Transcript advocacy vs. impartiality the problem is quite complex on one side

Paolo Vineis
University of Torino and ISI Foundation, Torino,
Italy
e-mail address: [email protected]
GENE-ENVIRONMENT INTERACTIONS
IN CANCER
LIMITATIONS OF GENETIC DETERMINISM
1. HUMANS AND MICE HAVE THE SAME ESTIMATED
NUMBER OF EXPRESSED GENES
2. HUMANS AND CHIMPANZEES SHARE 98% OF THE
GENOME
3. THE SEQUENCE INFORMATION IN DNA IS
INSUFFICIENT TO DETERMINE HOW GENE
PRODUCTS INTERACT TO PRODUCE AN ORGANISM
4. GENETIC PATHWAYS COMPLETELY SPECIFY
ORGANISMAL FUNCTION ONLY IN RARE CASES, I.E.
MONOGENIC DISEASES (SICKLE CELL ANEMIA,
MUSCULAR DISTROPHY), WHEN THE CELL HAS NO
COMPENSATORY MECHANISM AND
ENVIRONMENTAL INFLUENCES ARE NIL – “ONE
MUTANT GENE – ONE DISEASE PARADIGM”
STROHMAN: “THE CELL IS STARTING TO LOOK
MORE LIKE A COMPLETE ADAPTIVE SYSTEM
RATHER THAN A FACTORY FLOOR OF ROBOTIC
MACHINE GENES”
SOME FIGURES
LIFETIME RISK OF BREAST CANCER IS 12.6% IN
WOMEN, OF PROSTATE CANCER IS 15.9% IN MEN,
AND OF COLON CANCER IS 5.6% IN BOTH SEXES
BRCA1 AND BRCA2 CONFER A RELATIVE RISK OF
BREAST CANCER OF 5-10
GENOTYPES AT MISMATCH REPAIR LOCI CONFER A
RR OF COLON CANCER OF 9.3
METABOLIC POLYMORPHISMS CONFER A RR FOR
SEVERAL TYPES OF CANCER OF LESS THAN 2
ABOUT 0.25% OF WOMEN CARRY BRCA1 OR BRCA2
SUSCEPTIBLE VARIANTS, AND 0.1% OF PEOPLE
HAVE SUSCEPTIBLE VARIANTS FOR MISMATCH
REPAIR LOCI
THESE GENOTYPES ACCOUNT FOR LESS THAN 5%
OF BREAST OR COLON CANCERS
20% OF THE GENERAL POPULATION HAVE THE
APOLIPOPROTEIN E4 ALLELE, WITH A RR OF
ALZHEIMER’S DISEASE OF ABOUT 2; THIS GENE
ACCOUNTS FOR 16.7% OF ALL CASES OF AD
HOW MANY CANCERS ARE ATTRIBUTABLE TO
GENETIC PREDISPOSITION?
LICHENSTEIN ET AL, N ENGL J MED 343: 78-85, 2000
44,788 PAIRS OF TWINS STUDIED IN SCANDINAVIAN
COUNTRIES
ESTIMATES:
PROSTATE 42% (95% CI 29-55)
COLORECTAL 35% (10-48)
BREAST 27% (4-54)
EDITORIAL BY R. HOOVER: GENE-ENVIRONMENT
INTERACTIONS ARE NOT ACCOUNTED FOR (THESE
ARE PROBABLY OVERESTIMATES)
ATTENTION TO DIFFERENT RISK MEASURES:
ABSOLUTE RISK, E.G. LIFE-TIME CUMULATIVE RISK (50-70%
OF BREAST CANCERS FROM BRCA1 MUTATIONS IN
MUTATION CARRIERS)
RELATIVE RISK (PENETRANCE)=5-10 TIMES
ARe =PROPORTION AMONG CARRIERS= 80% OF BREAST
CANCERS IN CARRIERS OF BRCA1 MUTATIONS ARE DUE TO
THIS GENE
ARp = PROPORTION IN THE POPULATION=5-10% OF ALL
BREAST CANCERS ARE ATTRIBUTABLE TO BRCA1
MUTATIONS
WHEN ESTIMATING THE EFFECTIVENESS OF
“SCREENING” WE HAVE TO CONSIDER:
(A) PREDICTIVE VALUE, THAT DEPENDS ON THE
PREVALENCE OF MUTATIONS
(B) PENETRANCE OF THE GENE (NNT)
(C) MOST IMPORTANT, THE AVAILABILITY OF
PREVENTIVE OR CURATIVE MEASURES
EXAMPLE:
MUTATION
YES NO
TEST
POSITIVE
15
10
NEGATIVE
5
115
PREVALENCE=20/145=13.8%
SENSITIVITY=15/20=0.75
SPECIFICITY=115/125=0.92
POSITIVE PREDICTIVE VALUE=15/25=0.60
I.E. OUT OF 100 POSITIVE TESTS 60 HAVE THE
MUTATION
EXAMPLE:
MUTATION
YES NO
TEST
POSITIVE
15
1000
NEGATIVE
5
11500
PREVALENCE=20/12520=0.16%
SENSITIVITY=15/20=0.75
SPECIFICITY=11500/12500=0.92
POSITIVE PREDICTIVE VALUE=15/1015=0.015
I.E. OUT OF 1015 POSITIVE TESTS ONLY 15 HAVE THE
MUTATION
WHAT IS THE EFFECT OF PENETRANCE?
IMAGINE WE HAVE AN EFFECTIVE PREVENTIVE
MEASURE THAT REDUCES THE RISK OF DISEASE IN
THE SCREENEES BY 58%
LET US IMAGINE THAT THE RISK OF DISEASE IS 1.4%
FOR A LOW PENETRANT MUTATION AND 37% FOR A
HIGHLY PENETRANT MUTATION
(REALISTIC FIGURES FOR A METABOLIC
POLYMORPHISM AND BRCA1, RESPECTIVELY)
WE COMPUTE THE NUMBER NEEDED TO TREAT
(SCREEN), I.E. THE NUMBER OF MUTATIONS
CARRIERS WHO NEED TO UNDERGO SCREENING TO
PREVENT A SINGLE CANCER
THE NNT(S) DEPENDS ON PENETRANCE PLUS THE
EFFECTIVENESS OF PREVENTIVE MEASURES
WITH 58% SUCCESSES AND A RISK OF DISEASE OF 37%,
THE NUMBER OF CASES DECREASES BY 22%.
THE RECIPROCAL OF 0.22 IS APPROXIMATELY 4.5
WITH A RISK OF DISEASE OF 1.4% AND 58% SUCCESSES, THE
RISK DECREASES BY 0.008 AND ITS RECIPROCAL IS 1/0.008=125
I.E. WE NEED TO TREAT 4.5 SUBJECTS IN THE FIRST CASE AND
125 IN THE SECOND (NNT)
NOW WE CAN COMBINE THE NNT AND PREVALENCE
TO OBTAIN THE NNS
CASE 1
IF WE SCREEN THE GENERAL POPULATION FOR A
LOW PENETRANT GENE (NNT=125,
PREVALENCE=13.8%, PPV=60%), IN ORDER TO
PREVENT A CANCER WE HAVE TO MULTIPLY THE
NNT BY THE RECIPROCAL OF PREVALENCE
125 X (1/0.138) = 906 SUBJECTS (WITHOUT
CONSIDERING SENSITIVITY AND SPECIFICITY)
CASE 2
IF WE SCREEN THE GENERAL POPULATION FOR A
RARE, HIGHLY PENETRANT GENE (NNT=4.5,
PREVALENCE=0.16%, PPV=1.5%), IN ORDER TO
PREVENT A CANCER WE HAVE TO MULTIPLY THE
NNT BY THE RECIPROCAL OF PREVALENCE
4.5 X (1/0.0016) = 2813 SUBJECTS (WITHOUT
CONSIDERING SENSITIVITY AND SPECIFICITY)
CASE 3
IF WE SCREEN FAMILIES FOR A RARE, HIGHLY
PENETRANT GENE (NNT=4.5, PREVALENCE IN THE
FAMILIES=0.50), IN ORDER TO PREVENT A CANCER
WE HAVE TO MULTIPLY THE NNT BY THE
RECIPROCAL OF PREVALENCE
4.5 X (1/0.50) = 90 SUBJECTS (WITHOUT CONSIDERING
SENSITIVITY AND SPECIFICITY)
Calculation of the Number Needed to Screen in the case of screening for a low penetrant gene (GSTM1
in smokers), and a highly penetrant gene (BRCA1), respectively in the general population or in families
(from Vineis et al, The Lancet, 357: 709-712, 2001)
Lung cancer
in workers exposed to PAH
GSTM1 null
GSTM1 wild
1.34 (1.21 - 1.48)
(a)
1.0
Cumulative risk
13%
Risk reduction
Breast cancer
BRCA1
BRCA1
general population
families
5
10
10%
40% (b)
80 %
50%
50% (c)
50% (Tamoxifen
or Raloxifene) (d)
50%
Cumulative risk after
intervention
6.5%
5%
20%
40%
Absolute risk
reduction
6.5%
5%
20%
40%
Relative risk
Lung cancer
Breast cancer
GSTM1 null
GSTM1 wild
BRCA1
general
population
NNS in mutation
carriers
15
20
5
2.5
Prevalence
50%
50%
0.2% (e)
50%
40
2,500
NNS in whole
target population 30
NNS in all
occupationally
exposed
BRCA1
families
5
35
(a) from Vineis et al, IARC Scie. Publ. No. 148, 1999 1999
(b) from Hopper et al, 1999
(c) theoretical maximum reduction in risk of lung cancer due to preventive action
(d) theoretical benefit, based on the BCPT trial with a 45% benefit, and the Raloxifene trial with a 76%
benefit
(e) Coughlin et al, 1999
An illustration of the principle of “one exposure - many diseases,
one disease - many low penetrant genes”.
Exposure
Proportion attributable
Tobacco smoke
Lung cancer
Bladder cancer
Occupational
exposure
to PAH
Lung cancer
Bladder cancer
(a) depending on the areas
90%
70% men
30% women
Larynx cancer
90%
CHD
12.5%
Chronic bronchitis 80%
4-20% (a)
1-10% (a)
Disease
Low penetrant genes
Odds Ratio (a)
Lung cancer
CYP1A1 MspI (asians)
CYP1A1 MspI (caucasians)
CYP1A1 Exon 7 (asians)
CYP1A1 Exon 7 (caucasians)
CYP2D6
GSTM1
1.73
1.04
2.25
1.30
1.26
1.34
Bladder cancer
NAT - 2 slow
GSTM1
1.37
1.57
Colon cancer
NAT - 2 rapid
1.19
(a) meta - analisys from Vineis et al, IARC Scie. Publ. No. 148, 1999
MANY GENES CONTRIBUTE TO MODULATE THE RISK, AND
THE CONTRIBUTION OF EACH IS MODEST ( EXCEPT IN HIGH
RISK FAMILIES)
ADOPTING PREVENTIVE MEASURES ONLY IN THE
HIGHLY SUSCEPTIBLE WOULD IMPLY VERY LITTLE
ADVANTAGES OVER PREVENTION
FOR ALL