Identification of Molecular and Biological Predictors of Genomic Complexity in Chronic Lymphocytic Leukemia.

Unbiased quantitative assessments of genomic complexity in CLL using SNP arrays have identified high genomic complexity (≥3 subchromosomal lesions per genome) as a strong negative independent prognostic factor for short time to first therapy (TTFT) and short time to subsequent therapy (TTST) in CLL.

In the present study, we have analyzed various variables as potential predictors for genomic complexity in CLL, including ZAP-70 status, CD38 status, IgV_H mutation status, p53 aberrations (exon 5–9 sequence mutations or absent expression), CLL FISH categories (del17p, del11q, trisomy 12, and del13q14 type I or del13q14 type II, based on SNP arrays), MDM2 SNP309 allele status and the fraction of CLL cells alive 40 hours post-irradiation using univariate (N=178 cases) and multivariate (N=138 to 166 cases, depending on the included variables) analysis. Data on irradiation-induced ATM autophosphorylation are nearing completion and will be analyzed once available. SNP array-based genomic complexity was analyzed either as a dichotomous variable (<3 versus ≥3 subchromosomal lesions per genome) or as a quantitative variable and was based on either losses only (~85% of all lesions in CLL) or as total complexity (losses, gains and UPD).

Results: Using the dichotomous genomic complexity score based on losses, the percentage of cells alive post-irradiation (odds ratio [OR] 3.26, p<0.0001), CD38 expression (OR=3.42, p=0.006), ZAP-70 expression (OR=4.06, p=0.005), IgV_H-unmutated (OR=3.54, p=0.008), p53 aberrations (OR=13.2, p<0.0001), del13q14 type II (OR=3.42, p=0.015), del17p (OR=21, p<0.0001) and del11q (OR=6.6, p=0.002) all emerged as significant predictors of genomic complexity in univariate analysis. For multivariate analysis, various models were built incorporating factors with significant univariate associations and alternatively incorporating either del17p, p53 aberrations or the fraction of CLL cells alive post-irradiation (all three variables are p53 status-dependent) into the model.

Using the dichotomous genomic complexity score based on losses, the fraction of cells alive post-irradiation, del17p, del11q, del13q14 type II, p53 aberrations and CD38 expression, emerged as significant independent predictors of genomic complexity in multivariate analysis. ZAP-70, IgV_H status and del13q14 type I were not independent predictors. Using quantitative genomic complexity based on losses, the fraction of cells alive post-irradiation, del17p, del11q, del13q14 type I or II, and p53 aberrations emerged as significant predictors of genomic complexity in multivariate analysis. ZAP-70, CD38 expression and IgV_H status were not independent predictors.

In summary, multiple markers independently predict for genomic complexity in CLL, facilitating identification of genes/mutations (p53, ATM, Rb and others) involved in conferring unstable genomes in CLL. This analysis identifies SNP array-based genome interrogations as a single test uniquely able to capture genomic complexity in CLL.