An epidemiological perspective on the biology of cancer

1
An epidemiological perspective on the biology of
cancer

• Julian Peto
• Collaborating in Cancer Research
• Cardiff 8th March 2006
• London School of Hygiene and Tropical Medicine
• and Institute of Cancer Research

2
Bradford Hill
Richard Doll
3
Tobacco

• Elimination of tobacco would prevent one third of
  all cancers worldwide
• Increased risk for cancers of lung, pancreas,
  bladder, kidney, larynx, mouth, pharynx,
  oesophagus
• Recent evidence links smoking to increased risk
  of stomach, liver and cervical cancers

4
Cancer rates in migrants become similar to those
in the local population
5
Viruses, bacteria parasites

• Infections cause about 4 of cancers in the UK
  but 20 of all cancers worldwide
• The most important are
• Helicobactor pylori (a chronic gastric bacterial
  infection) causes stomach cancer
• Human papillomaviruses (HPV) cause cervical
  cancer
• Hepatitis B and C viruses cause liver cancer

6
Cancers linked to obesity (Calle et al., NEJM
April 24 2003)Oesophagus, colon, rectum, liver,
gallbladder, pancreas, kidney, non-Hodgkins
lymphoma, myelomaProstate, breast, uterus,
ovary, cervix
7
Breast cancer and breastfeeding collaborative
reanalysis of individual data from 47
epidemiological studies in 30 countries,
including 50 302 women with breast cancer and 96
973 women without the disease Collaborative
Group on Hormonal Factors in Breast Cancer
Lancet (2002)
360187-195Reduction in risk 3.0 per year
for earlier age at first birth7.0 per
birth4.3 per year of breastfeeding
8
Predicted reduction in Western breast cancer
rates if women had 6 or 7 children and breastfed
each child for 2 years Lancet (2002) 360 187-95
9
Percentage of US cancer deaths that would be
avoided by eliminating known risks (Peto (2001)
Nature 411 390-395)
Cause
Current
Non-
smokers
smokers
Smoking
60
-
Known infections
2
5
Alcohol
0.4
1
Sunlight
0.4
1
Air pollution
0.4
1
Occupation
0.4
1
Lack of exercise
0.4
1
Diet
Overweight
4
10
2
(BMIgt25kg m
)
Other dietary factors
4 - 12?
10 - 30?
Presently unavoidable
About a quarter
At least half
10
Collaborative cancer research

• Epidemiology Avoidable causes
  Prevention
• Genetics Mechanisms
  Better treatment
• Biology
  Better screening
• Epidemiology,
  genetics and biology
• Clinical research
• Evaluation of
  treatment,
• screening and
  prevention

11
Pre-molecular cancer epidemiology1950s and
1960s Cancer rates in adults suggest a series
of rate-limiting heritable steps in multi-stage
carcinogenesis.Armitage and Doll (1954) Br J
Cancer 8 1-12The age distribution of cancer
and a multi-stage theory of carcinogenesisReprin
ted in Int J Epid (2004) 33 1174-79
12
The tumour suppressor gene hypothesis for
polyposis coli and familial retinoblastomaCancer
s in apparently autosomal dominant syndromes such
as polyposis coli and familial retinoblastoma
appear as a result of subsequent somatic
mutations in which individual cells become
homozygous for a recessive neoplasm-causing
gene.DeMars 1969
13
Colon cancer incidence is proportional to the
cube of age in familial polyposis coli, and to
the fifth power of age in the general population
Ashley (1969) J Med Genet 6,378
14
Age distribution of breast cancer incidence
15
Model of rate of breast tissue ageingPike et al
(1983) Nature 303, 767
Menarche
Start of perimenopausal period
First birth
Last menstrual period
16
Stem cells and cancer

• Cancer is caused by the accumulation of several
  genetic or epigenetic changes in a single cell.
• Teenage exposure (e.g. smoking or cervical HPV
  infection) greatly increases your cancer risk 50
  years later.
• The only cells that survive for 50 years are stem
  cells.
• Therefore most cancers develop in very long-lived
  stem cells.
• Stem cells must have evolved to minimise the
  accumulation of multiple mutations in a single
  cell.
• To understand cancer we have to understand stem
  cell biology.
• Shackleton et al (2006) Nature 439 84-88.
  Generation of a functional mammary gland from a
  single stem cell
• Frank and Nowak (2004) Bioessays 26 291-9.
  Problems of somatic mutation and cancer

17
Stem cell mechanisms to prevent mutationCairns
(1975) Nature 255 197-2001. The immortal
strand. When a stem cell divides, the original
DNA stays in the stem cell daughter, and the new
copy (containing copying errors) goes into the
differentiating daughter and is discarded within
a week or two.

• Copied DNA in differentiating
  daughter
• Original
  DNA in stem cell
• daughter

18
Stem cell mechanisms to prevent mutation2. DNA
damage causes stem cell death rather than
error-prone repair 3. Mutant stem cells cannot
escape from the stem cell niche.
19
Stem cell mechanisms to prevent
mutation4. Stem cell hierarchy. With four
successive layers of stem cells that divide 12
times and then die, the maximum number of
divisions in any cells ancestry is reduced from
20,000 to 48 in a human lifespan

• 12 divisions in 20 days (every 40 hours)
• 12 divisions in 8 months (every 20 days)
• 12 divisions in 8 years (every 8 months)
• 12 divisions in 100 years (every 8 years)

20
Stem cell theorem

• How many times should a stem cell divide,
  contributing a daughter to the next level in a
  hierarchical stem cell compartment, to minimise
  the number of divisions any cell can undergo?
• Answer e ( 2.7, the base of natural logs)

21
Intestinal crypt Leedham et al (2005) J Cell Mol
Med 111-24
22
Lung cancer mortality in continuing smokers and
ex-smokers. Halpern et al (1993) JNCI 85, 457
Age stopped
60-64
55-59
50-54
40-49
23
The stem cell perspective

• What is the last step in lung cancer? It cannot
  be caused by smoking, as the incidence rate does
  not fall when smokers give up. The division of a
  long-lived mutant stem cell seems plausible.
• Do most mutagens cause cancer mainly by killing
  stem cells, causing stem cell proliferation,
  rather than by direct mutagenesis?
• Cairns (2002) PNAS 99 10567-70
• Are stem cell proliferation (hormones, childhood
  cancers) and chromosomal damage (Blooms
  syndrome, asbestos) more important causes of most
  cancers than point mutation (XP)?
• Cairns (1998) Genetics 148 1433-40

24
British cervical cancer mortality before
screening began (1920 birth cohort)
25
CIN3 and cervical cancer

• Almost 10 of British women develop CIN3 by age
  50.
• The prevalence of CIN3 in unscreened middle-aged
  women is never much more than 1 even populations
  where 5 will eventually develop cancer.
• The cervical cancer incidence rate in unscreened
  populations is constant from ages 50 to 85.
• Therefore most undiagnosed CIN3s must eventually
  regress cytologically but remain at constant risk
  of malignant progression for at least 35 years.

26
Cumulative lifetime cancer mortality for general
population, unilateral sporadic retinoblastoma
survivors and genetic retinoblastoma survivors.
Fletcher et al. (2004) J Natl Cancer Inst
96357-63
Cumulative mortality ()
Genetic Rb
Sporadic Rb
General population
Age (years)
27
Cancer mortality after age 25 in hereditary
retinoblastoma survivors Fletcher et al. (2004)
J Natl Cancer Inst 96357-63
28
Apart from adrenocortical carcinoma in p53
carriers and retinoblastoma in Rb1 carriers, the
same early onset cancers are associated with
germline p53 or Rb1 mutation (osteosarcoma, soft
tissue sarcomas, breast cancer and brain cancer).
Most of the adult-onset cancers in Rb1 carriers
are caused by DNA-damaging agents
29

• Breast cancer and percent mammographic density.
  Boyd et al. (1995).
• Adjusted
• Density Cases Controls odds ratio
• None 10 25 1.0
• lt10 29 61 1.2
• 10- 65 73 2.2
• 25- 94 97 2.4
• 50- 90 67 3.4
• 75 66 31 5.3

30
Correlates of breast density

• Breast cancer risk
• Sister with dense breasts
• HRT
• Late age at first birth
• No breast feeding
• Nulliparity
• IGF-I

31
Polygenic model for breast cancer risk Pharoah
et al (2002) Nat Genet 31, 33 Peto (2002)
Cancer Cell 1, 411
3 10 20
Lifetime risk (log scale)

• Lifetime risk General population
  Breast cancer patients
• lt3 0.50 0.12
• 3-10 0.38 0.38
• 10-20 0.09 0.25
• gt20 0.03 0.25

32
Cumulative breast cancer risks for first degree
relatives of CHEK21100delC heterozygotes with
bilateral breast cancer, first degree relatives
of CHEK2 wild types with bilateral breast cancer,
and the expected rate in the general population
Johnson et al. (2005) Lancet 366 1554-7
33
Familial site-specific cancer risks in Sweden
Dong and Hemminki (2001) Int J Cancer 92 144

• Parent or sibling
  affected Parent and sibling affected
• SIR
  (N) SIR
  (N)
• Breast 1.9 (1577) 2.4 (42)
• Colon 2.0 (182) 27.6 (13)
• Rectum 1.8 (47) 32.8 (2)
• Ovary 2.6 (88) 25.5 (4)
• Prostate 2.7 (134) 23.7 (6)
• Thyroid 6.9 (69)
  292.0 (18)
• Other endocrine 3.3 (84) 70.3 (8)
• Lung 1.7 (112) 13.7 (3)
• Kidney 1.8 (45) 28.4 (2)
• Melanoma 2.7 (256) 10.4 (7)
• Brain CNS 1.8 (188) 11.4 (6)
• Leukaemia 2.3 (74) 33.3 (4)
• Lymphoma 1.8 (76) 12.7 (2)

34
The damage done to British medical research by
influential ethical experts
35
In the late 1980s Ian Kennedy told a Select
Committee that HIV testing of the anonymised
discarded residue of blood samples from pregnant
women was unethical because a test must confer
some benefit on the patient.This illogical
assertion was accepted by the Select Committee
and delayed the introduction of HIV monitoring,
despite the pleas of many eminent scientists
including Black, Cox, Doll, Bodmer, Hoffenberg
and Weiss. Black et al. (1987) Lancet ii 1277
36
Data can be used for any medical research
purpose under the Data Protection Act, without
the need for the consent of individuals. So
Professor Julian Peto is simply wrong when he
states that the Data Protection Act is preventing
data from being passed to medical researchers.
(Lord Falconer. Letter to The Times, May 17th
2001.)
37
The Data Protection Act in the real worldThe
General Medical Council instructed doctors that
they might face litigation under the Data
Protection Act if they notified their patients to
cancer registries without obtaining fully
informed consent.
38
The increasing bureaucratic burden imposed by
ethicists causes harm to patients and does
enormous damage to British cancer research.
Legislation is needed to restore the following
principle 'Consent is not required for access
to medical records for non-commercial medical
research that has no effect on the individuals
being studied and has been approved by an
accredited research ethics committee. 93 of
the audience voted for this proposed law at a
Parliamentary Group on Cancer public meeting(Nov
5th 2002)
39
(No Transcript)
40

• Olivia Fletcher
• Nikki Johnson
• Clare Palles
• Maribel Almonte
• Isabel dos Santos Silva
• John Cairns
• Richard Doll

41
(No Transcript)