This document summarizes Paolo Vineis' presentation on measuring the exposome. It discusses: 1. Defining the exposome as the totality of environmental exposures from conception onward, including measuring internal exposures through biomarkers in biological samples. 2. Challenges in exposome research like limited biobanked samples, single spot samples, lack of life-course cohorts, and feasibility of extensive exposure assessment and omics measurements. 3. The "meet-in-the-middle" approach which integrates epidemiology, exposure assessment, omics, and bioinformatics to study cancer risk factors using samples from existing cohorts.

1. Paolo Vineis
Imperial College
Towards the Exposome
Manchester 10 September 2013

2. Environmental PAFs for cancer (Global
burden of disease, WHO)
2
SM Rappaport
Data from Ezzati et al., “Comparative Quantification of Mortality and Burden of Disease Attributable to
Selected Risk Factors,” Global Burden of Disease and Risk Factors, Chapter 4, WHO, 2006.

3. A self‐fulfilling prophecy: are we underestimating the role
of the environment in gene‐environment interaction
research?
( P Vineis Int J Epidemiol 2004)
According to estimates, the common genotyping method Taqman
has 96% sensitivity and 98% specificity, thus allowing little error in
classification. On the contrary, sensitivity in environmental
exposure assessment is quite often lower than 70% and specificity
even lower.
Genotype is stable, measured accurately (sens, spec=90-100%),
frequency of alleles is high
Environmental exposures are changing (life-course events), often
measured inaccurately, frequency may be too low

4. Some environmental exposures can be studied by epidemiology with
confidence , i.e. measurement error is relatively low and has little
impact on estimates (e.g. smoking). Advancement in exposure
assessment due e.g. to GIS techniques for air pollution.
When measurement error is too high we need biomarkers (e.g. number
of sexual partners, OR for cervical cancer around 2; HPV strains, OR
around 100‐500).

5. Discoveries that support the original model of molecular epidemiology
Marker linked to exposure or disease
Internal dose
Urinary metabolites (NNK, NNN)
Biologically effective dose
DNA adducts
Albumin adducts
Hemoglobin adducts
Preclinical effect
Chromosome aberrations
HPRT
Glycophorin A
Gene expression
Genetic susceptibility
Phenotypic markers
SNPs
NAT2, GSTM
CYP1A1
Vineis and Perera, 2007
Exposure
Nitrosocompounds in tobacco
PAHs , aromatic compounds
AFB 1
Acrylamide, Styrene,
1,3-Butadiene
Exposure and/or cancer
Lung, Leukemia,
Benzene
PAHs, 1,3-Butadiene
PAHs
Cisplatin
DNA repair capacity in head
and neck cancer
Bladder
Lung

6. Exposome - definition
The exposome concept refers to the totality of environmental
exposures from conception onwards, The internal exposome is based
on measurements in biological material of complete sets of biomarkers
of exposure, using repeated biological samples especially during
critical life stages.
Biomarkers which can be measured in this context cover a wide range
of molecules, ranging from xenobiotics and their metabolites in blood
(metabolomics) to covalent complexes with DNA and proteins
(adductomics).
The term omics generally refers to the rigorous study of a complete set
of biological and non-biological molecules with high-throughput
techniques (Rappaport and Smith 2010).

7. SM Rappaport
7

8. Prospective cohorts provide an ideal context in which to bring
together the best laboratory science, epidemiology, biostatistics
and bioinformatics to investigate cancer risk factors.
To realize this potential, however, requires a commitment to
develop and adapt laboratory tools for application to the bio‐
specimens being collected.
In addition, emphasis is needed on the collection and processing
of biological specimens in a manner, as far as is predictable,
consistent with the future laboratory analyses and avoid biases
at the time of sampling.

9. Challenges:
1. precious and limited biobanked material, not easily released
by PIs
2. single (spot) biological samples
3. usually blood, not urine (which may be better e.g. for
metabolomics)
4. no cohorts allow life‐course epidemiology
5. in‐depth exposure assessment is limited by feasibility (for
cancer you need large sample sizes)
6. lab measurements and omics have the same limitations
related to sample size and feasibility
7. biostatistical approaches and causal interpretation
8. ethical issues

10. The “meet‐in‐the‐middle” approach

11. Schematic representation of the implementation of the ‘meet-in-the middle’ approach
(Chadeau-Hyam et al, Biomarkers 2011).

12. Main finding in pilot study from EPIC-Italy on colon
cancer - Role of gut microbiota? Concept of “gut
health”
No markers found in association with breast cancer, 8 signals
found in association with colon cancer (Chadeau-Hyam et al,
2011)
Dietary fibers intake was found to be associated to four putative
markers out of 235 (with corresponding p-values ranging from
0.003 to 0.02).
One marker indicates a possible link with gut microbial
fermentation of plant phenolics in the colon (Nicholson et al.,
2005, Phipps et al., 1998, Aura, 2007), a process also plausibly
linked to higher dietary fibers exposure and lower colon cancer
risk.

13. Second study nested in EPIC on colon
cancer
•194 cases, 194 controls (age and sex‐matched) from EPIC‐Italy
•134 non‐polar metabolites measured using LC‐MS (Imperial
Lab) – e.g. lipids
•Investigation of metabolite levels in relation to colorectal
cancer risk
•Association of metabolites with colorectal cancer risk factors (in
particular obesity and dietary factors)

14. Metabolomics and Colon Cancer Risk:
Multivariate Model
Metabolite
Decanoylcholine
PC(18:4(6Z,9Z,12Z,1
5Z)/P‐16:0
PC(22:4(7Z,10Z,13Z,
16Z)/22:5(7Z,10Z,13
Z,16Z,19Z))
PE(20:5(5Z,8Z,11Z,1
4Z,17Z)/P‐16:0)
10,11‐dihydro‐
leukotriene B4
Tetracosanoic Acid
1OR
Class/Pathway
Fatty Acid
Phospholipid
OR1 (95% CI)
2.44 (0.009)
0.63 (0.047)
Phospholipid
0.55 (0.016)
Phospholipid
0.62 (0.039)
Arachidonic Acid
2.76 (0.02)
Arachidonic Acid
0.43 (0.04)
for 1-log unit change in metabolite level

15. Choline Metabolism

16. Nature 2008

17. Supervised analysis
defined profiles robustly
associated with known
dietary factors

18. AHRR
1x10‐7
1x10‐5

19. 1.0
0.0
MI - Validation
1.0
0.8
0.6
0.4
0.2
0.0
1500
2500
MI - Test
0 500
ng/mL
0.6
0.4
0.2
Sensitivity
0.8
Cotinine vs Smoking Status
Never
Former
Current
Specificity
Exposure marker - AHRR methylation is strongly associated with former smoking
(first marker of past smoking). (Shenker et al, Human Molecular Genetics 2013;
Epidemiology, in press)

20. Basic components of the EXPOsOMICS project
1. Select and integrate subjects, samples and data from 3 types of existing
studies that collectively reflect all life stages from conception to old age
(WP2):
Experimental Short‐Term Studies (STS) including Oxford Street study
Mother‐Child Cohorts
(MCO)
Adult Long‐Term Studies
(ALTS)
2. Measure the external exposome component for air and water
contaminants by performing extensive, repeated Personal Exposure
Monitoring (PEM) (WP3‐4).
3. Measure the internal exposome in fresh and archived samples (WP5‐7).
Fresh blood samples will be collected from the individuals undergoing PEM,
i.e. individuals in STS and those from a representative subsamples from MCO
and ALTS.

22. A conceptual model of life-course disease risk
Population studies of chronic diseases have traditionally recruited middleaged subjects. However, there is strong evidence that (a) the risk of disease is
influenced by early exposures, including in utero; (b) life-stages include critical
periods (during which changes in exposure have long-term effects on disease
risks or related, intermediate markers) and sensitive periods (during which an
exposure has stronger effect on development and, hence, disease risk than at
other times).
The idea of a sequence of critical and sensitive periods leads to the concept of
"chain of risk", i.e. the interplay of early exposures and late exposures. To use
this concept in practice implies having access to multiple life-stages in
exposure assessment and epidemiological studies, and repeated measurements
of biomarkers at different time windows. This approach requires an intergenerational epidemiological study design and novel statistical analyses.

23. Critical stages of life
RAPTES
PISCINA
20
SAPALDIA
EPIC‐
ESCAPE
Age
30
ALSPAC
10
PISCINA
INMA
OXFORD ST
0
RHEA
50
PICCOLI+
Birth
60
MCC
Mid‐ and late‐life
EPICURO

24. PEM device
Sensor Pack
MicroAeth
(BC)
UFP Sensor
SmartPhone
EXPOsOMICs App
USB
USB
USB
Rechargeable
Battery Pack
(use 5V o/p.)
‐GPS position
‐Accelerometer (user activity)
‐Altimeter
‐Compass
‐user I/O (questionnaire)
‐Download of logged data (above) via USB
‐Application setup
USB hub
‐ “Dedicated” smart phone in a pouch on sensor pack to enable user input/output (i.e. we do not intend to
“leverage” the user’s personal phone, at this stage).
‐Rechargable Li battery pack supplies power to instruments and smartphone via USB hub for 36 hr autonomy.
‐Each Sensor and Smartphone log data independently (synched in time during initial setup).

25. Oxford Street: ‘active’ exposure

26. Hyde Park: ‘control’ exposure

27. exposures; ‘cross‐over’
•
•
gentle walking for two hours (6km)
Oxford Street and Hyde Park
– random order
– separated by >3 weeks
•
contemporary measurements of exposures
PM10‐2.5
PM2.5‐0.1
PM0.1
black carbon
ultrafine particles
NO2/x
temperature
relative humidity.

28. Need for new biostatistical tools and causal interpretation
- repeat samples and intra-individual variation
- validation of omics: Hebels et al, EHP
- quality controls (e.g. nuisance parameters: Chadeau-Hyam et al, submitted)
- “cross-omics”
- longitudinal models of causality
Chadeau-Hyam M, et al. Deciphering the complex: Methodological overview
of statistical models to derive OMICS-based biomarkers.
Environ Mol Mutagen. 2013 Aug;54(7):542-57.
Hebels et al. Performance in omics analyses of blood samples in long-term
storage: opportunities for the exploitation of existing biobanks in
environmental health research. Environ Health Perspect. 2013 Apr;121(4):4807.
Vineis P, et al. Advancing the application of omics-based biomarkers in
environmental epidemiology. Environ Mol Mutagen. 2013 Aug;54(7):461-7.

29. Longitudinal HMM for smoking‐induced lung cancer (Chadeau‐Hyam et
al., Epidemiology in press)
• Exposome concept: risk of chronic diseases is not only driven
by the exposure level itself, but also by its evolution in time and
by potential temporal patterns in the exposure history.
Include a temporal component in causal inferences
• Definition of a longitudinal compartmental model (SIR-type)
S
m
PS-I
I
m
PI-R
R
M
S: healthy; I: growing and undiagnosed tumor; R: diagnosed;
M: other cause mortality. S and I are hidden (SUI is observed)

30. The end