Research Advances

Research in Genetic Epidemiology and Risk Assessment (GERA) is bringing scientists closer to the day of 'personalized medicine' when physicians will routinely order genetic profiles to determine treatment and to even recommend prevention strategies for patients based on this information.

A brief overview of selected advances in GERA research is listed below. Other research activities are described in Ongoing Research.

Prostate Cancer Gene Discovery
The longest-standing research program at Mayo in gene discovery is for prostate cancer, and is just one example of the many intra-programmatic and inter-programmatic collaborations that exist within GERA and between it and the other Mayo Clinic Cancer Center Programs. Daniel Schaid, Ph.D. and Stephen Thibodeau, Ph.D. assembled 162 Mayo families having a minimum of three men affected with prostate cancer, and much of their recent work has focused on pooling data and fine mapping candidate regions through collaborative projects, and on identifying genes associated with aggressive disease. Drs. Schaid and Thibodeau are actively involved in pooling their data with other prostate cancer linkage studies as members of the International Consortium for Prostate Cancer Genetics (ICPCG). High profile results from the ICPCG include the initial pooling of families leading to suggestive linkage targets at 5q12, 8p21, 15q11, 17q21, and 22q12 (Xu et al., 2005) and more recent compelling evidence for a prostate cancer gene at 22q12.3 (Camp et al., 2007). Further, combining data from Mayo and a study from Seattle (R01 CA80122), the interval on this region was narrowed to 1.36 Mb (Johanneson et al., 2007). Aggressiveness genes have been reported at several loci (Slager et al., 2006; Stanford et al., 2006), and Dr. Schaid led a pooled analysis in the ICPCG that implicated 6p22.3, 11q14.1-14.3, and 20p11.21-q11.21 for aggressive disease (Schaid et al., 2006). In other prostate research, two common variants on chromosome 8q24 were associated with increased risk of prostate cancer (Wang et al., 2007).

Susceptibility Genes in Colon Cancer
GERA studies found that people whose colon cancers have a specific genetic alteration – microsatellite instability, have better prognoses than those who do not. The studies also give information about the genetic susceptibility of patients and their family members to cancer. The information derived from these studies can help surgeons weigh options to better plan each patient's course of treatment, and to provide more accurate cancer risk assessment.

Minnesota Breast Cancer Family Study
More than 9,000 Minnesota women at risk for breast cancer were studied, resulting in several important observations. One was that the use of oral contraceptives marketed prior to 1975 was associated with a threefold risk of breast cancer among first-degree relatives in high-risk breast cancer families. The study also found that smoking increased the risk of breast cancer, and that increased risk of breast cancer due to alcohol use may be limited to women with a family history of the disease. Finally, the studies found that most reproductive factors influence breast cancer risk similarly in women with and without a family history of the disease.

Tissue Markers: Benign Breast Disease
In a strong interprogrammatic interaction with Lynn Hartmann, M.D. from the Women's Cancer Program, V. Shane Pankratz, Ph.D., Celine Vachon, Ph.D. and Ellen Goode, Ph.D. have collaborated on a cohort of nearly 10,000 women with benign breast disease. Recent findings include that benign breast disease was associated with a 56 percent increased risk of breast cancer, this risk persisted for at least 25 years after biopsy, and risk was strongest for lesions with atypia or proliferative changes with atypia (Hartmann et al., 2005). This group also reported that lobular involution was associated with a reduced risk of breast cancer, suggesting that aberrant involution may be an important biologic mechanism in breast carcinogenesis (Milanese et al., 2006).

Metabolic Pathways
The role of inter-individual genetic variation in the synthesis and metabolism of hormones in the etiology of several hormone-dependent cancers, including breast, ovarian and prostate cancer, has been an important focus of the research programs involving Janet Olson, Ph.D., M.P.H.; Julie Cunningham, Ph.D.; James Cerhan, M.D., Ph.D.; and Drs. Pankratz and Schaid, and presents a strong inter-programmatic interaction with the Developmental Therapeutics Program and the laboratory of Richard Weinshilboum, M.D., the Women's Cancer Program (Dr. Hartmann and Fergus Couch, Ph.D.), and the Prostate Cancer Program (Dr. Thibodeau and Michael Blute, M.D.). In ovarian cancer, seven SNPs in four genes involved in catechol estrogen formation (CYP1A1 and CYP1B1) or conjugation (COMT and SULT1A1) were genotyped in 503 ovarian cases and 609 controls (combined studies from Mayo Clinic and Duke University). While none of the individual genotypes were significantly associated with risk, an oligogenic model that modeled the joint effects of the four candidate genes provided evidence for an association (p=0.015), providing preliminary support for the hypothesis that low-penetrance susceptibility alleles from this pathway influence ovarian cancer risk (Sellers et al., 2005).

Dr. Weinshilboum's lab resequenced CYP19 through the Pharmacogenetic Research Network, providing data on 88 SNPs (Ma et al., 2005). CYP19 encodes aromatase, which is a major pathway for conversion of androgens to estrogens, and a strong candidate gene for breast cancer. Haplotype and tagSNPs were selected and genotyped, but none of the individual SNPs were associated with breast cancer (Olson et al., 2007) or mammographic breast density (Olson et al., 2007b). In prostate cancer, 46 polymorphisms from the estrogen and androgen metabolic pathways were evaluated in familial (N=438 from 178 high-risk families), sporadic (N= 499 with no family history), and screen-negative controls (N=493). Suggestive findings were found with AKR1C3, HSD17B1, NQO1, and GSTT1, although none were significant after adjustment for multiple comparisons (Cunningham et al., 2007). Further, in the process of genotyping SULT1A1 in the prostate cancer study, suspicious genotyping results across platforms lead to further study of the gene and the finding that copy number variation is very common, and that variability in the level of enzyme activity was best explained by gene number differences when all sources of genetic variability were considered (copy number, deletions, haplotypes) (P<0.0001) (Hebbring et al., 2007). This has led to the need to reassess findings for this important gene in drug and hormone metabolism.

Medical History and Patient Record Advances
Drs. Cerhan, Olson and Vachon, in collaboration with investigators from the Cancer Prevention and Control Program (Jon Ebbert, M.D. and Paul Limburg, M.D.) reported that aspirin use, but not non-aspirin NSAID use, was associated with lower risks of cancer incidence, cancer morality, and coronary heart disease mortality, which was more pronounced among former and never smokers than current smokers (Bardia et al., 2007). The population attributable risk calculated from these data suggest that aspirin use could potentially prevent 4.7 percent of the cancer incidence burden, 3.5 percent of the cancer mortality burden, and 7.6 percent of the coronary heart disease burden in the general population.

An automated system for assigning diagnosis codes to patient encounters was developed and implemented, resulting in a reduction in staff engaged in manual coding from 34 coders to seven verifiers (Pakhomov et al., 2006).

Gene Expression
In a collaboration of Alexander Parker, Ph.D.; George Vasmatzis, Ph.D.; Bradley Leibovitch, M.D.; John Cheville, M.D.; Dr. Blute and others; a combination of genomic profiling and validation by quantitative RT-PCR was used to identify a panel of candidate biomarkers for determining aggressiveness of clear cell renal cell carcinoma (RCC); 34 of 35 of the most significant transcripts were validated in an independent sample using RT-PCR (Kosari et al., 2005). Furthermore, protein expression of one of the top candidate markers (BIRC5 or survivin) was inversely associated with cancer-specific survival (p=0.017) in a cohort of 183 Mayo RCC patients. Dr. Parker validated this observation in an independent cohort of Mayo Mayo RCC patients, showing that survivin expression was associated with disease progression (HR=3.9; 2.4-6.2) after adjusting for clinical prognostic factors (Parker et al., 2005). This research was conducted through the support of a number of Cancer Center Shared Resources, for example, the expression work was conducted in the Gene Analysis Shared Resource, and the tissue processing and staining for survivin was conducted in the TACMA Shared Resource.

Association Studies
The evaluation of the association of single genetic loci with a trait may be greatly enhanced by the evaluation of haplotypes. Analysis of haplotypes can also provide critical information regarding the function of a gene; however when unrelated subjects are sampled, haplotypes are often ambiguous because of unknown linkage phase of the measured loci along a chromosome. Dr. Schaid has had funding to develop innovative statistical methods and software to model haplotypes and other complex genetic mechanisms in case-control studies. The rationale and application of haplotype approaches for the analysis of data from genetic epidemiology studies were published in a series of papers (Lake et al, 2003; Schaid DJ, 2004a, 2004b, 2004c) that highlight the many advantages of the regression framework, including the ability to directly model the effects of haplotypes, control for non-genetic covariates, use step-wise selection to screen for a subset of markers that explain most of the association, evaluate haplotype-environment interactions, and utilize regression diagnostics.

On the same note, Dr. Schaid derived analytic methods to determine sample size and power to test the association of haplotypes with either a quantitative trait or disease status (e.g., a case-control study design), assuming that all subjects are unrelated. These derivations covered both phase-known and phase-unknown haplotypes, allowing evaluation of the loss of efficiency due to unknown phase, and included an extension to when a subset of tag-SNPs is chosen, allowing investigators to explore the impact of tag-SNPs on power. Dr. Schaid has also developed an exact stratified test of Hardy-Weinberg equilibrium (HWE) for diallelic markers and an exact test for homogeneity of HWE, which is particularly useful for samples composed of multiple ethnic groups (Schaid et al., 2006). These various methods have been made widely available through the internet-based Mayo SPLUS/R library, including haplo.stats, haplo.glm and haplo.power.

In another development, researchers determined multi-locus association analyses, including haplotype-based analyses, can sometimes provide greater power than single locus analyses for detecting disease susceptibility loci. Drs.Yu and Schaid presented a sequential haplotype scan method to search for combinations of adjacent markers that are jointly associated with disease status. When evaluating each marker, markers are added close to each other in a sequential manner: a marker is added if its contribution to the haplotype association with disease is warranted, conditional on current haplotypes. The proposed sequential haplotype scan algorithm is more powerful than single-locus method and it is more computationally efficient than sliding window methods.

Pathway Analysis
While major developments have been made in identifying the genetic causes of rare Mendelian disorders, slower progress has been made in the discovery of common gene variations that predispose to complex diseases. The single gene variants that have been shown to associate reproducibly with complex diseases typically have small effect sizes or attributable risks. However, the joint actions of common gene variants within pathways may play a major role in predisposing to complex diseases. Some recent GERA accomplishments in this area include development of a robust methodology for combining at risk genotypes in a common carrier model which was applied to a study evaluating the role of multiple immune SNPs in predicting survival in follicular lymphoma patients (Cerhan et al., 2007); and development by Terry Therneau, Ph.D., and colleagues of a new class of semi-parametric regression models, termed partially linear tree-based regression models, which yields a parsimonious summary of the joint effects of multiple SNPs from a pathway.

Microarray and Proteomics Analysis
The purpose of normalization is to minimize the systematic variations in the measured gene expression levels among different array hybridizations to allow the comparison of expression levels across arrays, so that biological differences can be more easily identified. Dr. Therneau in collaboration with Ann Oberg, Ph.D. and Karla Ballman, Ph.D. have developed a method of normalization for high-density oligonucleotide arrays that yields results similar to cyclic loess normalization with speed comparable to quantile normalization (Ballman et al., 2004). For normalization of two-channel microarray experiments, they also developed a semiparametric approach to account for the systematic biases in the measured expression levels related to experimental factors such as spot location (often referred to as a print-tip effect), arrays, dyes, and various interactions of these effects for two-channel microarrays (Eckel et al., 2005). They pointed out that normalization is a critical initial step in the analysis of a microarray experiment, where the objective is to balance the individual signal intensity levels across the experimental factors, while maintaining the effect due to the treatment under investigation.

Future Studies
GERA members have received substantial funding for their research from the National Institutes of Health and the Department of Defense. In the next five years, the program plans to strengthen its work in the areas of kidney/bladder cancer, leukemia, brain cancer and melanoma and to aggressively pursue new initiatives in proteomics and genomics. The program will also expand its interaction with Mayo's Cancer Prevention and Control Program, including studies of the genetics of tobacco addiction, behavioral research in cancer genetic risk, and screening.