1. IRCCS San Raffaele Pisana, Rome, Italy, 28 February - 2 March 2018
Why do we need epidemiology and statistcs in human studies with biomarkers ?
IRCCS San Raffaele Pisana, Rome, Italy, 28 February - 2 March 2018

3. The past ...
I received my degree in Biological Sciences in 1981 at the University of Genoa and the specialty in Epidemiology and Biostatistics 6 years later, at the University of Pavia.
My career started as occupational and environmental epidemiologist.
In early 1990’s I started working on the design and analysis of human studies based on biomarkers of chromosomal damage, entering the realm of Molecular Epidemiology.

4. The present ... My current interests are:
- Validation of biomarkers of chromosome damage as predictors of cancers (and other diseases) in humans
Standardization and enhancement of the Micronucleus test in PBL (the HUMN project) and exfoliated buccal cells (The HUMNxl project)
Improvement of statistical methods in human biomonitoring studies
Statistical analysis of microarrays and other high-throughput techniques
- Creation of platforms of systems medicine (Respiratory and psychiatric diseases)

5. The future ... I am planning researches on
Use of high throughput technologies in human studies
Design and Statistical analysis
Application in public health of biomarkers validated as risk predictors
Validation of various indexes of individual genetic susceptibility as predictors of adverse outcomes
Molecular epidemiology of neurodegenerative disorders
Molecular epidemiology of respiratory and psychiatric diseases
Geriatrics: Markers of (well) aging
Systems Biology and Systems Medicine

6. Epidemiology
Laboratory

7. External Exposure
Cancer

8. Genetic Polymorphisms
External Exposure
Endogenous
Exposure
Genetic Polymorphisms
Target Tissue
SurrogateTissue
Cancer

9. Deterministic model Stocastic model
The major reason for using epidemiology in biomarkers studies is that this allows us to move from a
Deterministic model
Good for Mutations
Good for Polymorphisms
Stocastic model

10. ENVIRONMENT GENES DISEASE
Biomarkers of Susceptibility
Biomarkers of exposure
Biomarkers of early effect
DISEASE

11. SNParray and individual to Malignant Mesothelioma
Susceptibility
to Malignant Mesothelioma

12. Study design A Molecular Epidemiology Case-Control Study
93 mesotheliomas
115 lung cancers
128 controls
Collected from Pneumology departments of hospitals located in Liguria, Italy

13. The Array
101 GENES, 247SNPs
ABCG2 (2), ADH1B(3), ADH1C, AHR, ALDH2(4), APEX/APE1(2), ATM(6), ATR,BARD1(2), BRCA1(5), BRCA2(3), CASP-10CASP-3, CASP-8, CASP-9,CCND1(2), CDA(2),CDK7,CDKN1B, CDKN2A(3), CDKN2B(2), CHEK2,COMT(4), CYP1A1(6), CYP1A2(4), CYP1B1(6), CYP2A6(4), CYP2C18, CYP2C19, CYP2C9(2), CYP2D6(9), CYP2E1(6), CYP3A4, DRD2(7), DRD4(5), EPHX1(4), ERCC1(5), ERCC2, ERCC2 / XPD(2), ERCC4 / XPF, ERCC5 / G(2), FANCD2, GADD45A, GRPR, GSTA2, GSTA4(2), GSTM1, GSTM3(3), GSTP1(3), GSTT2(2), IRS2(3), LIG1(4), LIG3, LIG4(2), MDM2(4), MDR1(3), MGMT/AGT(4), MLH1, MnSOD2, MSH2, MSH3(3), MSH6(2), MTHFR(3), MYH(2), NAT1(7), NAT2(6), NBS1(2), NOD2/CARD15(4), NQO1/DIA4(2), NT5E,OGG1, p21/ KN1A(2), PARP/ADPRT(6), PCNA(3), PMS2, POLB(2), RAD23, RAD51(2), RAD52, RAD54B, RAD9, RB1(2), RECQL, SLC6A3/DAT1(2), SULT1A1(2), SULT1A2(2), TERC, TERT, TP53(3), TP53BP1(2), TP53BP2(2), TPMT(3), UGT1A7(2), XPA, XPC, XRCC1(3), XRCC2(3), XRCC3(3), XRCC4, XRCC5(2), XRCC9

14. SNP Mutated Odds Ratio P
CYP1B <0.001
CYP2D <0.01
MnSOD <0.001
GSTM ~ 0.02
NAT2 fast ~ 0.02
mEH low-activity <0.05
XRCC <0.01

15. A continuum of biomarker categories reflecting the
carcinogenic process resulting from xenobiotic exposures
Exposure
Biologically
Early
Internal
Effective
Biological
dose
dose
Effect
Altered
Structure/
Function
Susceptibility
Disease
Committee on biological markers of the National Research Council. EHP, 1987.

16. European Study Group on Cytogenetic Biomarkers and Health - ESCH

17. Multivariate Cox Regression Analysis of CA Frequency
Hagmar et al., Cancer Res, 1998

18. Kaplan-Meier curves for total cancer incidence and mortality
Survival refers to time from CA test to first cancer diagnosis or cancer death
Hagmar et al., Cancer Res, 1998

19. The p value for a hypothesis test is the probability of obtaining, when H0 is true, a value of the test statistic as extreme as or more extreme than the one actually computed.
using p values is more informative than just saying significant or not significant, but

20. The common epidemiologic approach to the study of health effects recognizes as the principal aim of a study that of quantifying exposures-effect relationships, rather then simply evaluate whether an effect is present.

21. A point estimate of the association between exposure and effect should be computed, along with its confidence interval, so as to provide a range of values within which the true association falls with a given probability.
and this concepts bring us to the measures of association !

22. Measures of association reflect the strength of the statistical relationship between a study factor and a disease

23. The most useful and widely used measures of association in Epidemiology and Clinical Investigation are:
Risk difference, Incidence Rate Difference, Risk Ratio, Incidence Rate Ratio, and Odds Ratio

24. Possible benefits to using biomarkers in human studies
Biologic phenomena not otherwise measurable
Reduce impact of differential exposure missclassification
Introduce individual parameters into the study design

25. Possible benefits to using biomarkers in human studies
Pre-disease Events/States
Pre-clinical disease
Link exposure to natural history
Shorter studies
Increase disease homogeneity (etiologic or prognostic)

26. Possible benefits to using biomarkers in human studies
Make better inferences ref. to disease mechanisms
Improve process of causal inference
Improve risk assessment
Improve prevention strategies
Improve disease treatment