Properties of CD34+ CML stem/progenitor cells that correlate with different clinical responses to imatinib mesylate

Abstract

Imatinib mesylate (IM) induces clinical remissions in chronic-phase chronic myeloid leukemia (CML) patients but IM resistance remains a problem. We recently identified several features of CML CD34(+) stem/progenitor cells expected to confer resistance to BCR-ABL-targeted therapeutics. From a study of 25 initially chronic-phase patients, we now demonstrate that some, but not all, of these parameters correlate with subsequent clinical response to IM therapy. CD34(+) cells from the 14 IM nonresponders demonstrated greater resistance to IM than the 11 IM responders in colony-forming cell assays in vitro (P < .001) and direct sequencing of cloned transcripts from CD34(+) cells further revealed a higher incidence of BCR-ABL kinase domain mutations in the IM nonresponders (10%-40% vs 0%-20% in IM responders, P < .003). In contrast, CD34(+) cells from IM nonresponders and IM responders were not distinguished by differences in BCR-ABL or transporter gene expression. Interestingly, one BCR-ABL mutation (V304D), predicted to destabilize the interaction between p210(BCR-ABL) and IM, was detectable in 14 of 20 patients. T315I mutant CD34(+) cells found before IM treatment in 2 of 20 patients examined were preferentially amplified after IM treatment. Thus, 2 properties of pretreatment CML stem/progenitor cells correlate with subsequent response to IM therapy. Prospective assessment of these properties may allow improved patient management.

Figure 1

Comparison of the effect of different concentrations of IM on colony formation by CFCs from IM responders and nonresponders. CD34⁺ cells from 11 patients classified clinically as imatinib mesylate (IM) responders and 14 as nonresponders were assayed in duplicate 1-mL methylcellulose cultures containing 0, 1, 5, or 10μM IM. Two weeks later, the colonies were enumerated to determine the numbers derived from BFU-E, CFU-GEMM, or CFU-GM. The total number of colonies obtained in the IM-supplemented cultures was then calculated as a percentage of those obtained in the absence of IM. (♦) Results obtained from the 3 patients who rapidly developed BC; and cross bars, mean plus or minus SEM of data for each patient group. Differences between the results for the IM responders and nonresponders at all doses of IM tested are significant (P < .001).

Figure 2

Comparison of the effect of IM in vitro on different subtypes of CD34⁺ cells from IM responders and nonresponders. (A) Comparison of the effect of 5μM IM on the ability of different subtypes of CFCs in unmanipulated CD34⁺ CML cells from IM responders and nonresponders to form colonies in vitro (same experiments and method of data calculation as shown in the middle panel of Figure 1). Differences between the results for BFU-E and CFU-GM from IM responders and nonresponders are significant (P ≤ .005). (B) The effect of 5μM IM on the patient-specific yield of CFCs in 3-week liquid suspension cultures initiated with the same cells as plated directly in methylcellulose in panel A. (♦) Results obtained from the 3 patients who rapidly developed BC. No significant differences were found between the effect of IM on CFC production in 3-week cultures initiated with CD34⁺ cells from the 2 patient groups (P = .33). In both panels, cross bars represent the mean plus or minus SEM of data for each patient group.

Figure 3

Comparison of BCR-ABL and transport gene transcript levels in CD34⁺ cells from IM responders and nonresponders. (A) Comparison of BCR (left panel) and BCR-ABL (right panel) transcript levels in CD34⁺ CML cells from IM responders (n = 9) and nonresponders (n = 14) relative to BCR transcript levels in CD34⁺ cells isolated from normal persons (n = 6). (B) Comparison of ABCG2 (left panel), ABCB1/MDR (middle panel), and OCT1 (right panel) transcript levels relative to GAPDH transcripts in CD34⁺ CML cells from IM responders and nonresponders. (♦) Results obtained from the 3 patients who rapidly developed BC. Each data point represents the average of a triplicate quantitative RT-PCR measurement of each transcript normalized as described in “Methods.” Cross bars represent the mean ± SEM of data for each patient group. The differences between the results for IM responders and nonresponders are not significant (P > .05).

Figure 4

Comparison of the frequency and characterization of mutant BCR-ABL transcripts in CD34⁺ cells from IM responders and nonresponders. (A) BCR-ABL kinase domain transcript fragments in extracts of patients CD34⁺ cells were cloned and sequenced (10-30 clones per sample), and the frequency of mutant clones detected in each patients' sample was then calculated. Cross bars indicate the mean plus or minus SEM of data for each patient group. The difference between the frequencies of mutant transcripts detected in the IM nonresponders' cells (10 patients) and the 10 responders' cells (10 patients) is significant (P < .003). No mutations were found in CD34⁺ BM cells from 6 normal persons. (B) Locations of the single amino acid substitutions identified in relation to subdomains of the BCR-ABL kinase domain. The prevalence of each mutation among the 20 patients studied is shown as the proportion of patients in which each mutation was seen. Red letters indicate amino acid changes previously associated with IM resistance in patients. P-Loop indicates phosphate-binding loop; C-helix, catalytic domain; SH3 contact, Src homology 3 domain contact; IM binding site, imatinib binding site; SH2 domain, Src homology domain contact; and A-loop, activation loop. (C) Localization of the residue V304 based on the crystal structure of the c-ABL kinase domain in complex with IM (PDB code 1IEP). Substitution of V304 with a polar residue aspartic acid (V304D) would perturb the conformation and/or structure of the helix that forms part of the IM-binding groove. (D) Frequency of various mutant BCR-ABL transcripts in CD34⁺ CML cells before and after IM treatment from 2 IM nonresponders calculated as described for panel A.

Figure 5

Detection of mutant ATP1A1 transcripts in CD34⁺ cells from IM nonresponders. (A) Structure of the ATP1A1 gene and location of 3 forward and reverse primer pairs used to generate 3 overlapping fragments for PCR amplification of the entire ATP1A1 transcript cDNA (3.7 kb). Frameshift mutations (in red) that generate new amino acid sequences and premature stop codons (*) within the ATP1A1 gene were identified in CD34⁺ CML cells from all 3 IM nonresponders studied. (B) Specific ATP1A1 mutations identified in the pretreatment CD34⁺ cells from the 3 IM nonresponders studied. The mutations detected within the BCR-ABL tyrosine kinase domain from the same patients' cells are also indicated.