Bhramar Mukherjee, PhD Professor of Biostatistics and Epidemiology University of Michigan School of Public Health bhramar@umich.edu SAMSI-SAVI Workshop, Mumbai, 2016 Working Group # 6
Working Group 6: Epigenomics Elena Colicino Sudha Ramalingam Bhramar Mukherjee
The Patriotic Peacocks Bhramar Tanujit Rajani Prakash Mohan Dimple Sharayu
Introduction What is interaction? Why measure it? -biology, sub-group identification, improving power How to measure it? -Choice of scale, method of analysis, coding When to report it? -public health relevance, biological significance, statistical significance
Interactions
Statistical Interaction “ Interaction as statisticians think of it is a Weasel parameter.” – Professor David Clayton, JSM 2012 Weasel Word: “an informal term for words and phrases aimed at creating an impression that a specific and/or meaningful statement has been made , when only a vague or ambiguous claim has been communicated, enabling the specific meaning to be denied if the statement is challenged ” (wikipedia)
Very few replicable interactions reported in human observational studies!
Me, 1978 Gene x Environment x Time Me, 2016
Lead exposure among children in India: determinants, neurobehavioral effects and genetic susceptibility Working Group 6: Data Example
Dataset Environmental Health Perspective, 2011
Dataset Neurotoxicology, 2013
World blood lead levels among children Burden of disease, 2010
Lead levels and lead in gasoline USA, NHANES II ( Annest et al. 1983)
Sources of lead exposure Leaded gasoline phased later than in US Leaded paint Occupational: • Garage workers • Smelting and metal working operations • Jewelery workers • Industrial activity • Mining Cultural practices • Ayurvedic medication • Cosmetics (surma, sindhur) • Holi colors • Spices
Cosmetics Religious powders Ayurvedic medication
Lead in paint (2009) Clark, C.S. et al, Lead levels in new enamel household paints from Asia, Africa and South America. Environ. Res. (2009), doi:10.1016/j.envres.2009.07.002.
Lead Paint NDTV 2010 New York Times 2007
Electronic waste 10-20,000 tonnes, employing 25,000 people, in New Delhi alone E waste management and handling Rule 2011 ( new law MOEF, India) Needs implementation Toxics link 2010
Determinants of blood lead levels among 3-7 year old children in Chennai, India (2005-2006)
India Lead Study (Chennai) Chennai Study population ( N= 756) High Low • Cross-sectional industry Industry • 12 schools (3 in 4 zones) High HT/HI HT/LI • 3-7 year old children traffic (3 schools) (3 schools) Low LT/HI LT/LI traffic (3 schools) (3 schools) • Blood lead levels assessed by • LeadCare™ Analyzer
3 0 N=756 Mean=11.5 g/dl Range=2.6-40.5 g/dl 2 5 55% > 10 µg/dl 2 0 2% > 10 µg/dl (NHANES III) P e r c 1 5 e n t 1 0 5 Distribution of blood lead levels ( g/dl) in children in 0 1 . 5 4 . 5 7 . 5 1 0 . 5 1 3 . 5 1 6 . 5 1 9 . 5 2 2 . 5 2 5 . 5 2 8 . 5 3 1 . 5 3 4 . 5 3 7 . 5 4 0 . 5 BL L Chennai
Assessment of Predictors Questionnaires (primary care givers : Tamil) • Socioeconomic status o Family income, parental education, occupation o Type of house • Possible sources of exposure o Residence (traffic and industry zone), parental occupation, presence of lead based industry, traffic level near house o Type of paint o Sources and storage of drinking water o Surma and ayurvedic medication use
Predictors of blood lead (>10µg/dl) 3.50 Variables Estimate 95% CI Partial R2 ** 3.00 Age (months) 0.002 -0.001 0.005 0.003 2.50 Sex -0.028 -0.094 0.039 0.001 Odds ratio 2.00 Average monthly family income (Rs)*** <2000 0.259 0.125 0.394 0.028 1.50 2000-4000 0.233 0.123 0.342 0.033 4000-6500 0.182 0.081 0.282 0.017 1.00 Drinking water storage vessel*** 0.50 Brass/ Bronze 0.210 0.061 0.359 0.010 0.00 Residence *** Other <2000 >6500 Brass/Bronze 2000-4000 4000-6500 High industry 0.074 -0.082 0.231 0.007 * accounting for clustering at school level using generalized estimating equations ** unadjusted for clustering using linear regression *** compared to >6500 Rs/ month, ** all other drinking water storage vessels, ***low industry area Income (Rs) DWV Total model R 2 = 5.8% DWV: Type of vessel used for storage of drinking water. Adjusted for age (months), sex p-values<0.05
Conclusions Predictors of blood lead • Blood lead levels o Lower socioeconomic status o Drinking water stored in brass or bronze vessels Residence in a high industry zone (<5 year old) • No effect of use of ayurvedic medication, surma , traffic, paint • Little variation in blood lead was explained o Need in depth exposure assessment
Lead exposure and behavior, IQ, Visual Motor skills children in Chennai, India
Lead and IQ o IQ is best characterized (Needleman 1979, • Bellinger 1983) o No threshold o Non-linear dose-response • (Schwartz 1994) Heated debate!! Lanphear et al. 2005 Lanphear et al 2005
Behavioral and cognitive assessment Behavior: Questionnaires administered to the class teachers Connors ADHD DSM IV Scales (CADS) ADHD Index o DSM IV: Hyperactivity o DSM IV: Inattention o Behavior Rating Inventory of Executive Function (BRIEF) Executive function composite o Behavioral regulation (inhibit, shift, emotional control) o Metacognition (Initiate, working memory, planning, organization of materials, o monitoring) Connors Teacher Rating Scales (CTRS-39) Anxiety, Sociability, (Aggression, Hyperactivity, Inattention) o
Behavioral and cognitive assessment (con ’ t) Intelligence • Binet - Kamat Intelligence scales ( Tamil) o mental age/ chronological age= IQ o administered to children Genotyping • Bioserve Hyderabad, India • Mass Array Iplex (Sequenom process) o PCR and mass spectrometry o Blood • Negative and positive controls o 24 DNA samples from the Coriell Discovery panel
Effect of lead and hemoglobin (Hb) on IQ Generalized estimating equations* Roy et al pending publication
Lead and Visual motor skills Pallaniapan & Roy et al 2011
Conclusions Lead and behavior • Blood lead levels are associated with poorer behavior and visual-motor skills o ADHD, internalizing behaviors and executive function • Executive function is most sensitive to lead ( 0.4 SD) o 4 IQ points (0.25 SD IQ) o In ADHD, inattention is most affected o No effect seen on hyperactivity • Dose-response relationships are linear for behavior • Blood lead levels are associated with poorer
Lead exposure, iron and intelligence: genetic susceptibility
Lead and IQ Wide variation in effect estimates • Residual confounding • Measurement error • Different dose ranges • Effect modification • Nutritional differences • Genetic differences Lanphear et al. 2005
Effect modification by Transferrin C2 polymorphsim
Effect modification by Transferrin C2 polymorphsim Roy et al Pending publication
Distribution of DRD2 Taq IA genotype
Effect of lead and Hb on IQ by DRD2 genotype Roy et al 2011
Hemoglobin, Lead & IQ: genetic susceptibility IQ IQ - * * * * - - - - -
Plan for Analysis Working Group Data consists of 159 variables, including genotype data on 18 genetic polymorphisms We will try to reproduce the published analysis with one marker at a time: -Choice of confounders -Transformation of Y and X -Dose response relationship -Interpreting interaction on the transformed scale -Reporting of findings -How robust are the conclusions - Extend to incorporate multiple markers, calculate a polygenic risk score. -Unexplored Associations (birth order related to IQ?)
OVERARCHING PARADIGM Blood lead Behavior Determinants ADHD Executive function SES Internalizing behavior Industrial activity Brass and bronze vessels Cognition Dopamine D2 receptor polymorphism Iron IQ Transferrin C2 polymorphism Hemoglobin
AKNOWLEDGEMENTS SRMC HSPH Research Team David C. Bellinger Joel Schwartz Kalpana Balakrishnan Robert Wright Kavitha Palaniapan Ananya Roy Padmavathi Ramaswamy University of Toronto Venkatesh S.M. Shankar K.M. Howard Hu BIOSERVE YSPH Rama Modali Adrienne Ettinger Funding : NIH (R01 ES007821) , Fogarty grant (R03 TW005914)
Study Participants!
How do we translate all these findings of reported associations and interactions into Public Health action? Why Should Francesca care?
Recommend
More recommend
