Model Selection Approach Suggests Causal Association between 25-Hydroxyvitamin D and Colorectal Cancer

Introduction Vitamin D deficiency has been associated with increased risk of colorectal cancer (CRC), but causal relationship has not yet been confirmed. We investigate the direction of causation between vitamin D and CRC by extending the conventional approaches to allow pleiotropic relationships and by explicitly modelling unmeasured confounders. Methods Plasma 25-hydroxyvitamin D (25-OHD), genetic variants associated with 25-OHD and CRC, and other relevant information was available for 2645 individuals (1057 CRC cases and 1588 controls) and included in the model. We investigate whether 25-OHD is likely to be causally associated with CRC, or vice versa, by selecting the best modelling hypothesis according to Bayesian predictive scores. We examine consistency for a range of prior assumptions. Results Model comparison showed preference for the causal association between low 25-OHD and CRC over the reverse causal hypothesis. This was confirmed for posterior mean deviances obtained for both models (11.5 natural log units in favour of the causal model), and also for deviance information criteria (DIC) computed for a range of prior distributions. Overall, models ignoring hidden confounding or pleiotropy had significantly poorer DIC scores. Conclusion Results suggest causal association between 25-OHD and colorectal cancer, and support the need for randomised clinical trials for further confirmations.

[1]  P. McKeigue,et al.  Instrumental Variable Estimation of the Causal Effect of Plasma 25-Hydroxy-Vitamin D on Colorectal Cancer Risk: A Mendelian Randomization Analysis , 2012, PloS one.

[2]  George Davey Smith,et al.  Using multiple genetic variants as instrumental variables for modifiable risk factors , 2012, Statistical methods in medical research.

[3]  J. Manson,et al.  The influence of health and lifestyle characteristics on the relation of serum 25-hydroxyvitamin D with risk of colorectal and breast cancer in postmenopausal women. , 2012, American journal of epidemiology.

[4]  C. Macera,et al.  Does the evidence for an inverse relationship between serum vitamin D status and breast cancer risk satisfy the Hill criteria? , 2012, Dermato-endocrinology.

[5]  John M. Winn,et al.  Causality with Gates , 2012, AISTATS.

[6]  I. Reid,et al.  Calcium and vitamin D supplements and health outcomes: a reanalysis of the Women's Health Initiative (WHI) limited-access data set. , 2011, The American journal of clinical nutrition.

[7]  F. Agakov,et al.  O1-2.2 Sparse instrumental variables: an integrative approach to biomarker validation , 2011, Journal of Epidemiology & Community Health.

[8]  F. Agakov,et al.  Diet, environmental factors, and lifestyle underlie the high prevalence of vitamin D deficiency in healthy adults in Scotland, and supplementation reduces the proportion that are severely deficient. , 2011, The Journal of nutrition.

[9]  R. Tibshirani,et al.  Regression shrinkage and selection via the lasso: a retrospective , 2011 .

[10]  M. Stampfer,et al.  Circulating Levels of Vitamin D and Colon and Rectal Cancer: The Physicians' Health Study and a Meta-analysis of Prospective Studies , 2011, Cancer Prevention Research.

[11]  G. Byrnes,et al.  Meta‐analysis of observational studies of serum 25‐hydroxyvitamin D levels and colorectal, breast and prostate cancer and colorectal adenoma , 2011, International journal of cancer.

[12]  E. Riboli,et al.  Meta-Analyses of Vitamin D Intake, 25-Hydroxyvitamin D Status, Vitamin D Receptor Polymorphisms, and Colorectal Cancer Risk , 2011, Cancer Epidemiology, Biomarkers & Prevention.

[13]  T. Frayling,et al.  C-reactive protein levels and body mass index: Elucidating direction of causation through reciprocal Mendelian randomization , 2010, International Journal of Obesity.

[14]  Felix Agakov,et al.  Inference of Causal Relationships between Biomarkers and Outcomes in High Dimensions , 2011 .

[15]  Amos J. Storkey,et al.  Sparse Instrumental Variables (SPIV) for Genome-Wide Studies , 2010, NIPS.

[16]  Jean-Baptiste Cazier,et al.  Meta-analysis of three genome-wide association studies identifies susceptibility loci for colorectal cancer at 1q41, 3q26.2, 12q13.13 and 20q13.33 , 2010, Nature Genetics.

[17]  John K Kruschke,et al.  Bayesian data analysis. , 2010, Wiley interdisciplinary reviews. Cognitive science.

[18]  J. Moan,et al.  Obesity and increased risk of cancer: does decrease of serum 25-hydroxyvitamin D level with increasing body mass index explain some of the association? , 2010, Molecular nutrition & food research.

[19]  Daniel L. Koller,et al.  Common genetic determinants of vitamin D insufficiency: a genome-wide association study , 2010, The Lancet.

[20]  P. Visscher,et al.  Common SNPs explain a large proportion of heritability for human height , 2011 .

[21]  S. Grant,et al.  Mendelian randomization in the era of genomewide association studies. , 2010, Clinical chemistry.

[22]  S. Wild,et al.  Bayesian methods for instrumental variable analysis with genetic instruments (‘Mendelian randomization’): example with urate transporter SLC2A9 as an instrumental variable for effect of urate levels on metabolic syndrome , 2010, International journal of epidemiology.

[23]  Murielle Bochud,et al.  Usefulness of Mendelian Randomization in Observational Epidemiology , 2010, International journal of environmental research and public health.

[24]  J. Pearl.  Causal inference in statistics: An overview , 2009 .

[25]  A. Wallace,et al.  A simple automated solid-phase extraction procedure for measurement of 25-hydroxyvitamin D3 and D2 by liquid chromatography-tandem mass spectrometry , 2009, Annals of clinical biochemistry.

[26]  W. Grant.  How strong is the evidence that solar ultraviolet B and vitamin D reduce the risk of cancer? An examination using Hill’s criteria for causality , 2009, Dermato-endocrinology.

[27]  Steven Gallinger,et al.  Meta-analysis of genome-wide association data identifies four new susceptibility loci for colorectal cancer , 2008, Nature Genetics.

[28]  Andrew B. Lawson,et al.  Bayesian Disease Mapping: Hierarchical Modeling in Spatial Epidemiology , 2008 .

[29]  M. Tobin,et al.  Mendelian Randomisation and Causal Inference in Observational Epidemiology , 2008, PLoS medicine.

[30]  Julian Peto,et al.  A genome-wide association study identifies colorectal cancer susceptibility loci on chromosomes 10p14 and 8q23.3 , 2008, Nature Genetics.

[31]  I. Deary,et al.  Genome-wide association scan identifies a colorectal cancer susceptibility locus on 11q23 and replicates risk loci at 8q24 and 18q21 , 2008, Nature Genetics.

[32]  George Davey Smith,et al.  Mendelian randomization: Using genes as instruments for making causal inferences in epidemiology , 2008, Statistics in medicine.

[33]  S. Horvath,et al.  Variations in DNA elucidate molecular networks that cause disease , 2008, Nature.

[34]  Florian Steinke,et al.  Bayesian Inference and Optimal Design in the Sparse Linear Model , 2007, AISTATS.

[35]  Steven Gallinger,et al.  Genome-wide association scan identifies a colorectal cancer susceptibility locus on chromosome 8q24 , 2007, Nature Genetics.

[36]  Oliver Sieber,et al.  A genome-wide association scan of tag SNPs identifies a susceptibility variant for colorectal cancer at 8q24.21 , 2007, Nature Genetics.

[37]  N. Sheehan,et al.  Mendelian randomization as an instrumental variable approach to causal inference , 2007, Statistical methods in medical research.

[38]  W. Willett,et al.  A nested case control study of plasma 25-hydroxyvitamin D concentrations and risk of colorectal cancer. , 2007, Journal of the National Cancer Institute.

[39]  R. Recker,et al.  Vitamin D and calcium supplementation reduces cancer risk: results of a randomized trial. , 2007, The American journal of clinical nutrition.

[40]  J. Kyle,et al.  Dietary Flavonoids and the Risk of Colorectal Cancer , 2007, Cancer Epidemiology Biomarkers & Prevention.

[41]  C. Robert,et al.  Deviance information criteria for missing data models , 2006 .

[42]  W. Grant,et al.  The association of solar ultraviolet B (UVB) with reducing risk of cancer: multifactorial ecologic analysis of geographic variation in age-adjusted cancer mortality rates. , 2006, Anticancer research.

[43]  J. Robins,et al.  Instruments for Causal Inference: An Epidemiologist's Dream? , 2006, Epidemiology.

[44]  Wiebe R. Pestman,et al.  Instrumental Variables: Application and Limitations , 2006, Epidemiology.

[45]  H. Philippe,et al.  Computing Bayes factors using thermodynamic integration. , 2006, Systematic biology.

[46]  L. Smeeth,et al.  Limits to causal inference based on Mendelian randomization: a comparison with randomized controlled trials. , 2006, American journal of epidemiology.

[47]  J. Castle,et al.  An integrative genomics approach to infer causal associations between gene expression and disease , 2005, Nature Genetics.

[48]  David V Conti,et al.  Commentary: the concept of 'Mendelian Randomization'. , 2004, International journal of epidemiology.

[49]  David J. C. MacKay,et al.  Information Theory, Inference, and Learning Algorithms , 2004, IEEE Transactions on Information Theory.

[50]  M. Lipkin,et al.  Chemoprevention of colon cancer by calcium, vitamin D and folate: molecular mechanisms , 2003, Nature Reviews Cancer.

[51]  S. Ebrahim,et al.  'Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? , 2003, International journal of epidemiology.

[52]  Bradley P. Carlin,et al.  Bayesian measures of model complexity and fit , 2002 .

[53]  Philip A. Trostel,et al.  Estimates of the economic return to schooling for 28 countries , 2002 .

[54]  M. Lipkin,et al.  Cellular Mechanisms of Calcium and Vitamin D in the Inhibition of Colorectal Carcinogenesis , 2001, Annals of the New York Academy of Sciences.

[55]  Radford M. Neal.  Annealed importance sampling , 1998, Stat. Comput..

[56]  James H. Stock,et al.  Instrumental Variables in Statistics and Econometrics , 2001 .

[57]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[58]  David Card.  The Causal Effect of Education on Learning , 1999 .

[59]  C. Jarzynski.  Equilibrium free-energy differences from nonequilibrium measurements: A master-equation approach , 1997, cond-mat/9707325.

[60]  R. Tibshirani.  Regression Shrinkage and Selection via the Lasso , 1996 .

[61]  R. Mclean,et al.  A Unified Approach to Mixed Linear Models , 1991 .

[62]  Richard E. Neapolitan,et al.  Probabilistic reasoning in expert systems - theory and algorithms , 2012 .

[63]  C. Garland,et al.  Do sunlight and vitamin D reduce the likelihood of colon cancer? , 1980, International journal of epidemiology.