Low 25(OH) vitamin D3 levels are associated with adverse outcome in newly diagnosed, intensively treated adult acute myeloid leukemia

Authors


Abstract

BACKGROUND

Several studies have suggested that low 25(OH) vitamin D3 levels may be prognostic in some malignancies, but no studies have evaluated their impact on treatment outcome in patients with acute myeloid leukemia (AML).

METHODS

Vitamin D levels were evaluated in 97 consecutive, newly diagnosed, intensively treated patients with AML. MicroRNA expression profiles and single nucleotide polymorphisms (SNPs) in the 25(OH) vitamin D3 pathway genes were evaluated and correlated with 25(OH) vitamin D3 levels and treatment outcome.

RESULTS

Thirty-four patients (35%) had normal 25(OH) vitamin D3 levels (32-100 ng/mL), 34 patients (35%) had insufficient levels (20-31.9 ng/mL), and 29 patients (30%) had deficient levels (<20 ng/mL). Insufficient/deficient 25(OH) vitamin D3 levels were associated with worse relapse-free survival (RFS) compared with normal vitamin D3 levels. In multivariate analyses, deficient 25(OH) vitamin D3, smoking, European Leukemia Network genetic group, and white blood cell count retained their statistical significance for RFS. Several microRNAs and SNPs were associated with 25(OH) vitamin D3 levels, although none remained significant after multiple test corrections; one 25(OH) vitamin D3 receptor SNP, rs10783219, was associated with a lower complete remission rate (P = .0442) and with shorter RFS (P = .0058) and overall survival (P = .0011).

CONCLUSIONS

It remains to be determined what role microRNA and SNP profiles play in contributing to low 25(OH) vitamin D3 level and/or outcome and whether supplementation will improve outcomes for patients with AML. Cancer 2014;120:521–529. © 2013 American Cancer Society.

INTRODUCTION

Epidemiologic studies suggest an association between low 25(OH) vitamin D3 levels and acute myeloid leukemia (AML). For example, a study from the United Arab Emirates (UAE)[1] indicated that AML is more common among adult women than among adult men, although the population of the UAE consists of more men than women, and although it is widely known that AML is more common in men. These findings suggest that low vitamin D3 levels secondary to the practice of women wearing extensive body coverage[2] may contribute to the higher incidence of AML.

In addition, in the early 1980s, it was demonstrated that vitamin D differentiated AML cells into mature myeloid cells.[3] That finding suggested that low serum 25(OH) vitamin D3 levels may be associated with enhanced clonal proliferation. It is noteworthy that low serum 25(OH) vitamin D3 levels were associated with inferior event-free survival and overall survival (OS) in patients with diffuse large B-cell and T-cell non-Hodgkin lymphoma (NHL),[4] and vitamin D insufficiency at diagnosis was associated with a decreased time to the initiation of treatment in patients with chronic lymphocytic leukemia (CLL).[5] Therefore, we hypothesized that the vitamin D level at diagnosis may be associated with outcome in intensively treated patients with AML.

Vitamin D predominantly exerts its effects through binding to the cognate nuclear vitamin D receptor (VDR). Ligand-bound VDR heterodimerizes with the retinoic X receptor (RXR) and binds to vitamin D-responsive elements in the promoter regions of target genes, such as cytochrome P450, family 24, subfamily A, polypeptide 1 (CYP24A1); bone γ-carboxyglutamate protein (BGLAP) (osteocalcin); and cyclin-dependent kinase inhibitor 1A (CDKN1A, p21Waf1/Cip1); and with several protein kinase C (PKC) isoforms,[6] including the p42 extracellular regulated kinase (p42 ERK) and c-Jun N-terminal kinase (JNK) families of mitogen-activated protein kinases (MAPKs), which are important in differentiation, metabolism, and the cell cycle.[7-11] We recently demonstrated that microRNA 106b (miR106b), 1 of the small noncoding RNAs (miRs) that are key controllers of cellular function, increases in response to vitamin D exposure.[12] Therefore, we also hypothesized that miR expression would differ among patients with AML who have low vitamin D levels compared with those who have normal vitamin D levels.

Finally, it has been hypothesized that, for individuals with similar vitamin D intake or status, those with VDR or other vitamin D pathway polymorphisms may have increased susceptibility to colorectal cancer risk. However, evidence to date is inconclusive. Phenotypic variations like these often are caused by genetic variations in the polymorphic genes encoding the proteins that biotransform vitamin D, leading to variations in serum 25(OH) vitamin D3 levels.[13, 14] A recent meta-analysis[15] of relevant vitamin D studies demonstrated an inverse association between vitamin D intake, vitamin D status, and the Bsml VDR polymorphism (rs1544410) and colorectal cancer risk. Therefore, we included the hypothesis that single nucleotide polymorphisms (SNPs) in the vitamin D pathway genes may play a role in AML.

MATERIALS AND METHODS

Patients and Treatment

Pretreatment bone marrow, peripheral blood, and serum samples were obtained from 97 patients with AML (excluding those with acute promyelocytic leukemia) ages 19 to 91 years (median age, 60 years) who received intensive first-line therapy with combined cytarabine (100 mg/m2 daily for 7 days), daunorubicin (90 mg/m2 daily for 3 days in patients aged <60 years and 60 mg/m2 daily for 3 days in patients aged ≥60 years), and etoposide (100 mg/m2 daily for 3 days) (ADE). Thirty patients who were in complete remission (CR) received consolidation with high-dose cytarabine; 8 patients received ADE (for 5 days, 2 days and 2 days for cytarabine, daunorubicin, and etoposide, respectively, at the same doses), and the others received miscellaneous regimens as consolidation. Seven patients proceeded to autologous stem cell transplantation, and 16 patients proceeded to allogeneic stem cell transplantation in first CR. All patients provided informed consent to treatment, sample procurement, and further testing; treatments were in accordance with the Declaration of Helsinki and were approved by the Roswell Park Cancer Institute institutional review board. The Roswell Park Cancer Institute Scientific Review Committee and institutional review board approved this study.

25(OH) Vitamin D3 Levels

Serum 25(OH) vitamin D3 levels were analyzed using a standard, commercially available 25-hydroxyvitamin D3-(I125) radioimmunoassay kit from DiaSorin Company (Stillwater, Minn).[16] The lower limit of normal for this assay is 32 ng/mL, which is based on maximum suppression of parathyroid hormone[17]; the normal range is from 32 to100 ng/mL (80-250 nmol/mL), insufficient levels are from 20 to 31.9 ng/mL; and deficient levels are <20 ng/mL.[18] Samples from the healthy volunteers were assayed in the laboratory of Dr. Bruce W. Hollis using the same radioimmunoassay.[17] Serum 25(OH) vitamin D3 measurements and normal ranges in both laboratories were the same.

MicroRNA Profiling

An exploratory analysis of 20 samples (10 with subnormal normal vitamin 25[OH] D3 levels [<32 ng/mL] and 10 with normal or above normal levels) was performed on miR arrays by our core facility using the Exiqon platform (Exiqon, Inc., Woburn, Mass). Samples were labeled with 3H-indocyanine dye (Cy3). Up-regulation of 14 miRs and down-regulation of 3 miRs were detected. An additional 58 samples were studied by reverse transcriptase-polymerase chain reaction (RT-PCR) for 16 miRs (1 miR sequence was not available for study).

Tag Single Nucleotide Polymorphism Selection and Genotyping

Tag SNPs were derived from vitamin D metabolism pathway genes, including 4 cytochrome P450 family members (CYP24A1, CYP27A1, CYP27B1 and CYP2R1) responsible for vitamin D metabolism (hydroxylation), the group-specific component GC (vitamin D binding protein) for transport, and VDR as the target of vitamin D. Briefly, SNP genotype data sets for Caucasians were selected from the National Center for Biotechnology Information and HapMap databases, in addition to resequencing genotype data for CYP24A1 and GC, which were generated similarly to previous reports,[19-21] using Caucasian DNA samples from the Coriell Cell Repository (Camden, NJ). The genotype data were then loaded into the Haploview program (MIT/Harvard Broad Institute, Cambridge, Mass) to derive both haplotype and linkage disequilibrium (LD) tag SNPs for the study.[22] In total, 90 tag SNPs were genotyped on the Sequenom MassARRAY platform (Sequenom, Inc., San Diego, Calif) by our core facility in accordance with the manufacturer's instructions. Controls were included to ensure genotyping accuracy in addition to genotyping approximately 10% of the samples in duplicate.

Statistical Analyses

Outcome analyses

CR, relapse-free survival (RFS), and OS were defined as previously described.[23] Descriptive statistics, such as frequencies and relative frequencies, were computed for categorical variables. Numeric variables were summarized using simple descriptive statistics, such as the mean, standard deviation, median, range, etc. The Fisher exact test was used to study the association between categorical variables. The Wilcoxon rank sum-test was used to compare the groups with regard to numeric variables. The distribution of OS and RFS was estimated using the Kaplan-Meier method. Patients who were alive at last follow-up were censored. By using this distributed estimate, summary descriptive statistics, such as the median survival and 95% confidence interval (CI) of the median survival, were obtained. Statistical assessment of observed differences in the survival distributions of different groups of interest was done using the log-rank test. A Cox proportional hazards model was used to assess the effect of study variables on survival in both univariate and multivariate analyses. The Cox hazard ratio is a standard assessment of the incremental increase in the odds of a patient having a given outcome. A logistic regression model was used to investigate the association between CR status and vitamin D groups. Computations of P values and 95% CIs for the odds ratios were based on the Wald test. A .05 nominal significance level was used in all testing. All statistical analyses were done using the SAS software package (version 9.3; SAS Institute, Inc., Cary, NC).

MicroRNA and single nucleotide polymorphism data analyses

All data analyses were performed using the R programming environment (R Foundation for Statistical Computing, Vienna, Austria; available at: www.r-project.org, accessed August 10, 2013). For miR analysis, we used the t test to calculate the level of differential miR expression between subnormal (<32 ng/mL) and normal (≥32 ng/mL) 25(OH) vitamin D3 levels. For SNP analysis, first, we considered the patients' 25(OH) vitamin D3 level as a continuous variable and used linear regression to examine whether the SNP genotypes were associated significantly with the patients' 25(OH) vitamin D3 level. Then, we separated the patients into 2 groups based on 25(OH) vitamin D3 level (<32 ng/mL vs ≥32 ng/mL) and used logistic regression to examine whether the SNP genotypes were associated significantly with the 25(OH) vitamin D3 group status. Multiple testing corrections were controlled using the approach of Benjamini and Hochberg.[24]

RESULTS

25(OH) Vitamin D3 Levels in Acute Myeloid Leukemia

There were 34 patients (35%) with normal 25(OH) vitamin D3 levels, 34 (35%) with insufficient levels, and 29 (30%) with deficient levels. A similar distribution was observed among a cohort of 100 healthy volunteers from western New York, in which 29 individuals (29%) had normal levels, 40 (40%) had insufficient levels, and 31 (31%) had deficient levels.[25]

Associations of 25(OH) Vitamin D3 Levels With Pretreatment Clinical and Molecular Characteristics

No differences in age or sex were observed between patients who had normal or low 25(OH) vitamin D3 levels; not surprisingly, nonwhite patients (n = 7) tended to have lower 25(OH) vitamin D3 levels (P = .02). Similarly, there were no differences in any of the other pretreatment clinical characteristics, as outlined in Table 1, except that the fms-related tyrosine kinase 3 (FLT3) internal tandem duplication (FLT3-ITD) was rarely present in patients who had normal 25(OH) vitamin D3 levels (P = .02).

Table 1. Patient Characteristics According to Vitamin D Level
 Vitamin D Level: No. of Patients (%) 
Patient CharacteristicDeficient: <20 ng/mLInsufficient: 20–31.9 ng/mLNormal: 32–100 ng/mLP
  1. Abbreviations: AML, acute myeloid leukemia; BM, bone marrow; BMI, body mass index; ELN, European LeukemiaNet; FAB, French-American-British; PB, peripheral blood; WBC, white blood cell.

  2. a

    These P values indicate a statistically significant difference.

  3. b

    Secondary includes both the presence of antecedent hematologic disorder and therapy-related AML.

Total29 (30)34 (35)34 (35) 
Age, y   .71
<6016 (55.2)15 (44.1)16 (47.1) 
≥6013 (44.8)19 (55.9)18 (52.9) 
Sex   .28
Men12 (41.4)21 (61.8)17 (50) 
Women17 (58.6)13 (38.2)17 (50) 
Race   .03a
White23 (82.1)32 (94.1)34 (100) 
Nonwhite5 (17.9)2 (5.9)0 (0) 
WBC count, ×109/L   .32
Median [range]21.8 [1.1–555.2]5.4 [0.7–292.6]15.7 [0.6–186.6] 
Percentage of PB blasts   .82
Median [range]25 [0–96]35 [0–92]36.5 [0–92] 
Percentage BM blasts   .94
Median [range]58 [21–92]68 [18–95]57.5 [20–97] 
FAB category   .15
M02 (7.1)1 (3.1)2 (6.7) 
M17 (25)9 (28.1)9 (30) 
M210 (35.7)6 (18.8)12 (40) 
M48 (28.6)8 (25)2 (6.7) 
M51 (3.6)7 (21.9)4 (13.3) 
M60 (0)0 (0)1 (3.3) 
M70 (0)1 (3.1)0 (0) 
BMI: Lee 2012[26]   .53
Underweight1 (3.4)1 (2.9)2 (5.9) 
Normal10 (34.5)7 (20.6)11 (32.4) 
Overweight6 (20.7)16 (47.1)12 (35.3) 
Obese5 (17.2)6 (17.6)5 (14.7) 
Very obese7 (24.1)4 (11.8)4 (11.8) 
Smoking   .17
Current smoker10 (34.5)10 (29.4)4 (11.8) 
Previous smoker10 (34.5)11 (32.4)11 (32.4) 
Nonsmoker9 (31)13 (38.2)19 (55.9) 
AML presentation   .47
De novo19 (65.5)27 (79.4)26 (76.5) 
Secondaryb10 (34.5)7 (20.6)8 (23.5) 
NPM1   .16
Mutated8 (27.6)8 (25)3 (9.4) 
Wild type21 (72.4)24 (75)29 (90.6) 
FLT3-ITD   .02a
Present7 (24.1)8 (23.5)1 (3) 
Absent22 (75.9)26 (76.5)32 (97) 
ELN genetic group   .45
Favorable4 (16)11 (34.4)7 (21.9) 
Intermediate-I11 (44)10 (31.3)11 (34.4).45
Intermediate-II4 (16)8 (25)6 (18.8) 
Adverse6 (24)3 (9.4)8 (25) 

Associations of 25(OH) Vitamin D3 Levels With Clinical Outcome

The level of 25(OH) vitamin D3 was not associated with the probability of attaining CR (Table 2). At a median follow-up of 15.6 months (range, 0.1-84.3 months) in the patients who remained alive, those who had insufficient and deficient 25(OH) vitamin D3 levels, compared with those who had normal levels, had significantly shorter RFS (P = .025) in Kaplan-Meier analysis; the median RFS was 8.7 months (95% CI, 5.9-64.1 months), 5 months (95% CI, 2.9-12.1 months), and 16.3 months (95% CI, 10.5-41.3 months) for those with insufficient, deficient, and normal levels, respectively (Fig. 1A). In a Cox hazard ratio analysis, a statistical difference was detected only between those who had deficient levels and those who had normal levels (Table 2). There was no significant difference in OS between those who had insufficient and deficient 25(OH) vitamin D3 levels compared with those who had normal levels in Kaplan-Meier analysis; the median OS was 12.5 months (95% CI, 6.7-64.1 months), 9.8 months (95% CI, 2.9-19.3 months), and 25.2 months (95% CI, 14.3 months to not reached) for those with insufficient, deficient, and normal levels, respectively (Fig. 1B). In a Cox hazard ratio analysis, patients who had deficient 25(OH) vitamin D3 levels had a significantly greater OS hazard compared with those who had normal levels (Table 2).

Table 2. Patient Outcome According to Vitamin D Level
 Vitamin D LevelP
EndpointDeficient: <20 ng/mLInsufficient: 20–31.9 ng/mLNormal: 32–100 ng/mLDeficient vs NormalInsufficient vs Normal
  1. Abbreviations: CI, confidence interval; HR, hazard ratio; OR, odds ratio.

  2. a

    These P values indicate a statistically significant difference.

Total no. of patients (%)29 (30)34 (35)34 (35)  
Complete remission: OR [95% CI]1.636 [0.598–4.48]1.267 [0.488–3.289]1.00.3379.6272
Relapse-free survival: HR [95% CI]2.094 [1.183–3.707]1.185 [0.66–2.127]1.00.0112a.5707
Overall survival: HR [95% CI]1.987 [1.094–3.607]1.395 [0.76–2.56]1.00.0241a.2822
Figure 1.

Kaplan-Meier survival curves illustrate (A) relapse-free survival and (B) overall survival according to vitamin D levels. CI indicates confidence interval.

The univariate analysis is presented in Table 3. In a multivariate model for RFS and OS (Table 4), the 25(OH) vitamin D3 level retained its association with outcome when the model was adjusted for white blood cell count, smoking status, age, and European Leukemia Network genetic group.

Table 3. Univariate Survival Analysis
 Relapse-Free SurvivalOverall Survival
GroupHR (95% CI)PHR (95% CI)P
  1. Abbreviations: CI, confidence interval; ELN, European LeukemiaNet; HR, hazard ratio.

  2. a

    These P values indicate a statistically significant difference.

  3. b

    Age was fit at 10-year increments to model the difference between every 10 years as opposed to annual increments.

Vitamin D level    
Insufficient vs normal1.185 (0.66–2.127).57071.395 (0.76–2.56).2822
Deficient vs normal2.094 (1.183–3.707).0112a1.987 (1.094–3.607).0241a
White blood cell count, ×109/L    
<100 vs ≥1000.164 (0.081–0.331)< .0001a0.294 (0.148–0.586).0005a
Smoking status    
Current smoker vs nonsmoker0.643 (0.341–1.215).17360.623 (0.322–1.206).1604
Previous smoker vs nonsmoker1.057 (0.625–1.789).83641.099 (0.64–1.886).7320
Age    
Each 10-y increaseb1.174 (1.016–1.356).0299a1.255 (1.076–1.464).0039a
ELN genetic groups    
Intermediate-I vs favorable1.398 (0.728–2.685).31371.422 (0.715–2.826).3155
Intermediate-II vs favorable0.802 (0.351–1.835).60190.993 (0.424–2.324).9875
Adverse vs favorable2.484 (1.204–5.126).0138a2.788 (1.297–5.991).0086a
Table 4. Multivariate Survival Analysis
 Relapse-Free SurvivalOverall Survival
GroupHR (95% CI)PHR (95% CI)P
  1. Abbreviations: CI, confidence interval; ELN, European LeukemiaNet; HR, hazard ratio.

  2. a

    These P values indicate a statistically significant difference.

Vitamin D level    
Insufficient vs normal1.645 (0.819–3.307).16202.227 (1.067–4.65).0330a
Deficient vs normal2.588 (1.283–5.22).0079a2.927 (1.388–6.171).0048a
White blood cell count, ×109/L    
<100 vs ≥1000.105 (0.044–0.247)< .0001a0.166 (0.071–0.386)< .0001a
Smoking status    
Current smoker vs nonsmoker0.315 (0.147–0.677).0031a0.303 (0.135–0.68).0038a
Previous smoker vs nonsmoker0.754 (0.414–1.373).35530.746 (0.405–1.374).3477
Age    
Each 10-y increase1.154 (0.99–1.347).06791.273 (1.082–1.497).0036a
ELN genetic groups    
Intermediate-I vs favorable1.171 (0.585–2.346).65601.209 (0.586–2.494).6071
Intermediate-II vs favorable0.893 (0.378–2.109).79701.199 (0.499–2.883).6852
Adverse vs favorable3.179 (1.428–7.079).0046a3.997 (1.707–9.364).0014a

Association of 25(OH) Vitamin D3 Levels With MicroRNA Expression Profile

To understand the biology of the effect of 25(OH) vitamin D3 on AML outcome, we analyzed miR expression. The initial screen with whole genome microarray, using an unadjusted P < .05 and at least 2-fold expression level change, revealed that 13 miRs were up-regulated and 4 were down-regulated in patients with subnormal (<32 ng/mL) 25(OH) vitamin D3 levels, as mentioned above (see Materials and Methods). No significant results were observed after multiple test corrections. It is interesting to note that miR144 also was up-regulated in samples from patients with prostate cancer who had subnormal 25(OH) vitamin D3 levels (Campbell M, unpublished data). Next, we analyzed 16 of these miRs using RT-PCR in an additional cohort of 58 patients who had samples available (44 with subnormal 25[OH] vitamin D3 levels and 14 with normal levels). However, none of the specific signature miRs retained a significant association with the serum 25(OH) vitamin D3 level (data not shown).

Association Between 25(OH) Vitamin D3 Levels and Single Nucleotide Polymorphisms in the Vitamin D Pathway

To evaluate the possible contribution of pharmacogenetics to variations in serum 25(OH) vitamin D3 status in patients with AML, 90 genotyped tag SNPs in the genes encoding the vitamin D pathway enzymes (CYP27A1, CYP2R1, CYP27B1, CYP24A1, GC, and VDR) were successfully analyzed. By using unadjusted P values < .05 for the first analysis, in which the 25(OH) vitamin D3 level was treated as continuous variable, 6 SNPs had genotypes that were associated significantly with the patients' 25(OH) vitamin D3 level. For the second analysis, in which the patients were separated into 2 groups according to 25(OH) vitamin D3 level (ie, <32 ng/mL vs ≥32 ng/mL), 6 SNPs had genotypes that were associated significantly with 25(OH) vitamin D3 levels. In total, 3 SNPs were significant after both linear and logistic model analyses, namely, GC SNPs rs4588 and rs2762934 and VDR SNP rs10783219. However, after adjusting for multiple comparisons, none retained significance.

Association Between Vitamin D Receptor Single Nucleotide Polymorphisms and Outcome

The 6 SNPs with genotypes that had a significant association with patients' 25(OH) vitamin D3 levels were analyzed for their correlation with outcome. In the VDR SNP rs10783219, the presence of the T allele was associated significantly with an inferior CR rate (P = .0442), shorter RFS (P = .0058), and shorter OS (P = .0011) (Fig. 2A,B). In multivariate analysis, this SNP retained statistical significance for RFS and OS (Tables 5 and 6). It is noteworthy that there was a significant association between the 25(OH) vitamin D3 level and rs10783219 genotype (P = .0132). The 25(OH) vitamin D3-deficient group more often had the TA genotype (62.5%), and the normal 25(OH) vitamin D3 group more often had the AA genotype (69.6%). This may explain why 25(OH) vitamin D3 was not significant in the multivariate model when adjusted for rs10783219.

Table 5. Multivariate Analyses: Effect of Markers on Clinical Outcome by Type 3 Tests
 P
EffectCRRFSOS
  1. Abbreviations: CR, complete remission; ELN, European LeukemiaNet; OS, overall survival; RFS, relapse-free survival; SNP, single nucleotide polymorphism;.

  2. a

    These P values indicate a statistically significant difference.

Reference SNP rs10783219.1458.0183a.0092a
Vitamin D level.0920.5076.3336
White blood cell count.0059a< .0001a< .0001a
Smoking.8191.0009a.0021a
Age.2310.1676.0197a
ELN genetic groups.0182a.0137a.0057a
Table 6. Multivariate Analyses: Effect of Markers on Clinical Outcome With Odds Ratios/Hazard Ratios and 95% Confidence Intervals
 Complete RemissionRelapse-Free SurvivalOverall Survival
GroupOR (95% CI)PHR (95% CI)PHR (95% CI)P
  1. Abbreviations: CI, confidence interval; ELN, European LeukemiaNet; HR, hazard ratio; OR, odds ratio; SNP, single nucleotide polymorphism.

  2. a

    These P values indicate a statistically significant difference.

Reference SNP rs10783219      
TA vs. AA0.344 (0.066–1.793).20512.55 (1.199–5.422).0150a3.113 (1.385–6.994).0060a
TT vs AA3.907 (0.224–68.148).35010.936 (0.297–2.95).90941.11 (0.349–3.532).8596
Vitamin D level      
Insufficient vs normal1.896 (0.336–10.699).46881.558 (0.688–3.531).28761.945 (0.807–4.689).1386
Deficient vs normal10.62 (1.202–93.8).0335a1.574 (0.642–3.862).32171.546 (0.599–3.987).3674
White blood cell count, ×109/L      
<100 vs ≥10025.161 (2.531–250.09).0059a0.087 (0.034–0.223)< .0001a0.133 (0.052–0.336)< .0001a
Smoking status      
Current smoker vs nonsmoker1.571 (0.268–9.205).61680.171 (0.068–0.432).0002a0.187 (0.073–0.476).0004a
Previous smoker vs nonsmoker1.619 (0.289–9.064).58350.495 (0.235–1.042).06420.548 (0.258–1.167).1188
Age      
Each 10-y increase0.788 (0.534–1.164).23101.142 (0.946–1.38).16761.265 (1.038–1.541).0197a
ELN genetic groups      
Intermediate-I vs favorable0.05 (0.005–0.457).0080a0.9 (0.404–2.003).79591.191 (0.518–2.738).6815
Intermediate-II vs favorable0.072 (0.006–0.801).0324a0.827 (0.314–2.183).70181.235 (0.457–3.339).6777
Adverse vs favorable0.013 (0.001–0.202).0020a4.556 (1.593–13.025).0047a6.548 (2.19–19.579).0008a
Figure 2.

Kaplan-Meier survival curves illustrate (A) relapse-free survival and (B) overall survival according to the vitamin D receptor single nucleotide polymorphism rs10783219. CI indicates confidence interval.

DISCUSSION

To our knowledge, this is the first report to associate low 25(OH) vitamin D3 levels with a worse outcome in patients with AML. This finding is in line with previous work in NHL[4] and CLL[5] indicating that low 25(OH) vitamin D3 levels were associated with a worse outcome.

We are also the first to report an association between AML molecular subgroups and 25(OH) vitamin D3 levels. The finding that normal 25(OH) vitamin D3 levels are rarely associated with FLT3-ITD is intriguing. It is of further interest when considering the miR data. We previously demonstrated that miR144 was up-regulated in AML with FLT3-ITD,[27] and the finding that miR144 also was up-regulated in patients who had AML with low vitamin D raises questions about its role in FLT3-ITD leukemogenesis. In that regard, Gocek et al[28] attempted to differentiate AML cells with 1,25-dihydroxyvitamin D3 and demonstrated that blasts from patients who had AML with FLT3-ITD failed to differentiate despite elevated VDR expression, suggesting that the failure lies downstream of the receptor.

Conversely, AML with monosomy 7 or partial loss of 7q appeared to be extremely sensitive to 1,25-dihydroxyvitamin D3 and demonstrated significant differentiation.[28] Considering that AML with monosomy 7 is a subset with an especially poor outcome to conventional chemotherapy, vitamin D may prove a potential treatment adjunct in patients who have AML with this abnormality. In addition, although our data did not support a correlation between vitamin D levels and French-American-British (FAB) AML category, others[29] have demonstrated a correlation between functional VDR and FAB subtypes.

One shortcoming of this study is the relatively small sample size, which does not provide us enough power for some of the molecular markers (eg, NPM1) in miR or SNP analysis. Therefore, our small study is exploratory in nature, and the data should be interpreted with caution. This is important, because others[30] have demonstrated that vitamin D activates miR26a, thereby inducing an anti-leukemic effect. Nevertheless, the observation that several SNPs examined here are associated with low 25(OH) vitamin D3 levels and that 1 was associated with outcome is novel; however, it but raises some concerns if confirmed by others. Specifically, if indeed an SNP signature is inherent in AML, then supplementing these patients with vitamin D may not affect their outcome. However, it may affect the microenvironment that nurtures leukemia cells. Further studies with a larger sample size and functional experiments for understanding this mechanism are warranted.

Therefore, measuring 25(OH) vitamin D3 levels may predict outcome in AML, and studies to supplement with vitamin D compounds are warranted. The strengths of our study include the intensively treated cohort of newly diagnosed, well characterized AML patients; and the limitation is the small number of patients. The other limitation of our study is that we do not provide a causal relation between low serum 25(OH) vitamin D3 levels and worse outcome in patients with AML. We have initiated a pharmacokinetic study of vitamin D supplementation and expect those results in 1 or 2 years.

In conclusion, our data provide the first evidence that serum 25(OH) vitamin D3 levels may be an important factor influencing AML outcome and that patients with FLT3-ITD rarely have normal vitamin D levels. Therefore, it appears that the serum 25(OH) vitamin D3 level is a modifiable and economically feasible patient risk factor. This raises the possibility that, when judiciously intervened, adjusting the serum 25(OH) vitamin D3 level may improve the outcome of patients with AML without significant toxicity or cost.

FUNDING SUPPORT

This work was supported in part by grants from the National Cancer Institute Grant (CA16056: H.J.L., J.R.M., W.T., Q.H., D.W., S.L., G.E.W., L.A.F., S.N.J.S., A.W.B., A.A.A., M.B., EAG, J.E.T., E.S.W., C.S.J., D.L.T., and M.W.); the Szefel Foundation, Roswell Park Cancer Institute (E.S.W.); the Leonard S. LuVullo Endowment for Leukemia Research (M.W.); the Nancy C. Cully Endowment for Leukemia Research (M.W.); the Babcock Family Endowment (M.W.); and the Heidi Leukemia Research Fund, Buffalo, New York (M.W.).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary