Ongoing Research

Current research activities by GERA members in the area of cancer etiology have been grouped into four major themes that cut across multiple cancer sites and the unifying aspect to these themes relates more to our methodological approach:

• Genetic epidemiology of cancer, pre-malignant conditions and risk markers using family studies
• Genetics of cancer risk and outcome using association studies to evaluate genome-wide genetic variation, candidate gene and gene pathways and gene by environment interactions
• Biomarkers of risk
• Non-genetic risk factors and descriptive epidemiology

The following are some of the ongoing research projects in GERA:

Micro-satellite instability/ HNPCC (Hereditary Nonpolyposis Coli Cancer)
Investigators: Stephen Thibodeau, Ph.D., Noralane Lindor, M.D.

Drs. Thibodeau, Lindor and colleagues are researching an interesting genetic pattern found in patients with colon cancer. This pattern is called micro-satellite instability and is due to gene defects that are acquired or inherited, and that are important in the later occurrence of colon cancer. The inherited disease is called HNPCC (hereditary nonpolyposis coli cancer). The connection between microsatellite instability and colon cancer has been extensively studied at Mayo. The studies have found that people whose colon cancers have microsatellite instability have a better prognosis compared to those who do not. Mayo investigators have started to use this information to help surgeons determine the type of surgery that colon cancer patients should have. The use of genetics, family history, and microsatellite instability testing has moved from research to clinical practice at Mayo. For example, surgeons and gastroenterologists ask about family history to help refine risk and improve patient management, and the Department of Pathology routinely performs the microsatellite instability test on colon cancers of any patients who are under the age of 50. This work shows that if there are two people of the same age and same sex, with the same disease (colon cancer), but we find through this test that their tumors have different microsatellite instability test results, the surgeon can consider different surgical strategies for each individual. This direct benefit to patients is an example of the translational aspect of the GERA Program and its role as an interface between the laboratory bench and the patient's bedside.

Prostate Cancer Research

One of the best examples of gene discovery and identification is a strong collaboration in GERA with the Prostate Cancer Program. Dr. Thibodeau and Daniel Schaid, Ph.D. have been funded since 1996 with a National Cancer Institute (NCI) grant to identify and characterize genes that cause prostate cancer. A total of 12,675 Mayo Rochester patients who underwent surgery to remove their prostate for localized prostate cancer or who received radiation therapy for prostate cancer were sent a family cancer-history survey. From those surveys, 199 high-risk families were identified. More detailed family histories were obtained over the telephone and three- to four-generation pedigrees were constructed. From this group, 162 families having a minimum of three men affected with prostate cancer were collected for studies to determine if there were specific genetic factors in those families related to the high incidence of prostate cancer.

The DNA of these families was studied by looking at markers (similar to mileposts) on the human genome to help locate the gene that might be responsible for prostate cancer. A milepost on the long arm of chromosome 20 (20q13) appears to be indicating the genomic location of the prostate cancer gene, which is called HPC20. Drs. Thibodeau and Schaid have also pooled their data with other prostate cancer linkage studies as part of the International Consortium for Prostate Cancer Genetics.

Drs. Schaid, Thibodeau and James Cerhan, M.D., Ph.D., have been interested in evaluating candidate genes for aggressive prostate cancer. In a recent report from their study of 1,121 prostate cancer cases (435 with familial, 491 with sporadic and 195 with aggressive disease) and 545 controls, they confirmed two polymorphic variants (rs1447295 and DG8S737) were risk factors for familial and aggressive prostate cancer, but not sporadic disease (Wang et al., 2007). To expand this study, Alexander Parker, Ph.D., has opened a protocol to add aggressive cases to the study from Jacksonville.

Minnesota Breast Cancer Family Study

The Minnesota Breast Cancer Family Study, funded through a National Cancer Institute grant, is a study of relatives of 426 women who were diagnosed with breast cancer between 1944 and 1952. These women form a general population-based sample as they were not selected for their family history, and they came from all parts of Minnesota. Ninety-eight percent of eligible families were recruited, and 93 percent of family members participated for a total of 9,073 women.

Several important observations from this study have been reported to date. One of the first and most important observations was that the use of oral contraceptives marketed prior to 1975 was associated with a threefold excess risk of breast cancer among first-degree relatives in high-risk breast cancer families. This risk was especially evident (11-fold increase) in those families with the greatest familial clustering of breast and ovarian cancer. This report received widespread media attention after its publication (along with an editorial) in JAMA, The Journal of the American Medical Association, because there were almost no data on indicating whether or not oral contraceptive use increased the risk of breast cancer in women with a strong family history. These results were particularly critical for the high-risk women, because some had considered using oral contraceptives to reduce ovarian cancer risk. Other observations from this study included findings that smoking increased the risk of breast cancer in families with multiple cases of breast or ovarian cancers, especially those with the strongest apparent familial predisposition. In addition, the study showed that the increased risk of breast cancer due to alcohol use may be limited to women with a family history of the disease. Finally, most reproductive factors influence breast cancer risk similarly in women with and without a family history of the disease.

In a related study, investigators looked at mammographic breast density - one of the strongest and most consistent risk markers for breast cancer. In the first genome-wide scan for mammographic breast density, Celine Vachon, Ph.D., identified a strong linkage peak on chromosome 5 (Vachon et al., 2007). This project included interprogrammatic collaborations with Dr. Schaid, V. Shane Pankratz, Ph.D., and Julie Cunningham, Ph.D., and inter-programmatic collaborations with Fergus Couch, Ph.D. (Women's Cancer Program). The project also utilized the Survey Research, Biostatistics, Biospecimens Accessioning and Processing, and Genotyping Shared Resources. Dr. Vachon has just received a new R01 to follow-up this chromosomal region. The study will evaluate genetic variation in all candidate genes in the region of interest on chromosome 5 among a family study of 889 people used for the initial linkage study. Those candidates that explain the linkage signal will be examined in two samples from the Mayo Mammography Health Study. Any candidates that replicate in the two samples will also be examined with breast cancer risk.

Pancreatic Cancer Research

The genetic epidemiology of pancreatic cancer has matured as a major research focus in GERA over the last renewal period, and includes a strong intraprogrammatic collaboration among Gloria Petersen, Ph.D., Lisa Boardman, M.D., Mariza de Andrade, Ph.D., and Robert McWilliams, M.D., and interprogrammatic collaborations with Dr. Couch (Cell Biology Program). In 2002, Dr. Petersen was initially funded as the principal investigator of the Pancreatic Cancer Genetic Epidemiology (PACGENE) Consortium (R01 CA97075); this grant was recently competitively renewed for another five years to identify and characterized pancreatic cancer susceptibility genes. The database was used to show that a prior report of linkage to 4q32-34 could not be replicated in the PACGENE families (Klein et al., 2007).

The Mayo collection of families assembled for PACGENE is a resource that has been used to address other research questions as well. Dr. McWilliams, in collaboration with Drs. de Andrade, and Petersen, and Janet Olson, Ph.D., M.P.H.; found that a history of familial cancer syndromes (breast, ovarian, colorectal, melanoma) was associated with a younger onset of pancreatic cancer, but a family history of pancreatic cancer was not associated with age at onset, suggesting differing etiologic pathways based on the type of family history (McWilliams et al., 2006). First degree relatives of probands with pancreatic cancer were at nearly a 2-fold risk of pancreatic cancer but not other malignancies, supporting the recommendation of not conducting targeted screening in families with a pedigree not suggestive of known familial cancer syndromes (McWilliams et al. 2005). In a collaborative effort by Drs. Couch, Petersen and de Andrade, the contribution of mutations in BRCA2 to familial pancreatic cancer was assessed. Probands from 151 high-risk families from Mayo and 29 families from Johns Hopkins were screened for germline mutations, and a total of 10 carriers were identified, suggesting that BRCA2 mutations account for 6 percent of moderate and high risk pancreatic cancer families (Couch et al., 2007).

Drs. Petersen and de Andrade were co-investigators on a study (R01 CA100685) lead by Suresh Chari, M.D., (Gastrointestinal Cancer Program) that evaluated whether diabetes as assessed by fasting blood glucose levels predicted pancreatic cancer risk. The study was nested in the Mayo medical practice, and all fasting glucose levels for up to 60 months before diagnoses (for 736 pancreatic cancer cases) or reference data (1,875 matched controls) were abstracted. Diabetes (as defined by a fasting blood glucose of 126 milligrams/deciliter or greater or treatment for diabetes) had a higher prevalence (40 percent) in cases compared to controls (19 percent, P<0.001) (Chari et al., 2008). The team is following up on this observation with a project in the competitive renewal of the Pancreatic Cancer SPORE by attempting to identify a specific biomarker for pancreatic cancer-induced diabetes that would allow for screening for pancreatic cancer in new-onset diabetes.

Hematologic Malignancies Research

During the last funding cycle, a major new initiative in family studies in hematologic malignancies was initiated. Susan Slager, Ph.D., was funded to develop the Genetic Epidemiology of CLL (GEC) Consortium, which is both an intraprogrammatic collaboration with Dr. Vachon and James Cerhan, M.D., Ph.D., an interprogrammatic collaboration with Timothy Call, M.D., Neil Kay, M.D., and Curtis Hanson, M.D., (Hematologic Malignancies Program). This is a collaborative study with six other institutions including the National Cancer Institute's intramural program, and is funded by a U01 that was converted from an R01 (CA118444). The goal is to ascertain a large number of high-risk CLL families in order to conduct a genomic screen in order to map the chromosomal position of susceptibility gene(s) for CLL. The study is also ascertaining a CLL precursor (monoclonal B-cell lymphocytosis) in the families to increase study power and better understand the genetics of this recently described condition. An early collaborative effort from the GEC consortium (Drs. Slager and Call) and a major European consortium (International CLL Consortium) was recently published (Sellick et al., 2007). In that study, a genome-wide linkage analysis of 101 new CLL pedigrees using a high-density single nucleotide polymorphism (SNP) array was completed and combined the results with data from a previously reported analysis of 105 families. The combined study found evidence for linkage at 2q21.2, 6p22.1, and 18q21.1.

Dr. Cerhan is evaluating genetic variation in host defense as a prognostic factor in non-Hodgkin lymphoma, specifically variation in genes important in host immune function (R01 CA96704). Recent gene-expression data suggest that host immune genetic signatures predict outcome in patients with follicular lymphoma, and the strongest predictors of follicular lymphoma survival appear to be the gene expression signatures of the nonmalignant tumor infiltrating immune cells ("microenvironment"). This raises the hypothesis that germline, common variation in immune genes may be associated with survival. In collaboration with Thomas Habermann, M.D., and Stephen Ansell, M.D., Ph.D., (Hematologic Malignancies Program), Dr. Cerhan evaluated 73 SNPs from 44 genes, and found that SNPs in IL8, IL2, IL12B and IL1RN were strong predictors of survival (Cerhan et al., 2007). A summary score of the number of deleterious genotypes from these four genes was strongly associated with survival (p=0.001), and a risk score that combined the four SNPs plus clinical and demographic factors was even more strongly associated with survival (P<0.001). The 5-year Kaplan-Meier survival estimates were 95 percent, 72 percent and 58 percent for low, intermediate, and high risk, respectively. Similarly, an IL10 haplotype (global p=0.03) and SNPs in IL8RB, IL1A, TNF, and IL4R were the strongest predictors of overall survival in diffuse large B-cell lymphoma, and a risk score that combined the latter four SNPs with clinical and demographic factors was strongly associated with survival in a Cox model (p=6.0 X 10^-11) (Habermann et al., 2008). The 5-year Kaplan-Meier survival estimates for low, intermediate-low, intermediate-high, and high risk patients were 94 percent, 79 percent, 60 percent, and 48 percent respectively.

MGUS (monoclonal gammopathy of undetermined significance) is defined by the presence of serum monoclonal protein at a concentration of 3 g per deciliter or less, and is a premalignant plasma cell disorder that can develop into smoldering and multiple myeloma. Dr. Vachon is leading an effort to examine familial clustering of MGUS in first-degree family members of a diagnosed case of MGUS in Southeastern Minnesota (R01 CA107476). This is an interprogrammatic collaboration with S. Vincent Rajkumar, M.D., (Hematologic Malignancies Program). She also has another study with Dr. Rajkumar to evaluate MGUS clustering in family members of persons with multiple myeloma (P01 CA62242). This study should help determine whether screening for MGUS will identify family members at greatest risk of myeloma.

In another study, among residents of Olmsted County, Minnesota, MGUS was found in 3.2 percent of persons over the age of 50 years and in 5.3 percent of persons over the age of 70 years (Kyle et al., 2006), providing the most accurate estimates of prevalence published to date from a population-based sample. In a major interprogrammatic interaction with Dr. Rajkumar and colleagues from the Hematologic Malignancies Program, Terry Therneau, Ph.D., and Drs. Vachon and Cerhan are collaborators on the epidemiologic components to understanding MGUS as part of a program project grant (P01 CA62242) and an R01 to Dr. Rajkumar (CA107476).

GERA members, including Ping Yang, M.D., Ph.D., and Drs. de Andrade and Petersen, continue active involvement and funding in the Genetic Epidemiology of Lung Cancer Consortium (U01 CA 76293). The long-term objective of the GELCC is to characterize lung cancer susceptibility gene(s) and their interaction with cigarette smoking and/or other environmental agents. In 2004, the first linkage analysis of 52 high-risk extended pedigrees was published, and a major lung cancer susceptibility locus at 6q23-25 was identified (Baily-Wilson et al., 2004). Parametric linkage and variance-components analysis that incorporated effects of age and smoking history were implemented by Dr. de Andrade.

Brain Tumor Research

Building out of the Mayo SPORE in Brain Cancer, Dr. Yang and Robert Jenkins, M.D., Ph.D. (Neuro-Oncology Program), lead Mayo's participation in the Genetic Epidemiology of Glioma International Consortium (GLIOGENE) (R01 CA1192515), which is an international, multi-center study to identify susceptibility genes in high-risk familial brain tumor pedigrees. This project is fully-integrated into the ongoing work on the genetics of glioma risk and prognosis at Mayo Clinic, and represents a major interprogrammatic with the Neuro-Oncology Program.

Genome-Wide Association Studies

Ellen Goode, Ph.D. is Primary Investigator of a genome-wide study of ovarian cancer (R01 CA114343). The study will interrogate over 500,000 SNPs throughout the genome in a multi-stage, consortia analysis of ovarian cancer cases and controls. Sample preparation and plating is occurring in the Biospecimens Accessioning and Processing Shared Resource. The genotyping of the 500,000 SNPs will be conducted in the Genotyping Shared Resource, coordinated by Dr. Cunningham. Dr. Goode will also lead the analysis in collaboration with investigators at the H. Lee Moffitt Cancer Center & Research Institute. Dr. Slager, in collaboration with Drs. Cerhan, Cunningham, Kay and Call, is leading a study of 1,000,000 SNPs for chronic lymphocytic leukemia, which includes controls and sporadic cases from Dr. Cerhan's case-control study (R01 CA 91253) and familial cases from Genetic Epidemiology of CLL consortium (U01 CA118444). Dr. Vachon is leading a similar study of 1,000,000 SNPs for multiple myeloma in collaboration with Dr. Rajkumar. The latter two studies are taking advantage of a large genotyping project on which Dr. de Andrade is the lead biostatistician (R01 HL087660).

Christopher Chute, M.D., Dr.PH., received funding to conduct genome-wide association studies using electronic medical record-defined phenotypes (U01 HL004599). While the initial study will focus on myocardial infarction and peripheral vascular disease, the methods developed for mining the electronic medical record will be directly applicable to GERA investigators. Dr. Goode is Primary Investigator of a study through the Colon Cancer Family Registries (U01 CA122839). She will coordinate data analysis and integrate the results from the genome-wide association study with her linkage study.

Linkage Studies

Linkage peaks from family studies can be very useful for suggesting candidate genes for association studies. For example, Dr. Yang in collaboration with Dr. Jenkins is pursuing candidate genes on chromosome 19q with risk of glioma as an aim in Project 4 of the Mayo SPORE in Brain Cancer. Dr. Vachon is following up several candidate gene leads from her linkage study of mammographic density as part of her new R01. Similar types of studies are occurring for prostate cancer (Drs. Cunningham and Schaid with Dr. Thibodeau), pancreatic cancer (Drs. de Andrade and Petersen), and CLL (Drs. Slager and Cerhan with Drs. Kay and Call).

Immune and Inflammatory Pathways

Dr. Goode has recently received R01 (CA122443) funding to expand an ovarian cancer clinic-based case-control study and evaluate the role of genetic variation in the NF-κB pathway. This includes intraprogrammatic collaborations with Drs. Cunningham and Pankratz and interprogrammatic collaboration with the Women's Cancer Program (Dr. Couch and Lynn Hartmann, M.D.). A pilot study that evaluated over 80 genes from the NF-κB genes and risk of breast cancer was recently completed. These genes were tagged using an in-house program initially developed with GERA funding support and now part of the Bioinformatics Core.

Xenobiotic Metabolizing Enzymes

The role of phase I and II enzymes in the etiology of cancer has been an important research focus in the molecular epidemiology community, mainly through their role in metabolizing tobacco products and other exogenous chemicals. Glutathione (GSH) pathway genes (γGCS, GPX1, GSTP1, GSTM1, and GSTT1) were associated with lung cancer development in young and old populations through differing interactions with cigarette smoking and family history (Yang et al., 2004).

Richard Weinshilboum, M.D., collaborates with Dr. Yang extensively. For example, his lab has been involved in describing the pharmacogenomics of glutathione S-transferase (GST) omega 1 and omega 2 (Mukherjee et al., 2006), which is the rate-limiting step in arsenic biotransformation. These genes are now being evaluated population studies. Dr. Yang was the senior author of a HuGE review showing that myeloperoxidase G-463A polymorphism is significantly protective in ever but not never-smokers (Taioli et al., 2007). In a Mayo-based case-control study of pancreatic cancer, Drs. Petersen, Olson, and de Andrade are evaluating the role of NAT2, CYP1A1, CYP2E1, GSTM1, GSTT1, and NQO1 with risk of pancreatic cancer, and the interaction of these genes with smoking and dietary carcinogens from meat preparation (P50 CA102701).

Medical Informatics

Most of the research in medical informatics is related to Dr. Chute's pioneering work in this area. Among recent activities are the development and deployment of the open-source suite of terminology services manifest in the LexGrid project (http://informatics.mayo.edu); authoring, coordinating and leading the publication of Common Terminology Services (CTS) specification in HL7 (www.hl7.org), now an ANSI and ISO standard; deployment of CTS as a key infrastructure component in caBIG; development of open standards-compliant export formats for clinical and genomic data, among many other activities. Dr. Chute's long-term research interest has been in natural language processing (NLP). His group has built a 15-element pipeline, including tokenizers, part of speech classifiers, lexical normalizers, synonym mappers, and entity recognition tools. The latter tools have been syntactically and semantically normalized to standard CHI terminologies, including SNOMED for diagnoses and symptoms and RxNorm for drugs. The Mayo NLP annotator pipeline receives clinical documents immediately after transcription, and processing them within seconds of transcription, giving near immediate access to detailed clinical data in the Mayo electronic medical record (which is completely electronic). GERA members are beginning to take advantage of these tools for study design, participant identification, phenotype definition, exposure assessment, and patient follow-up/outcome. These tools are also being integrated into the Cancer Center informatics infrastructure.

Dr. Chute also has a subcontract for his work on the National Center for Biomedical Ontology (N01 HG04028), which is focused on developing tools and methods for assimilating, archiving, and accessing machine-processable representations of biomedical domain objects, processes, and relations to assist in the management, integration, visualization, analysis and interpretation of extremely large, distributed datasets. Finally, Dr. Chute also coordinates the medical informatics component of the Rochester Epidemiology Project (R01 AR30582). A new U01 (HG004599) awarded to Dr. Chute from NHGRI (EMR Phenotypes and Community Engaged Genomic Associations) which will develop medical informatics approaches to phenotype definition from medical records to facilitate genome-wide association studies. While the initial grant is focused on cardiovascular disease, the methods developed will be directly applicable to cancer genetic epidemiology studies. Drs. de Andrade and Cerhan are co-investigators on the grant

Survey Research Methodology

Timothy Beebe, Ph.D., who also leads the Survey Research Shared Resource, has an active research program in survey research methodology. He has recently compared the effects of paper versus web-based data collection modes on response rates and other properties in adolescents (Beebe et al., 2006) and physicians (Beebe et al., 2007), and has been recently funded to conduct a systematic analysis of survey nonresponse bias due to the Health Insurance Portablity and Accountability Act (HIPAA) authorization form and the use of mixed-mode data collection design. Dr. Beebe is also actively researching approaches for recruiting population-based samples through the Rochester Epidemiology Project (R01 AR30582) and in the clinic (part of Mayo Clinic Controls Initiative).

Dr. Yang's group has used her large cohort of lung cancer patients to evaluate additional prognostic factors for lung cancer prognosis beyond traditional clinical characteristics and tumor markers. For example, male sex was confirmed to be an independent predictor of unfavorable prognosis for non-small cell lung cancer (Visbal et al., 2004) and greater duration of smoking abstinence post-lung cancer diagnosis was associated with better survival among women, but this was not observed among men or for small-cell lung cancer in either sex (Ebbert et al., 2005). Continued cigarette smoking after lung cancer diagnosis negatively impacted quality of life (QOL) scores as measured by the Lung Cancer Symptom Scale (Garces et al., 2004), change in QOL correlated with several baseline variables including disease stage, histology, Karnofsky scale, weight loss and gender (Svobonik et al., 2004). Another report found that vitamin and mineral supplementation is associated with better survival in non-small cell lung cancer (HR=0.74; 95 percent CI 0.44-0.65) and a better quality of life (Jatoi et al., 2005). Dr Yang used these preliminary data to develop an R01 (CA115857) entitled Health and QOL among Long-Term Lung Cancer Survivors, which focuses on both a descriptive study of clinical outcomes and general health and an analytic study of predictors of a longer and better QOL among lung cancer survivors.