Search

Recruitment of minority participants to clinical trials, especially studies without therapeutic intent, has been historically challenging. This study describes barriers to and successes of recruitment and retention strategies to dietary studies. A flaxseed study was conducted in healthy, postmenopausal women of African ancestry (AA) and European ancestry (EA) to assess associations between gut microbial community composition and host metabolism (NCT01698294). To ensure equitable participation by AA and EA women, multiple forms of recruitment were utilized, including advertisements, posters, e-mail, word of mouth, and community outreach. Successful recruitment and retention of AA women to the intervention depended upon the specific methods used. AA women compared with EA women were more likely to respond to direct recruitment and community-based methods, rather than general advertisements. However, once women expressed interest, similar rates of consent were observed for AA and EA women (AA and EA: 51.6% vs. 55.7%, respectively; P  > 0.05), supporting the willingness of minority populations to participate in clinical research. Retention, however, was lower among AA compared with EA women (AA and EA: 57.6% vs. 80.9%, respectively; P  < 0.01), which may be related to multiple factors, including health reasons, intolerance to flaxseed, noncompliance with study requirements, time constraints, and nonspecified personal reasons. This study confirms the utility of direct community-based strategies for recruitment of diverse populations into nontherapeutic dietary intervention studies. The methods used successfully identified eligible women who expressed willingness to consent to the trial and were able to achieve >70% of recruitment goals for AA women. Future efforts are warranted to improve retention to complex studies. This trial was registered at www.clinicaltrials.gov as NCT01698294., (© The Author(s) 2022. Published by Oxford University Press on behalf of the American Society for Nutrition.)

Motivation: High-dimensional genomic data can be analyzed to understand the effects of variables on a target variable such as a clinical outcome. For understanding the underlying biological mechanism affecting the target, it is important to discover the complete set of relevant variables. Thus variable selection is a primary goal, which differs from a prediction criterion. Of special interest are functional modules, cooperating sets of variables affecting the target which can be characterized by a graph. In applications such as social networks, the concept of balance in undirected signed graphs characterizes the consistency of associations within the network. This property requires that the module variables have a joint effect on the target outcome with no internal conflict, an efficiency that may be applied to biological networks., Results: In this paper, we model genomic variables in signed undirected graphs for applications where the set of predictor variables influences an outcome. Consequences of the balance property are exploited to implement a new module discovery algorithm, balanced Functional Module Detection (bFMD), which selects a subset of variables from high-dimensional data that compose a balanced functional module. Our bFMD algorithm performed favorably in simulations as compared to other module detection methods. Additionally, bFMD detected interpretable results in an application using RNA-seq data obtained from subjects with Uterine Corpus Endometrial Carcinoma using the percentage of tumor invasion as the outcome of interest. The variables selected by bFMD have improved interpretability due to the logical consistency afforded by the balance property., Supplementary Information: Supplementary data are available at Bioinformatics Advances online., (© The Author(s) 2021. Published by Oxford University Press.)

Purpose: GOG-0218, a double-blind placebo-controlled phase III trial, compared carboplatin and paclitaxel with placebo, bevacizumab followed by placebo, or bevacizumab followed by bevacizumab in advanced epithelial ovarian cancer (EOC). Results demonstrated significantly improved progression-free survival (PFS), but no overall survival (OS) benefit with bevacizumab. Blood samples were collected for biomarker analyses., Experimental Design: Plasma samples were analyzed via multiplex ELISA technology for seven prespecified biomarkers [IL6, Ang-2, osteopontin (OPN), stromal cell-derived factor-1 (SDF-1), VEGF-D, IL6 receptor (IL6R), and GP130]. The predictive value of each biomarker with respect to PFS and OS was assessed using a protein marker by treatment interaction term within the framework of a Cox proportional hazards model. Prognostic markers were identified using Cox models adjusted for baseline covariates., Results: Baseline samples were available from 751 patients. According to our prespecified analysis plan, IL6 was predictive of a therapeutic advantage with bevacizumab for PFS ( P = 0.007) and OS ( P = 0.003). IL6 and OPN were found to be negative prognostic markers for both PFS and OS ( P < 0.001). Patients with high median IL6 levels (dichotomized at the median) treated with bevacizumab had longer PFS (14.2 vs. 8.7 months) and OS (39.6 vs. 33.1 months) compared with placebo., Conclusions: The inflammatory cytokine IL6 may be predictive of therapeutic benefit from bevacizumab when combined with carboplatin and paclitaxel. Aligning with results observed in patients with renal cancer treated with antiangiogenic therapies, it appears plasma IL6 may also define those patients with EOC more or less likely to benefit from the addition of bevacizumab to standard chemotherapy., (©2020 American Association for Cancer Research.)

Purpose: Bladder cancer is one of the top five cancers diagnosed in the U.S. with a high recurrence rate, and also one of the most expensive cancers to treat over the life-course. However, there are few observational, prospective studies of bladder cancer survivors., Methods: The Bladder Cancer Epidemiology, Wellness, and Lifestyle Study (Be-Well Study) is a National Cancer Institute-funded, multi-center prospective cohort study of non-muscle-invasive bladder cancer (NMIBC) patients (Stage Ta, T1, Tis) enrolled from the Kaiser Permanente Northern California (KPNC) and Southern California (KPSC) health care systems, with genotyping and biomarker assays performed at Roswell Park Comprehensive Cancer Center. The goal is to investigate diet and lifestyle factors in recurrence and progression of NMIBC, with genetic profiles considered, and to build a resource for future NMIBC studies., Results: Recruitment began in February 2015. As of 30 June 2018, 1,281 patients completed the baseline interview (774 KPNC, 511 KPSC) with a recruitment rate of 54%, of whom 77% were male and 23% female, and 80% White, 6% Black, 8% Hispanic, 5% Asian, and 2% other race/ethnicity. Most patients were diagnosed with Ta (69%) or T1 (27%) tumors. Urine and blood specimens were collected from 67% and 73% of consented patients at baseline, respectively. To date, 599 and 261 patients have completed the 12- and 24-month follow-up questionnaires, respectively, with additional urine and saliva collection., Conclusions: The Be-Well Study will be able to answer novel questions related to diet, other lifestyle, and genetic factors and their relationship to recurrence and progression among early-stage bladder cancer patients.

Background: Metabolism and excretion of the phytoestrogen enterolactone (ENL), which has been associated with breast cancer risk, may be affected by variation in steroid hormone and xenobiotic-metabolizing genes., Methods: We conducted a randomized, crossover flaxseed intervention study in 252 healthy, postmenopausal women [137 European ancestry (EA) and 115 African ancestry (AA)] from western New York. Participants were randomly assigned to maintain usual diet or consume 10 g/day ground flaxseed for 6 weeks. After a 2-month washout period, participants crossed over to the other diet condition for an additional 6 weeks. Urinary ENL excretion was measured by gas chromatography-mass spectrometry and 70 polymorphisms in 29 genes related to steroid hormone and xenobiotic metabolism were genotyped. Mixed additive genetic models were constructed to examine association of genetic variation with urinary ENL excretion at baseline and after the flaxseed intervention., Results: SNPs in several genes were nominally ( P < 0.05) associated with ENL excretion at baseline and/or after intervention: ESR1, CYP1B1, COMT, CYP3A5, ARPC1A, BCL2L11, SHBG, SLCO1B1 , and ZKSCAN5 . A greater number of SNPs were associated among AA women than among EA women, and no SNPs were associated in both races. No SNP-ENL associations were statistically significant after correction for multiple comparisons., Conclusions: Variation in several genes related to steroid hormone metabolism was associated with lignan excretion at baseline and/or after flaxseed intervention among postmenopausal women., Impact: These findings may contribute to our understanding of the differences observed in urinary ENL excretion among AA and EA women and thus hormone-related breast cancer risk., (©2019 American Association for Cancer Research.)

Objective: This study evaluated single nucleotide polymorphisms (SNPs) associated with progression free (PFS) and overall survival (OS) in patients with advanced stage serous EOC., Methods: Patients enrolled in GOG-172 and 182 who provided specimens for translational research and consent were included. Germline DNA was evaluated with the Illumina's HumanOMNI1-Quad beadchips and scanned using Illumina's iScan optical imaging system. SNPs with allele frequency>0.05 and genotyping rate>0.98 were included. Analysis of SNPs for PFS and OS was done using Cox regression. Statistical significance was determined using Bonferroni corrected p-values with genomic control adjustment., Results: The initial GWAS analysis included 1,124,677 markers in 396 patients. To obtain the final data set, quality control checks were performed and limited to serous tumors and self-identified Caucasian race. In total 636,555 SNPs and 289 patients passed all the filters. The pre-specified statistical level of significance was 7.855e -08 . No SNPs met this criteria for PFS or OS, however, two SNPs were close to significance (rs10899426 p-2.144e - 08 ) (rs6256 p-9.774e -07 ) for PFS and 2 different SNPs were identified (rs295315 p-7.536e - 07 ; rs17693104 p-7.734e - 07 ) which were close to significance for OS., Conclusions: Using the pre-specified level of significance of 1×10 -08 , we did not identify any SNPs of statistical significance for OS or PFS, however several were close. The SNP's identified in this GWAS study will require validation and these preliminary findings may lead to identification of novel pathways and biomarkers., (Copyright © 2017 Elsevier Inc. All rights reserved.)

Multiple candidate gene-association studies of non-HLA single-nucleotide polymorphisms (SNPs) and outcomes after blood or marrow transplant (BMT) have been conducted. We identified 70 publications reporting 45 SNPs in 36 genes significantly associated with disease-related mortality, progression-free survival, transplant-related mortality, and/or overall survival after BMT. Replication and validation of these SNP associations were performed using DISCOVeRY-BMT (Determining the Influence of Susceptibility COnveying Variants Related to one-Year mortality after BMT), a well-powered genome-wide association study consisting of 2 cohorts, totaling 2888 BMT recipients with acute myeloid leukemia, acute lymphoblastic leukemia, or myelodysplastic syndrome, and their HLA-matched unrelated donors, reported to the Center for International Blood and Marrow Transplant Research. Gene-based tests were used to assess the aggregate effect of SNPs on outcome. None of the previously reported significant SNPs replicated at P < .05 in DISCOVeRY-BMT. Validation analyses showed association with one previously reported donor SNP at P < .05 and survival; more associations would be anticipated by chance alone. No gene-based tests were significant at P < .05. Functional annotation with publicly available data shows these candidate SNPs most likely do not have biochemical function; only 13% of candidate SNPs correlate with gene expression or are predicted to impact transcription factor binding. Of these, half do not impact the candidate gene of interest; the other half correlate with expression of multiple genes. These findings emphasize the peril of pursing candidate approaches and the importance of adequately powered tests of unbiased genome-wide associations with BMT clinical outcomes given the ultimate goal of improving patient outcomes., (© 2017 by The American Society of Hematology.)

The incidence and mortality rates of B-cell acute lymphoblastic leukemia (B-ALL) differ by age and sex. To determine if inherited genetic susceptibility contributes to these differences we performed 2 genome-wide association studies (GWAS) by age, sex, and subtype and subsequent meta-analyses. The GWAS included 446 B-ALL cases, and 3027 healthy unrelated blood and marrow transplant (BMT) donors as controls from the Determining the Influence of Susceptibility Conveying Variants Related to One-Year Mortality after BMT (DISCOVeRY-BMT) study. We identified 1 novel variant, rs189434316, significantly associated with odds of normal cytogenetic B-ALL (odds ratio from meta-analysis [OR meta ] = 3.7; 95% confidence interval [CI], 2.5, 6.2; P value from meta-analysis [ P meta ] = 6.0 × 10 -9 ). The previously reported pediatric B-ALL GWAS variant, rs11980379 ( IKZF1 ), replicated in B-ALL pediatric patients (OR meta = 2.3; 95% CI, 1.5, 3.7; P meta = 1.0 × 10 -9 ), with evidence of heterogeneity ( P = .02) between males and females. Sex differences in single-nucleotide polymorphism effect were seen in those >15 years (OR = 1.7; 95% CI, 1.4, 2.2, P Males = 6.38 × 10 -6 /OR = 1.1; 95% CI, 0.8, 1.5; P Females = .6) but not ≤15 years (OR = 2.3; 95% CI, 1.4, 3.8; P Males = .0007/OR = 1.9; 95% CI, 1.2, 3.2; P Females = .007). The latter association replicated in independent pediatric B-ALL cohorts. A previously identified adolescent and young-adult onset ALL-associated variant in GATA3 is associated with B-ALL risk in those >40 years. Our findings provide more evidence of the influence of genetics on B-ALL age of onset and we have shown the first evidence that IKZF1 associations with B-ALL may be sex and age specific., Competing Interests: Conflict-of-interest disclosure: T.H. owns stock in Novartis Pharmaceuticals Corporation. D.J.W. provided consulting services to, and served on advisory boards for, Kadmon and Alexion, and received research funding from Alexion. S.S. received compensation for travel, accommodations, and expenses from Astellas Pharma. M.C.P. received honoraria, as well as compensation for travel, accommodations, and expenses, from Baxalta and Atara Biotherapeutics. P.L.M. received honoraria from Celgene, Bristol-Myers Squibb, Janssen Pharmaceutical, Sanofi, and Karyopharm Therapeutics Inc; research funding from Celgene; and compensation for travel, accommodations, and expenses from Celgene and Sanofi. The remaining authors declare no competing financial interests.

Clinical trials commonly use adjudication committees to refine endpoints, but observational research or genome-wide association studies rarely do. Our goals were to establish definitions of cause-specific death after unrelated-donor allogeneic hematopoietic cell transplantation (URD-HCT), to estimate discordance between reported and adjudicated cause-specific death, and to identify factors contributing to inconsistency in cause-specific death determination. A consensus panel adjudicated cause-specific death in 1484 patients who died within 1 year after HCT, derived from 3532 acute leukemia or myelodysplasia patients after URD-HCT from 2000 to 2011 reported by 151 US transplant centers to the Center for International Blood and Marrow Transplant Research. Deaths were classified as disease-related or transplant-related. The panel agreed with >99% of deaths reported by centers as disease-related and 80% reported as transplant-related. Year of transplant (cohort effect) and disease status significantly influenced agreement between the panel and centers. Sensitivity analysis of deaths < 100 days post-transplant yielded the lowest agreement between the panel and centers for myelodysplastic syndrome patients. Standard predefined criteria for adjudicating cause-specific death led to consistent application to similar clinical scenarios and clearer delineation of cause-specific death categories. Other studies of competing events such as cancer-specific versus treatment-related mortality would benefit from our results. Our detailed algorithm should result in more consistent reporting of cause-specific death by centers., (Copyright © 2015 American Society for Blood and Marrow Transplantation. Published by Elsevier Inc. All rights reserved.)

Proteinuria is the most important predictor of outcome in glomerulonephritis and experimental data suggest that the tubular cell response to proteinuria is an important determinant of progressive fibrosis in the kidney. However, it is unclear whether proteinuria is a marker of disease severity or has a direct effect on tubular cells in the kidneys of patients with glomerulonephritis. Accordingly we studied an in vitro model of proteinuria, and identified 231 "albumin-regulated genes" differentially expressed by primary human kidney tubular epithelial cells exposed to albumin. We translated these findings to human disease by studying mRNA levels of these genes in the tubulo-interstitial compartment of kidney biopsies from patients with IgA nephropathy using microarrays. Biopsies from patients with IgAN (n = 25) could be distinguished from those of control subjects (n = 6) based solely upon the expression of these 231 "albumin-regulated genes." The expression of an 11-transcript subset related to the degree of proteinuria, and this 11-mRNA subset was also sufficient to distinguish biopsies of subjects with IgAN from control biopsies. We tested if these findings could be extrapolated to other proteinuric diseases beyond IgAN and found that all forms of primary glomerulonephritis (n = 33) can be distinguished from controls (n = 21) based solely on the expression levels of these 11 genes derived from our in vitro proteinuria model. Pathway analysis suggests common regulatory elements shared by these 11 transcripts. In conclusion, we have identified an albumin-regulated 11-gene signature shared between all forms of primary glomerulonephritis. Our findings support the hypothesis that albuminuria may directly promote injury in the tubulo-interstitial compartment of the kidney in patients with glomerulonephritis.

Background: Multifactorial diseases such as cancer and cardiovascular diseases are caused by the complex interplay between genes and environment. The detection of these interactions remains challenging due to computational limitations. Information theoretic approaches use computationally efficient directed search strategies and thus provide a feasible solution to this problem. However, the power of information theoretic methods for interaction analysis has not been systematically evaluated. In this work, we compare power and Type I error of an information-theoretic approach to existing interaction analysis methods., Methods: The k-way interaction information (KWII) metric for identifying variable combinations involved in gene-gene interactions (GGI) was assessed using several simulated data sets under models of genetic heterogeneity driven by susceptibility increasing loci with varying allele frequency, penetrance values and heritability. The power and proportion of false positives of the KWII was compared to multifactor dimensionality reduction (MDR), restricted partitioning method (RPM) and logistic regression., Results: The power of the KWII was considerably greater than MDR on all six simulation models examined. For a given disease prevalence at high values of heritability, the power of both RPM and KWII was greater than 95%. For models with low heritability and/or genetic heterogeneity, the power of the KWII was consistently greater than RPM; the improvements in power for the KWII over RPM ranged from 4.7% to 14.2% at for α = 0.001 in the three models at the lowest heritability values examined. KWII performed similar to logistic regression., Conclusions: Information theoretic models are flexible and have excellent power to detect GGI under a variety of conditions that characterize complex diseases.

In high-dimensional studies such as genome-wide association studies, the correction for multiple testing in order to control total type I error results in decreased power to detect modest effects. We present a new analytical approach based on the higher criticism statistic that allows identification of the presence of modest effects. We apply our method to the genome-wide study of rheumatoid arthritis provided in the Genetic Analysis Workshop 16 Problem 1 data set. There is evidence for unknown bias in this study that could be explained by the presence of undetected modest effects. We compared the asymptotic and empirical thresholds for the higher criticism statistic. Using the asymptotic threshold we detected the presence of modest effects genome-wide. We also detected modest effects using 90th percentile of the empirical null distribution as a threshold; however, there is no such evidence when the 95th and 99th percentiles were used. While the higher criticism method suggests that there is some evidence for modest effects, interpreting individual single-nucleotide polymorphisms with significant higher criticism statistics is of undermined value. The goal of higher criticism is to alert the researcher that genetic effects remain to be discovered and to promote the use of more targeted and powerful studies to detect the remaining effects.

Evaluation of the association between single-nucleotide polymorphisms (SNPs) and disease outcomes is widely used to identify genetic risk factors for complex diseases. Although this analysis paradigm has made significant progress in many genetic studies, many challenges remain, such as the requirement of a large sample size to achieve adequate power. Here we use rheumatoid arthritis (RA) as an example and explore a new analysis strategy: pathway-based analysis to search for related genes and SNPs contributing to the disease.We first propose the application of measure of explained variation to quantify the predictive ability of a given SNP. We then use gene set enrichment analysis to evaluate enrichment of specific pathways, where pathways, are considered enriched if they consist of genes that are associated with the phenotype of interest above and beyond is expected by chance. The results are also compared with score tests for association analysis by adjusting for population stratification.Our study identified some significantly enriched pathways, such as "cell adhesion molecules," which are known to play a key role in RA. Our results showed that pathway-based analysis may identify other biologically interesting loci (e.g., rs1018361) related to RA: the gene (CTLA4) closest to this marker has previously been shown to be associated with RA and the gene is in the significant pathways we identified, even though the marker has not reached genome-wide significance in univariate single-marker analysis.

Background: Prior to cluster analysis or genetic network analysis it is customary to filter, or remove genes considered to be irrelevant from the set of genes to be analyzed. Often genes whose variation across samples is less than an arbitrary threshold value are deleted. This can improve interpretability and reduce bias., Results: This paper introduces modular models for representing network structure in order to study the relative effects of different filtering methods. We show that cluster analysis and principal components are strongly affected by filtering. Filtering methods intended specifically for cluster and network analysis are introduced and compared by simulating modular networks with known statistical properties. To study more realistic situations, we analyze simulated "real" data based on well-characterized E. coli and S. cerevisiae regulatory networks., Conclusion: The methods introduced apply very generally, to any similarity matrix describing gene expression. One of the proposed methods, SUMCOV, performed well for all models simulated.

Large scale genomic studies with multiple phenotypic or genotypic measures may require the identification of complex multivariate relationships. In multivariate analysis a common way to inspect the relationship between two sets of variables based on their correlation is canonical correlation analysis, which determines linear combinations of all variables of each type with maximal correlation between the two linear combinations. However, in high dimensional data analysis, when the number of variables under consideration exceeds tens of thousands, linear combinations of the entire sets of features may lack biological plausibility and interpretability. In addition, insufficient sample size may lead to computational problems, inaccurate estimates of parameters and non-generalizable results. These problems may be solved by selecting sparse subsets of variables, i.e. obtaining sparse loadings in the linear combinations of variables of each type. In this paper we present Sparse Canonical Correlation Analysis (SCCA) which examines the relationships between two types of variables and provides sparse solutions that include only small subsets of variables of each type by maximizing the correlation between the subsets of variables of different types while performing variable selection. We also present an extension of SCCA--adaptive SCCA. We evaluate their properties using simulated data and illustrate practical use by applying both methods to the study of natural variation in human gene expression.

Many clustering methods require that the number of clusters believed present in a given data set be specified a priori, and a number of methods for estimating the number of clusters have been developed. However, the selection of the number of clusters is well recognized as a difficult and open problem and there is a need for methods which can shed light on specific aspects of the data. This paper adopts a model for clustering based on a specific structure for a similarity matrix. Publicly available gene expression data sets are analyzed to illustrate the method and the performance of our method is assessed by simulation.

Unlabelled: We evaluated the characteristics of 1,142 women and men who attended Canadian osteoporosis clinics and had T-score < or = -2.0 and no prior fractures to determine the predictors of first clinical fracture. Greater age and failure to start osteoporosis drug treatment increased the risk of first clinical fracture., Introduction: Although risk factors for osteoporotic fractures are well-known, it is unclear which factors predict poor prognosis in patients with primary osteoporosis. The purpose of this study was to determine prognostic factors for first clinical fracture in patients with T-score < or = -2.0 and no previous clinical fracture., Methods: We examined prospectively collected data from 1,142 patients aged 40 and over in the Canadian Database of Osteoporosis and Osteopenia. We used prognosis methodology and performed survival analysis to determine factors that increase the risk of first clinical fracture., Results: Our inception cohort (mean age = 60.6 years, 91% females) had a cumulative fracture incidence of 5.1% (incidence rate: 2.53/100 person-years). Age and osteoporosis drug use predicted incident clinical fractures in multivariable regression analyses. The risk of first fracture increased by 3% per year. Failing to initiate osteoporosis treatment increased fracture risk by 2.4 times. In addition, low physical activity, high body mass index and low T-scores were found to predict fracture risk in certain patient subgroups using tree-structured survival analysis. These findings were robust and did not change with most sensitivity analyses., Conclusion: Age and osteoporosis drug treatment are the main prognostic predictors of first clinical fracture in patients with T-score < or = -2.0.

There is a growing interest in studying natural variation in human gene expression. Studies mapping genetic determinants of expression profiles are often carried out considering the expression of one gene at a time, an approach that is computationally intensive and may be prone to high false-discovery rate because the number of genes under consideration often exceeds tens of thousands. We present an exploratory method for investigating such data and apply it to the data provided as Problem 1 of Genetic Analysis Workshop 15 (GAW15). In multivariate analysis, canonical correlation analysis is a common way to inspect the relationship between two sets of variables based on their correlation. It determines linear combinations of all variables from each data set such that the correlation between the two linear combinations is maximized. However, due to the large number of genes, linear combinations involving all single-nucleotide polymorphism (SNP) loci and gene expression phenotypes lack biological plausibility and interpretability. We introduce sparse canonical correlation analysis, which examines the relationships of many genetic loci and gene expression phenotypes by providing sparse linear combinations that include only a small subset of loci and gene expression phenotypes. These correlated sets of variables are sufficiently small for biological interpretability and further investigation. Applying this method to the GAW15 Problem 1 data, we identified groups of 41 loci and 150 gene expressions with the highest between-group correlation of 43%.

This paper summarizes contributions to group 12 of the 15th Genetic Analysis Workshop. The papers in this group focused on multivariate methods and applications for the analysis of molecular data including genotypic data as well as gene expression microarray measurements and clinical phenotypes. A range of multivariate techniques have been employed to extract signals from the multi-feature data sets that were provided by the workshop organizers. The methods included data reduction techniques such as principal component analysis and cluster analysis; latent variable models including structural equations and item response modeling; joint multivariate modeling techniques as well as multivariate visualization tools. This summary paper categorizes and discusses individual contributions with regard to multiple classifications of multivariate methods. Given the wide variety in the data considered, the objectives of the analysis and the methods applied, direct comparison of the results of the various papers is difficult. However, the group was able to make many interesting comparisons and parallels between the various approaches. In summary, there was a consensus among authors in group 12 that the genetic research community should continue to draw experiences from other fields such as statistics, econometrics, chemometrics, computer science and linear systems theory., ((c) 2007 Wiley-Liss, Inc.)

Using the Problem 1 data set made available for Genetic Analysis Workshop 15, we assessed sensitivity of linkage results to a correlation-based feature extraction method as well as to different normalization procedures applied to the raw Affymetrix gene expression microarray data. The impact of these procedures on heritability estimates and on expression quantitative trait loci are investigated. The filtering algorithm we propose in this paper ranks genes based on the total absolute correlation of each gene with all other genes on the array and has the potential to extract features that may play role in functional pathways and gene networks. Our results showed that the normalization and filtering algorithms can have a profound influence on genetic analysis of gene expression data.

Background: With the advent of high-throughput proteomic experiments such as arrays of purified proteins comes the need to analyse sets of proteins as an ensemble, as opposed to the traditional one-protein-at-a-time approach. Although there are several publicly available tools that facilitate the analysis of protein sets, they do not display integrated results in an easily-interpreted image or do not allow the user to specify the proteins to be analysed., Results: We developed a novel computational approach to analyse the annotation of sets of molecules. As proof of principle, we analysed two sets of proteins identified in published protein array screens. The distance between any two proteins was measured as the graph similarity between their Gene Ontology (GO) annotations. These distances were then clustered to highlight subsets of proteins sharing related GO annotation. In the first set of proteins found to bind small molecule inhibitors of rapamycin, we identified three subsets containing four or five proteins each that may help to elucidate how rapamycin affects cell growth whereas the original authors chose only one novel protein from the array results for further study. In a set of phosphoinositide-binding proteins, we identified subsets of proteins associated with different intracellular structures that were not highlighted by the analysis performed in the original publication., Conclusion: By determining the distances between annotations, our methodology reveals trends and enrichment of proteins of particular functions within high-throughput datasets at a higher sensitivity than perusal of end-point annotations. In an era of increasingly complex datasets, such tools will help in the formulation of new, testable hypotheses from high-throughput experimental data.

Background: Mammographic density is a strong risk factor for breast cancer that is heritable and associated with blood levels of growth hormone and insulin-like growth factor-I (IGF-I). We tested single nucleotide polymorphisms (SNP) in pituitary growth hormone (GH1) and growth hormone-releasing hormone receptor (GHRHR) genes for an association with mammographic density, hormones of the growth hormone/IGF-I axis, and anthropometric variables., Methods: Mammograms from 348 women were measured using a computer-assisted method, blood collected, and DNA extracted. The SNPs genotyped were GH1 -57G>T, GH1 -75G >A, and GHRHR A57T. ANOVA and covariance were used to examine associations, adjusted for age, body mass index, ethnicity, and menopausal status, between each SNP and three measures of the mammogram: percent density, total dense area, and total nondense area. Similarly, the SNPs were tested for associations with serum growth hormone, IGF-I, IGFBP3, prolactin, and anthropometric variables., Results: GH1 -57G >T and GH1 -75G >A were both associated with percent density and total nondense area. GH1 -57T homozygotes had 5.2 more mean adjusted percent density than other subjects combined (P = 0.03) and 16.2 cm(2) (14.6%) less nondense area (P = 0.01). GH1 -75A homozygotes had 3.4 more percent density than subjects with at least one G allele (P = 0.04) and also had 32% higher serum growth hormone levels (P = 0.02)., Conclusion: We have found associations between mammographic density and two SNPs in the pituitary growth hormone gene, one of them also associated with serum growth hormone levels. These findings suggest that the GH1 gene may also influence breast cancer risk.

Emerging clinical and experimental evidence strongly implicates proteinuria in the progression of kidney disease. One pathway involves the activation of NFkappaB by albumin, and it has been demonstrated that the activation of NFkappaB induced by albumin is dependent on mitogen-activated protein kinase ERK1/ERK2. To study the effect of albumin on gene expression, primary human renal tubular cells were exposed in vitro to albumin (1%) for 6 h, and gene expression profiling was performed with the human oligonucleotide microarray, U133A Affymetrix Gene Chip. In all, 223 genes were differentially regulated by albumin, including marked upregulation of the EGF receptor (EGFR) and IL-8. Accordingly, the authors sought to delineate the signaling pathway linking albumin to the EGFR and activation of ERK1/ERK2. It was found that albumin led to a dose- and time-dependent activation of ERK1/ERK2. Treatment with albumin led to EGFR phosphorylation, but the activation of ERK1/ERK2 was prevented by pretreatment of the cells with AG-1478, the EGFR kinase inhibitor, at a dose that inhibited EGF-induced ERK1/ERK2 activation. Exogenously administered reactive oxygen species (ROS) were found to activate ERK1/ERK2 via the EGFR and src tyrosine kinase activity and pretreatment of cells with the antioxidant N-acetylcysteine (NAC) and the NADPH oxidase inhibitor DPI abrogated albumin-induced activation of ERK1/ERK2. The src tyrosine kinase inhibitor, PP2, also inhibited the albumin-induced activation of ERK1/ERK2. Finally, pretreatment with AG-1478, the MEK inhibitor UO126, and NAC prevented the albumin-induced increase in IL-8 expression. The authors conclude that the EGF receptor plays a central role in the signaling pathway that links albumin to the activation of ERK1/ERK2 and increased expression of IL-8. Gene profiling studies suggest that there may be a positive feedback loop through the EGFR that amplifies the response of the proximal tubule cell to albumin. Taken together, these results suggest that the EGFR may be an important treatment target for kidney disease associated with proteinuria.

Mammographic density is associated with increased breast cancer risk and is influenced by sex hormones. A T27C polymorphism (alleles A1 and A2, respectively) in the 5' promoter region of CYP17 may be associated with elevated sex hormone levels. In a cross-sectional study of 181 pre- and 173 postmenopausal women, we examined the relationship of this polymorphism with mammographic density and other risk factors for breast cancer. Subjects were recruited across five categories of density. Risk factor and dietary information, anthropometric measures, and blood samples were obtained. Sex hormone, lipid, growth factor levels, and CYP17 genotypes were determined. CYP17 genotype was not associated with mammographic density levels before or after adjusting for risk factors for breast cancer. In premenopausal women, the A2 allele was associated with higher levels of dehydroepiandrosterone sulfate, and in postmenopausal women, with higher levels of total estradiol and lower levels of follicle stimulating hormone. Among premenopausal women, interactions were observed between CYP17 genotype and endogenous insulin levels as well as dietary variables associated with mammographic density. Our findings suggest that the CYP17 A2 allele is associated with hormone levels, and interacts with insulin levels and diet to affect breast density levels and potentially breast cancer risk.

Introduction: Mammographically dense breast tissue is a strong predictor of breast cancer risk, and is influenced by both mitogens and mutagens. One enzyme that is able to affect both the mitogenic and mutagenic characteristics of estrogens is cytochrome P450 1A2 (CYP1A2), which is principally responsible for the metabolism of 17beta-estradiol., Methods: In a cross-sectional study of 146 premenopausal and 149 postmenopausal women, we examined the relationships between CYP1A2 activity, malondialdehyde (MDA) levels, and mammographic density. In vivo CYP1A2 activity was assessed by measuring caffeine metabolites in urine. Levels of serum and urinary MDA, and MDA-deoxyguanosine adducts in DNA were measured. Mammograms were digitized and measured using a computer-assisted method., Results: CYP1A2 activity in postmenopausal women, but not in premenopausal women, was positively associated with mammographic density, suggesting that increased CYP1A2 activity after the menopause is a risk factor for breast cancer. In premenopausal women, but not in postmenopausal women, CYP1A2 activity was positively associated with serum and urinary MDA levels; there was also some evidence that CYP1A2 activity was more positively associated with percentage breast density when MDA levels were high, and more negatively associated with percentage breast density when MDA levels were low., Conclusion: These findings provide further evidence that variation in the activity level of enzymes involved in estrogen metabolism is related to levels of mammographic density and potentially to breast cancer risk.

Introduction: Breast cancer risk may be determined by various genetic, metabolic, and lifestyle factors that alter sex hormone metabolism. Cytochrome P450 1A2 (CYP1A2) is responsible for the metabolism of estrogens and many exogenous compounds, including caffeine., Methods: In a cross-sectional study of 146 premenopausal and 149 postmenopausal women, we examined the relationships between CYP1A2 activity and known or suspected risk factors for breast cancer. Blood levels of sex hormones, lipids, and growth factors were measured. In vivo CYP1A2 activity was assessed by measuring caffeine metabolites in urine. Stepwise and maximum R regression analyses were used to identify covariates related to CYP1A2 activity after adjustment for ethnicity., Results: In both menopausal groups CYP1A2 activity was positively related to smoking and levels of sex hormone binding globulin. In premenopausal women, CYP1A2 activity was also positively related to insulin levels, caffeine intake, age, and plasma triglyceride levels, and negatively related with total cholesterol levels and body mass index. In postmenopausal women CYP1A2 activity was positively associated with insulin-like growth factor-1, and negatively associated with plasma triglyceride, high-density lipoprotein cholesterol, and age at menarche., Conclusion: These results suggest that CYP1A2 activity is correlated with hormones, blood lipids, and lifestyle factors associated with breast cancer risk, although some of the observed associations were contrary to hypothesized directions and suggest that increased CYP1A2 function may be associated with increased risk for breast cancer.

Extensive mammographic density is heritable, strongly associated with increased breast cancer risk, and is influenced by sex hormone exposure. In a cross-sectional study of 181 pre- and 171 postmenopausal women without breast cancer, we examined the relationship of a functional polymorphism in catechol-O-methyltransferase (COMT; VAL-->MET) to mammographic density and other risk factors for breast cancer. We hypothesized that individuals who inherited the low-activity form of COMT (COMT*2 allele) would have higher levels of breast density, presumably because of reduced inactivation/detoxification of catecholestrogens. Subjects were recruited across five categories of breast density. Risk factor information, anthropometric measures, and blood samples were obtained; sex hormone and growth factor levels were measured, and COMT genotypes determined. Mammograms were digitized and measured using a computer-assisted method. After adjustment for age and ethnicity, among pre- but not postmenopausal subjects, each low-activity COMT*2 allele was associated with lower levels of percentage breast density. The statistical significance of this association was lost after further adjustment for serum growth factors [growth hormone, insulin-like growth factor-1 (IGF-1), and insulin-like growth factor binding protein-3 (IGFBP-3)], hormones [follicle-stimulating hormone (FSH) and progesterone], and body size (body mass index and waist:hip ratio). The low-activity COMT*2 allele was also associated, after adjustment for age and ethnicity in premenopausal women, with lower serum levels of IGF-1, higher levels of FSH and progesterone, and with a larger waist:hip ratio, body mass index, and subscapular skinfold. After adjustment for body size, the associations of genotype with IGFBP-3 and FSH were no longer significant. These findings indicate that COMT genotype is associated with several risk factors for breast cancer and suggest that the low-activity COMT*2 allele is associated with a reduced risk of breast cancer among premenopausal women.

This article presents a score test for genetic linkage in nuclear families which applies to any trait having a distribution belonging to the exponential family, which includes binary and normal distributions, and distributions which are skewed or have nonnormal kurtosis. The specific distribution need not be specified and the method applies to sibships of arbitrary size. Tests of complex genetic effects are given, including unspecified mode of inheritance or additive, dominant, overdominant, and recessive modes of inheritance, covariates, multiple-locus models, including gene-gene interactions, and gene-environment interactions. The relation of our method to the Haseman-Elston methods is studied theoretically and by simulation.

Background: Women with extensive dense breast tissue visible on a mammogram have a risk of breast cancer that is 1.8 to 6.0 times that of women of the same age with little or no density. Menopausal status, weight, and parity account for 20 to 30 percent of the age-adjusted variation in the percentage of dense tissue., Methods: We undertook two studies of twins to determine the proportion of the residual variation in the percentage of density measured by mammography that can be explained by unmeasured additive genetic factors (heritability). A total of 353 pairs of monozygotic twins and 246 pairs of dizygotic twins were recruited from the Australian Twin Registry, and 218 pairs of monozygotic twins and 134 pairs of dizygotic twins were recruited in Canada and the United States. Information on putative determinants of breast density was obtained by questionnaire. Mammograms were digitized, randomly ordered, and read by a blinded investigator., Results: After adjustment for age and measured covariates, the correlation coefficient for the percentage of dense tissue was 0.61 for monozygotic pairs in Australia, 0.67 for monozygotic pairs in North America, 0.25 for dizygotic pairs in Australia, and 0.27 for dizygotic pairs in North America. According to the classic twin model, heritability (the proportion of variants attributable to additive genetic factors) accounted for 60 percent of the variation in density (95 percent confidence interval, 54 to 66) in Australian twins, 67 percent (95 percent confidence interval, 59 to 75) in North American twins, and 63 percent (95 percent confidence interval, 59 to 67) in all twins studied., Conclusions: These results show that the population variation in the percentage of dense tissue on mammography at a given age has high heritability. Because mammographic density is associated with an increased risk of breast cancer, finding the genes responsible for this phenotype could be important for understanding the causes of the disease., (Copyright 2002 Massachusetts Medical Society)

This paper presents a unified framework for transmission-disequilibrium tests for discrete and continuous traits. A conditional score test is derived that maximizes power to detect small effects for any exponential family distribution, which includes binary and normal distributions, and distributions that are skewed or have non-normal kurtosis. The specific distributional form need not be specified, and the method applies to sibships of arbitrary size. Formulas for the distribution of the test statistic are given for models including complex genetic effects (additive, dominant, and recessive gene action), covariates, multiple gene models including gene-gene interactions or heterogeneity, and gene-environment interactions. We develop refinements of our method for trait-based sampling designs and multiple siblings that can have dramatic effects on power., (Copyright 2002 Wiley-Liss, Inc.)

We randomly chose replicates 28 and 29 of the simulated data sets of Genetic Analysis Workshop 12 to model the dependence of affection status on covariates, quantitative traits, and genes using all living pedigree members. First we explored the relationship of affection status to demographic and environmental factors using logistic regression and the Cox proportional hazards models. In the second stage of our analyses the generalized transmission disequilibrium test (GTDT) was applied to nuclear families with at least two affected siblings to select single markers and high-risk alleles, which were tested in the population association analyses including all pedigree members. Multiple logistic regression models were fitted to investigate the joint contributions of genetic and nongenetic factors and a block-recursive modeling approach was adopted to study inherent hierarchical dependence structure in the data. We found that allele 2 on marker 35 of chromosome 6 is associated with higher risk compared with the other 3 alleles of this marker. In addition to this significant genetic effect, age at exam and four of the five quantitative traits (QT1, QT2, QT4, and QT5) had a significant association with the disease. Our results were obtained without knowledge of the true disease generating models.

When summarizing studies in a meta-analysis, the methodological quality of the studies can vary widely. This variation is often described by assigning quality scores to studies. This paper presents a probability model for the effect of quality on a summary effect measure. The model motivates criteria used to assess quality, and can be used to incorporate quality into the calculation of a summary effect. We derive a number of summarization methods and compare them for simulated data and an actual set of studies., (Copyright 1999 John Wiley & Sons, Ltd.)

This paper constructs graphical models for two important analysis problems to demonstrate how graphs can clearly represent complex relationships. A powerful property of graphical models called d-separation describes the statistical associations in a given graphical model. This allows the effects of unmeasured variables to be predicted, and suggests conditional analyses which can distinguish conjectured models. If an assumption called quasi-linearity is made, further conclusions can be drawn based on the structure of the graph and d-separation. Making the quasi-linearity assumption explicit also contributes to our understanding of which aspects of our causal intuition are based on linearity assumptions. The examples that we consider are the scientific interpretation of interventions and the evaluation of the validity of candidate surrogate endpoints for clinical trials. Intervention studies can be plagued by non-compliance with assigned treatment and ambiguity in interpreting results when treatment assignment manipulates multiple factors. We show graphically the conclusions that can be drawn under various assumptions. The topic of surrogate endpoints is addressed in causal terms and modelled graphically., (Copyright 1999 John Wiley & Sons, Ltd.)

Using randomly selected sib pairs from a subset of the GAW11 simulated data in Problem 2, we compared the results of some linkage analysis methods based on allele sharing. One method was the Haseman-Elston test for a binary disease outcome (unaffected vs. mild or severe). The other methods, which analyzed the trinary ordered outcome unaffected/mild/severe were the Haseman-Elston test, an extended Haseman-Elston incorporating sib-pair sums, variance components analysis, and regression analysis. Our analysis was done without knowledge of the generating model.

Background: The appearance of breast tissue on mammography varies according to its composition. Fat is radiolucent and appears dark on mammography, while stromal and epithelial tissue has greater optical density and appears light. Extensive areas of radiologically dense breast tissue seen on mammography are associated with an increased risk of breast cancer., Purpose: The purpose of the present study was to determine whether the adoption of a low-fat, high-carbohydrate diet for 2 years would reduce breast density., Methods: Women with radiologic densities in more than 50% of the breast area on mammography were recruited and randomly allocated to an intervention group taught to reduce intake of dietary fat (mean, 21% of calories) and increase complex carbohydrate (mean, 61% of calories) or to a control group (mean, 32% of calories from fat and 50% of calories from carbohydrates). Mammographic images from 817 subjects were taken at baseline and compared with those taken 2 years after random allocation by use of a quantitative image analysis system, without knowledge of the dietary group of the subjects or of the sequence in which pairs of images had been taken. The effects of the intervention on the mammographic features of breast area, area of dense tissues in the breast, and the percent of the breast occupied by dense tissue were examined using t tests. Multiple regression was used to examine these effects while accounting for age at trial entry, weight change, and menopausal status., Results: After 2 years, the total area of the breast was reduced by an average of 233.7 mm2 (2.4%) (95% confidence interval [CI] = 106.9-360.6) in the intervention group compared with an average increase of 26.3 mm2 (0.3%) (95% CI = -108.0-160.5) in the control group (P = .01). The area of density was reduced by 374.4 mm2 (6.1%) (95% CI = 235.1-513.8) in the intervention group compared with an average of 127.7 mm2 (2.1%) (95% CI = 8.6-246.7) in the control group (P = .01). Weight loss was associated with a reduction in breast area. The effect of the intervention on breast area was only marginally statistically significant after weight change, menopausal status, and age at trial entry were taken into account (P = .06). Greater weight loss and becoming postmenopausal were associated with statistically significant reductions in the area of density on the mammographic image at 2 years (P = .04 and P<.001, respectively). Age at entry into the trial was marginally significant in the same direction (P = .06). The effect of the intervention on area of density remained statistically significant after controlling for weight loss, age at entry, and menopausal status (P = .03). The change in the percentage of dense tissue in the mammographic image was not significantly different between the two groups (P = .71)., Conclusions and Implications: These results show that after 2 years, a low-fat, high-carbohydrate diet reduced the area of mammographic density, a radiographic feature of the breast that is a risk factor for breast cancer. Longer observation of a larger number of subjects will be required to determine whether these effects are associated with changes in risk of breast cancer.

In a randomly chosen replicate of extended pedigrees from GAW10, we conducted robust multipoint genome scans for linkage using a dense marker map. For analysis of the quantitative traits, we selected sibships from the pedigrees, and for analysis of disease status, small families of affected relatives were selected. Lod-score likelihood analyses were conducted in the full pedigrees and in the affected relative families for selected regions. We located a flanking marker for MG1 on chromosome 5, and identified marker regions including MG2, MG4, and MG5 on chromosomes 8 and 9. The analytic methods were consistent for the major gene with a strong effect; false positive errors on chromosomes 1 and 10 could have been eliminated by requiring evidence from more than one method.

This paper considers randomized interventions which do not completely determine an intended determinant of response, and which may also manipulate additional, possibly unobserved, variables influencing response. The example we use throughout this paper is counseling for a low-fat diet for breast cancer prevention, where the intervention is counseling and dietary fat is hypothesized to reduce breast cancer risk. We use additive linear models to derive conditions and assumptions for considering fat to be the sole explanation of an observed treatment effect. A modified experimental design which supports stronger conclusions about causality is proposed.

There is strong evidence that the risk of breast cancer in populations is influenced by environmental factors. Plasma lipids and lipoproteins are known to be under environmental control and to have epidemiological and/or biological characteristics that suggest they may be relevant to breast cancer risk. The purpose of the study described here was to determine whether plasma lipids, lipoproteins, and the urinary excretion of the mutagen malondialdehyde (MDA) are associated with differences in breast cancer risk. We measured plasma lipids, lipoproteins, and urinary MDA in women without breast cancer but with different degrees of density of the breast parenchyma on mammography, a strong risk factor for breast cancer. Mammograms from 273 premenopausal women were digitized to high spatial resolution by a scanning densitometer, and images were analyzed to quantify the extent of density. The percentage of the breast occupied by mammographic densities was found, after controlling for the effects of age and the Quetelet index of obesity, to be significantly associated with plasma levels of high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglyceride, apoprotein B, and urinary excretion of MDA. A multivariate model comprised of the Quetelet index of obesity, alcohol consumption, apoprotein B, parity, daily MDA excretion, and the skinfold thickness sum accounted for 36% of the variation in breast density. These results suggest that differences in lipid metabolism are associated with differences in breast cancer risk as defined by mammographic densities. These findings are consistent with several other observations that show a relationship between plasma lipids, lipoproteins, and risk factors for breast cancer.

The most convincing evidence that a factor such as dietary fat is causally related to breast cancer would be obtained from a randomised controlled trial in which exposure to dietary fat intake was systematically varied. A limitation of randomised controlled trials of breast cancer prevention, however, is the large sample size required to detect plausible reductions in risk resulting from the intervention. We describe here experience over a period of 9 years with the use of one risk factor for breast cancer as a criterion for entry to a clinical trial of breast cancer prevention. The risk factor used was the presence of extensive densities in the breast tissue on mammography, which has been found by several investigators to be strongly associated with risk of breast cancer. Using this criterion for selection, 1800 subjects of mean age 46 years were enrolled between 1982 and 1986, and again between 1988 and the present. Throughout this period, the point estimate of annual invasive cancer incidence was approximately 6 per 1000 per year. The observed cancer incidence has been consistently 4-5 times the incidence expected from age-specific breast cancer incidence data for women living in Ontario. These data show that the selection of subjects for a clinical trial of breast cancer prevention using the criterion of extensive breast parenchymal densities does identify a group at substantially increased risk of breast cancer. Use of this criterion for the selection of subjects can substantially reduce the sample size required for a clinical trial of a preventive strategy.

We have examined the relationship between plasma lipids, lipoproteins, and a family history of breast cancer. We measured the plasma lipids and lipoproteins in unaffected female members of the nuclear family of women with familial breast cancer and compared them with those of the female members of the nuclear family of women with sporadic breast cancer. A mean number of 3.3 relatives of mean age 35 years were studied in 23 pairs of familial and sporadic breast cancer families. After adjustment in multivariate analysis for variables that either differed between high and low risk families, or were significantly associated with plasma levels or lipoproteins, statistically significant differences were found in plasma levels of total cholesterol, low density lipoprotein cholesterol, and apoprotein B, all of which were lower in familial breast cancer than in sporadic breast cancer families. These data suggest that inherited factors associated with breast cancer risk may play a role in determining plasma lipid and lipoprotein levels and that lipid regulatory genes should be considered in this context.

An interpretation of the standardized difference is given which assists in judging the importance of such a difference. This is useful for designing two sample studies when information on mean difference or standard deviation is unavailable.

We assessed the validity of information reported by patients with breast cancer on cancer in first- and second-degree relatives. In Toronto, Canada, 165 patients completed mailed questionnaires about cancer in relatives and were then interviewed in person. Their reports were compared with relatives' hospital records, cancer registry or death records for presence of cancer, site and age at diagnosis. Questionnaire and interview reports agreed with records for 82-96% of reports on first-degree and 48-80% on second-degree relatives. Proband reports of cancer sites in first-degree relatives were generally accurate (breast 99%, ovary 100%, prostate 85%, colon 93%). Reports of cancer sites in second-degree relatives were accurate for prostate cancer but only for 85% of breast and 72% of colon cancers. Age at diagnosis of breast cancer was correct in 92% of cases in first-degree and 54% in second-degree relatives. The interview contributed additional information about the presence of cancer in second-degree relatives, and the site and age at diagnosis in first- and second-degree relatives. In a similar population the questionnaire alone should yield adequate data for identifying families that warrant further investigation.

Animal experimental evidence and human ecological data suggest that dietary fat intake is related to breast cancer risk. Epidemiological studies within countries have given inconsistent results but are limited by the restricted range of dietary intake found in Western populations and by error in the measurement of fat consumption. Experimental evidence, derived from controlled clinical trials in which the range of fat intake is increased beyond that seen in most Western populations, is capable of overcoming this limitation of observational epidemiology, and would provide the strongest evidence available concerning the relationship of dietary fat intake to breast cancer risk. Further, such trials are the only means likely to answer the question of whether breast cancer risk in high-risk subjects can be modified by changing dietary fat intake. We describe here several aspects of the feasibility of an experimental approach to this problem, including the identification of subjects at increased risk for breast cancer, and the demonstration that such subjects will enter a clinical trial of dietary fat reduction and comply with a low-fat diet. It is shown that subjects can be recruited and retained in such trials, that satisfactory dietary compliance can be achieved over at least 24 mon and that the subjects selected are at demonstrably increased risk of breast cancer. This finding indicates that it is feasible to test the dietary fat-breast cancer hypothesis experimentally by means of a clinical trial.

The appearance of the breast parenchyma as seen on mammography is known to vary between individuals. The evidence is reviewed that some of the sources of this variation, namely, the densities referred to as "dysplasia", are related to breast cancer risk. We have carried out a meta-analysis of the published literature to determine the magnitude of the risk of breast cancer associated with mammographic densities by calculating summary odds ratios for studies grouped according to their design and the method used to classify mammographic parenchymal patterns. This has shown that subjects with mammographic densities have an increased risk of breast cancer relative to those without densities. Of the studies that used Wolfe's method of classification, cohort studies (n = 8) had a summary odds ratio of 5.19 (95% CI, 3.6 to 7.48); case control studies (n = 13) had a summary odds ratio of 1.8 (95% CI, 1.5 to 2.13); and prevalence surveys had a summary odds ratio of 0.54 (95% CI, 0.4 to 0.7). Other methods of classification, using quantitative estimates of the proportion of the breast occupied by densities, gave substantially higher odds ratios than Wolfe's system. The ability to recognize individuals within the population at different risks for breast cancer could be exploited in studies of potential etiological factors.

One hundred and sixty-six patients between the ages of 12 and 48 years with acute myelogenous leukemia (AML), acute lymphoblastic leukemia (ALL) or chronic myelogenous leukemia (CML) underwent allogeneic bone marrow transplantation following single fraction total body irradiation (TBI) of 500 cGy from a cobalt source. Patients also received one of three chemotherapeutic regimens according to their diagnosis or disease status at time of transplant. The median follow-up was 67 months with a range of 33-120 months. The actuarial 5-year event-free survival (EFS) for the subgroup of patients with good risk disease (first complete remission AML and ALL or first chronic phase CML) was 43% with an actuarial relapse rate at 5 years of 26%. Patients with poor risk disease (other than first remission AML and ALL or other than first chronic phase CML) had an EFS at 5 years of 15% with a relapse rate of 62%. Disease status at the time of transplantation was the most important factor predicting outcome in this patient population. We conclude that preparation of good risk patients with chemotherapy and single fraction TBI of 500 cGy at a dose rate of 42-91 cGy/min resulted in EFS and relapse rates similar to those observed by centers using fractionated radiotherapy schedules, without a concomitant increase in toxicity, in particular interstitial pneumonitis and cataracts.