Comprehensive Genetic Characterization of 17p Deleted CLL Identifies Predictors of Overall Survival

Many studies have shown that deletion at chromosome 17p targeting the TP53 gene, or del(17p), is associated with poor prognosis in chronic lymphocytic leukemia (CLL). Despite this, not all del(17p) CLL progresses rapidly to treatment and has short survival. We hypothesized that other coexisting genetic aberrations may contribute to the poor clinical outcome and heterogeneity of del(17p) cases. To assess this, we analyzed copy number alterations (CNAs) using Affymetrix SNP array data from 200 CLL patients (55 with del(17p)), and somatic mutation profile by whole exome sequencing (WES; Illumina) in 168 patients (53 with del(17p)). Ninety-nine patients were studied with both SNP arrays and WES, 39 of whom had del(17p).

Analysis of copy number showed that del(17p) CLL had a median of 11 CNA events, mostly copy number losses, compared to 2 events in WT CLL (p=1.1E-12). Both the number of CNA events (p=7.9E-10) and the total length of copy number gain (p=1.6E-4) or loss (p=1.3E-8) were associated with shorter overall survival (OS), even when controlling for co-existing del(17p). Using GISTIC analysis, we discovered three significantly deleted regions specific to del(17p), namely large chromosomal deletions at 3p, 4p, and 9p. These novel recurrent deletions were rarely seen in wild type CLL and the presence of any of these deletions with del(17p) was strongly associated with shorter OS compared to del(17p) alone. We were able to evaluate complex karyotype (CKT) determined by stimulated metaphase cytogenetics in a subset of the cohort, and found that 17/27 (63%) evaluable del(17p) CLLs had CKT, while only 13/58 (22%) WT CLLs did (p=0.0005). Analysis of OS showed that del(17p) and CKT together had worse OS than either individually, although individually they each conferred OS worse than WT (p<0.0001).

Next we tested whether the number and size of CNAs predicts progression to treatment in 138 patients who were treatment naïve at sampling (n=23 for del(17p)), 65 of whom progressed to treatment after sampling (n=16 for del(17p)). We found that increasing number of CNA events (p=9E-6), total length of losses (p=4E-4), and total length of gains (p=2E-4) were all predictive of need for future treatment in WT as well as del(17p) CLL. Those treatment naive del(17p) patients who remained untreated had a median of 4 CNAs (n=7), compared to 12.5 for those who went on to treatment (p=0.013).

Turning our attention to somatic mutation analysis, del(17p) CLLs had higher numbers of total somatic mutations (21 vs 18, p=0.0046), and nonsynonymous mutations (16 vs 13, p=0.0059) than WT CLL, with no difference in subclonal mutations (12 vs 10, p=0.34). Increasing number of total mutations (p=0.0017) and nonsynonymous mutations (p=0.0003) were both associated with shorter OS, even when controlling for 17p deletion. No significant correlation was observed between number of CNAs and number of somatic mutations in a given CLL, suggesting different mechanisms involved in their causation. In the del(17p) CLLs, as expected, TP53 was the most commonly mutated gene, seen in 43/53 (81%) patients. Interestingly, most TP53 mutations (78%) were clonal, suggesting they occurred early in CLL development, while SF3b1 assessed for comparison was most commonly subclonal (68%). Patients with subclonal TP53 mutations had longer OS than patients with clonal TP53 mutations (p=0.041).

We also explored the outcome of patients with loss of both TP53 alleles via 17p deletion and TP53 mutation. We found that biallelic loss was associated with more CNAs (median = 15 vs 9) and longer total length of CNAs than monoallelic loss. Patients with biallelic loss of TP53 showed a trend towards worse OS than the limited cohort with monoallelic loss (p=0.07).

In addition to TP53 and other previously reported CLL drivers such as NOTCH1 (n=8, q=8E-5) and DDX3X (n=4, q=0.03), we report RPS15 (n=6, q=6E-5) and GPS2 (n=3, q=0.03) as novel significantly mutated genes in del(17p) CLL but not in WT CLL. All RPS15 mutations were clustered in a 15-amino acid region in the far carboxyl terminus of the gene, suggesting conserved functions. RNA sequencing (n=2) confirmed that the mutant alleles of RPS15 were expressed in the CLLs. We conclude that the genetic profile of del(17p) CLL differs significantly from WT CLL and show that worse OS in del(17p) CLL is associated with complex karyotype and biallelic inactivation of TP53 as well as increasing number of somatic mutations and novel CNAs.