Supplementary Methods
DMET analysis
Mononuclear cells were isolated from peripheral blood or bone marrow aspirates of 94 AML patients by
Ficoll gradient centrifugation and genomic DNA was extracted using the DNA Blood Mini Kit (Qiagen,
Valencia, CA, USA). DNA quality was evaluated by agarose gel electrophoresis, while double strand DNA
concentration was assessed with Quant-iT™ PicoGreen® dsDNA Reagent and Kits (Invitrogen™); Tecan
Infinite® 200 PRO reader was then used to detect the fluorescence emission, and λDNA serial dilutions
were used to create the standard curve.
Genotype identification of 1,931 Single Nucleotide Polymorphisms (SNPs) and 5 Copy Number Variations
(CNVs) were performed using the DMET™ platform and DMET™ Plus GeneChip (Affymetrix) according to
manufacturer’s instructions. Briefly, samples were diluted at a final concentration of 60 ng/μl and
processed together with three gDNA controls: 2 μl of each gDNA was first amplified in a multiplex PCR step,
to increase concentration of the interested genomic regions and to overcome wrong signals due to
pseudogenes or close homologs. Subsequently, amplified and non-amplified gDNA were combined for the
following annealing and amplification steps where the molecular inversion probes (MIP) technology was
used to genotype in a single reaction all the interrogated genomic sites.1 After purification, fragmentation,
labeling and array hybridization, samples were scanned with the Gene Chip Scanner 3000.
SNPs genotyping and quality control
The DMET Console version 1.1 (Affymetrix) was used to perform genotype calls from .cel files and to
generate quality control (QC) metrics. Only individuals with average call rates for the genotyped SNPs
higher than 96% were retained for the analysis. According to such QC filter, usable genotypes were
produced for 91 samples out of 94 (97%).
Goodness of fit to the Hardy-Weinberg equilibrium (HWE) expectation were computed for each SNP using
the exact test statistic described by Wigginton et al.2 to compare observed and expected genotype
frequencies in the described patients’ sub-groups. All the examined subjects were disease individuals (i.e.
no controls were included in the analysis) and no errors in the genotyping procedures were identified, thus
is not recommended that polymorphisms deviating from HWE have to be excluded from the analysis.3-4
However, the lack of HWE for some SNPs in the groups of individuals without adverse or not effective
reaction to the treatment may be due to population stratification, so that we decided to discard them in
order to avoid confounding results. According to this, four polymorphisms (rs737497, rs1065411,
rs34815109, rs7436962) in all the performed comparisons, two (rs6729738 and rs6457816) in the
comparison of responses to the induction cycle, two (rs4149057 and rs6909523) in the comparison of
reactions to MY and four (rs7169, rs1049434, rs1202283 and rs2223477) in the comparison of toxicity
profiles, showed significant deviations from HWE in the identified AML sub-groups and were removed prior
to the analyses.
Moreover, of the 1,931 variants included in the DMET™ panel, 993 showing a null minor allele frequency (
i.e. monomorphic variants), 46 belonging to the X chromosome and 6 showing >20% missing data were
dropped from the analysis.
SNPs validation
To evaluate genotyping accuracy, the genomic regions containing ADH1A rs6811453, FMO2 rs2020863 and
SULT2B1 rs2302948 were amplified and sequenced by conventional direct Sanger method. 50 ng of gDNA
from 15 patients was amplified using the following primers: forward ttacagaacaagtgctgatggaa and reverse
ctctttctggctcacattgct for ADH1A; forward taaatgatgatgtcccaagtcg and reverse agggttgacttgtccaggtg for
FMO2; forward gcacagctagttagacccatga and reverse ggtgtgagcagagggtatgg for SULT2B1. Samples were
incubated at 95°C for 5 min, followed by 35 cycles of 95°C for 30s, 62°C for 30s, 72°C for 30s, and a final
step of 72°C for 7min. PCR products were purified using the QIAquick PCR purification kit (Qiagen) and then
sequenced using the Big Dye Terminator DNA sequencing kit (Applied Biosystems) and the ABI PRISM 3730
automated DNA sequencer (Applied Biosystems).
Statistical analyses
Goodness
of
fit
to
the
HWE
expectation
and
genotypic
association
analysis
between
heterozygotic/homozygotic status of the examined SNPs’ minor alleles with effectiveness and toxicity of
the treatment were performed using the R 2.12.2 package (http://www.r-project.org/). In particular, the
SNPassoc library, implementing binary logistic regression methods under five different genetic models
(codominant, dominant, recessive, overdominant and log-additive), was used. For each of the tested
model, a matrix with sample size and percentages for each genotype, Odds Ratios (ORs) and their 95%
confidence intervals (CIs) were calculated, together with p-values corresponding to the likelihood ratio test
and obtained from comparison with the null model. The Akaike Information Criterion (AIC)5 was computed
in order to select the most parsimonious genetic model for each polymorphism, with the best one showing
the lowest AIC value.
Allelic
association
analysis
was
performed
using
the
PLINK
1.07
package
(http://pngu.mgh.harvard.edu/purcell/plink/)6, by computing Fisher's exact test and obtaining an exact pvalue, as well as estimated ORs and their 95% CIs, for each SNP.
Although multiple comparisons were implicated in the association analyses mentioned above, according to
the exploratory and hypothesis-generating purpose of this study a conventional 1% threshold for statistical
significance was set, so that results of potential interest were limited to those in which the p-value was <
0.01.
Differences in the distributions of prognostic factors among the examined AML sub-groups were evaluated
by Fisher’s exact test.
Multivariate analysis was performed applying a logistic regression model7 and results were expressed as
ORs  95% CIs.
Definitions of response
Complete response, or complete remission (CR) is defined by a non-blastic marrow aspirate (blast cells
<5%) with a WBC of more than 2 x 109/L, a platelet count of more than 50 x 109/L and no clinical or
imaging evidence of leukaemia; partial response, or partial remission (PR) is defined by a cellular marrow
aspirate with 5 to 20% blast cells, with a WBC of more than 2 x 109/L, a platelet count of more than 50 x
109/L and no clinical or imaging evidence of leukaemia; resistance or not response (NR) is defined by any
marrow aspirate with more than 20% blast cells for more than 21-28 days after the treatment. Side or toxic
effects are evaluated and recorded using the World Health Organization (WHO) grading scale.
References
1.
Deeken J: The Affymetrix DMET platform and pharmacogenetics in drug development. Curr
Opin Mol Ther 11:260-8, 2009
2.
Wigginton JE, Cutler DJ, Abecasis GR: A note on exact tests of Hardy-Weinberg equilibrium.
Am J Hum Genet 76:887-93, 2005
3.
Balding DJ: A tutorial on statistical methods for population association studies. Nat Rev
Genet 7:781-91, 2006
4.
Jorgensen AL, Williamson PR: Methodological quality of pharmacogenetic studies: issues of
concern. Stat Med 27:6547-69, 2008
5.
Akaike H: A New Look at the Statistical Model Identification. IEEE Transaction on Automatic
Control AC-19:716-23 1974
6.
Purcell S, Neale B, Todd-Brown K, et al: PLINK: a tool set for whole-genome association and
population-based linkage analyses. Am J Hum Genet 81:559-75, 2007
7.
Hosmer DW, Lemeshow S: Applied logistic regression. New York, NY: Wiley, 1989
Supplementary Table 1. ADH1A rs6811453 and rs1826909 genotypes and other clinical relevant factors
for predicting Disease Free Survival (multivariate analysis).
Parameter
C/C or C/T vs T/T (rs6811453)
G/G or G/A vs A/A (rs1826909)
Age:
as continuous variable
Risk:
Standard/intermediate vs. high
Sex:
F vs. M
FLT3 mutations:
Mutated vs wild -type
NPM1 mutations:
Mutated vs wild -type
Pr > ChiSq
Odds Ratio
95% C.I.
0.0194
9.504
(1.438-62.819)
0.7237
0.990
(0.937-1.046)
0.1576
0.196
(0.021-1.878)
0.0487
5.808
(1.010-33.407)
0.5399
1.729
(0.300-9.957)
0.3582
2.353
(0.379-14.609)