Abstract
B-cell chronic lymphocytic leukemia (CLL), the most common adult leukemia, is a clonal disease manifested by an absolute lymphocytosis. Previously we identified genes capable of sorting and predicting the B-cells of patients with CLL or monoclonal B-cell Lymphocytosis (MBL) from aged matched controls. Using gene expression profiling, we and others have identified genes transcribed at significantly different levels between CLL patients and normal subjects. From these genes RT-QPCR verification was performed, in triplicate or more, on 22 gene candidates. A gene panel of seven genes, significantly different between CLL and normal individuals was chosen: FMOD, CKAP4, PI3Kc2b, LEF1, PFTK1, Bcl2 and GPM6a (in order of Receiver Operating Characteristic [ROC] concordance). This RT-QPCR verification step was based upon the comparison of freshly isolated PBMCs from CLL patients and age matched normal subjects after enriching for B cells with negative selection. The panel was then used to blindly categorize and predict the relationship of RNA from apparently normal subjects, some with and some without MBL, to the mRNA of patients with CLL as well as normal subjects, presumably without MBL. Since the numbers of cells within the expanded clones from individuals with MBL, after positive selection was limiting, RNA was amplified utilizing an RNA whole genome amplification approach and compared to similarly amplified RNA after negative selection; CLL cells, B-cells from normal age matched controls, and cord blood-derived B1 cells, enriched further by CD5 expression with positive selection. Using the Bayesian relevance determination method, a novel computer-learning tool to provide a probabilistic model, 56 samples were identified correctly at the rate of 100%. FMOD and PI3Kc2b were determined to be optimal at predicting blinded samples. In order to make the panel more clinically feasible, a universal reference RNA standard was used as a control, eliminating the need for pairing samples with age-matched controls. Analysis of this new model led to the discovery that a single gene; lymphoid enhancer binding factor 1 (LEF1), was not only capable of sorting CLL B cells from age-matched controls at a 100% rate (n=40) but it was also capable of predicting CLL B cells from other B cell malignancies (n=15). Verification of the protein expression levels for these genes is ongoing. The RNA expression levels of the genes in this panel provides a novel way to identify cell populations and gene expression patterns that change during the transition from B-cell clonal expansions that occur physiologically from those that occur among pre-leukemic and leukemic B-cell populations. The addition of a standard reference RNA enhances the clinical application of this gene panel. These data strongly suggest a unique role for LEF1 in CLL.
Homozygous transcobalamin II deficiency maintained on oral hydroxocobalamin
