Sarah Berry, Nicola Segata, Jose Ordovas and Tim Spector reveal novel findings from the world's largest ongoing nutrition study, PREDICT. The presentation shares learnings on how we metabolize food, the importance of food sequencing and combining, the gut microbiome and inflammation. These findings are some of the most cutting edge in the field of nutrition science, highlighting the need for precision nutrition. Learn more at www.joinzoe.com/science

1. Findings from PREDICT:
The Personalized Responses
to Dietary Composition Trial
1

2. Speakers
Tim Spector, Ph.D
Professor of Genetic Epidemiology
King’s College London, UK
Sarah Berry, Ph.D
Senior Lecturer, Department
of Nutritional Sciences
Nicola Segata, Ph.D
Associate Professor,
Computational Metagenomics
Jose Ordovas, Ph.D
Director and Senior Scientist,
Nutrition and Genomics Lab
2

3. The PREDICT Programme
Tim Spector, Ph.D
Professor of Genetic Epidemiology
King’s College London, UK
3

4. Largest ongoing program to measure
postprandial responses to food in nutritional science
Nutrition
academia
Tech
companies
4Tim Spector King’s College London, UK

5. Jun 2018 - May 2019 Jan 2019 - May 2019
n=1,100
PREDICT - Carbs
n=100
Healthy
60% TwinsUK
Healthy Compliant
individuals from
PREDICT 1
Postprandial
responses
Meal order
Time of day
Clinic & Home
(UK/US)
Home (UK)
STATUS:
Complete
STATUS:
Complete
Validation of remote
study delivery
Postprandial responses
Microbiome profiling
Jun 2019 – Mar 2020
n=1,000
Healthy
Ethnicity
Home (US)
STATUS:
Complete
PREDICT - Cardio
Sep 2019 - Ongoing
n=50
High/Low CVD
risk subset of
PREDICT 1 Plus
Cardiometabolic
end-points
Liver fat
Vascular function
Clinic & Home
(UK)
Continuation of
PREDICT 1
Postprandial
responses
PREDICT 1 Plus
n=900
Jun 2019 - Ongoing
Healthy
TwinsUK
Clinic & Home
(UK)
STATUS:
n=250
Postprandial responses
Dietary assessment
Microbiome profiling
July 2020
n=10,000
Healthy
Ethnicity
Chronic disease
Home (US)
STATUS:
Pending enrolment
P R E D I C T 2
PREDICT
Postprandial
responses
P R E D I C T 3
5Tim Spector King’s College London, UK
P R E D I C T 1
STATUS:
Complete

6. Nutritional advice in the past and present…evidence is constantly evolving
Guidelines take on ‘one-size fits all’ approach
6Tim Spector King’s College London, UK
Lack of consensus
even amongst
expert nutritional
scientists

7. Even the world leading experts don’t agree…
7Tim Spector King’s College London, UK
With perfect agreement
1
2
3
1
2
3
4
5
6
7
8
9
10
11
12
13
In reality
13 independent experts
105 food items tested
Why?
We are complicated
and food
is complicated

8. Whole almonds Ground Almonds
0
50
100
150
200
Kcal/serving
• Thousands chemicals/nutrients in each food;
not just the nutrients listed on a label
• Nutrient-nutrient interactions
• Food matrix
• Processing
Why is food complicated?
Meals/foods containing same ingredients and/or
composition but differing in matrix can have
completely different effects
What does this mean?
That’s because food is complicated…
8Tim Spector King’s College London, UK

9. How meaningful is the mean?
9Tim Spector King’s College London, UK
Large inter-individual variability
in responses to food
Nutritional studies typically
show the mean response
Manipulation of lipid bioaccessibility of almond seeds influences postprandial
lipemia in healthy human subjects. Berry et al (2007)
The need for personalisation
Nutritional advice is based
on population averages
A difference of 106 kcal per serving
~742 kcal per week
4.6
2.3 kcal - 6.0 kcal range
62 kcal 168 kcal
“Poor”
metabolisers
“Good”
metabolisers
Average serving
28g
Metabolizable Energy from Nuts
kcal
Discrepancy between the Atwater factor predicted and empirically measured
energy values of almonds in human diets. Novotny et al (2012)
But how many of
us are average?

10. Quality&Precision
Quantity
Digital devices
• Mobile phone apps
• Clinical devices, continuous
glucose monitors, activity monitors
Remote clinical testing
DNA, microbiome, blood tests
Citizen science
Today we can look beyond the mean
10
Traditional
Randomized
Controlled Trials
Big Data
Opportunities
Tim Spector King’s College London, UK
Duplicate
Diet Records
Epidemiological
Studies
Food
Frequency
Questionnaires

11. Nutrition
academia
Tech
companies
Quality
and quantity
11
Largest ongoing program to measure
postprandial responses to food in nutritional science
11Tim Spector King’s College London, UK

12. What explains
these differences?
Can we PREDICT
individual responses using
machine learning?
MEAL
CONTEXT
GENETICSMICROBIOME
AGE /
SEX / BMI
MEAL
COMPOSITION
2. 3.
How much variability
between people?
1.
By using genetic, metabolomic, metagenomic and meal-context
information to predict individuals’ response to food
12Tim Spector King’s College London, UK

13. What makes our work so unique….we are looking at the integrated response and
interrelated multi-directional pathways….
13Tim Spector King’s College London, UK

14. How we respond to food… using postprandial responses
Single Meal
2
1.5
1.0
0.5
0 100 200 300 400 500
Minutes since breakfast
7
6
5
Traditional measures of disease risk
Typical Day
50g fat, 85g carb. AJCN. 2011. 94, 1433-41. n=50
2.0
1.5
1.0
06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 00:00 02:00 04:00Time
7
6
5
Breakfast Lunch DinnerSnack Snack Snack
PLASMA TRIACYLGLYCEROL PLASMA GLUCOSE INSULIN
PlasmaTriacylglycerol(mmol/l)
PlasmaTriacylglycerol(mmol/l)
PlasmaGlucose(mmol/l)
PlasmaGlucose(mmol/l)
14Tim Spector King’s College London, UK

15. Time since breakfast
Astley, C. M. et al. Clin Chem 64, 192-200, doi:10.1373/clinchem.2017.280727 (2018). Bansal, S. et al. JAMA 298, 309-316, doi:10.1001/jama.298.3.309 (2007). Lindman, A. S., . et al. . Eur J Epidemiol 25, 789-798, doi:
10.1007/s10654-010-9501-1 (2010). Nordestgaard, B. G., . et al. . JAMA 298, 299-308, doi:10.1001/jama.298.3.299 (2007). Levitan, . et al. . Arch Intern Med 164, 2147-2155, doi:10.1001/archinte.164.19.2147 (2004).
Why do post-prandial peaks matter?
15
PLASMA TRIACYLGLYCEROL
PLASMA GLUCOSE
INSULIN
• Lipoprotein
Re-modelling
• Oxidative Stress
Inflammation
• Endothelial
Dysfunction
• Raised Insulin
Secretion
• Hunger & Appetite
Impact…
• Cardiovascular
Disease
• Metabolic Disease
(Type 2 Diabetes,
Fatty Liver, Insulin
Resistance)
• Weight gain
Increased risk for…
Tim Spector King’s College London, UK

16. Controlled Time (Mins)
BLOOD
Fasting 30 6015 120 180 240 2700 300 360
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
Questionnaires
FFQ, Lifestyle &
Medical
Anthropometry
DEXA, waist/hip &
BMI
Blood pressure
and heart rate
Genetics,
Clinical assays &
Metabolomics
Metabolomics,
Saliva & Urine
SALIVA
SALIVA
Metabolic challenge
Other test meals
Aims
Use genetic, metabolomic, metagenomic and meal-context information to predict individuals’ postprandial responses to food.
Validation Cohort
n=100
Main Cohort
n=1,002
Home Phase (Days 2-14)
Dietary
Assessment
Standardised
meals
Blood
Spot tests
Digital
devices
Stool
samples
Study app; weighed records; in-study support
Nutritionally varied test breakfasts and lunches
TAG, C-peptide assays
Continuous glucose, physical activity and sleep monitoring
Microbiome profiling
Clinic
Day 2 3 4 5 6 7 9 10 13 1411 128
Baseline Clinic Visit (Day 1)
16Tim Spector King’s College London, UKJun 2018 – May 2019 IRAS 236407 IRB 2018P002078 NCT03479866

17. 2,022,000
CGM glucose
readings
Mean (SD)
Age (yr) 45.7 (12.0)
BMI (kg/m2) 25.6 (5.0)
Sex (%) 72 F/ 28 M
Triacylglycerol (mmol/L) 1.1 (0.5)
Insulin (IU/mL) 6.1 (4.3)
Glucose (mmol/L) 5.0 (0.5)
Total cholesterol (mmol/L) 5.0 (1.0)
n=1,002
MZ Twins 479
DZ Twins 172
Non-Twins 351
Drop-out 2.5%
The scale of the PREDICT 1 study
32,000
Muffins consumed
28,000
TAG readings
132,000
Meals logged
17Tim Spector King’s College London, UK

18. Decoding human responses
to food for precision nutrition
18
Sarah Berry, Ph.D
Senior Lecturer, Department of Nutritional
Sciences
King’s College London

19. Significant variability between healthy individuals
Baseline 6h rise
CV 50% 103%
Triacylglycerol Glucose Insulin
19Sarah Berry King’s College London, UK
Baseline 2h iAUC
CV 10% 68%
Baseline 2h iAUC
CV 69% 59%
TAG(mmol/L)
Glucose(mmol/L)
Insulin(mmol/L)
Clinic day data, n = 1,002

20. Intra-individual variability is lower than inter-individual variability
(6h rise, n=1018 meals at home and in clinic)
Interindividual CV is calculated for identical meals, between random pairs of individuals.
Intraindividual CV is calculated between pairs of nutritionally identical meals for the same individual
Triacylglycerol Glucose
(iAUC 0-2h, n=7898 meals at home)
20Sarah Berry King’s College London, UK
Differences
between
individuals are
repeatable
Intra-individual Inter-individual
CV 36% 68%
Intra-individual Inter-individual
CV 24% 40%

21. Identical twins have very different responses
Height Glucose
(iAUC 0-2h)
Triacylglycerol
(6h iAUC)
ACE Heritability Modelling Key:
21Sarah Berry King’s College London, UK
Genetics do not
explain most
nutritional
differences
Genetics Upbringing Environment
48%
5
%
48% 48% 52%46% 50% 4%
TwinMZ2GLU0-2hiAUC
Twin MZ 1 GLU 0-2h iAUC Twin MZ 1 Log scaled 6h TAG iAUCTwin MZ 1 Height
TwinMZ2Height
TwinMZ2Logscaled6hTAGiAUC

22. What causes the variability in responses?
22Sarah Berry King’s College London, UK
Glucose
iAUC 0-2h
R2 adjusted
Triacylglycerol
6hr rise
R2 adjusted
* P<0.05, ** P<0.01, *** P<0.001

23. Machine Learning can predict individual responses
Machine learning
model correlates
77%
to measured
glucose responses
23
Machine Learning model
uses test results to predict
responses to new meals
Individual takes test
Sarah Berry King’s College London, UK
Pearson R = 0.77; p = 0

24. High resolution and high density measures allows deep dive
24Sarah Berry King’s College London, UK
Meal context
Time of day, Meal sequence,
Exercise
Lipoprotein re-modelling
Oxidative stress
Inflammation
Endothelial dysfunction
Raised insulin secretion
Meal composition
Genetics
Microbiome
Age
Blood pressure
Serum measures
Anthropometry
Habitual diet
Sex
Hunger & appetite

25. 0
2000
4000
6000
8000
10000
12000
14000
MealiAUC(mmol/l*2h)
Breakfast Lunch
Time of day influences the postprandial response,
but as we age, this largely disappears
13:00
25Sarah Berry King’s College London, UK
9:00
+4h

26. Home-based Intervention (Days 2-13)
Study app; weighed diet records; in-study support
Set-up
Day 1 2 3 4 5 6 7 9 10 118
Continuous glucose, physical activity and sleep monitoring
TAG TAG
12 13
Nutritionally varied test meals: breakfast, lunch, snack, sweetener preloads
Daily intervention timeline
Free-livingFasted 3-4h fast 3h fast 2h fast
Breakfast Lunch Snack
Microbiome profiling
Dietary
Assessment
Standardised
meals
Blood
Spot tests
Digital
devices
Stool
samples
1. Effects on glycaemic responses of:
i. carbohydrate staple foods
ii. composite meals
iii. sweetener preloads
iv. time-of-day
v. meal-sequence
2. Collect free-living dietary intake and energy
expenditure data to validate the Zoe app
Aims
Single-staple breakfasts
White bread Rye bread Pasta Mashed potatoes Rice
Composite staple lunches
White bread
& cheese
Rye bread
& cheese
Spaghetti
bolognaise
Cottage pie Chickpeas
& chicken
Staple snacks & sweeteners
Sucralose, Aspartame,
Stevia preloads
Biscuits and juicePotato
crisps
Chocolate
bar
Dietary Intervention
n=100
26Sarah Berry King’s College London, UK 26Jan 2019 – May 2019 IRAS 236407 NCT03479866

27. 27INTERIM UNPUBLISHED DATA
Using carbohydrate-rich test meals at breakfast/lunch
Type of breakfast can impact glycaemic response at lunch eaten 4 hours later
Sarah Berry King’s College London, UK
Lunchtime response affected by breakfast meal

28. Breakfasts
Time since breakfast (min)
Glucose dip
Divergent responses seen 2-3 hours after a meal
28Sarah Berry King’s College London, UKINTERIM UNPUBLISHED DATA

29. High resolution and high density measures allows deep dive
29Sarah Berry King’s College London, UK
Lipoprotein re-modelling
Oxidative stress
Inflammation
Endothelial dysfunction
Raised insulin secretion
Hunger & appetite
Meal context
Time of day, Meal sequence,
Exercise
Meal composition
Genetics
Microbiome
Age
Blood pressure
Serum measures
Anthropometry
Habitual diet
Sex

30. Impact of glucose ‘dips’ on hunger and calorie intake
30
Individuals with smallest 25% of Glucose Drop (Q1),
versus largest 25% of Glucose Drop (Q4)
Sarah Berry King’s College London, UK
Glucose dip UK average meal (mmol/L)
Large inter-individual variability
in ‘dippers’ – even in response
to the same meal
Frequency
+9 (+/-4) -2 (+/-2) -25 mins(+/-10) +79 Kcal(+/-28) +321 Kcal(+/-87)
Hunger+2-3hpost-meal–pre-meal
Alertness+2-3hpost-meal
Minutessincefirstmeal(mins)
Calories(Kcal)
Calories(Kcal)
Q1 Q4 Q1 Q4 Q1 Q4 Q1 Q4 Q1 Q4
Increase in
hunger
Alertness
level
Time until
next meal
Calories
eaten 3-4h
Calories eaten
in full day
n=689, meals=5693

31. iAUC?
Increase
from fasting?
Duration?
Peak
concentration?
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0 100 200 300 400 500
PlasmaTriacylglycerol(mmol/l)
Minutes since breakfast
Late postprandial rise?
The response curve is complicated – what feature matters most?
31Sarah Berry King’s College London, UK
• Peak Concentration?
• iAUC?
• Increase from fasting?
• Duration?
• Late postprandial rise?
AJCN. 2011. 94, 1433-41

32. Postprandial Lipoprotein remodelling associated with CVD
Inflammatory marker associated with CVD
Outcomes from the metabolomic analysis include measures
of cardiovascular inflammation and atherogenic lipoprotein remodelling
Lipoprotein subclasses Lipoprotein compositional re-modelling
32Sarah Berry King’s College London, UK
HDL VLDL
CE
TAG
PL
CE
TAG
PL
Cholesteryl ester
Transfer protein
Cholesteryl ester
Transfer protein
Diameter (nm)
Density(g/ml)
Postprandial-induced inflammation
IL-6 and GlycA
Incident CVD; FINRISK (n=7256), SABRE (n=2622), British Woman's Health Study (n=3563); Circulation, 2015.

33. The 6h increase in TAG from fasting values is most strongly
associated with intermediary cardiometabolic risk factors
33Sarah Berry King’s College London, UK
Atherogenic
lipoproteins and
inflammatory markers
most closely associated
with a prolonged
postprandial response
0.1
0.2
0.3
0.4
0.5
n=1,002
Remnant-C
HDL-C
Gp
XL-VLD-C
XL-VLD-P
XXL-VLD-C
XXL-VLD-P
4h
TAG iAUC
6h
TAG iAUC
4h
TAG
Concentration
4h
TAG
increase
6h
TAG
Concentration
6h
TAG increase

34. Food-induced inflammation is highly variable, clinically relevant
and determined mainly by postprandial lipaemia
IL-6 GlycA
• Significant, clinically
relevant increase
postprandially
• Correlated with peak in
glucose (r=0.239) and TG
(r=0.832)
• ML shows lipaemia is
stronger determinant,
specifically TG 6h-rise
• Increases
postprandially
• Not correlated with
glycaemica or lipaemia
Time (min)
0
(Breakfast)
240
(Lunch)
360
10
15
20
25
30
Interleukin-6(mmol/l)
10
15
20
25
30
Glycoproteinacetylation(mmol/l)
Time (min)
0
(Breakfast)
240
(Lunch)
360
34Sarah Berry King’s College London, UK
Doubled ASCVD risk for
individuals with >90th
percentile GlycA increase

35. Neck-to-knee XMR imaging
Liver lipid quantification
PREDICT 1 PlusPREDICT - Cardio
Pulse wave velocity Plaque grading
Carotid intima-media thickness
1. Measure clinical intermediary cardiometabolic
outcomes, in fasted PREDICT 1 Plus participants
2. Determine the link between postprandial glycaemic/
lipaemic responses and cardiometabolic disease
Clinic
Day 1
2 3 4 5 6 7 9 10 118
Aims
35Sarah Berry King’s College London, UK
Sub-cohort of PREDICT 1 Plus
n=50
Sep 2019 – Feb 2020 IRAS 236407 NCT03479866

36. Summary
36Sarah Berry King’s College London, UK
• Everyone is unique in food response
– even identical twins
• Genetics explains only a fraction of
how we respond to foods
• The composition of the meal
explains
<30% of our responses
• How and when we eat can impact
our response to food
• Dissecting the integrated, inter-
related multidirectional pathways
with large scale, high resolution data
makes precision nutrition achievable

37. The hidden human microbiome
diversity and personalized host-
microbiome interaction
37
Nicola Segata, Ph.D
Associate Professor, Computational Metagenomics,
CIBIO, University of Trento
@nsegata

38. Species-level
features
#samples 232 121 110 253 344 96
Cirrhosis Colorectal IBD Obesity T2D WT2D
Pasolli et al., PLoS Comput Biol, 2016
Human Microbiome Project, MetaHIT
Thomas, Manghi et al., Nat Med, 2019
Nine total (and
concordant) datasets
for Colorectal cancer
But what about the microbiome in “health” levels and pre-disease?
What is health and what is disease for the microbiome?
38Nicola Segata University of Trento, Italy
The microbiome in disease The microbiome in health

39. On the road to link the gut microbiome with food and metabolic health
39Nicola Segata University of Trento, Italy

40. Truth is, we are all unique
Even identical twins have very different responses to food
40Nicola Segata University of Trento, Italy

41. Subjects from around the world (~3000 sbj from 4 continents)
Subjects from US (from two universities)
Subjects from EU (6 countries)
Samples from same subjects collected at ~6 months
Subjects from EU
Subjects from US
Truong et al., Genome Research, 2017
Truth is, also our microbiome is unique
0.250
0.275
0.300
0.325
0.350
0.400
0.425
0.375
Commonspecies(%)
Twin pair Unrelated
41Nicola Segata University of Trento, Italy

42. The PREDICT 1 study at ZOE
42
1,002 samples
Predict 1 (UK Cohort)
100 samples
Predict 1 (US Cohort)
Food Frequency Questionnaire
Clinic data
Serum metabolomics
Continuous Glucose Monitor
Stool metagenomics
The microbiome data
769 species
29 spp. 90%
prevalent
91 spp. 50%
prevalent
174 spp. 20%
prevalent
UniRef90
1,910,069
UniRef50
878,520
KEGG KOs
6,163
Pathways
445
Taxonomic Functional Assembly
48,181 MAGs
29,035 MQ
19,146 HQ
8.8 avg. 2.2 sd. Gb/sample
58.3 avg 14.6 sd. Mreads/sample
Metadata
Age, BMI, weight, waist/hip
ratio, visceral fat, antibiotic
usage, blood pressure
Personal
Tot. 18
Foods, Food groups,
Nutrients, Nutrients % kcal,
Dietary patterns
Habitual diet
Tot. 275
Lipoproteins, ApoLipoproteins,
Risk scores, Glucose
mediated, FA’s metabolism
Fasting
Tot. 247
All fasting measures up to
7 timepoints including max
values and rises
Post prandial
Tot. 484
42Nicola Segata University of Trento, Italy

43. The workflow
of shotgun
metagenomics
43Nicola Segata University of Trento, ItalyQuince, Walker, Simpson, Loman, and Segata. Nature Biotechnology, 2017

44. The PREDICT 1 study at ZOE
4444Nicola Segata University of Trento, Italy

45. How strongly is the microbiome linked with habitual diet?
45Nicola Segata University of Trento, Italy

46. And the associations re reproducible in the US cohort!
aMed scoreHFD
46Nicola Segata University of Trento, Italy

47. What are the microbes most associated with foods?
47
Beef,
poultry,
sausages,
pork, ham
Shellfish,
oily fish,
eggs
Red wine
YogurtJam Dark
Chocolate
Tomato
Ketchup
Sauces
Support for health-association for the two clusters: p-value
< 1e-20
Baked
beans
47Nicola Segata University of Trento, Italy

48. What are the microbes most associated with foods?
48
Healthy plant-based Less healthy plant-based Unhealthy animal-based Healthy animal-based
48Nicola Segata University of Trento, Italy

49. What are the microbes most associated with habitual diet?
4949Nicola Segata University of Trento, Italy

50. From microbe-food links to microbe-obesity links
50Nicola Segata University of Trento, Italy

51. Postprandial Lipoprotein remodelling associated with CVD
Inflammatory marker associated with CVD
Metabolomics for measures of cardiometabolic health
Incident CVD; FINRISK (n=7256), SABRE (n=2622), British Woman's Health Study (n=3563); Circulation, 2015.
Lipoprotein subclasses Lipoprotein compositional re-modelling
HDL VLDL
CE
TAG
PL
CE
TAG
PL
Cholesteryl ester
Transfer protein
Cholesteryl ester
Transfer protein
Diameter (nm)
Density(g/ml)
Postprandial-induced inflammation
IL-6 and GlycA
51Nicola Segata University of Trento, Italy

52. The gut microbiome and fasting cardio-metabolic markers
52Nicola Segata University of Trento, Italy

53. 53
The gut microbiome and fasting cardio-metabolic markers
53Nicola Segata University of Trento, Italy

54. Fasting and postprandial level are more predictable than rises
Total lipid concentrationsInflammatory measures Lipoprotein particle sizeLipoprotein concentrations Glycaemic mediatedApoLipoproteins
54Nicola Segata University of Trento, Italy

55. 5555Nicola Segata University of Trento, Italy

56. An overall signature of the “healthy” microbiome “in health”
56Nicola Segata University of Trento, Italy

57. • Thousands of
unknown species
• Millions of
unsampled genes
• Missing links with
diseases and conditions
…but a lot of the human microbiome diversity is still unexplored
Mapping reads (%)
Yes
No
0 20 40 60 80 100
Average mapping reads (%)
Skin
Oral Cavity
Stool
Vagina
0 20 40 60 80 100
Westernized
57Nicola Segata University of Trento, Italy

58. The future of
precision Nutrition
58
Jose Ordovas, Ph.D
Director and Senior Scientist,
Nutrition and Genomics Lab, Tufts University

59. International Food Information Council Foundation: 2018 Food and Health Survey. 2018.
https://foodinsight.org/wp-content/uploads/2018/05/2018-FHS-Report-FINAL.pdf. 59Jose Ordovas Tufts, USA
Before moving into the future let’s peek the present:
Cardiovascular Health Top Desired Benefit from Food
Weight loss, energy, and brain function also rank as top benefits consumers are interested in getting from food.
24% of African Americans
ranked weight loss as a
top three health benefit,
compared to 41% of
non-Hispanic whites
More older adults (65+)
ranked bone health and
diabetes management in
top 3 benefits from food
Interest in health benefits from food and nutrients
59

60. Source: https://www.foodnavigator.com/Article/2019/02/06/Food-confusion-prevalent-in-Europe-says-Arla-Foods#
Eighty per cent of consumers
report finding information about
food and nutrition conflicting.
Fifty-nine per cent say that
conflicting information makes them
doubt their choices.
This significant consumer segment
also experiences heightened
stress while shopping
Jose Ordovas Tufts, USA
International Food Information Council Foundation: 2018 Food and Health Survey. 2018.
https://foodinsight.org/wp-content/uploads/2018/05/2018-FHS-Report-FINAL.pdf.
The Concern:
Consumer confusion about food and nutrition is a prevalent issue
60

61. Jose Ordovas Tufts, USA
International Food Information Council Foundation: 2018 Food and Health Survey. 2018.
https://foodinsight.org/wp-content/uploads/2018/05/2018-FHS-Report-FINAL.pdf.
Conflicting information creates “Confusion”
61

62. Jose Ordovas Tufts, USA
International Food Information Council Foundation: 2018 Food and Health Survey. 2018.
https://foodinsight.org/wp-content/uploads/2018/05/2018-FHS-Report-FINAL.pdf.
Level of trust vs. Reliance as a source
Health professionals trusted and used by consumers to guide health and food decisions
Relation between trust and reliance
62

63. With limited access/knowledge to clear, science-based, unbiased nutrition
information, public trust in generalized nutrition guidelines is compromised.
Jose Ordovas Tufts, USA
International Food Information Council Foundation: 2018 Food and Health Survey. 2018.
https://foodinsight.org/wp-content/uploads/2018/05/2018-FHS-Report-FINAL.pdf.
Familiarity with the MyPlate graphic
59% Have seen
the MyPlate graphic
69% of parents with
children under 18 have
seen the MyPlate graphic
Younger consumers. Those in better health, parents and women are particularly familiar with the icon
More education is needed: Three in ten know a lot/fair amount about MyPlate
63

64. • In 2017 ~11 million global deaths and 255 million
disability-adjusted life years (DALYs) could be
attributed to dietary risk factors.
• Treating Chronic Diseases within the current
healthcare model accounts for a staggering 90% of
the United States $3.3 trillion healthcare costs.
• Global nutrition recommendations have failed to
reduce the incidence of chronic disease.
Jose Ordovas Tufts, USA
Why is precision nutrition needed?
64
1995 2000 2005 2010
2003
2015
2011
Dietary guidelines for Americans
USDA

65. A “Proof of Principle” study of Personalised
Nutrition across Europe:
The Food4Me intervention study
This project has received funding from the European Union’s Seventh Framework Programme
for research, technological development and demonstration. (Contract n°265494) Jose Ordovas Tufts, USA
Does precision nutrition work?
65

66. Is personalised nutrition advice more
effective than general healthy eating
guidelines?
Is phenotypic or genotypic information
more effective than diet-based advice alone?
Is the internet a successful
delivery method?
Research questions
Jose Ordovas Tufts, USA 66

67. Level 0: Generic dietary advice (Control)
Level 1: Personalization based on DIETARY analysis
Level 2: Personalization based on DIETARY + PHENOTYPIC analysis
Level 3: Personalization based on DIETARY + PHENOTYPIC + GENOMIC analysis
Randomized to 4 Arms
Jose Ordovas Tufts, USA 67

68. Celis-Morales C et al. (2017) Int. J. Epidemiol. 46, 578-588 Jose Ordovas Tufts, USA
Personalized nutrition improved dietary behaviour (All participants)
RedMeatIntake
(g.day-1)
HealthyEating
index
Control
(LO)
Personalized
nutrition
(Mean L1, L2, L3)
L1 L2 L3
68

69. Celis-Morales C et al. (2017) Int. J. Epidemiol. 46, 578-588 Jose Ordovas Tufts, USA
PN improved dietary behaviour (participants receiving targeted advice)
SaturatedFatIntake
(%fromtotalenergy)
FolateIntake
(μg.day-1)
Control
(LO)
Personalized
nutrition
(Mean L1, L2, L3)
L1 L2 L3
69

70. Challenge - Nutrition research:
Data management, study design and translation.
Meeting the challenge:
Precision Nutrition research and data (PREDICT)
Large Datasets
Nutrition trials
Integrated responses
Dynamic responses
Jose Ordovas Tufts, USA
Meeting the Challenges of Nutrition with Precision Nutrition
70

71. Summary
Tim Spector, Ph.D
Professor of Genetic Epidemiology
King’s College London, UK
71

72. Traditional targets for
personalisation
Integrating multiple factors for
comprehensive personalisation
Past, present, and future of precision nutrition
The past (1950s – present) The present (2020)
Meal composition
Genetics
Meal context
Serum glycemic markers
Microbiome
Age
Serum lipid markers
Blood pressure
Anthropometry
Other serum markers
FFQ
Sex
New Studies
• PREDICT 2
• PREDICT 3
New Technologies
• Non-invasive Real time Biosensors
• POC Lab-on-a-chip
• Artificial Intelligence 2.0
New Knowledge for Professionals and
Individuals
• Empowering the Health Professional
• Empowering the Individual
• Precision Public Health
• Insurance Coverage
The future
72Tim Spector King’s College London, UK

73. 73Tim Spector King’s College London, UK
PREDICT - Carbs
PREDICT - Cardio
PREDICT 1 Plus
P R E D I C T 2
PREDICT
P R E D I C T 3
PREDICT1
J F M A M J J A S O N D J F M A M J J AJ J A S O N D
2018 2019 2020
Completed Ongoing Pending enrolment
The PREDICT program
PREDICT Ongoing

74. King’s College London
Tim Spector
Sarah Berry
Deborah Hart
ZOE
Richard Davies
Jonathan Wolf
George Hadjigeorgiou
University of Trento, Italy
Nicola Segata (PI)
Francesco Asnicar
Massachusetts General Hospital
& Harvard University
Linda Delahanty
Oxford University
Leanne Hodson
Mark McCarthy
Massachusetts General Hospital
Andrew T. Chan
David Drew
Lund University, Sweden
Paul Franks
Acknowledgements
Harvard University
Curtis Huttenhower
University of Leeds
John Blundell
University of Nottingham
Ana Valdes
Tufts University
Jose Ordovas