A polymorphism in the XPD gene predisposes to leukemic transformation and new nonmyeloid malignancies in essential thrombocythemia and polycythemia vera

Essential thrombocythemia (ET) and polycythemia vera (PV) are myeloproliferative neoplasms (MPNs) characterized by an indolent clinical course, a tendency to develop thrombohemorrhagic complications, and a risk of transformation into acute myeloid leukemia (AML) sometimes preceded by a phase of myelofibrosis or myelodysplastic syndrome.¹  Leukemic transformation of ET and PV is associated with an ominous prognosis and has been reported to occur in 5%-10% of patients 10 years from diagnosis, with its incidence increasing with disease duration.^2,,–5  A recent study by the French Polycythemia Study Group identified AML as the first cause of death in patients in whom PV was diagnosed before the age of 65 and who were therefore followed up for a long period of time.⁶ 

Although the pathogenic mechanisms underlying the progression to AML remain largely unknown, there is considerable evidence supporting the idea that some cytoreductive therapies used in ET and PV can contribute to this complication.^3,5,7,8  Nevertheless, a significant number of patients who develop AML have either never been exposed to cytoreductive agents or the cumulative doses that they have received are too low to be considered as leukemogenic.⁷  This observation emphasizes the hypothesis that individual factors not related to the treatment could play a significant role in the leukemic transformation of ET and PV.

Recently, concern has been raised on the increased risk of new nonmyeloid cancers in MPN patients.^9,–11  In a nationwide registry-based study from Denmark, the standardized incidence ratio of nonhematologic cancer was 1.2 for ET patients and 1.4 for PV patients.⁹  A study from Italy reported a 3.4-fold increased risk of developing lymphoid malignancies in patients with ET or PV compared with the general population.¹¹  Interestingly, the risk was significantly higher in the JAK2V617F-mutated patients. Although the increased risk of new malignancies may be related to the cytoreductive drugs used for the treatment of ET and PV, the evidence supporting such a relationship is not conclusive and patient-related factors might also be involved.

Hereditary genetic defects in the cellular mechanisms of DNA repair increase the susceptibility to cancer.¹²  The nucleotide excision repair (NER) and the base excision repair (BER) pathways play a major role in cell protection against genotoxic damage by repairing DNA lesions such as those induced by UV irradiation or chemical carcinogens.^13,14  Inherited variations in DNA-repair efficiency have been implicated in the predisposition to de novo¹⁵  and therapy-related AML^16,17  and increased susceptibility to a variety of nonhematologic cancers.^18,19  Therefore, it is reasonable to hypothesize that common polymorphisms in genes encoding for proteins of the NER and BER pathways could influence on the likelihood of developing cancer in patients with ET and PV. To test this hypothesis, in the present study, we investigated the association between 5 single nucleotide polymorphisms (SNPs) of 4 critical genes involved in NER and BER mechanisms and the risk of either leukemic transformation or development of new nonmyeloid malignancies in patients with ET and PV.

Case-control studies for the predisposition to either leukemic transformation or new primary nonmyeloid malignancies were nested within a cohort of white patients diagnosed with ET or PV. The study population consisted of 64 patients who progressed to AML and 358 ET/PV patients who did not progress. Patients were accrued from several Spanish hospitals on the basis of the availability of DNA samples for genotyping, and were followed up for a median time of 8.6 years (range, 1-37). Clinical and hematologic features at diagnosis of ET and PV, as well as treatment-related data and the incidence of new nonmyeloid malignancies, were obtained from the clinical records. In every patient, the diagnosis of ET and PV was reassessed using the updated criteria of the World Health Organization.²⁰  The main characteristics of these patients at diagnosis of ET and PV are summarized in Table 1. The study was approved by the ethics committees of all participating institutions and conducted according to the Declaration of Helsinki.

An analysis of 5 SNPs located in 4 DNA-repair genes involved in the NER (ERCC2 [also known as XPD], ERCC5 [XPG], and XPC) and BER (XRCC1) pathways was performed. The selected SNPs have high heterozygosity and are located on codifying regions involving an amino acid change. The 5 candidate SNPs were: XPD Lys751Gln (rs13181), ERCC5 Asp1104His (rs17655), XPC Ala499Val (rs2228000), XPC Lys939Gln (rs2228001), and XRCC1 Arg399Gln (rs25487; Table 2).

The DNA samples of the MPN patients with leukemic transformation were obtained from the chronic phase of the disease (n = 17), the myelofibrotic phase (n = 4), and the leukemic phase (n = 43). Genotyping analysis of the SNPs was performed by real-time PCR using the TaqMan SNP Genotyping on Demand Assays, which are commercially supplied by Applied Biosystems (Life Technologies). Assays were performed according to the manufacturer's instructions. Briefly, each sample reaction was composed of 2.5 μL of TaqMan Genotyping Master Mix, 0.12 μL of TaqMan probe assay 40×, and 2.5 μL of DNA sample at 5 μg/mL. Thermal cycling and detection was performed in a Fast-Real time PCR system 7900HT from Applied Biosystems (Life Technologies). Thermal cycler conditions were: a first stage of 50°C for 2 minutes, a second stage of 95°C for 10 minutes, and a third stage consisting on 45 cycles of 95°C for 15 seconds, 60°C for 1 minute.

Genotyping of intronic JAK2 SNP (C/G; rs12340895) was performed using the TaqMan SNP genotyping assay referenced as C_31941686_10, supplied by Applied Biosystems (Life Technologies). The JAK2 46/1 haplotype was tagged by the G allele at the JAK2 SNP, as described previously.²¹  The detection of the JAK2V617F mutation was based on quantitative allele-specific PCR technology with the JAK2 MutaQuant kit (IPSOGEN). A sample was considered JAK2V617F positive when the mutated DNA rate was higher than 0.21. JAK2 exon 12 mutation screening was performed by direct sequencing using standard methods.

The statistical tools for genotype analysis of SNPs (Hardy-Weinberg equilibrium, allele and genotype distributions, and association tests) were provided by SNPStats.²²  This web-based application generates odds ratios (ORs), 95% confidence intervals (CIs), and P values for multiple inheritance models (codominant, dominant, recessive and log-additive). The Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) were calculated to select the best inheritance model for each specific polymorphism, with the preferred model being the one with the lowest AIC/BIC value. Because SNPStats calculates unconditional ORs, logistic regression conditional on the matched groups of cases and controls was performed when appropriate.

For each case patient who progressed to AML, 2-5 ET/PV control patients were randomly selected from the subgroup that did not progress and matched to the case by type of MPN (ET or PV) and duration of follow-up. The latter was done by constraining the matching criteria so that dates of diagnosis and last follow-up of the controls selected for a given case anteceded ET/PV diagnosis and postdated AML transformation, respectively, of that given case. This matching criterion guaranteed that controls had the same opportunity as cases to progress to AML and also that every case and his/her controls were followed up roughly during the same calendar years.

The association between candidate SNPs and risk of leukemic transformation, adjusted for other potentially predisposing factors, was assessed by logistic regression conditional on the set of matched cases and controls. Clinicohematologic data at diagnosis of ET/PV and treatment-related covariates investigated for their association with progression to AML were selected on the basis of having been identified as predictors of leukemic transformation in previous studies.^2,–4,23  These included age (categorized by quintiles), sex, leukocytosis (dichotomized at 10 × 10⁹/L or 15 × 10⁹/L), cytogenetic abnormalities, JAK2 mutational status, and exposure to hydroxyurea (HU) and/or other cytoreductive agents.

Controls (n = 2-5) for each patient diagnosed with a new primary nonmyeloid malignancy were randomly selected on the basis of having a follow-up time longer than the time elapsed from the diagnosis of ET/PV to the new primary cancer in the case patient. Matching was based on dates as detailed in the preceding paragraphs and ensured that controls had the opportunity of developing a new primary cancer at the time when the incidence case occurred. The association between candidate SNPs and risk of new primary cancer was then adjusted through conditional logistic regression for age, sex, type of MPN (ET/PV), JAK2 mutational status, JAK2 SNP genotype (rs12340895), and previous exposure to cytoreductive therapies.

Potential synergisms among risk factors (eg, SNPs and exposure to cytoreductive agents) were investigated by including the corresponding interaction terms in the regression models. To safeguard against associations occurring by chance because of multiple simultaneous tests, the cutoff values for the z test were Bonferroni adjusted by dividing 0.05 by the number of covariates included in each regression model. The descriptive comparison of variables between cases and controls was done by univariate conditional logistic regression. Statistical analyses were performed using the SPSS Version 15.0 software package and the Stata Version 11.1 software (www.stata.com).

The case-control study included the 64 patients with ET or PV who evolved into AML and 271 who did not. The characteristics of the case and control patients are listed in Table 3. Median follow-up time to diagnosis of AML in case patients or to death or last follow-up in control patients was 9.3 years (range, 1.2-29) and 12.3 years (range, 2.5-37), respectively. As can be seen in Table 3, case patients were older than control patients, had higher WBC counts at presentation, more frequently displayed a JAK2-mutated gene, and were more exposed to cytoreductive agents. Nevertheless, they accumulated a significantly lower dose of HU than control patients, mainly because of the shorter treatment duration on this drug.

Among the presenting clinicohematologic features, only age, categorized by quintiles, was independently associated with an increased risk of AML (OR = 1.4; 95% CI, 1.1-1.7; P = .008 adjusted for sex and leukocytosis). Because there were relatively few patients who received radioactive phosphorus (³²P), busulfan, melphalan, or cytotoxic agents other than HU, which precluded to accurately ascertain the individual contribution of each agent to the risk of AML, exposure to cytoreductive therapies was graded as “no exposure,” “HU only,” and “other agents alone or in combination” (Table 3). For the purposes of the present study, patients who received IFN or anagrelide as the only cytoreductive treatment were included into the “no exposure” group because these drugs are widely regarded as nonleukemogenic. Taking the “no exposure” category as the baseline risk, the OR for the association with AML increased by 3.2 (95% CI, 1.7-5.8; P < .001) through the other 2 categories, reflecting the increasing frequency of AML across the 3 levels of exposure to cytoreductive agents (Table 4). Age lost its predictive value for leukemic transformation when it was adjusted for the categories of exposure to cytoreductive agents. Further investigation of this finding unveiled a strong association between both covariates, with exposure to cytoreductive agents being more frequent in the older age groups (data not shown).

All candidate SNPs in DNA-repair genes (XPD Lys751Gln, ERCC5 Asp1104His, XPC Ala499Val, XPC Lys939Gln, and XRCC1 Arg399Gln) were successfully genotyped in more than 95% of the study samples. Genotypic distribution of the 5 SNPs was found to be in Hardy-Weinberg equilibrium. The only DNA-repair gene polymorphism independently associated with leukemic risk was XPD Lys751Gln (Table 3). Although the inheritance model with the lowest AIC/BIC at the unmatched analysis was the log additive, at the conditional logistic regression analysis, the best fit was obtained by the recessive model. Therefore, patients homozygous for the C minor allele (Gln/Gln genotype) in XPD had a significantly higher risk of AML (OR = 4.4; 95% CI, 1.8-10.9; P = .001 adjusted for the other candidate SNPs) than carriers of the A wild-type allele (Lys/Lys and Lys/Gln genotypes). Interaction between the XPD SNP and other SNPs did not modify the risk of progression to AML significantly. No definite baseline clinicohematologic profile was noted in patients carrying the minor allele of XPD in homozygosis. Moreover, the median time from the diagnosis of ET/PV to leukemic transformation did not differ between the different XPD genotypes.

JAK2 SNP (rs12340895) was also tested with regard to its putative influence on the risk of progression to AML and no association between this SNP and the risk of leukemic transformation was found.

At the multivariate analysis, both the XPD genotype and the category of exposure to cytoreductive agents emerged as independent predictors of the progression to AML (OR = 4.9; 95% CI, 2.0-12; P = .001, and OR = 3.5; 95% CI, 2.0-6.2; P < .001, respectively). No significant interaction was observed between the covariates, and the higher risk driven by the Gln/Gln genotype in XPD SNP grew in parallel across the 3 categories of increasing exposure to cytoreductive agents (Table 4).

Finally, a similar case-control analysis including the 29 patients who progressed to myelofibrosis and 139 control patients who did not was performed. Neither the clinicohematologic features at diagnosis of ET/PV nor the SNP genotypes predicted for the risk of progression to myelofibrosis.

Median patient age at the time of leukemic transformation was 71 years (range, 31-86). Acute leukemia occurred abruptly in 47 patients (73%), whereas a preceding diagnosis of myelodysplasia or myelofibrosis was made in 5 and 12 instances, respectively. All patients diagnosed with myelodysplasia developed AML soon after (median period, 5 months; range, 2-7). In contrast, the interval between diagnosis of myelofibrosis and leukemic transformation was more variable, ranging from 3-99 months (median, 12).

Information on the karyotype at diagnosis of AML was available in 44 patients. Among them, only 6 (14%) had no cytogenetic abnormalities, whereas 28 (64%) exhibited a complex karyotype (defined as 3 or more unrelated abnormalities). The most common individual cytogenetic aberrations involved total or partial deletions of chromosomes 5 [del(5), n = 5; del(5q), n = 16] and 7 [del(7), n = 13; del(7q), n = 4]. Thirteen patients had abnormalities involving chromosome 17p. Three of the 4 patients with del(7q) had the Gln/Gln genotype in the XPD SNP (P = .014). There was no significant association between the XPD genotype and any other cytogenetic feature.

Median survival after AML transformation was 2.3 months, with 82% of the patients dying within the first 6 months (supplemental Figure 1A, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Three patients were censored at the time of allogeneic hematopoietic stem cell transplantation. Survival after leukemic transformation was not influenced by the XPD SNP (median of 2.2 months for Gln/Gln vs 2.7 months for the Lys/Lys and Lys/Gln genotypes, P = .9; supplemental Figure 1B).

The case-control study included the 50 ET/PV patients who received a diagnosis of new primary nonmyeloid malignancy during follow-up and 221 control patients who did not. The characteristics of the cases and controls are summarized in Table 5. Cases were on average a decade older than controls (64 vs 53 years; P < .001), and more frequently harbored the JAK2 mutation and the minor allele (G) of JAK2 SNP in homozygosis. Median follow-up from diagnosis of ET/PV to recognition of the new cancer in case patients or to death, AML, or last follow-up in control patients was 7 years (range, 0.02-29) and 11 years (range, 1.9-37), respectively.

A significant association was found between XPD SNP and the incidence of new nonmyeloid cancer for the recessive model, in which homozygous patients for the C minor allele (Gln/Gln genotype) had a higher risk of cancer than carriers of the A wild-type allele (Lys/Lys and Lys/Gln genotypes; OR = 3.9; 95% CI, 1.5-10.6; P = .007 adjusted for the other candidate SNPs). Interaction between the XPD genotype with other candidate SNPs did not modify the risk of new cancer significantly.

Multivariate analysis of the clinical, therapy-related, and genetic data, showed that only age, categorized by quintiles (OR = 2.0; 95% CI, 1.4-2.8; P < .001) and the Gln/Gln genotype in XPD (OR = 4.2; 95% CI, 1.5-12; P = .007) were independent predictors of the risk of developing a new primary nonmyeloid malignancy. Exposure to cytoreductive therapies neither predicted the occurrence of new cancer nor potentiated the risk associated with the Gln/Gln genotype in XPD SNP. We also failed to show any significant interaction between age and the XPD genotype with regard to the predisposition to develop a new cancer. Presence of the JAK2 mutation retained a trend that was associated independently with the development of a new primary cancer (OR = 2.5; 95% CI, 1.08-5.9; P = .04) after adjustment for age and XPD genotype.

In the present study, a polymorphism in the XPD gene (Lys751Gln) was identified for the first time as an independent risk factor for leukemic transformation in ET and PV. Specifically, homozygous carriers for the minor allele (Gln/Gln) of XPD SNP had a nearly 5-fold higher risk of progression to AML compared with the other XPD genotypes. In addition, the Gln/Gln XPD variant was associated with a 4-fold increased risk of developing a new primary nonmyeloid malignancy during follow-up, regardless of the type of treatment given to the patients. These findings provide further evidence supporting a pathogenetic role for inherited genetic factors in determining the clinical features of the MPNs.^21,24,,–27 

There is little information on the biologic factors predisposing to AML in ET/PV. Because AML is a relatively rare event that often appears late in the course of ET and PV and can be influenced by therapeutic exposure, studies must include large numbers of patients with long follow-up times to allow for the predisposing factor to become apparent. To overcome this limitation, we conducted a nested case-control study in which case patients with ET or PV who developed AML over a 35-year period were compared with control patients who did not progress to AML despite having been monitored for at least as long as the cases. Such a study design has been proposed as an efficient methodology to ascertain genetic predisposition to second neoplasms and its interaction with therapeutic exposures,²⁸  and has been applied recently to the investigation of clinical and therapy-related factors predicting AML transformation in MPNs.⁷ 

Patient age, leukocytosis at diagnosis, and exposure to cytoreductive treatments have been identified previously as risk factors for progression to AML in both ET^2,4,8  and PV.^3,23,29  Nevertheless, the relative contribution of each of these factors is difficult to ascertain because they are often closely interrelated. For example, most elderly patients had been exposed to cytoreductive agents in our series, so this latter covariate abrogated the predictive value of age. Patients with leukocytosis at presentation are likely to start cytoreductive treatment earlier, and therefore the predictive value of each factor is difficult to separate. Persistent leukocytosis, however, has been identified recently as a risk factor for leukemic transformation in PV patients on HU treatment, probably reflecting a higher myeloproliferative potential.³⁰ 

In the present study, we found a significant association between the level of exposure to cytoreductive agents and the risk of AML transformation, with the OR increasing by 3-fold from the group of patients never exposed to cytoreductive agents to the group who received HU alone and to those treated with agents other than HU (mainly ³²P, busulfan, and melphalan). Previous studies have reported an increased risk of AML in MPN patients treated with ³²P,^5,7  pipobroman,^3,6  and alkylating agents,^3,5,7  especially when several cytoreductive drugs are used in combination or sequentially.^{3,5,7,8,31,32}  In contrast, the leukemogenic potential of long-term HU therapy in ET and PV remains controversial. Although there is compelling experimental evidence that HU reduces DNA repair, its mutagenic and carcinogenic potential seems to be low when assessed by in vitro assays.^33,34  In the clinical setting, some studies have failed to find an association between exposure to HU alone and AML transformation,^2,3,7  whereas in others the risk of AML associated with this drug was higher than expected.^6,8  Regarding our present data, patients who progressed to AML accumulated a significantly lower dose of HU than control patients, mainly because they were exposed to the drug for a shorter time. This observation is more consistent with a patient selection bias driven by the indication for starting on HU than with a drug-related, cumulative leukemogenic effect. Alternatively, it could be hypothesized that patients carrying the Gln/Gln genotype in the XPD SNP would be more susceptible to low cumulative doses of HU because of reduced NER function.

Consistent with the above hypothesis, it has been postulated that perhaps only a subset of MPN patients, such as those susceptible to the effects of HU on DNA repair, might be predisposed to leukemic transformation when exposed to this agent.³³  We found an increased risk of AML in patients carrying the Gln/Gln XPD genotype after adjustment for other predisposing factors. Although we could not demonstrate a synergistic effect between the Gln/Gln XPD variant and the degree of cytoreductive exposure on the risk of developing AML, the possibility of such interaction cannot be totally ruled out. First, this genetic variant has been associated previously with chemotherapy-induced AML in patients with solid neoplasms treated with alkylating agents.¹⁶  Second, our study may have lacked enough statistical power to unveil a subtle yet clinically relevant interaction between the XPD gene polymorphism and exposure to specific chemotherapy agents, mainly alkylators, because of the limited number of patients who received these drugs.

With regard to the increased risk of new primary nonmyeloid cancer in patients with MPNs,^9,–11  a possible pathogenic role has been attributed to the cytoreductive agents used in the management of these diseases.^31,32,35  Therefore, studies conducted in the late 1990s reported an excess risk of carcinoma in PV patients treated with ³²P and HU,³²  as well as in ET patients who were sequentially exposed to HU and busulfan.³¹  In addition, long-term use of HU has been associated with the development of nonmelanoma skin cancer in sun-exposed areas,³⁵  which suggests a role for this drug in the defective nucleotide excision repair of DNA damage induced by UV radiation.³⁶  However, no association was observed between cytoreductive exposure and subsequent nonmyeloid cancer in the present study. Although our study may have been underpowered to unfold a weak association, it is worth mentioning that the available evidence supporting a causal role for cytoreductive therapies in the development of nonmyeloid cancer in ET and PV is weak. Indeed, in the abovementioned reports the associations were either scarcely significant^31,32  or based on case-reports³⁵  and the more recent, population-based studies did not investigate the potential causative role of cytoreductive treatments.^9,–11  In contrast, we found in the present study that older age and the Gln/Gln XPD SNP were significant risk factors for the occurrence of new primary nonmyeloid cancer, reinforcing the pathogenic role of this SNP in determining the clinical evolution of ET and PV. We also found that the JAK2 mutation retained a weak association with the risk of nonmyeloid cancer after adjustment for age and the XPD genotype. This finding is consistent with recent data showing an increased rate of lymphoid malignancies among MPN patients with the JAK2V617F mutation,¹¹  as well as a higher incidence of cancer in the general population carrying this mutation.³⁷  Because our study was not population-based, we were unable to ascertain whether the predisposition to nonmyeloid cancer driven by the Gln/Gln variant of XPD gene, and to some extent by the JAK2V617F mutation, is stronger in patients with ET or PV than in the general population.

Both biologic plausibility and epidemiologic data give support to the XPD genotype (Lys751Gln) as a predisposing factor for leukemic transformation and cancer development in patients with MPNs. XPD gene encodes for a DNA helicase involved in at least 3 crucial cellular mechanisms, DNA repair by NER, transcription initiation, and cell-cycle regulation.³⁸  In this process, ATP binding and hydrolysis are critical for the function of the NER helicases because they cause conformational changes that drive the directional movement of the helicase on DNA.³⁹ XPD SNP rs13181 results in a lysine-to-glutamine transition at position 751, which is predicted to induce a major change in the interaction domain between the XPD protein and its helicase activator, the p44 protein.⁴⁰  Although such change does not seem to affect the transcriptional activity of XPD,⁴¹  it is important for nucleotide excision repair.⁴²  It is biologically plausible that individuals carrying the minor allele (Gln) in XPD SNP rs13181 could be more sensitive to DNA damage and therefore more prone to cancer. Indeed, such susceptibility to malignant transformation might be more pronounced in the setting of the MPNs because of their inherent tendency toward leukemic transformation and the frequent use of cytoreductive agents to manage patients with these disorders. A recent meta-analysis of 56 case-control studies concluded that the minor allele of such SNP is associated with cancer susceptibility regardless of environmental factors.¹⁸  Moreover, the Gln/Gln XPD genotype has also been associated with an increased risk of chemotherapy-induced AML,¹⁶  and a higher frequency of AML with adverse cytogenetic features has been noted among carriers of the Gln allele.⁴³ 

In summary, the present study shows that an SNP in the XPD gene predisposes individuals with ET and PV to develop AML and new nonmyeloid malignancies. These results, if confirmed in other series, could allow the identification of high-risk patients who would benefit from close surveillance and individualized therapeutic approaches.

The authors thank the following investigators for sending DNA samples and clinical information for their study: Dr Ismael Buño (Hospital Gregorio Marañón, Madrid, Spain), Dr Joaquín Martínez-López (Hospital 12 de Octubre, Madrid, Spain), Dr Javier López (Hospital Ramón y Cajal, Madrid, Spain), Dr José Román-Gómez (Hospital Reina Sofía, Córdoba, Spain), Dr Francisca Ferrer-Marín (Hospital Morales Meseguer, Murcia, Spain), Dr María Teresa Gómez-Casares (Hospital Dr Negrín, Las Palmas, Spain), and Dr Josefa Marco (Hospital General, Castellón, Spain).

This work was supported by the grants 09/02324 and 10/00236 from the Fondo de Investigaciones Sanitarias, Spanish Ministry of Health. P.A. is currently supported by a research grant from the Fundación Científica of the Asociación Española contra el Cáncer.

Contribution: J.-C.H.-B. designed the research, collected the data, interpreted the results, and wrote the manuscript; A.P. performed the statistical analysis, interpreted the results, and wrote the manuscript; F.C. and A.A.-L. collected the data, interpreted the results, and revised the manuscript; M.C., B.B., I.M., and P.A. performed the molecular studies and approved the final version of the manuscript; E.S., M.J.A., C. Boqué, B.X., and M.M. collected the data and approved the final version of the manuscript; C. Besses interpreted the results and revised the manuscript; and V.G. designed the research, performed the SNP genotyping, interpreted the results, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Vicent Guillem, Hematology Department, Hospital Clínico Universitario, Avd Blasco Ibáñez 17, 46010 Valencia, Spain; e-mail: vguillem@uv.es.