Abstract
Abstract 1774
Introduction:
Chronic Lymphocytic Leukemia (CLL) follows a variable, difficult to predict course. Biomarkers (IGHV mutations, CD38 and ZAP-70) have improved prognostication but accuracy is only ∼80%, not sufficient to initiate preemptive therapy. To address this, we have incorporated another molecular biological parameter, hypervariably expressed genes (HVEGs) from global gene expression profiles (GEPs) [Dozmorov, 2011], along with IGHV mutation status to develop a more robust method of prognostication.
Methods:
RNA was purified from B cells of 65 CLL patients and 25 normal subjects. GEPs were defined with Illumina HumanHT12 beadchips and analyzed to identify HVEGs fluctuating coordinately in CLL patients but not expressed in normal circulating human B cells. A Cox proportional hazards regression model was used to compare expression levels with clinical outcome.
Results:
Unmutated CLL (U-CLL) and mutated CLL (M-CLL) samples were sorted based on the averaged normalized gene expression level in the largest HVEG cluster consisting of 45 genes. Samples were split into subgroups based on high and low expression of these genes. Four subgroups were identified: U-CLL with high (U-HVEhi), U-CLL with low (U-HVElo), M-CLL with high (M-HVEhi), and M-CLL with low (M-HVElo) expression of HVEGs. When these subgroups were analyzed for time-to-first treatment (TTFT), the findings were remarkably different. Using a Cox proportional hazards regression model we found that for M-CLL the presence of low levels of HVEGs denoted a subgroup with a prolonged median TTFT compared to M-CLL patients with high levels of HVEGs (23.46 yrs in M-HVElo versus 9.05 yrs in M-HVEhi; P=0.0075). This association was reversed in U-CLL, where low levels of HVEGs pinpointed a subgroup with a shortened median TTFT compared to U-CLL patients with high levels of HVEGs: median TTFT was 8.01 yrs in U-HVEhi and decreased by 50% to 3.12 yrs in U-HVElo (P=0.0151). Compared to the TTFT for M-HVElo, the TTFT for U-HVElo was dramatically shorter (3.12 vs. 23.46 years; P<0.0001).
Discussion:
Combining the HVEG approach with IGHV mutation status enabled us to define intra-group heterogeneity, and thereby improve prognostication for apparently homogeneous subgroups of U-CLL and M-CLL patients. This combination was superior to IGHV mutations alone in that it pinpointed a subset of M-CLL patients with shorter TTFTs and a subset of U-CLL patients with longer TTFTs, indicating that the combined approach selects those patients that IGHV mutation analysis misclassified. Therefore, combining IGHV mutation status and HVEG expression provides a more precise indicator with potentially valuable clinical implications. The data suggest that the HVEGs are responsible for the different outcomes in the two groups and could provide mechanistic insights into key aspects of CLL B-cell biology as well as therapeutic targets.
Disclosures:
No relevant conflicts of interest to declare.
Variation in the number of activated torso receptors correlates with differential gene expression