Nutrient and Physical Activity Biomarker Studies in the WHI 544 women completed two-week DLW protocol with urine and blood collection and with FFQ and.

Nutrient and Physical Activity Biomarker Studies

in the WHI

• 544 women completed two-week DLW protocol with urine and blood 544 women completed two-week DLW protocol with urine and blood collection and with FFQ and other questionnaire data collection collection and with FFQ and other questionnaire data collection (50% intervention, 50% control). A 20% reliability subsample (50% intervention, 50% control). A 20% reliability subsample repeated protocol, separated by about 6 months from original data repeated protocol, separated by about 6 months from original data collection. collection. (NBS)(NBS)

• Biomarker study among 450 women in the WHI Observational Study Biomarker study among 450 women in the WHI Observational Study for evaluating and comparing measurement properties of dietary for evaluating and comparing measurement properties of dietary and physical activity assessment approaches (frequencies, records, and physical activity assessment approaches (frequencies, records, and recalls) and their combination. A 20% reliability subsample and recalls) and their combination. A 20% reliability subsample again included again included (NPAAS)(NPAAS)

. . Human feeding study currently enrolling 150 WHI women in Seattle, Human feeding study currently enrolling 150 WHI women in Seattle, toward developing biomarkers for additional nutrients/foods toward developing biomarkers for additional nutrients/foods (NPAAS2)(NPAAS2)

Acknowledgments

NBS and NPAAS Coordinators: NBS and NPAAS Coordinators: Marian Neuhouser, Lesley Tinker, FHCRCMarian Neuhouser, Lesley Tinker, FHCRC

NBS and NPAAS Clinical Center PIs:NBS and NPAAS Clinical Center PIs:Shirley Beresford - SeattleShirley Beresford - Seattle Judy Ockene - U MassJudy Ockene - U Mass

Bette Caan - OaklandBette Caan - Oakland Gerardo Heiss - UNCGerardo Heiss - UNC

Linda Van Horn - ChicagoLinda Van Horn - Chicago Lewis Kuller* - PittsburghLewis Kuller* - Pittsburgh

Cynthia Thomson - ArizonaCynthia Thomson - Arizona Marcia Stefanick* - Marcia Stefanick* - StanfordStanford

Yasmin Mossavar-RahmaniYasmin Mossavar-Rahmani†† - NYC - NYC Ellen Smit* - PortlandEllen Smit* - Portland

Karen Johnson - MemphisKaren Johnson - Memphis Annlouise Assaf* - PawtucketAnnlouise Assaf* - Pawtucket

Gloria Sarto - WisconsinGloria Sarto - Wisconsin*NBS only*NBS only††NPAAS onlyNPAAS only

Biomarker PIs:Biomarker PIs: Statistical Collaborators:Statistical Collaborators:

Dale Schoeller - WisconsinDale Schoeller - Wisconsin Pamela Shaw - NIAIDPamela Shaw - NIAID

Sheila Bingham - CambridgeSheila Bingham - Cambridge Ying Huang – FHCRCYing Huang – FHCRC

NPAAS2NPAAS2-Johanna Lampe, Marian, Lesley,Shirley,…-Johanna Lampe, Marian, Lesley,Shirley,…

Characteristic Coefficient (SD)Log Total Energy

Coefficient (SD)Log Protein

Coefficient (SD)Log % Energy from Protein

Intercept 7.61 (0.13) 4.28 (0.024) 2.66 (0.01)

Log FFQ 0.062 (0.018) 0.212 (0.032) 0.439 (0.058)

BMI 0.013 (0.001) 0.012 (0.002) -0.004 (0.002)

Age -0.005 (0.001) -0.008 (0.002) -0.005 (0.002)

Black -0.016 (0.017) -0.130 (0.047) __

Hispanic -0.004 (00.30) -0.021 (0.056) __

Other race -0.093 (0.027) -0.100 (0.058) __

Regression Calibration Coefficients for Log-Regression Calibration Coefficients for Log-Transformed Total Energy, Total Protein and Transformed Total Energy, Total Protein and

Percent Energy from Protein Percent Energy from Protein (Neuhouser et al, AJE, (Neuhouser et al, AJE, 2008)2008)

Geometric Mean

(95 % Confidence Interval)

Energy (kcal/day) Protein (g/day) % of Energy from Protein

Uncalibrated Calibrated* Uncalibrated Calibrated Uncalibrated Calibrated

DM1480.3

(675.9, 3242.2)

2143.1

(1796.5, 2556.5)

60.9

(26.2, 141.6)

78.3

(58.3, 105.0)

16.6

(11.4, 24.1)

14.4

(12.1, 17.2)

OS1380.2

(642.6, 2964.3)

2059.1

(1726.1, 2456.3)

58.6

(247.0, 138.7)

74.4

(55.5, 99.9)

16.9

(11.5, 25.1)

14.4

(11.9, 17.6)

Geometric means and 95% confidence intervals for uncalibrated dietary intakes as estimated by the Women’s Health Initiative (WHI) Food Frequency Questionnaire, and for calibrated intakes using nutritional biomarker data in the WHI Dietary Modification trial comparison group (DM)

and Observational Study (OS)(Prentice et al, AJE, 2009)

Body fat deposition and energy and protein associations with disease risk

Increase in body mass fat deposition (e.g. increased Increase in body mass fat deposition (e.g. increased BMI) over time may be the principal mediator of energy BMI) over time may be the principal mediator of energy and cancer, CVD, and diabetes associations.and cancer, CVD, and diabetes associations.

But, analyses that include BMI in calibration equation But, analyses that include BMI in calibration equation but not in disease risk model require specialized but not in disease risk model require specialized methods (Prentice and Huang, 2011, Can J Stat)methods (Prentice and Huang, 2011, Can J Stat)

NBS calibration equations also used for other NBS calibration equations also used for other purposes(e.g. protein in relation to both frailty and purposes(e.g. protein in relation to both frailty and kidney function (Jeannette Beasley)kidney function (Jeannette Beasley)

Calibration Equation Coefficients (β), Standard Errors (SE), and Percent of Biomarker Variation Explained (R2) from Regression of

Log(biomarker) on Log(self-report), and Other Factors among 450 Observational Study Women

Energy

Prentice RL, Mossavar-Rahmani Y, Huang Y et al (2011, AJE)

Variable Food Frequency 4DFR 24HR

β SE R2 Adj R2 β SE R2 Adj R2 β SE R2 Adj R2

ENERGY

Intercept 7.614a 0.009 7.597a 0.009 7.607a 0.009 FFQ 0.054a 0.017 3.8 6.5 4DFR 0.161a 0.028 7.8 13.3 24HR 0.101a 0.026 2.8 4.8 BMI 0.013a 0.001 26.9 45.9 0.013a 0.001 27.0 46.0 0.013a 0.001 28.7 48.9 Age -0.010a 0.001 9.7 16.5 -0.009a 0.001 8.4 14.3 -0.009a 0.001 9.1 15.5 Black -0.023 0.019 -0.024 0.018 -0.024 0.018 Hispanic -0.062a 0.021 1.3 2.2 -0.065a 0.020 1.5 2.6 -0.063a 0.020 1.5 2.6 Other minority

-0.041 0.040 -0.039 0.038 -0.038 0.039

(Total)b 41.7 71.1 44.7 76.2 42.1 71.8

Calibration Equation Coefficients (β), Standard Errors (SE), and Percent of Biomarker Variation Explained (R2) from Regression of

Log(biomarker) on Log(self-report), and Other Factors among 450 Observational Study Women

Protein

Prentice RL, Mossavar-Rahmani Y, Huang Y et al (2011, AJE)

Variable Food Frequency 4DFR 24HR

β SE R2 Adj R2 β SE R2 Adj R2 β SE R2 Adj R2

PROTEIN

Intercept 4.263 0.017 4.235 0.016 4.269 0.016 FFQ 0.135a 0.021 8.4 16.4 4DFR 0.465a 0.045 22.6 44.2 24HR 0.404a 0.046 16.2 31.7 BMI 0.012a 0.002 5.8 11.4 0.012a 0.002 5.1 10.0 0.012a 0.002 5.8 11.4 Age -0.012a 0.002 4.1 8.0 -0.009a 0.002 2.2 4.3 -0.011a 0.002 3.4 6.7 Black -0.120a 0.038 -0.138a 0.034 -0.145a 0.035 Hispanic -0.078 0.040 2.0 3.9 -0.067 0.036 2.7 5.3 -0.069 0.037 3.0 5.9 Other minority

-0.018 0.076 0.012 0.070 -0.026 0.072

(Total)b 20.3 39.7 32.7 63.8 28.4 55.6

Lessons from Studies of Dietary Consumption Effects on Chronic Disease

Control of dietary assessment measurement errors key to Control of dietary assessment measurement errors key to reliable nutritional associations with chronic disease.reliable nutritional associations with chronic disease.

Simple calibration equations that incorporate any of the of Simple calibration equations that incorporate any of the of the dietary assessments considered may be adequate for the dietary assessments considered may be adequate for epidemiologic purposes, for energy, protein and protein epidemiologic purposes, for energy, protein and protein density. Biomarker data across a time period relevant to density. Biomarker data across a time period relevant to disease risk are needed. disease risk are needed.

Development of objective measures for other Development of objective measures for other nutrients/aspects of physical activity, and methodologic nutrients/aspects of physical activity, and methodologic research as to how to use existing nutritional measures, research as to how to use existing nutritional measures, should have a high priority in nutritional epidemiology should have a high priority in nutritional epidemiology research agenda.research agenda.

Measurement Models for Nutritional Epidemiology

(Carroll, Freedman, Kaaks, Kipnis, Spiegelman, Rosner, Prentice…)

Recovery Biomarkers:

Wbiomarker = Z + eQself-report = a0 + a1Z + a2V + (r + ε)

Can estimate odds ratios (Sugar et al, 2007, Biometrics), or hazard ratios (Shaw and Prentice, 2011, to appear Biometrics), corresponding to Z from cohort data on W and subcohort data on X.

Hazard Ratio Estimation

Under a joint normality assumption for (Z, r + e) given V, conditional expectation of Z of the form

E (Z│Q,V) = b0 + b1Q + b2V and

E(Z I Q,V) = E (W │Q,V)

•Calibrated estimates of Z from linear regression of W on (Q,V) in the biomarker subsample

•Regression calibration estimation of hazard ratios by inserting calibrated consumption estimates in Cox regression, and using a bootstrap procedure for standard error estimation

Hazard Ratio Estimates for a 20% Increment in Nutrient Consumption in Women’s Health Initiative

(Prentice et al, AJE, 2009; Prentice et al, Epidemiol, 2011)

Hazard Ratio Estimates for a 20% Increment in Calibrated FFQ Energy, Protein, and Protein Density in Relation to

Various Cardiovascular Disease Outcomes, from 80,330 Women Enrolled in the WHI Dietary

Modification Trial Comparison Group or Observational Study