Chronic myeloid leukemia (CML) is a myeloproliferative disease originating in hematopoietic stem-cells. It develops from the oncogenic activity of a 210kD protein (p210^BCR-ABL), which is the product of a chimeric gene BCR-ABL1 with high activity tyrosine kinase (TK). One of the treatments of choice is based on the use of tyrosine kinase inhibitors (TKIs), such as imatinib, which blocks the ATP binding site in the catalytic pocket of the protein BCR-ABL1 [123]. Imatinib prevents the phosphorylation of proteins involved in different signaling pathways which promote cell proliferation and inhibit apoptosis [45]. Despite being a highly specific and effective molecular treatment for inhibiting TK, its therapeutic efficacy can be limited in some patients who develop a refractory phenotype at disease onset or during progression to both the accelerated phase and/or blast crisis [678]. An escalation in the dose of imatinib from 400 mg/d to 600-800 mg/d can be permitted in patients who had an initial suboptimal response or treatment failure, as defined by the European LeukemiaNet [9]. However these therapeutic schedules are not recommended for patients with imatinib intolerance, or who develop bone marrow aplasia as a secondary complication to imatinib and/or interferon treatment [1011].

An alternative is to use other TKIs, such as dasatinib or nilotinib [312]. These are more commonly used in patients with mutations in the BCR-ABL gene which could generate failure of therapeutic response to imatinib. Another proposed TKI drug-resistance mechanism is that this can be generated by an insufficient drug concentration at the intracellular level, as the result of active transport being mediated by the over-expression of the ABC-transporters P-glycoprotein (P-gp) and/or BCRP (breast cancer resistant protein), encoded by the MDR1/ABCB1 and ABCG2 genes respectively [1314]. Variations in single-nucleotide polymorphisms (SNPs) of MDR1/ABCB1 have been described as potential factors related to the clinical-therapeutic evolution of several diseases [1516]. To date, 216 SNPs have been described in the ABCB1 gene (SNPer: http://snpper.chip.org), and several studies have identified polymorphic variants of TT in Exon 21 (T2677T) and in Exon 26 (T3435T) in Caucasian patients, which could also correlate with a phenotype of increased sensitivity to treatment with TKIs [171819]. Other groups have described that ABCB1-SNP: C1236T on Exon 12 could also be related to a higher frequency of major molecular response in patients with CML treated with imatinib [20]. Taking together the above, it was suggested that association of the genotypes from these SNPs of the ABCB1 gene may configure a specific haplotype as a genetic biomarker of prognosis for therapeutic response in CML-patients treated with TKIs [2122]. Here, we studied the SNPs C1236T, G2677T/A, and C3435T of the ABCB1 gene in CML-patients as potential predictive biological markers for therapeutic response and illness evolution.

Twenty four CML-patients, 22 (91.7%) with chronic phase and 2 (8.3%) under blast crisis, were studied. Thirteen (54.2%) were males, and 11 (45.8%) were females. The Sokal risk was low in 7 (29.2%), intermediate in 7 (29.2%), high in 6 (25%), and not evaluated in 4 (16.6%). Age, gender, phase of disease, TKI treatment and doses, molecular responses (at 6, 12, and 18 months) and haplotypes are listed in Table 1.

A total of seventeen different haplotypes were identified in CML-patients and controls. The frequency distribution indicates that 6 were only detected in CML-patients, 7 only in controls, and 4 in both groups. Heterozygous T-variant genotypes were observed in all 25 controls (100%), and in 18 CML-patients (75%). In this last group, 9 different T-variant haplotypes were identified as follows: full mutated-homozygote (TT-TT-TT) in 3 (12.5%), full mutated-heterozygous (CT-GT/A-CT) in 6 (25%), and other combinations of T-variants in 9 cases (37.5%). The remaining 6 (25%) CML-patients were wt-homozygous CC-GG-CC (Fig. 1). No associations between Sokal risk with haplotypes or therapeutic response were observed. TT-variant distributions for each exon indicate a predominant expression in exon 12 for controls but in exons 21 and 26 for CML-patients (Fig. 2).

The therapeutic responses to different treatments were evaluated at the molecular level at three different end points (6, 12 and 18 months) (Table 1). However, for statistical comparison, only the end point at 12 month was evaluated. A total of 11 cases were treated with imatinib, 6 with nilotinib, and one with dasatinib as first and sole TKI; the other 6 cases were treated with the addition of dasatinib (4) or nilotinib (2) after initial administration of imatinib. Eleven patients received imatinib or imatinib+hydroxyurea (HU), and one of them who had a wt-haplotype (CC-GG-CC), developed bone marrow aplasia. This patient had a complex cytogenetic variant [46XX, t(3q2.5;9q3.4;22q1.1)] described only in a minority (2-10%) of CML-patients. The other ten cases also treated with imatinib were T-variants, and reached an MMR at six months. Similar responses were observed in the six patients (4 T-variants and 2 wt) treated with nilotinib. An earlier MMR response at 3 months was observed in the sole patient treated with dasatinib alone, however this patient relapsed with a minor molecular response (Mi) at 12 months; he also had wt-haplotype CC-GG-CC. The remaining 6 patients were treated with a combination of 2 TKIs as imatinib+dasatinib (N=4) or imatinib+nilotinib (N=2). One of these cases developed a breast cancer, another case in BC died within the first month, and another case that also had the wt-haplotype CC-GG-CC remained with a null response at 18 months (Fig. 3).

From a total of 24 CML-patients studied, 2 were in blast crisis at the time of diagnosis, and presented with haplotypes showing T-variants in all studied exons (CT-GT-TT and CT-GA-CT). This last case, who was the only patient with an allelic variant "A" in exon 21, was in BC and died during the first month after treatment was started. Excluding the BC cases, 16 of 22 CML-patients in the chronic phase were T-variants (heterozygotes and/or homozygotes). One of these cases developed a breast cancer, and so their therapeutic response was not evaluated. The remanding 15 patients were all responders (P<0.001). In the 6 cases with wt-haplotype only 3 (50%) were responders; of the others two had a null molecular response at 12 months and one developed bone marrow aplasia mentioned above (Fig. 4).

CML affects mainly adults with a peak incidence between 46-55 years of age and a homogeneous distribution between genders, and in our study, the age and gender distribution was coincident with previous reports [328]. In our study, the frequency of heterozygous variants on exon 12 (C1236T) was similar to the previous description of Dulucq et al. [20]; however, we observed a higher percentage of these variants for exon 21 (2677G>/TA) (50% vs. 39.5%) and exon 26 (C3435T) (36.4% vs. 18.9%) [20]. Considering the relationship between allelic variants and therapeutic responses, 16/22 cases with CP were mutated (homozygous or heterozygous) and 15 of them (93.8%) reached a good molecular response at 6-18 months after treatment. The remaining case developed breast cancer and her therapeutic response was not evaluated. In the 6 cases in CP with full wt-homozygous haplotypes, 5 had an apparently good cytogenetic response; however, at the molecular level, 2 were refractory to treatment, and in another patient who developed bone marrow aplasia, the therapeutic response at the molecular level could not be evaluated. All this data reinforces the higher sensitivity of molecular methods for the measurement of minimal residual disease, according to the latest international recommendations for the management of CML [2930].

To date, there are few studies of SNPs of MDR1/ABCB1 gene in patients with CML suggesting a potential relationship between genotype/haplotype studies and therapeutic responses when evaluated at the molecular level. Dulucq et al. [20] reported for the first time the association of mutated variants of exons 12 and 21 with an increased number of cases reaching a major molecular response (MMR). Additionally, they observed that wt-haplotypes (CC/GG/CC) had a lower frequency of MMR. Our results were in agreement with these observations. Maffioli et al. [21] state that the presence of a T allele only in exon 26 was associated with a better therapeutic response, while the haplotype characterized by the presence of a T allele in exon 12 and accompanied by mutant variants in exons 21 and 26 (1236T-2677G-3435C) was associated with primary resistance to imatinib. A different conclusion was recently reached by Angelini et al. [22], indicating that the wt genotype (CC) in exon 26 was associated with a complete molecular response, while T-variant haplotypes in all three exons did not have any relation with therapeutic responses. However, our study appears to indicate a better therapeutic response in patients with T-variants present in all these exons, because 93.75% of them reached MMR at 18 months. On the other hand, we observed a potential association between the genotypes and/or haplotypes detected with the risk of developing the disease (Fig. 1 and 2). In this regard, our study demonstrated that the full wt-haplotype CC-GG-CC (6/24 cases) was only present in patients with CML, and this haplotype was related in 3 of them with poor prognoses showing minor (one case) or null (one cases) responses at 18 months, and developing bone marrow aplasia in one case. Interestingly one other case had a null response during the first 12 months, but reached an MMR at 18 months.

In conclusion, we suggest that combined evaluation of haplotypes and genotypes of the SNPs of the MDR1/ABCB1 gene on exons 12, 21 and 26 could be a useful predictive tool for both therapeutic response as well as the risk of developing and the clinical evolution of CML.