Supplementary Material (doc 30K)

advertisement
Supplemental Material to:
A genome-wide approach identifies that the aspartate metabolism pathway
contributes to asparaginase sensitivity
Shih-Hsiang Chen, MD1,2, Wenjian Yang, PhD1, Yiping Fan, PhD3, Gabriele Stocco, PhD1,
Kristine R. Crews, PharmD1,5, Jun J. Yang, PhD1, Steven W. Paugh, PhD1, Ching-Hon Pui,
MD4,5, William E. Evans, PharmD1,5, Mary V. Relling, PharmD1,5
1
Department of Pharmaceutical Sciences, St Jude Children's Research Hospital, Memphis,
Tennessee, USA
2
Division of Hematology/Oncology, Department of Pediatrics, Chang Gung Memorial
Hospital, College of Medicine, Chang Gung University, Taoyuan, Taiwan
3
Hartwell Center for Bioinformatics and Biotechnology, St Jude Children's Research Hospital,
Memphis, Tennessee, USA
4
Department of Oncology, St Jude Children's Research Hospital, Memphis, Tennessee, USA
5
Department of Clinical Pharmacy, University of Tennessee, Memphis, Tennessee, USA
Methods
Multivariate model to predict association of asparaginase IC50 with genotypes. Random
forests is an ensemble tree-based regression algorithm for performing regressions in pathwaybased analyses. Pathway analyses allow one to cluster associated SNPs in relevant genes in
biologically meaningful pathways, but such analyses of grouped SNPs are subject to bias.
Random forests has several positive features. One is that it uses a random selection of features
(SNPs in our case) to split each node, yielding error rates that compare favorably to other
regression methods and are more robust with respect to noise. It provides an unbiased
estimate of the regression error as the forest is built. We used the library package
RandomForest v4.5-16 from the R program (http://www.r-project.org/)1 for our analysis. The
algorithm grows a large number of unpruned trees (n=10,000 in our case) using bootstrap
samples from the original data and predicts out-of-bag data (the original data that are not in
the bootstrap sample, approximately one-third of the original data) in each tree. The overall
prediction is then calculated by averaging the predictions of all the trees. The results of this
approach for the highest ranked pathway (aspartate) are presented in the main manuscript.
Comparing within-trio variability to that expected from unrelated trios of individuals.
We compared the observed child-parent variability in asparaginase IC50 within trios to the
variability observed in randomly-chosen child-parent trios. For each family, we estimated the
intra-trio variability using the following formula:
(|Child IC50 – father IC50| + |Child IC50 – mother IC50|)/ (2*|mother IC50 – father IC50|). We
estimated the observed average of the child-parent variability within trios across the 30
families. To estimate “child-parent” variability had the trios not consisted of families (e.g.
been “random” trios), we kept the parents unchanged for each family, but randomly selected
their “child” from other families. We repeated this 1000 times, i.e. we created 1000 datasets,
each dataset had 30 families, and in each family the child-parents relationship was not
biological. Out of 1000 permuted estimates of such “child-parent” variability, 45 had
variability less than that observed, indicating that the variability in IC50 we observed within
parent/child trios is somewhat lower than that expected among unrelated “parent/child” trios
(p = 0.045).
References:
1.
Liaw A, Wiener M. Classification and regression by randomForest. RNews.
2002;2:18-22.
2.
Scherf U, Ross DT, Waltham M, et al. A gene expression database for the molecular
pharmacology of cancer. Nat Genet. 2000;24:236-244.
3.
Stams WA, den Boer ML, Beverloo HB, et al. Sensitivity to L-asparaginase is not
associated with expression levels of asparagine synthetase in t(12;21)+ pediatric ALL. Blood.
2003;101:2743-2747.
4.
Stams WA, den Boer ML, Holleman A, et al. Asparagine synthetase expression is
linked with L-asparaginase resistance in TEL-AML1-negative but not TEL-AML1-positive
pediatric acute lymphoblastic leukemia. Blood. 2005;105:4223-4225.
5.
Fine BM, Kaspers GJ, Ho M, Loonen AH, Boxer LM. A genome-wide view of the in
vitro response to l-asparaginase in acute lymphoblastic leukemia. Cancer Res. 2005;65:291299.
6.
Holleman A, Cheok MH, den Boer ML, et al. Gene-expression patterns in drugresistant acute lymphoblastic leukemia cells and response to treatment. N Engl J Med.
2004;351:533-542.
7.
Krejci O, Starkova J, Otova B, et al. Upregulation of asparagine synthetase fails to
avert cell cycle arrest induced by L-asparaginase in TEL/AML1-positive leukaemic cells.
Leukemia. 2004;18:434-441.
8.
Iwamoto S, Mihara K, Downing JR, Pui CH, Campana D. Mesenchymal cells regulate
the response of acute lymphoblastic leukemia cells to asparaginase. J Clin Invest.
2007;117:1049-1057.
9.
Estes DA, Lovato DM, Khawaja HM, Winter SS, Larson RS. Genetic alterations
determine chemotherapy resistance in childhood T-ALL: modelling in stage-specific cell lines
and correlation with diagnostic patient samples. Br J Haematol. 2007;139:20-30.
10.
Su N, Pan YX, Zhou M, Harvey RC, Hunger SP, Kilberg MS. Correlation between
asparaginase sensitivity and asparagine synthetase protein content, but not mRNA, in acute
lymphoblastic leukemia cell lines. Pediatr Blood Cancer. 2008;50:274-279.
Download