Polygenic Risk Score and Risk of Chronic Lymphocytic Leukemia, Monoclonal B-Cell Lymphocytosis (MBL), and MBL Subtypes

Background

MBL is a precursor to chronic lymphocytic leukemia (CLL) and is subclassified into low-count (LC) MBL (absolute B-cell count<0.5x10⁹/L) and high-count (HC) MBL (absolute B-cell count between 0.5 and 5x10⁹/L). We previously reported that a polygenic risk score (PRS) based on a weighted average of 41 CLL-susceptibility variants was associated with risk of both MBL and CLL among a cohort of individuals from CLL families. Here we evaluate this PRS in an independent cohort of MBL and CLL individuals of European ancestry (EA), all of whom were ascertained agnostic to CLL family-history status. We also evaluate the PRS by MBL subtype (LC/HC), and in African American (AA) CLL cases and controls.

Methods

We genotyped 535 EA MBLs (139 HC-MBL, 396 LC-MBLs), 735 CLLs (640 EA, 95 AA), and 2,866 controls (2,631 EA, 235 AA) from the Mayo Clinic CLL Resource, Duke University, and Weill Cornell Medical College. We computed the CLL-PRS for each individual and used logistic regression to estimate odds ratios (OR) and 95% confidence intervals, adjusting for age and sex. To assess discriminatory accuracy, we computed the c-statistic. Among EA individuals, we calculated a trend test among LC-MBL, HC-MBL, and CLL risk using the P-value for heterogeneity from a polytomous logistic regression analysis. Moreover, we plotted a boxplot for the PRS among controls, LC-MBL, HC-MBL, and EA CLL, as well as for AA CLL cases and controls, and tested the statistical difference using the Kruskal Wallis test and Mann-Whitney test, respectively.

Results

We found a significant association of PRS with overall MBL risk (OR=1.87, P=1.1x10^-28) with good discrimination (c-statistic=0.72). Significant associations were also found for LC-MBL (OR=1.75, P=7.5x10^-19, c-statistic=0.72), HC-MBL (OR=2.22, P=1.4x10^-17, c-statistic=0.74), and CLL of EA (OR=2.60, P=1.2x10^-62, c-statistic=0.78), with a significant difference among these cohorts (Figure 1.A) and a significant positive trend across these cohorts (P_{heterogeneity}=8.4x10^-6). Although we observed a 33% increased risk of CLL in AA (c-statistic=0.57), the PRS was borderline significant (P=0.07, Figure 1.B).

Conclusion

The CLL-PRS is a strong prediction-tool for risk of CLL and MBL among individuals of EA. Future studies are needed to improve the PRS for AAs including performing GWAS of AA in order to identify CLL-susceptibility SNPs that are more representative within known CLL loci and to discover novel CLL loci that are unique for AAs.