Genetic Epidemiology of Pancreatic Cancer: New Insights

1
Genetic Epidemiology ofPancreatic Cancer New
Insights

• Gloria M. Petersen, Ph.D.
• Department of Health Sciences Research
• Mayo Clinic
• Rochester, Minnesota
• April 19, 2004

2
Mayo Clinic Rochester, MN Jacksonville, FL
Scottsdale, AZ

• Organized as a charitable, not-for-profit
  foundation
• 3,200 physicians and scientists
• Over 6 million patients since inception in 1889
• 80 are treated as outpatients 80 in MN are
  regional
• In 2002

Rochester, Minnesota

• 501,109 unique patients
• 2.2 million outpatient visits
• 1.87 million diagnostic X-ray procedures
• 18.5 million lab tests
• 2,418 licensed beds
• 128,881 surgical cases
• 124,663 admissions

3
PANCREATIC CANCER A BLACK BOX

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?
Only interdisciplinary, focused effort will work.
4
Pancreas
5
Scope of Problem
?

• Pancreatic cancer is devastating and poorly
  understood.

• In 2004, 31,860 new cases in the U.S. and
  31,270 deaths

• Approximately 95 of pancreatic neoplasms are
  ductal adenocarcinomas. 80 present with
  metastatic disease.

• 5-year survival rate is 4, the worst of any
  major cancer fifth leading cancer killer in the
  U.S.

• Numbers of new cases in U.S. among males and
  females were nearly equal .

• U.S. incidence (8.5/105). African Americans
  (11.5/105) compared to whites (8.5/105) in the
  U.S.

• Average onset late 60s. Rarely seen in
  individuals under the age of 45.

6
Pancreatic Cancer has been ignored for years TOP
FIVE CANCERS BY MORTALITY NCI FUNDING, 1998-2002
Millions of Dollars
Source NCI Facts Book
7
Pancreatic Cancer An Agenda for Action Report
of the Pancreatic Cancer Progress Review
Group February 2001 Co-Chairs Scott Kern, M.D.
Margaret Tempero, M.D. Executive Director
Barbara Conley, M.D.
http//prg.nci.nih.gov/pancreatic/finalreport.html
8
PRG Risk/Prevention/Screening/Diagnosis

• Critical resources include new and expanded
  registries for
• Identification of high-risk patients and
  kindreds.
• Linkage analysis.
• Tissue and specimen resources.
• Identification of screening and surveillance
  cohorts.
• Epidemiologic assessment of gene-environment
  interactions.

9
PRG Risk/Prevention/Screening/Diagnosis

• The three most important research priorities are
  to
• Identify genetic factors, environmental factors,
  and gene-environment interactions that contribute
  to pancreatic cancer development.
• Develop, implement, and evaluate approaches to
  prevent pancreatic cancer in high-risk cohorts
  (e.g., familial pancreatic cancer, hereditary
  pancreatitis, older age). Studies should be
  performed in humans and in animal models of early
  neoplasia (e.g., PanIN-3).
• Identify and develop surveillance and diagnosis
  methods for the early detection of pancreatic
  cancer and its precursors.

10

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Infrastructure at Mayo Clinic
• Establish an ultra-ultra rapid patient registry
• Rapid biospecimen acquisition
• Multi-station web-based data entry
• Pathology, laboratory medicine, epidemiology,
  biostatistics, gastroenterology, oncology

11
Mayo Biospecimen Resource for Pancreas Research
Patient Recruitment (in person)
Registry Coordination
Data Management Biostatistics
Mayo Biospecimen Cores DNA/serum from
blood Archival tissue Frozen tissue
Clinical expertise Gastroenterologist Oncologist P
athologist
12
Mayo Biospecimen Resource for Pancreas Research

• Started in October, 2000
• Patients identified in Pancreatology Clinic and
  Oncology
• To date, 1,103 potentially eligible patients
  approached, 998 have given definitive response
• 868/998 (87) have consented to participate in
  research registry
• 696 with histologically confirmed adenocarcinoma
• 85 of pts with confirmed adenoca consented
• 55 male, 95 Caucasian, 45 nonsmokers
• Patients consent to blood sample, questionnaire,
  use of archival tissue, and medical records,
  family studies

13
(No Transcript)
14

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Analytic methods
• Familial aggregation
• Segregation analysis
• Linkage analysis
• Association studies

15

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Analytic methods
• Familial aggregation
• Segregation analysis
• Linkage analysis
• Association studies

16
Carter's two sisters, his brother, and his father
died of pancreatic cancer. His mother had
pancreatic, bone, and breast cancer when she died
at age 85.
17
Is there a genetic basis for familial pancreatic
cancer?
Sporadic 80
?
Familial pancreatic cancer 10 or more
Known genetic syndromes (5)
18
Hereditary Syndromes and Genes Associated with
Pancreas Cancer
Disorder Hereditary Pancreatitis Hereditary
Nonpolyposis Colorectal Cancer Lynch II Variant
Familial Adenomatous Polyposis BRCA2 Familia
l Atypical Multiple Mole Melanoma Syndrome
(FAMMM) Peutz-Jeghers Syndrome
Gene PRSS1 (Cationic trypsinogen) hMSH2,
hMLH1 APC BRCA2 p16 STK11/LKB1
Possible role of Fanconi anemia genes in
Pancreatic Cancer?
19
Is there a genetic basis for familial pancreatic
cancer?
Sporadic 80
?
Familial pancreatic cancer 10 or more
Known genetic syndromes (5)
20
Prospective Pancreatic Cancer Study NFPTR

• Established in 1994
• 5,179 individuals included from 853 families, (at
  baseline 381 FPC families and 472 SPC families)
  who were followed for a total of 14,128
  person-years

A. Klein et al., Cancer Res 642634, 2004
21
Study Population

• 22 incident pancreatic cancers have developed in
  NFPTR kindreds
• 21 in blood relatives, only one in a spouse
• 21 of the 22 presented with metastatic disease
• One early cancer was detected serendipitously
  during imaging for a kidney stone
• 19 of the 22 incident pancreatic cancers were
  included in the analysis

22
Klein A et al., Cancer Research 2004
23
Klein A et al., Cancer Research 2004
24
Klein A et al., Cancer Research 2004
25
(No Transcript)
26

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Analytic methods
• Familial aggregation
• Segregation analysis
• Linkage analysis
• Association studies

27
Segregation Analysis

• A statistical methodology aimed at determining
  if a major gene could cause the observed familial
  aggregation of a trait. Segregation analysis can
  also provide a model for how this gene is
  inherited (AD vs. AR, frequency, penetrance)

28
Klein A et al. Evidence for a Major Gene
Influencing Risk of Pancreatic Cancer. Genetic
Epidemiology 23133-49, 2002.

• 287 Hopkins patients, 1994-1999, residents of
  Maryland
• Segregation analysis, which looks at patterns
  of transmission of pancreatic cancer in families
• Age at onset, susceptibility to pancreatic
  cancer separately evaluated

29
RESULTS

• Non-genetic transmission models were strongly
  rejected, whether inheritance was modeled as
  age-at-onset or susceptibility
• Modeling age-at-onset provided a better fit to
  the observed data
• Autosomal dominant inheritance of a rare allele
• 0.7 of the population appears to be at high risk
  of developing pancreatic cancer due to this
  putative gene
• Inclusion of smoking as a covariate did not
  significantly improve the fit of these models.

Klein A et al. Genetic Epidemiology 23133-49,
2002
30

• Mayo Segregation Analysis
• Data set constructed from
• 510 family histories were extracted from the Mayo
  Clinic records of pancreatic cancer patients seen
  in 1998-1999
• 210 probands (affected index cases through whom
  family was ascertained) patients who were
  registered in a prospective pancreatic cancer
  registry in 2001
• The data included in the analysis consisted of
  all members over age 40 (probands, siblings, and
  parents).

31
Table 1. 720 Mayo Clinic families included in
segregation analysis (parents, siblings, and
probands over age 40).
of Total Total of
YEAR Families of persons Affected Relatives
1998 294 1,874 13 1999 216
1,303 14 2001 210
1,287 17 TOTAL 720 4,464 44

• Proband mean age 66.7 10.8 (SD), range 40-100
  years
• Affected relative mean age 68.2 14.6, range
  43-94 years
• 58 of the probands had ever smoked

32
Table 3. Segregation analysis of pancreatic
cancer in 720 families Model Likelihoods and
Interpretation
Model -2lnL Interpretation 1. Dominant
478.53 Reject 2. Dominant, fixed p(A)
474.58 Cannot reject 3. Recessive 478.53
Reject 4. Additive 474.80 Cannot reject 5.
Codominant 474.56 Cannot reject 6. Sporadic
(nongenetic) 478.53 Reject 7. General
467.15 ------
33

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Analytic methods
• Familial aggregation
• Segregation analysis
• Linkage analysis
• Association studies

34
Pancreatic Cancer Genetic Epidemiology (PACGENE)
Consortium

• R01 CA 97075 (8/1/02- 7/31/07)
• Multi-center consortium
• Seven data collection sites
• Biostatistics Core at Mayo Clinic
• Tissue Genotyping Core at JHU
• Free genome scanning at CIDR

35
Pancreatic Cancer Genetic Epidemiology (PACGENE)
Consortium Sites

• Mt. Sinai /Toronto

• Mayo Clinic

• Karmanos CI

• Dana Farber

• Creighton Univ

• Johns Hopkins

• MD Anderson CC

36
PACGENE Specific Aims
Aim 1 To identify high risk pedigrees for
genetic linkage analysis utilizing established
pancreatic cancer family research resources. Six
centers will screen 8,900 new cases over 5 years
to accrue biospecimens, tumor tissue, and risk
factor data (including smoking history) from
available relevant family members of 75 FPC
pedigrees suitable for genetic linkage studies.
Aim 2 To genotype informative individuals in
high-risk FPC pedigrees with 400 evenly spaced
markers (10 centimorgan intervals) throughout
the genome. Genotyping with genome-wide
microsatellite markers will be done on an
estimated 1,500 individuals in the 75 FPC
families identified in Aim 1. Aim 3 To map a
pancreatic cancer susceptibility gene(s) by
genetic linkage analysis of the high-risk FPC
pedigrees.
37
PACGENE Consortium Structure
Steering Committee (Petersen, PI and Chair)
External Advisory Committee
Data Collection Sites
CORE 1 Data Management Statistical Genetics
(de Andrade, Mayo)
CORE 2 Pathology Archival Tissue
Genotyping (Goggins, Johns Hopkins)
Johns Hopkins University (Hruban)
Karmanos Cancer Institute (Korczak)
Mayo Clinic (Petersen)
MD Anderson Cancer Center (Bondy)
Mt. Sinai/Univ of Toronto (Gallinger)
SERVICE High Throughput Genotyping at Center
for Inherited Disease Research (CIDR)
Creighton University (Lynch)
Dana Farber (Syngal)
38
Familial Pancreatic Cancer
Dx 45
Dx 87
Dx 61
Dx 55
Mayo Pedigree 251
39
PACGENE Consortium Protocol
Yes
Confirmed pancreatic cancer case
2 affected blood relatives?
Detailed family hx qnaire (5-10 of cases)
No
Stop
3 affected FDRs or SDRs?
AND/OR Linkage simulation ELOD0.3?
No
Reserve for future model- free linkage studies
Yes
Extend family recruit all family members for
linkage study
Genotype 75 best pedigrees
Linkage Analysis
40
PANCREATIC CANCER GENETIC EPIDEMIOLOGY (PACGENE)
CONSORTIUM ACCRUAL AS OF 3/6/04
41
Rulyak SJ et al. Risk factors for the development
of pancreatic cancer in familial pancreatic
cancer kindreds. Gastroenterology 1241292-9,
2003

• Nested case-control study of 251 members of 28
  FPC families
• Smoking was an independent risk factor for
  familial pancreatic cancer (OR3.7 95 CI
  1.8-7.6)
• Smokers developed cancer 1 decade earlier than
  nonsmokers (59.6 vs. 69.1 years P 0.01)
• Number of affected first-degree relatives
  increased risk (OR, 1.4 95 CI, 1.1-1.9 for each
  additional family member).
• Diabetes was not a risk factor for pancreatic
  cancer
• 1/3 of families demonstrated genetic
  anticipation, as the mean age of onset decreased
  by 2 decades between generations.

42

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?

• Analytic methods
• Familial aggregation
• Segregation analysis
• Linkage analysis
• Association studies

43
Mayo Clinic SPORE in Pancreatic Cancer Project
period 9/29/03 - 8/31/08
Proj PI 1. Petersen/Olson 3.
Billadeau/ Erlichman 4. Couch/Alberts Cor
e PI 2. Lingle/Smyrk 4. De
Andrade/ Sargent
Title Molecular epidemiology of pancreatic
cancer The role of Vav1 protooncogene in
pancreatic cancer Characterization of the role
of BRCA2 in pancreatic cancer Patient Registry
Biospecimens Biostatistics
44
Molecular Epidemiology of Pancreatic
Cancer Study Goal To characterize the risk of
pancreatic cancer conferred by candidate genes
and cigarette smoking, family history of
pancreatic cancer, dietary carcinogens from meat
and its preparation, and NSAIDs.
45
Specific Aims 1. To assemble questionnaire
data, clinical data, and DNA samples from
leukocytes on 1200 ultra-rapidly recruited cases
of pancreatic adenocarcinoma and 1200 unrelated
primary care clinic controls. 2. To genotype
cases and controls for candidate gene
polymorphisms related to smoking 3. To estimate
risk of pancreatic adenocarcinoma associated with
candidate genes, environmental exposures, and
gene-environment interactions.
46
ASCERTAINMENT BIAS IN PANCREATIC CANCER
EPIDEMIOLOGY STUDIES

• Hoppin et al. (2002)
• 1,130 pancreatic cancer cases prospectively
  identified through a population-based SEER
  registry
• Only 30 were eligible for interview (70 had
  died)
• Interviewed 86 (26 of identified cases)
• Ultimate retrieval of 46 archival tissue blocks
  represented 16 of those interviewed, and 4 of
  all cases

47
ASCERTAINMENT BIAS IN PANCREATIC CANCER
EPIDEMIOLOGY STUDIES

• Porta et al. (2002)
• Retrospectively identified 149 pancreatic cancer
  cases in two hospital registries
• 58 had pathology records amenable to retrieve
  archival tissue
• Ultimately, they studied the tumors of only 34
  of the original 149 cases

48
ASCERTAINMENT BIAS IN PANCREATIC CANCER
EPIDEMIOLOGY STUDIES

• Anderson K (UM experience)
• Minnesota Cancer Surveillance System
• 56 notification of cases
• 18 were deceased when the research team was told
  about them and 7 of the cases could not be
  approached because their physician would not
  allow them to be contacted.
• 75 of approached patients enrolled.
• Total participation of eligible cases 30.

49
Allele/phenotype frequencies in pilot study of 62
pancreatic cancer cases and 61 normal colonoscopy
controls
Candidate genes Controls Cases CYP2E1 / G1293C
/ (C) .017 .016 CYP2E1 / C1053T /
(C) .967 .984 GSTT1 / (Null) .180 .145 GSTM1/
(Null) .525 .484 NQO1 / (C) .776 .879 Gen
omic control polymorphism HTR2A C1354T
(C) .926 .919
50
Allele/phenotype frequencies in pilot study of 62
pancreatic cancer cases and 61 normal colonoscopy
controls
Candidate genes Controls Cases All
Short Long (N61) (N62)
(N31) (N31) CYP2E1 / G1293C / (C)
.017 .016 .017 .016 CYP2E1 / C1053T /
(C) .967 .984 .984 .984 GSTT1 /
(Null) .180 .145 .161
.129 GSTM1/ (Null) .525 .484
.452 .516 NQO1 / (C) .776 .879
.817 .946 Genomic control polymorphism HTR2A
C1354T (C) .926 .919 .936
.903 Short died within 4 months of
recruitment Long survived 9 months after
recruitment
51
Allele/phenotype frequencies in pilot study of 62
pancreatic cancer cases and 61 normal colonoscopy
controls
Candidate genes Controls Cases
All No
Surg Surgery (N61) (N62)
(N52) (N10) CYP2E1 / G1293C / (C)
.017 .016 .012 -- CYP2E1 / C1053T / (C)
.967 .984 .981 1.00 GSTT1 / (Null)
.180 .145 .154 .100 GSTM1/ (Null)
.525 .484 .462 .600 NQO1 / (C)
.776 .879 .867 .944 Genomic control
polymorphism HTR2A C1354T (C) .926 .919
.914 .95
52

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?
53

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?
54

• Epidemiology
• Molecular epidemiology
• Genetic epidemiology

?
55
Acknowledgments

• Mariza de Andrade, Ph.D.
• Curt Olswold
• Kari Rabe, M.S.
• Que Luu
• Rob McWilliams, MD
• Suresh Chari, M.D.
• Janet Olson, Ph.D.
• Cindy Nixa
• Tammy Miller
• Tamela Dahl, RN
• Kathy Liffrig
• Debb Hare
• Mary Jo Eversman
• Diane Batzel
• PACGENE collaborators
• SPORE collaborators

56
(No Transcript)