Nutrition and cancer
Week 0: Prerequisite knowledge
Descriptive epidemiology = study the frequency/amount of disease or other characteristics in a
population. You are interested in person, time and place.
Analytical epidemiology = study of the cause of disease. You are interested in exposure-disease
associations. You ask yourself why these people, why this moment. Analytical epidemiology can be
divided into observational studies and experimental studies. Observational studies can be divided
into cross-sectional studies, cohort studies and case-control studies. The main experimental study is
the randomized control trial.
Study designs:
- Cross-sectional studies = you have a given population. By a population based survey you
measure the exposures and diseases/health outcomes. You can divide the population those
you have the disease + are exposure, who are exposed + don’t have the disease, who have
the disease + are not exposed or those who don’t have the disease + are not exposed (f.e.
smoking and lung cancer or BMI and physical activity). Cross-sectional studies can make
associations.
- Cohort study = you have a given population and divide it into those who are exposed and
those who are not exposed to a certain factor (f.e. smoking). You can follow these people for
a time (forward in time = longitudinal) and you can see who develops the health outcome
(f.e. lung cancer) in the exposed and the non-exposed people. By this, you can also study the
association between the determinant and the outcome.
o What is a cohort? Group of people who show common characteristics, f.e.
geographical, occupation, birth, patients.
o What is the purpose of a cohort study? To study etiological factors/causes of disease
and to study prognostic factors, to identify predictors of disease. If you want to study
etiological factors, people should be free of the disease at the beginning of the study,
if you want to study prognostic factors, people should already have the disease.
o At the beginning/baseline of the cohort study, you determine how you will measure
the exposure (f.e. smoking: how many cigarettes per day, how long per day) and the
covariates = variables that are possibly related with the exposure/outcome (f.e.
physical activity, blood pressure).
o During the follow up you want to measure the disease outcome (f.e. lung cancer,
cardiovascular disease), but also the updates about the exposure and the covariates
should be measured.
- Case control study (case reference study) = there is a defined population, case control is a
retrospective study (back in time), here you study the disease. F.e. some people have lung
, cancer and others don’t. You study if the people did in the past; so if they smoked or not.
Case control starts with the outcome and studies back in time. So, you are interested in the
exposure.
o When do you use this study? For rare diseases or when the disease has a long latency
time (long time to diagnose) or the disease develops slowly.
o In a case-control you can only focus on 1 single health outcome (difference with
cohort study). But you have more time (big advantage), so you can investigate the
potential confounders and do more detailed measurements, f.e. via interview or take
biomarkers. Other advantages are less burden for participants and case control
studies can be fast and cheap.
o What is the proper case group? Who are you going to select for your study? Incident
cases = newly diagnosed with the disease (this is the best) or prevalent cases = both
old and new occurrences. With prevalent cases, be aware of reverse causations; the
disease can influence the measurements. Also be aware of the survivor effect
(selection bias) = the people who are healthy, will longer stay in your study
compared to the people with the highest exposure, who may die before you can
measure.
o What is the proper control group? The control group should be represented for the
study group. Controls can be selected from the general population, the hospital,
relatives, colleagues, friends etc. So, you need a sample from the total population
which is representable for the cases, the control group should also be identified as a
case when they develop the disease. So, you can find your control population in
population registries, random digit dialling, GP registrations, postal codes or hospitals
(advantage is you interview the controls in the same environments as the cases, but
the disadvantage is that the risk factor distributions of the hospital controls may not
reflect that of the source population from which the cases emerge).
- Randomized controlled trial = also called the intervention study. You have a given population,
the investigator assigns the exposure randomly: investigator decides who is exposed to the
intervention/treatment and who doesn’t. So, we speak of random allocation. So, each
subject has equal probability of assignment to the intervention or control group. By this, you
balance the distribution of outcome predictors at the start of the study. In time (prospective
design) you can see who develops the disease and who doesn’t in both groups (group which
is randomly given the exposure and the control group). The RCT has more proof of causality
than observational study.
o Exposure can be: dietary patterns, lifestyle intervention, dietary supplements,
medication, change in behaviour, medical care etc.
o Outcome can be:
, Continuous = follow the level of the outcome in time. F.e. changes in blood
pressure during treatment, serum cholesterol, weight loss, lung function,
quality of life score etc.
Dichotomous = also called 0/1, yes/no, dummy or binary outcome. This
happens once in time. F.e. hip fracture, incident stroke, smoking cessation.
o Control treatment in RCT can be a regular treatment (commonly used in medical
practice), no intervention or a placebo (inactive substance). Placebo means ‘I shall
please’. The placebo effect is the measurable, observable or felt improvement in
health or behaviour not attributable to a medication or treatment that is given.
o The treatment effect in a RCT is measured as: effect of intervention – effect of
control treatment.
o Two types of RCT:
Parallel = people are either in the control group or the intervention group
and get this treatment the entire study.
Cross-over (two-period study or AB-AB study) = people get both treatments
(both control and intervention), half-way the study they switch. In this RCT,
you randomize the treatment order. Cross-over can only be used if the
exposure has no long lasting effect. Can also only be used for outcomes that
are reversible and that can change within a short time period. And
treatments may be separated by a wash-out period (get rid of treatment
during the study, to clean the body). Question: is a cross-over trial a good
idea to study the effect of the vaccination on influenza? No, because the
vaccination has a long time effect.
- Community trial = trial in population, often focused on primary disease prevention. Unit of
randomization are communities (towns, schools, not individuals), intervention is provided by
e.g. GPs, community health centers, local outpatient facilities. Suitable design for testing
lifestyle interventions that cannot be allocated to individuals.
- Quasi-experimental study = trial without random assignment to the treatment, so there is no
control group involved. Still, the investigator assigns the exposure. You test people before a
treatment and after the treatment measure again (pretest / posttest comparison). Often
used in hospital or public health setting. This is not as good as RCT, but still used when you
are interested in studying cause and effect, you can manipulate the independent variable
and you are not able to use a stronger experimental design because it is unfeasible or
unethical.
The trial design (RCT, community and quasi-experimental) cannot be applied to all research
questions, because:
Relatively high costs
Rare disease outcome cannot be studied because of limited duration
Cannot be used for non-modifiable risk factors (genes, birthweight)
May be unethical, e.g. virus infection, smoking, sexual behaviour
Wrong time window of exposure (f.e. it is unethical to study young children when you want
to study a disease in young children).
Not the full range of exposure can be studied
DNA = 2 meters in a cell, packing and folding the DNA is important to fit in the nucleus
Histone = protein that package and orders the DNA into structural units called nucleosome. Histones
are positively charged, DNA is negative, therefore DNA and histones attract each other. Histones are
responsible for the maintenance of chromatin shape and structure.
Nucleosome = DNA wrapped around a core of 8 histones.
Chromatin = multiple nucleosomes
, Heterochromatin = part of DNA that is condensed these genes are transcriptionally
silenced.
Euchromatin = part of DNA which is open genes can be transcribed.
Constitutive heterochromatin = when all cell types have the same closed histone state
Facultative heterochromatin = if all cells have a different histone state
Chromosome = the chromatins together; proteins with DNA wrapped around it.
Genetics = heritability of the DNA sequence
Genome = all our genes together, the genome of every individual is unique, except for monozygotic
twins.
HPG = human genome project; aims to map the complete DNA sequence to make a human reference
genome. It started in 1984 and completed in 2003.
Epigenetics = ‘over’ genetics = how can one genotype give rise to multiple phenotypes?
Week 0: Prerequisite knowledge
Descriptive epidemiology = study the frequency/amount of disease or other characteristics in a
population. You are interested in person, time and place.
Analytical epidemiology = study of the cause of disease. You are interested in exposure-disease
associations. You ask yourself why these people, why this moment. Analytical epidemiology can be
divided into observational studies and experimental studies. Observational studies can be divided
into cross-sectional studies, cohort studies and case-control studies. The main experimental study is
the randomized control trial.
Study designs:
- Cross-sectional studies = you have a given population. By a population based survey you
measure the exposures and diseases/health outcomes. You can divide the population those
you have the disease + are exposure, who are exposed + don’t have the disease, who have
the disease + are not exposed or those who don’t have the disease + are not exposed (f.e.
smoking and lung cancer or BMI and physical activity). Cross-sectional studies can make
associations.
- Cohort study = you have a given population and divide it into those who are exposed and
those who are not exposed to a certain factor (f.e. smoking). You can follow these people for
a time (forward in time = longitudinal) and you can see who develops the health outcome
(f.e. lung cancer) in the exposed and the non-exposed people. By this, you can also study the
association between the determinant and the outcome.
o What is a cohort? Group of people who show common characteristics, f.e.
geographical, occupation, birth, patients.
o What is the purpose of a cohort study? To study etiological factors/causes of disease
and to study prognostic factors, to identify predictors of disease. If you want to study
etiological factors, people should be free of the disease at the beginning of the study,
if you want to study prognostic factors, people should already have the disease.
o At the beginning/baseline of the cohort study, you determine how you will measure
the exposure (f.e. smoking: how many cigarettes per day, how long per day) and the
covariates = variables that are possibly related with the exposure/outcome (f.e.
physical activity, blood pressure).
o During the follow up you want to measure the disease outcome (f.e. lung cancer,
cardiovascular disease), but also the updates about the exposure and the covariates
should be measured.
- Case control study (case reference study) = there is a defined population, case control is a
retrospective study (back in time), here you study the disease. F.e. some people have lung
, cancer and others don’t. You study if the people did in the past; so if they smoked or not.
Case control starts with the outcome and studies back in time. So, you are interested in the
exposure.
o When do you use this study? For rare diseases or when the disease has a long latency
time (long time to diagnose) or the disease develops slowly.
o In a case-control you can only focus on 1 single health outcome (difference with
cohort study). But you have more time (big advantage), so you can investigate the
potential confounders and do more detailed measurements, f.e. via interview or take
biomarkers. Other advantages are less burden for participants and case control
studies can be fast and cheap.
o What is the proper case group? Who are you going to select for your study? Incident
cases = newly diagnosed with the disease (this is the best) or prevalent cases = both
old and new occurrences. With prevalent cases, be aware of reverse causations; the
disease can influence the measurements. Also be aware of the survivor effect
(selection bias) = the people who are healthy, will longer stay in your study
compared to the people with the highest exposure, who may die before you can
measure.
o What is the proper control group? The control group should be represented for the
study group. Controls can be selected from the general population, the hospital,
relatives, colleagues, friends etc. So, you need a sample from the total population
which is representable for the cases, the control group should also be identified as a
case when they develop the disease. So, you can find your control population in
population registries, random digit dialling, GP registrations, postal codes or hospitals
(advantage is you interview the controls in the same environments as the cases, but
the disadvantage is that the risk factor distributions of the hospital controls may not
reflect that of the source population from which the cases emerge).
- Randomized controlled trial = also called the intervention study. You have a given population,
the investigator assigns the exposure randomly: investigator decides who is exposed to the
intervention/treatment and who doesn’t. So, we speak of random allocation. So, each
subject has equal probability of assignment to the intervention or control group. By this, you
balance the distribution of outcome predictors at the start of the study. In time (prospective
design) you can see who develops the disease and who doesn’t in both groups (group which
is randomly given the exposure and the control group). The RCT has more proof of causality
than observational study.
o Exposure can be: dietary patterns, lifestyle intervention, dietary supplements,
medication, change in behaviour, medical care etc.
o Outcome can be:
, Continuous = follow the level of the outcome in time. F.e. changes in blood
pressure during treatment, serum cholesterol, weight loss, lung function,
quality of life score etc.
Dichotomous = also called 0/1, yes/no, dummy or binary outcome. This
happens once in time. F.e. hip fracture, incident stroke, smoking cessation.
o Control treatment in RCT can be a regular treatment (commonly used in medical
practice), no intervention or a placebo (inactive substance). Placebo means ‘I shall
please’. The placebo effect is the measurable, observable or felt improvement in
health or behaviour not attributable to a medication or treatment that is given.
o The treatment effect in a RCT is measured as: effect of intervention – effect of
control treatment.
o Two types of RCT:
Parallel = people are either in the control group or the intervention group
and get this treatment the entire study.
Cross-over (two-period study or AB-AB study) = people get both treatments
(both control and intervention), half-way the study they switch. In this RCT,
you randomize the treatment order. Cross-over can only be used if the
exposure has no long lasting effect. Can also only be used for outcomes that
are reversible and that can change within a short time period. And
treatments may be separated by a wash-out period (get rid of treatment
during the study, to clean the body). Question: is a cross-over trial a good
idea to study the effect of the vaccination on influenza? No, because the
vaccination has a long time effect.
- Community trial = trial in population, often focused on primary disease prevention. Unit of
randomization are communities (towns, schools, not individuals), intervention is provided by
e.g. GPs, community health centers, local outpatient facilities. Suitable design for testing
lifestyle interventions that cannot be allocated to individuals.
- Quasi-experimental study = trial without random assignment to the treatment, so there is no
control group involved. Still, the investigator assigns the exposure. You test people before a
treatment and after the treatment measure again (pretest / posttest comparison). Often
used in hospital or public health setting. This is not as good as RCT, but still used when you
are interested in studying cause and effect, you can manipulate the independent variable
and you are not able to use a stronger experimental design because it is unfeasible or
unethical.
The trial design (RCT, community and quasi-experimental) cannot be applied to all research
questions, because:
Relatively high costs
Rare disease outcome cannot be studied because of limited duration
Cannot be used for non-modifiable risk factors (genes, birthweight)
May be unethical, e.g. virus infection, smoking, sexual behaviour
Wrong time window of exposure (f.e. it is unethical to study young children when you want
to study a disease in young children).
Not the full range of exposure can be studied
DNA = 2 meters in a cell, packing and folding the DNA is important to fit in the nucleus
Histone = protein that package and orders the DNA into structural units called nucleosome. Histones
are positively charged, DNA is negative, therefore DNA and histones attract each other. Histones are
responsible for the maintenance of chromatin shape and structure.
Nucleosome = DNA wrapped around a core of 8 histones.
Chromatin = multiple nucleosomes
, Heterochromatin = part of DNA that is condensed these genes are transcriptionally
silenced.
Euchromatin = part of DNA which is open genes can be transcribed.
Constitutive heterochromatin = when all cell types have the same closed histone state
Facultative heterochromatin = if all cells have a different histone state
Chromosome = the chromatins together; proteins with DNA wrapped around it.
Genetics = heritability of the DNA sequence
Genome = all our genes together, the genome of every individual is unique, except for monozygotic
twins.
HPG = human genome project; aims to map the complete DNA sequence to make a human reference
genome. It started in 1984 and completed in 2003.
Epigenetics = ‘over’ genetics = how can one genotype give rise to multiple phenotypes?
