Pharmacogenomic markers of glucocorticoid response in the initial phase of remission induction therapy in childhood acute lymphoblastic leukemia

Acute lymphoblastic leukemia (ALL) is the most common hematological and overall malignancy in pediatrics, accounting for around 30% of all childhood cancers and around 80% of all childhood leukemias. It is one of the pediatric malignancies with the highest cure rate, exceeding 80%, when treated with standardized protocols like the European standard, the Berlin-Frankfurt-Munster (BFM) protocol.1, 2, 3 However, there is still more than 10% of patients with unfavorable outcome. The treatment of childhood ALL is based on risk stratification. Patients can be classified into groups according to the features that have been shown to affect prognosis and risk of treatment failure. In time, more elements are considered in order to modulate the treatment protocols and make them more efficient. Implementation of pharmacogenomics in the childhood ALL therapeutic strategy is the most promising approach to improve the outcome of childhood ALL.4

The four main components of ALL therapy are remission induction, consolidation, maintenance, and central nervous system-directed therapy. According to the BFM protocol, in the initial phase of the remission induction treatment of childhood ALL, glucocorticoid (GC) monotherapy is administered during the first 8 days. Its goal is to lower the number of lymphoblasts since GC have the ability to induce apoptosis in leukemic cells mediated through the glucocorticoid receptor (GR).5 The lymphoblast count on the day 8 is one of the stratification criteria important for therapy regime and survival.6 If the blast count in blood is below 1000/microL, the patient is declared as a GC sensitive patient or a prednisone good responder (PGR). If the peripheral blast count of a patient remains over 1000/microL, the patient is declared as GC resistant patient or a prednisone poor responder (PPR) and this is associated with a poor prognosis.

The mechanism of GC resistance in childhood ALL is still poorly understood, but genetic factors might play an important role.7, 8, 9 Therefore, it is of great importance for better treatment of childhood ALL to investigate, understand and overcome the problems related to pharmacogenomics profile of patients with a poor response to the initial GC treatment.

The glucocorticoid receptor gene (NR3C1) codes the GR, which is essential for the effects of glucocorticoids to manifest. Several NR3C1 variants, leading to altered sensibility of GR to glucocorticoids have been studied in pediatric diseases. Most frequently studied variants, like rs6189/rs6190 (ER22/23EK) and rs56149945 (N363S) have not shown significant association with the response on the day 8, when it comes to the therapeutic response to glucocorticoids in ALL.10, 11 One extensively studied variant, rs41423247 (BclI polymorphism), has shown association with the therapeutic response.12

Three variants in the NR3C1 gene, rs33389 (c.1185-6766C>T), rs33388 (c.1185-3562A>T) and rs6198 (c.^*3833A>G) have not been widely studied as pharmacogenomics markers in childhood ALL. The first two variants are located in intron 2, where they can alter consensus recognition sites for RNA splicing factors.13 If the minor rs33389 T allele and the major rs33388 A allele are present, alternate splicing occurs and an isoform of GR with lower affinity for glucocorticoids is expressed in a higher degree.14 In the pediatric nephrotic syndrome, the steroid response was affected by the presence of these two alleles in intron 2.15 The rs6198 variant is located in the 3′ UTR region exon 9β, in the “ATTTA” motif of an isoform of GR with drastically lower affinity for glucocorticoids.16 If the minor rs6198 G allele is present, the mRNA becomes more stable and it leads up to greater translation of the isoform of GR with lower affinity for glucocorticoids.17

Three glutathione S-transferase (GST) genes (GSTP1, GSTT1, and GSTM1) code the GST proteins, which are essential for GC elimination by making its first step, conjugation, possible.18, 19 Null-allele variants of GSTM1 and GSTT1 caused by a deletion of the gene, result in the absence of activity of these enzymes. Additionally, it was reported that GSTP1 gene variants rs1695 (c.313A>G, p.Ile105Val) and rs1138272 (c.341C>T, p.Ala114Val) influence the activity and the structure of GSTP1 and alter the efficiency of GC conjugation, if the minor alleles are present.20 An association between the rs1695 variant and GC response was found in ALL.21

The multidrug resistance 1 gene (MDR1, also known as ABCB1), encodes for a membrane transporter P-glycoprotein (P-gp), responsible for the efflux of chemotherapeutic agents used in leukemia therapy.22 Glucocorticoids are substrates of P-gp, which transports glucocorticoids out of cells. Overexpression of P-gp could mediate GC resistance.23 When considering ABCB1, three variants were often analyzed as pharmacogenomics markers for GC response (rs1128503 (c.1236C>T, p.Gly412=), rs2032582 (c.2677G>A/T, p.Ser893Ala) and rs1045642 (c.3435TC>T, p.Ile1145=)). The rs2032582 variant is a missense mutation, while rs1045642 is a synonymous mutation which leads to decreased expression of ABCB1 gene on the intestinal cell membranes.24 It was found that the steroid response in children with nephrotic syndrome varied based on the expression of ABCB1 gene.25

There have been a few reports which dealt with the topic of pharmacogenomics of GC resistance in adult leukemias, but they lacked conclusive evidence of a single contributing mechanism.26 The topic of pharmacogenomics of GC resistance in ALL, when it comes to the pediatric population, has not been sufficiently studied. In the reported results, only tendencies towards association with GC response for certain genotypes27 have been found, while most of the genetic variants, shown to be relevant for GC response, have never been studied in childhood ALL.

The aim of this study is to investigate the association between variants in NR3C1, GSTP1, GSTT1, GSTM1 and ABCB1 genes and GC therapeutic response in the initial phase of remission induction therapy of pediatric ALL patients. Also, we aimed to investigate if the analyzed pharmacogenomics markers could be helpful to achieve improved personalization of GC therapy, leading to more individualized approach. Namely, other values than 1000 of blast number on day 8 might be potentially used as a marker of therapy efficacy. For example, it has been reported that childhood ALL patients who has zero blasts on day 8 (blast negative), have longer disease-free survival than patients with detectable blasts (blast positive).28 In order to better characterize GC response on day 8 related to analyzed genetic variants, we carried out additional analysis in which cut-off value for GC response was 100 or 0 blasts in peripheral blood. By understanding the factors which contribute to GC resistance or good response, predictions could be made for an individual patient before the initial treatment, in order to use the adequate treatment regime and increase the chances of more efficient GC response.

Pharmacogenomics is dealing with the fact that the efficacy of the drug depends on the patient’s ability to absorb and metabolize the drug, which influences the effectiveness of the treatment. Furthermore, the toxicity of drug depends on the patient’s genome. Pharmacogenomics testing is already incorporated as a dosage-calibrating tool in the maintenance phase of childhood ALL treatment in order to minimize the occurrence of serious toxicities during 6-MP treatment.4, 32

Glucocorticoids are an essential component to induction remission phase of childhood ALL therapy. A poor response to the standard initial GC treatment and the persistence of blast count over 1000 per microliter on the day 8, puts a patient in a higher risk group with a poor prognosis. The following phases of treatment are dependent on risk-directed stratification of patients. However, many children experience severe toxicity associated with treatment with dangerous side effects, while some of them are not cured.33 So, it could be argued that these groupings are not yet comprehensive enough.34 As for induction remission phase of ALL treatment, it is essential to find as many potential markers of GC resistance as possible. By analyzing the associations between the pharmacogenetic variants and GC resistance or good response, this study was meant to contribute to individualization of GC treatment, so that the patients could be in future adequately treated according to their genetic background.

A few studies dealt with variants in NR3C1, GSTs and ABCB1 gene in relation with GC toxicity or disease-free survival in childhood ALL patients, often with conflicting results.10, 35, 36, 37, 38 Although toxicity and survival are the most important therapy outcome signifiers, still, they cannot be associated solely with GC response. On the contrary, we believe that GC response on day 8 assessed by blast count in blood is probably the best measure of GC efficacy in childhood ALL, because no other chemotherapeutic drug is given systemically beforehand. Low blast count (< 100 blasts/microL) or blast negative status could also be important to reveal patients with particularly good response to GC therapy. Those patients might require adjustment of GC dose to achieve remission.

In this study we focused on variants in non-coding region of NR3C1 gene, rather than on the most extensively studied variants of NR3C1 gene that were earlier analyzed in regard to GC response on day 8 in childhood ALL. One of those studies found only BclI variant to be associated with GC response in relatively small Chinese cohort of ALL patients.12 However, other studies did not find significant association with GC response on day 8.10, 11

Concerning the variant rs6198 in the N3RC1 gene, we have found an increased risk of PPR (> 1000 blasts/microL) in the initial stage when the carrier has the rare GG genotype. This variant is important for GRβ mRNA stabilization. Moreover, the GG genotype leads to greater expression of the GRβ isoform.17 And the increased level of GRβ isoform leads to the dominant negative inhibition of the GRα isoform.39 The GRβ isoform provides enhanced resistance to the biological and pharmacological effects of glucocorticoids.14, 16, 40, 41 The level of isoform GRβ was shown to influence glucocorticoid response in childhood ALL. Namely, glucocorticoid sensitivity was negatively correlated with GRβ/GRα ratio in leukemic blast cells.42 Our study is the first to report any result concerning association between rs6198 variant and response to GC treatment on day 8. The association of this variant and glucocorticoid response was shown in patients suffering from other diseases. In the pediatric nephrotic syndrome, it was found that carriers of the GG genotype had a worse treatment outcome,43 which is in line with our findings. Also, in the major depressive disorder, a haplotype (rs10482605-rs6198) containing the G allele of rs6198 was associated with GRβ mRNA stability. This haplotype contributed to the hyperactivity of the hypothalamus-pituitary-adrenal axis.44

Our results have shown that carriers of minor NR3C1 rs33389 T allele tended towards higher blast count (≥ 100 blasts/microL), while carriers of NR3C1 rs33388 T allele tended towards lower blast count (< 100 blasts/microL) at day 8 of GC treatment. It has been shown that the variants rs33389 and rs33388, T and A alleles respectively, are located in intron 2 of NR3C1, in a region where alternate splicing occurs, resulting in increased expression of isoform GRγ.14 GRγ has an affinity for the ligand similar to the standard isoform GRα, but it lacks the stability of GRα in binding to the glucocorticoid response element.45 On the other hand, rs33389 C allele and rs33388 T allele are parts of ACT (rs41423247-rs33389-rs33388) haplotype which is strongly associated with glucocorticoid sensitivity.46 Also, in the pediatric nephrotic syndrome, a significant association was shown between this haplotype and a good response to GC treatment.15 Moreover, CTA (rs33389-rs33388-rs6198) haplotype consisting of alleles found to be favorable for GC response on day 8 in our study, was associated with longer survival time in acute leukemia patients who underwent hematopoietic stem cell transplantation.47 Interestingly however, in our cohort, CAA haplotype was associated with PGR. This result further points out favorable association of rs33389 C and rs6198 A alleles with lower blast count.

Carriers of GSTP1 GC (rs1695-rs1138272) haplotype had decreased risk of PPR, were more likely to have low blast count (< 100 blasts/microL) and to be blast negative on day 8 of GC treatment. It was shown, while investigating the activity and the structure of GSTP1, that this haplotype codes the substrate binding region, H-site, of the GSTP1 protein, turning it into a protein with a much smaller Michaelis constant, leading to less efficient conjugation of agents.20 Consequently, glucocorticoid agents are capable of acting for a longer period of time. Our results have also associated GSTP1 rs1138272 T allele carriers and GSTP1 GT haplotype with blast positive status. Two studies that dealt with variants in GSTP1 gene and GC response on day 8 of ALL treatment, did not find significant association, but they enrolled relatively small number of patients.21, 48

Carriers of the GSTT1 null-genotype were more likely to have higher blast count on day 8 in our childhood ALL cohort. In contrast to our result, one study did not find any association,21 while the other observed statistical trend towards a PGR in childhood ALL.48 Also, Meissner and coworkers found that in subgroup of childhood ALL patients who were in higher risk for PPR, GSTT1 null allele is correlated with decreased risk of PPR.27 When it comes to risk of relapse and outcome in relation with GSTT1 null genotype, conflicting results were noted in two studies that enrolled large number of childhood ALL patients.36, 38

Regarding ABCB1 gene, we found that carriers of rare CGT (rs1128503-rs2032582-rs1045642) haplotype are more likely to be blast positive. Higher expression of ABCB1 was associated with steroid resistance.25, 49 Mayor alleles were found to lead to higher ABCB1 expression or higher ABCB1 activity,50 making them more likely to be associated with poor GC response.51 Our study is the first to deal with GC response on GC treatment day 8 of childhood ALL patients regarding ABCB1 haplotypes. In a large cohort of idiopathic thrombocytopenic purpura patients, various haplotype combinations of the same variants we analyzed were associated with GC response.51 However, no association was found in regard to CGT haplotype.

Despite the promising results, the limitations of the study need to be affirmed. The sample size is not big, since this is a single centric study enrolling patients suffering from a rare disease. Moreover, certain alleles of genetic variants we studied are not frequent, meaning that in some cases there are only a few carriers of certain genotypes. As a consequence, conclusions drawn analyzing such small groups of patients need to be taken with caution. Considering the shortcomings mentioned, it would be of great benefit to validate the results gained in this study on a larger sample preferably using prospective approach.

Association studies on the pharmacogenomic profile of patients and data on the toxicity of drugs are the most promising directions on the road to personalized medicine. The ultimate goal of the ongoing multicentric clinical trials is to optimize the use of known antileukemic drugs in the context of individual pharmacogenomic profile of each patient and molecular markers of the leukemic cells and modulate the treatment resulting in less toxicity and adverse reactions, and a higher survival rates.52 Personalized medicine approach of tailoring treatment to the individual characteristics of each patient has been a great success in several diseases. One thing that we have learnt from those successful examples is that a personalized childhood ALL approach implementation may be difficult. Our study pointed out the association between several variants in NR3C1, GSTP1, GSTT1, GSTM1 and ABCB1 genes and GC therapeutic response in the initial phase of remission induction therapy of pediatric ALL patients. We have shown that NR3C1 rs6198 variant and GSTP1 rs1695-rs1138272 haplotype are the most promising pharmacogenetic markers of GC response in ALL patients. However, studies including more childhood ALL patients, as well as more comprehensive analysis of personal “pharmacomics” profiles are needed for discovery of novel potential genetic markers for targeted therapy53 and for a design of modulations of the existing treatment protocols, leading to more individualized and more successful childhood ALL treatment.