Leukemic Blasts With The PNH Phenotype: Correlation With Cytogenetics In ALL

Abstract

It has been proposed that hypermutability is essential for the generation of malignancies that require more than a few oncogenic mutations and may also promote the emergence of drug-resistant clones by contributing to intratumoral diversity. However, most genes are not suitable as a reporter for quantitating the frequency of spontaneous somatic inactivating mutations. A notable exception is the PIG-A gene, which encodes an enzyme catalyzing an initial step in the biosynthesis of glycosylphosphatidylinositol (GPI), which links certain membrane proteins to the cell surface. Of note, the PIG-A-mutant phenotype can be readily demonstrated by flow cytometry, using antibodies specific for GPI-linked proteins (e.g., CD55 and CD59), as well as the FLAER reagent (which binds to GPI directly), enabling the rapid screening of large populations for rare variants that are GPI-negative. It is known from the condition paroxysmal nocturnal hemoglobinuria that a very broad spectrum of PIG-A mutations can produce this “PNH phenotype”, and because the gene is X-linked, this requires only a single mutation. Apart from patients with PNH, there is evidence that PIG-A mutations are growth-neutral in most all other circumstances. We have previously demonstrated rare spontaneously arising cells with the PNH phenotype and genotype in blood cells of normal donors and in B-lymphoblastoid cell lines (BLCLs); similar populations can be demonstrated in mice, with marked increases after mutagen exposure. We have also shown that malignant cell lines often demonstrate an elevated frequency (f) of spontaneously arising cells with the PNH phenotype. Recently, we have reported similar findings in ex vivo leukemic blasts derived from patients with ALL: in samples from 10 of 19 patients, the median frequency of blasts with the PNH phenotype was low, with a median f value of 13 per million, which was similar to that of BLCLs from normal donors. The remaining 9 samples demonstrated an elevated f value (median 566 per million). Based on our preliminary subset analysis, along with published data demonstrating a higher mutation burden in leukemias with the ETV6-RUNX1 (TEL-AML1) translocation compared with hyperdiploid samples, we hypothesized that we would find a higher mutant frequency in the former group. Therefore, we have now analyzed in a blinded manner a new panel of blast cells derived from ficolled pre-treatment marrow samples from patients with B-lineage ALL obtained from the Children's Oncology Group ALL Cell Bank. To determine f, samples were stained sequentially with an Alexa-488 conjugate of the FLAER reagent and then a mixture of mouse anti-CD55 and anti-CD59 antibodies, followed by FITC-conjugated rabbit anti-mouse immunoglobulin secondary antibody (registering on FL1), and then PE-conjugated anti-CD45 antibody (registering on FL2). The blast population was identified based on forward and side scatter, expression of CD45, and exclusion of propidium iodide. GPI-negative cells were defined as expressing <4% of the mean level of FL1 fluorescence as the GPI-positive population, and >10% of its mean FL2 fluorescence. Overall, among 26 ALL samples, the f values spanned 5 orders of magnitude and had a median value of 49 per million. For the 8 ETV6-RUNX1 samples, the median f value was 2243 per million, compared with 18 per million for the 11 hyperdiploid samples (p <0.003). Among 7 samples with neither abnormality, the median f value was 60 per million. The one sample with a t(1;19) translocation that we analyzed demonstrated an f value of 330 per million. In comparison, the median f value among 11 BLCLs from normal donors was ∼ 5 per million, and f was ∼1850 per million for HBL2A, a mantle cell lymphoma line that exhibits hypermutability. Based on the data for BLCLs, we used a value of 50 per million as a cutoff: 7 of 8 ETV6-RUNX1 samples demonstrated hypermutabilty, compared with 2 of 11 hyperdiploid samples, and 4 of 7 samples with neither abnormality. These data demonstrate that hypermutability is frequently seen in ALL and is associated with, but not restricted to samples with the ETV6-RUNX1 fusion. These data also confirm that hypermutability—at least as detected by this assay, is not always a requirement for leukemogenesis. We predict that it will be possible to find associations between other cytogenetic subsets and hypermutability, which may correlate with the number of mutations required to produce the malignancy. Disclosures:No relevant conflicts of interest to declare.