Chemicals, risk and cancer

1
Chemicals, risk and cancer

• David R. Bell
• 13210
• david.bell_at_nottingham.ac.uk

Introduction to the cellular and molecular
biology of cancer Ed Franks Teich
2
Do chemicals cause cancer ?

• Epidemiology
• The study of illness in populations
• Correlate with influences on the population, eg
  diet, work, geography
• 1775
• The surgeon, Percival Potts, noted that coal
  sweeps had a high incidence of cancer, scrotal
  cancer. He hypothesised that the cancer was due
  to their intimate and prolonged exposure to coal
  dust.

3
Epidemiological methods

• Case studies
• Essentially anecdotal, eg John was treated with
  X and developed cancer
• Poor evidence
• the sample group is not defined, and may be
  biased
• More rigorous approach needed for agents which
  cause a subtle effect

4
Comparison with controls

• Untreated people develop disease and die
• Must control by comparison with an appropriate
  group
• Enormous variation within population
• Social class
• Sex
• Age
• Other factors, e.g. smoking

5
Epidemiology methods 2

• Retrospective studies
• Study a group of people who have a particular
  cancer
• Retrospectively match with people who are matched
  for e.g. age, sex, smoking history, etc
• Sample bias is inherent in the study

6
Epidemiology methods 3

• Prospective studies
• Set out what you are going to do before you do
  it, and what you are looking for
• Randomise subjects to treated or control groups
• Identify and match control treated groups
  before onset of illness
• Least sample bias
• Highly labour intensive

7
Statistics Epidemiology

• Statistical design is paramount in epidemiology
• Sample size must be large (ngtgt1)
• If you compare control vs. treated, and look at
  20 organs, then at least one will be different at
  Plt0.05
• Set out your hypothesis before you start
• Salami slicing of treated groups to get a
  positive result

8
Causation is not correlation

• Many epidemiological approaches simply measure
  the association of a variable with a particular
  endpoint
• This is an uncontrolled experiment, where there
  are many other variables
• Low socio-economic status is associated with
  lung, liver cancer, bad health
• Association does not prove causation

9
Is cancer increasing ?
10
Cancer rates

• The total incidence of cancer is not always
  useful
• i.e. number of deaths from cancer
• Consider a population of rats 50 die from heart
  failure, and 50 from cancer. If they are given a
  drug which cures heart failure, what will be the
  outcome ?
• Incidence rates are more informative, eg
  age-specific incidence rates
• Sensitive to improvements in cancer diagnosis

11
Factors affecting cancer

• Sex
• Lung cancer
• Male 2-5 1 Female
• Breast cancer
• Male 0.01 1 Female
• Racial origin
• White skin and increased UV-induced skin cancer
• Standardise for age, sex, racial origin, etc

12
Latency
Cancer takes 10-50 years to develop in man
1900- Cigarette smoking an accepted habit
1940- rise in male lung cancer- alarm
1960- smoking is shown to be the cause of lung
cancer
1960-2000. People who started smoking before
1960 die of lung cancer Lung cancer is the most
frequent site of cancer in UK men
13
Epidemiology- issues

• Long latency period- up to 20 years
• Poor for prediction
• Sensitive against low background
• e.g. haemangiosarcoma is vanishingly rare in the
  general population, vs high levels in workers
  exposed to vinyl chloride monomer
• Insensitive against common cancers
• Frequently no measurement of chemical exposure
• Difficult to isolate the cause of cancer
• Unless you are testing for that cause

14
Causes of cancer
Note that diet, tobacco, occupation, alcohol and
pollution are principally chemically-induced
cancers.