitant trisomy 12, suggesting a specific pathobiological mechanism. Interestingly, five samples showed a novel combination of gain on chromosome 2p (overlapping region 73 Mbp) and the poor-prognostic loss of 11q (Figure 1b). The gain of 2p has been described in a recent study, in which the CDR covered 2p24.3, which contains MYCN.3 The potential overexpression of MYCN in combination with downregulation of ATM through del(11q) may lead to an even more aggressive course. In fact, all five patients with del(11q) and gain 2p had received treatment.

In CLL, the existence of small recurrent alterations, which include genes linked to the pathogenesis of the disease, still remain to be determined. Excluding the known recurrent alterations, we detected 146 CNAs that were <1 Mbp, with an average size of 0.5 Mbp. In this study, with the level of resolution offered by Affymetrix 250K arrays, the majority of alterations <1 Mbp were non-overlapping, and detected only in individual samples. Moreover, the small alterations were common in all prognostic groups (Tables 1 and 2). These results indicate that small CNAs occur as random secondary events in early CLL cells, perhaps involving genes with influential effect on the disease. Novel small aberrations have recently been reported in CLL, although with low recurrence either within or between studies.4, 9 This discrepancy between different array studies is probably highly influenced by the differential design of microarrays applied.8

Overall survival was investigated in relation to the hierarchal model of known recurrent alterations, which verified the clinical impact of these prognostic markers (Figure 2a). However, one exception to earlier studies was the similar disease course for del(11q) and trisomy 12 (Figure 2a). Moreover, patients with homozygous del(13q) showed a tendency to better survival compared with patients with heterozygous del(13q) (log-rank test, P=0.09), which, at least partly, can be due to the fact that a higher frequency of patients with homozygous del(13q) (95%) showed mutated IGHV genes than patients with heterozygous del(13q) (73%). Subsequently, samples were grouped according to genomic complexity (number of alterations) by applying various size cutoffs for alterations (all, <1, >1–5, and >5 Mbp) to investigate the impact of increasing complexity on overall survival. This analysis revealed that a higher number of large CNAs correlated to worse outcome (Figure 2b–d).

Figure 2

Survival analysis according to known recurrent aberrations and an increasing number of genomic aberrations (Kaplan–Meier plots). Kaplan–Meier survival analysis and the log-rank test were performed using the Statistica Software 8.0 (Stat Soft Inc., Tulsa, OK, USA). Overall survival was defined as time from diagnosis until date of last follow-up or death. Survival according to (a) known recurrent aberrations, overall P<0.00001. (bd) Overall survival according to the degree of genomic complexity showed (b) no impact on survival by the number of CNAs <1 Mb (P=0.88), (c) a statistically significant difference between patients carrying an increasing number of CNAs between 1 and 5 Mbp (P=0.008), and (d) a worse outcome for patients carrying an increasing number of CNAs >5 Mbp (P=0.00002). Tick marks in Kaplan–Meier plots indicate censored follow-up.

However, investigation of the genomic complexity in samples grouped according to the hierarchal model showed that the general frequency of CNAs was considerably higher in patients with del(17p), carrying an average of 7.3 CNAs/sample, followed by patients with del(11q), trisomy 12 and del(13), in which an average of 3.6, 3.7, and 2 CNAs/sample was detected, respectively. Patients without any of the known recurrent alterations had an average of 1.28 CNA/sample (Table 2). Moreover, the patients with del(17p) carried a higher frequency of large genomic aberrations (>5 Mbp) than all other clinical subgroups with an average size 15.6 Mbp (Table 2; Figure 1a). In fact, all patients in the subgroup with 4 aberrations >5 Mbp carried either del(17p) or del(11q). Furthermore, 84% (21/25) of the patients with 2–3 CNAs >5 Mbp showed del(11q), del(17p), or trisomy 12, whereas patients carrying 0–1 CNA >5 Mbp were overrepresented in subgroups showing no recurrent CNA or del(13q). Accordingly, the significant association between high complexity and poor survival was lost when patients with del(11q) and del(17p) were removed from the evaluation of overall survival. The poor-prognostic del(11q) and del(17p) covers the ATM and TP53 genes, respectively, which have essential functions in the cell-cycle control and DNA-repair mechanisms. Naturally, loss of function of these genes will lead to deficiencies in genome maintenance, and thus promote a higher genomic complexity. To exclude that the four samples with 2–3 CNAs, which did not display intermediate- or poor-risk genomic markers, carried mutations of the TP53 gene, we sequenced exons 4–8, which did not reveal any mutation. However, this analysis does not exclude the presence of other p53/ATM defects.

When investigating genomic complexity in relation to time to treatment, a significantly shorter time to initiation of treatment was shown in patients with increasing number of CNAs >5 Mbp (data not shown). Similar results have been presented by Kujawski et al.5 who reported a correlation between genomic complexity and a significantly shorter time to first and second treatment and presented the number of CNAs as an independent factor for prognosis. However, time to first treatment is subjective and time to second treatment is a poor end point, as it is highly dependent on type and intensity of firstline treatment. In contrast, our study had the potential to evaluate impact on overall survival, and we show that genomic complexity is a marker of poor outcome, although patients with a complex genome most often also carry del(17p) and del(11q) in these cases.

The identification and quantification of copy-number neutral loss of heterozygosity, which represents loss of heterozygosity without a change in copy number, was performed according to a newly developed method.10 Interestingly, we found that eight samples (3.9%) had recurrent copy-number neutral loss of heterozygosity covering large parts of chromosome 13q, which all carried a homozygous deletion of 13q14 (data not shown). The finding that all copy-neutral regions included a homozygous del(13q) supports the idea that loss of 13q14 preceded the loss of heterozygosity event as suggested by Pfeifer et al.3 The recurrence of copy-number neutral loss of heterozygosity on chromosome 13 observed by others and us suggests that this copy-neutral event is important in CLL biology.3, 6

One potential drawback with this study would at the first sight be that we did not perform cell sorting of the CLL cells, which might hamper detection of aberrations in a small proportion of malignant cells. However, as the investigated samples had a high tumor cell content (>70%, median 86%), we do not believe that the normal cell content per se would mask genomic aberrations with clinical relevance. In addition, cell sorting would not necessarily avoid the problem with detection of CNAs in subclones of tumor cells. In addition, one could argue that normal cells from the patients should have been investigated in parallel to properly exclude patient-specific copy-number variations. However, we believe that our chosen approach is more applicable in the clinical setting, in which cell sorting and parallel analysis of normal cells will probably not be an option.

In conclusion, whole-genome screening with SNP arrays revealed a high frequency of known recurrent alterations in newly diagnosed CLL patients. Moreover, the genome-wide analysis allowed detection of a novel combination of gain of 2p and del(11q), and additional large and small CNAs, which are important for the evaluation of overall complexity in CLL patients. In the survival analysis, we identified genomic complexity as a poor-prognostic marker; however, noted that this feature was strongly linked to established poor-risk molecular markers. As the small alterations were mostly non-overlapping, it seems unlikely that there are hitherto unknown recurrent CNAs >200 kbp involved in the CLL pathophysiology detectable in this setting. Upcoming high-resolution array studies may allow detection of even smaller genomic regions and reveal important genetic lesions. In addition, other newly developed methods such as genome-wide sequencing may reveal new recurrent mutations linked to CLL pathogenesis.

Conflict of interest

The authors declare no conflict of interest.

Accession codes




  1. 1

    Dohner H, Stilgenbauer S, Benner A, Leupolt E, Krober A, Bullinger L et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med 2000; 343: 1910–1916.

  2. 2

    Finn WG, Kay NE, Kroft SH, Church S, Peterson LC . Secondary abnormalities of chromosome 6q in B-cell chronic lymphocytic leukemia: a sequential study of karyotypic instability in 51 patients. Am J Hematol 1998; 59: 223–229.

  3. 3

    Pfeifer D, Pantic M, Skatulla I, Rawluk J, Kreutz C, Martens UM et al. Genome-wide analysis of DNA copy number changes and LOH in CLL using high-density SNP arrays. Blood 2007; 109: 1202–1210.

  4. 4

    Gunn SR, Bolla AR, Barron LL, Gorre ME, Mohammed MS, Bahler DW et al. Array CGH analysis of chronic lymphocytic leukemia reveals frequent cryptic monoallelic and biallelic deletions of chromosome 22q11 that include the PRAME gene. Leuk Res 2008; 33: 1276–1281.

  5. 5

    Kujawski L, Ouillette P, Erba H, Saddler C, Jakubowiak A, Kaminski M et al. Genomic complexity identifies patients with aggressive chronic lymphocytic leukemia. Blood 2008; 112: 1993–2003.

  6. 6

    Lehmann S, Ogawa S, Raynaud SD, Sanada M, Nannya Y, Ticchioni M et al. Molecular allelokaryotyping of early-stage, untreated chronic lymphocytic leukemia. Cancer 2008; 112: 1296–1305.

  7. 7

    Smedby KE, Hjalgrim H, Melbye M, Torrang A, Rostgaard K, Munksgaard L et al. Ultraviolet radiation exposure and risk of malignant lymphomas. J Natl Cancer Inst 2005; 97: 199–209.

  8. 8

    Gunnarsson R, Staaf J, Jansson M, Ottesen AM, Goransson H, Liljedahl U et al. Screening for copy-number alterations and loss of heterozygosity in chronic lymphocytic leukemia-a comparative study of four differently designed, high resolution microarray platforms. Genes Chromosomes Cancer 2008; 47: 697–711.

  9. 9

    Grubor V, Krasnitz A, Troge JE, Meth JL, Lakshmi B, Kendall JT et al. Novel genomic alterations and clonal evolution in chronic lymphocytic leukemia revealed by representational oligonucleotide microarray analysis (ROMA). Blood 2009; 113: 1294–1303.

  10. 10

    Goransson H, Edlund K, Rydaker M, Rasmussen M, Winquist J, Ekman S et al. Quantification of normal cell fraction and copy number neutral LOH in clinical lung cancer samples using SNP array data. PloS One 2009; 4: e6057.

Download references


We thank the Swedish CLL group for support during the collection of clinical data. This research was supported by the Nordic Cancer Union, the Swedish Cancer Society, the Swedish Research Council, Lion's Cancer Research Foundation, Uppsala, Sweden and the Svend Andersen Foundation, Denmark. A fellowship (2006/18) was awarded to R Rosenquist by the European Hematology Association.

Author information

Correspondence to R Gunnarsson.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Gunnarsson, R., Isaksson, A., Mansouri, M. et al. Large but not small copy-number alterations correlate to high-risk genomic aberrations and survival in chronic lymphocytic leukemia: a high-resolution genomic screening of newly diagnosed patients. Leukemia 24, 211–215 (2010).

Download citation

Further reading