Emily L. Harris

Replicating genotype-phenotype associations.

What constitutes replication of a genotype–phenotype association, and how best can it be achieved?

Management of the clinically inapparent adrenal mass ("incidentaloma").

The adrenals are triangular glands that sit atop each kidney. They influence or regulate the bodys metabolism, salt and water balance, and response to stress by secreting various hormones. Based on autopsy studies, adrenal masses are among the most common tumors in humans. At autopsy, an adrenal mass is found in at least 3% of persons older than age 50 years. Most adrenal masses cause no health problems. A small proportion, however, can lead to many serious hormonal diseases; approximately 1 of every 4000 adrenal tumors is malignant. Clinically inapparent adrenal masses are discovered inadvertently during diagnostic testing or treatment for other clinical conditions that are not related to suspicion of adrenal disease; thus, they are commonly known as incidentalomas. The definition of incidentaloma excludes patients undergoing imaging procedures as part of staging and work-up for cancer. Improvements in abdominal imaging techniques and technologies have increased detection of adrenal incidentalomas. Increasing clinical and scientific interest is reflected in a 20-fold increase in publications about this condition over the past three decades. When detected, clinically inapparent adrenal masses raise challenging questions for physicians and their patients. Diagnostic evaluation determines whether the lesion is hormonally active or nonfunctioning and whether it is malignant or benign. The test results will influence whether the mass is removed surgically or treated nonsurgically. Because the prevalence of these masses increases with age, appropriate management of adrenal tumors will be a growing challenge in our aging society. Over the past three decades, new information has become available regarding the epidemiology, biology, screening, treatment, and follow-up of adrenal tumors. For example, recent refinements in the field of minimally invasive surgery have made laparoscopic adrenalectomy a more frequently used method for removing adrenal masses. Recent reports suggest that up to 20% of patients with adrenal incidentaloma have some form of subclinical hormonal dysfunction and may represent a population at higher risk for metabolic disorders and cardiovascular disease. It is important to determine whether groups of patients with subclinical disease benefit from treatment. The psychological effect of the patients knowing that he or she harbors an adrenal incidentaloma, an incompletely understood clinical problem, merits investigation. A 2.5-day National Institutes of Health (NIH) state-of-the-science conference, Management of the Clinically Inapparent Adrenal Mass (Incidentaloma), was convened on 46 February 2002 to explore and assess the current knowledge regarding adrenal incidentalomas, so that health care providers and the general public can make informed decisions about this important public health issue. After 1.5 days of expert presentations and questions and public discussion by members of the panel and the audience of interested attendees on incidental adrenal masses, an independent, nonfederal panel weighed the evidence and drafted a statement that was presented on the third day of the conference. Expert presentations and the panels statement addressed the following questions: 1) What are the causes, prevalence, and natural history of clinically inapparent adrenal masses? 2) Based on available scientific evidence, what is the appropriate evaluation of a clinically inapparent adrenal mass? 3) What criteria should guide the decision on surgical versus nonsurgical management of these masses? 4) If surgery is indicated, what is the appropriate procedure? 5) What is the appropriate follow-up for patients for each management approach? and 6) What additional research is needed to guide practice? The panels draft statement was posted on the NIH Consensus Program Web site (consensus.nih.gov) on 6 February 2002. The primary sponsors of this meeting were the National Institute of Child Health and Human Development and the NIH Office of Medical Applications of Research. Cosponsors included the National Cancer Institute and the National Institute of Diabetes and Digestive and Kidney Diseases. 1. What are the Causes, Prevalence, and Natural History of Clinically Inapparent Adrenal Masses? Clinically inapparent adrenal masses are detected incidentally with imaging studies conducted for other reasons. They may be clinically important because some are caused by adrenal cortical carcinomas (estimated prevalence, 4 to 12 per million), which have a high mortality rate. The other clinical concern is hormone overproduction from pheochromocytomas, aldosteronomas, and subclinical hypercortisolism, which may be associated with morbidity if untreated. Prevalence of Clinically Inapparent Adrenal Masses In autopsy series, the prevalence of clinically inapparent adrenal masses is about 2.1%. Because of increased use of noninvasive high-resolution imaging technology, clinically inapparent adrenal masses are being recognized more often. Estimates range from 0.1% for general health screening with ultrasonography to 0.42% among patients evaluated for nonendocrinologic symptoms to 4.3% among patients who have a previous diagnosis of cancer. In addition to source of data (autopsy versus clinical series) and reasons for imaging (cancer work-up, nonendocrinologic symptoms, and general health screening), the prevalence of clinically inapparent adrenal masses varies with age. The prevalence of clinically inapparent adrenal masses detected at autopsy is less than 1% for patients younger than 30 years of age and increases to 7% in patients 70 years of age or older. Many of these lesions detected at autopsy are very small. More patients with clinically inapparent adrenal masses are women. This probably reflects the sex distribution of the population undergoing imaging procedures. Autopsy studies or general health examinations show no evidence of difference in prevalence between men and women. There is insufficient information to determine whether the prevalence of clinically inapparent adrenal masses differs by the initial diagnostic test. Causes of Clinically Inapparent Adrenal Masses Clinically inapparent adrenal masses can be benign or malignant. These include adenomas, pheochromocytomas, myelolipomas, ganglioneuromas, adrenal cysts, hematomas, adrenal cortical carcinomas, metastases from other cancers, and other rare entities. The distributions of the pathologic origins of clinically inapparent adrenal masses vary with several clinically important factors, including cancer history and mass size. Three fourths of clinically inapparent adrenal masses among patients with cancer are metastatic lesions. In contrast, two thirds of clinically inapparent adrenal masses in populations with no history of cancer are benign tumors. Prevalence of primary adrenal cortical carcinoma is clearly related to the size of the tumor. Adrenal cortical carcinoma accounts for 2% of tumors that are 4 cm or less, 6% of tumors that are 4.1 to 6 cm, and 25% of tumors that are greater than 6 cm. Among unselected patients and those with nonendocrinologic symptoms, clinically inapparent adrenal masses are most often nonfunctioning tumors (approximately 70%). Approximately 5% to 10% of patients being evaluated for nonendocrinologic symptoms have subclinical hypercortisolism (sometimes called subclinical Cushing syndrome). The percentage of patients with subclinical hypercortisolism depends on the testing methods and cortisol levels achieved after dexamethasone suppression. The distribution of clinically inapparent adrenal masses derived from surgical series will overestimate the prevalence of adrenal cortical carcinoma, since suspicion of adrenal cortical carcinoma is an indication for surgery. Moreover, the reported frequency of adrenal cortical carcinomas is derived from highly selected patient populations and does not reflect the prevalence rates seen in population-based studies. The age and sex of the patient do not seem to help predict the presence of adrenal cortical carcinoma. Distribution estimates from autopsy studies are not biased by surgical indications but may not reflect the risk for adrenal cortical carcinoma among the subset of people undergoing abdominal imaging studies. A precise estimate of the risk for adrenal cortical carcinoma that could guide clinical decision making may not be possible. Almost all the reported large studies used imaging equipment that is now considered obsolete. The use of contemporary equipment may increase the prevalence of detected clinically inapparent adrenal masses and enhance our ability to differentiate adrenal cortical carcinomas from adenomas. In addition, the literature comprises mainly small, retrospective studies with variable definitions of clinically inapparent adrenal masses, which cause variation in the relative proportions of adrenal pathologic classifications. Natural History of Clinically Inapparent Adrenal Masses The observed natural history of clinically inapparent adrenal masses varies, depending on the composition of the study sample and the size and pathologic classification of the adrenal mass. Patients with or without a previous cancer diagnosis found to have adrenal gland metastatic lesions will have a clinical course defined by the stage, grade, and site of the primary tumor. Usually, large clinically inapparent adrenal masses (>6 cm) are treated surgically. Approximately 25% of masses greater than 6 cm in diameter are adrenal cortical carcinomas, and these patients have poor clinical outcomes. Most studies report less than 50% 5-year overall survival for adrenal cortical carcinoma, and several report less than 50% 2-year overall survival. Inconclusive evidence suggests that adrenalectomy at stage 1 or 2 may improve the survival rate. Follow-up of patients with nonfunctioning adrenal masses suggests that 5% to 25% of masses increase in size by at least 1 cm. The threshold for clinically significant

Genetic Risk Assessment and BRCA Mutation Testing for Breast and Ovarian Cancer Susceptibility: Systematic Evidence Review for the U.S. Preventive Services Task Force

Clinically significant, or deleterious, mutations of BRCA1 and BRCA2 genes are associated with increased susceptibility for breast and ovarian cancer (1, 2). These mutations increase a womans lifetime risk for breast cancer to 60% to 85% (3, 4) and risk for ovarian cancer to 26% (BRCA1) and 10% (BRCA2) (5-8). Specific BRCA mutations are clustered among certain ethnic groups, such as Ashkenazi Jews (9-11), and in the Netherlands (12), Iceland (13, 14), and Sweden (15). Additional germline mutations associated with familial breast or ovarian cancer have been identified, and others are suspected (16, 17). BRCA1 and BRCA2 mutations are also associated with increased risk for prostate cancer, and BRCA2 mutations are associated with increased risk for pancreatic and stomach cancer and melanoma (18). Screening for inherited breast and ovarian cancer susceptibility is a 2-step process: assessment of risk for clinically significant BRCA mutations followed by genetic testing of high-risk individuals. Guidelines recommend testing for mutations only when an individual has personal or family history features suggestive of inherited cancer susceptibility, when the test result can be adequately interpreted, and when results will aid in management (19, 20). Several characteristics are associated with an increased likelihood of clinically significant BRCA mutations, including young age at breast cancer diagnosis, bilateral breast cancer, history of both breast and ovarian cancer, multiple cases of breast cancer in a family, both breast and ovarian cancer in a family, and Ashkenazi Jewish heritage (21-24). Risk status requires reevaluation when personal or family cancer history changes. Genetic counseling is recommended before mutation testing (25). Several approaches are in practice, including educational; decision-making; and psychosocial support (26, 27) provided by genetic counselors (28-30), nurse educators (31-33), or other professionals. The type of mutation analysis required depends on family history. Individuals from families or ethnic groups with known mutations can be tested specifically for them. Several clinical laboratories in the United States test for specific mutations or sequence-specific exons. Individuals without linkages to others with known mutations undergo direct DNA sequencing. In these cases, guidelines recommend that testing begin with a relative who has known breast or ovarian cancer to determine whether a clinically significant mutation is segregating in the family (19). Myriad Genetic Laboratories provides direct DNA sequencing in the United States and reports analytic sensitivity and specificity exceeding 99% (34). Approximately 12% of high-risk families without a BRCA1 or BRCA2 coding-region mutation may have other clinically significant genomic rearrangements (34, 35). Test results include not only positive (denoting a deleterious mutation) and negative (no mutation found) interpretations but also variants of uncertain clinical significance; this last group represents up to 13% of results (21). The results of genetic testing could lead to prevention interventions for reducing risk or mortality in mutation carriers. Experts recommend earlier and more frequent cancer screening, chemoprevention, and prophylactic surgery (Table 1) (36-40). Table 1. Detection and Prevention Recommendations Although clinically significant BRCA mutations are estimated to occur in 1 in 300 to 500 persons in the general population (41-44), public interest in testing is growing, and physicians are increasingly faced with this issue while providing primary health care. Women often overestimate their risks for breast cancer or BRCA mutations (32, 45, 46), and most women responding to surveys, including women at average and moderate risk, report a strong desire for genetic testing (27, 47), even though only those at high risk would potentially benefit. Concerns about cancer, publicized scientific advances, incomplete understanding of testing and interventions, and direct-to-consumer advertising probably influence these perceptions, increasing demand for genetic testing services (47). The objective of this systematic evidence review is to determine the benefits and harms of screening for inherited breast and ovarian cancer susceptibility in the general population of women presenting for primary health care in the United States. This review was prepared for the U.S. Preventive Services Task Force (USPSTF) and examines a chain of evidence about genetic risk assessment in primary care settings; impact of genetic counseling; ability to predict cancer risk in women with average, moderate, and high risks for clinically significant mutations; benefits of prevention interventions; and potential adverse effects. A review of studies about Ashkenazi Jewish women specifically is reported elsewhere (48). Methods The analytic framework in Figure 1 outlines the patient population, interventions, and health outcomes. This report focuses on the following key questions: Figure 1. Analytic framework. KQ BRCA BRCA BRCA BRCA1 BRCA2 1. Do risk assessment and BRCA mutation testing lead to a reduction in the incidence of breast and ovarian cancer and cause-specific or all-cause mortality? 2A. How well does risk assessment for cancer susceptibility by a clinician in a primary care setting select candidates for BRCA mutation testing? 2B. What are the benefits of genetic counseling before testing? 2C. Among women with family histories predicting an average, moderate, or high risk for a deleterious mutation, how well does BRCA mutation testing predict risk for breast and ovarian cancer? 3. What are the adverse effects of risk assessment, genetic counseling, and testing? 4. How well do interventions reduce the incidence and mortality of breast and ovarian cancer in women identified as high risk by history, positive genetic test results, or both? 5. What are the adverse effects of interventions? We identified relevant papers from multiple searches of MEDLINE (1966 to 1 October 2004) and the Cochrane Library databases; we obtained additional papers by reviewing reference lists of pertinent studies, reviews, editorials, and Web sites and by consulting experts (Appendix Figure). Investigators reviewed all abstracts and determined eligibility by applying inclusion and exclusion criteria specific to key questions (Appendix Table). We then reviewed full-text papers of included abstracts for relevance. Studies about patients with current or past breast or ovarian cancer were excluded unless they addressed genetic testing issues in women without cancer. Data were extracted from each included study, entered into evidence tables, and summarized by using descriptive or statistical methods or both. Two reviewers independently rated the quality of studies using criteria specific to different study designs developed by the USPSTF (Appendix 1) (49). When reviewers disagreed, a final rating was determined by reevaluations by the 2 initial reviewers and a third reviewer if needed. Only studies rated good or fair in quality were included, although studies with designs that do not have quality rating criteria, such as descriptive studies, were also included if relevant to the key questions. To estimate risks for breast and ovarian cancer due to clinically significant BRCA mutations, the screening population was stratified into groups at average, moderate, and high risk for being a mutation carrier based on history of breast or ovarian cancer in first- and second-degree relatives. This approach allows use of published data that describe risks in similar terms. The following definitions were used: average riskno first-degree relatives and no more than 1 second-degree relative on each side of the family with breast or ovarian cancer; moderate risk1 first-degree relative or 2 second-degree relatives on the same side of the family with breast or ovarian cancer; and high riskat least 2 first-degree relatives with breast or ovarian cancer. On the basis of pooled data from more than 100000 women without breast cancer from 52 epidemiologic studies, approximately 92.7% of the screening population would be expected to be average risk, 6.9% moderate risk, and 0.4% high risk according to these definitions (50). Risks for breast and ovarian cancer in mutation carriers have been primarily calculated from families of women with existing breast and ovarian cancer. To determine benefits and adverse effects of genetic testing in average-, moderate-, and high-risk groups, we estimated mutation prevalence as well as the probability of developing cancer given the presence of the mutation (penetrance) for each risk group. Penetrance was calculated from data about the prevalence of BRCA mutations in women with and without breast and ovarian cancer; the probability of breast or ovarian cancer in the U.S. population estimated from Surveillance, Epidemiology, and End Result (SEER) data (51) by using DevCan software (52); and relative risks for breast and ovarian cancer in moderate- and high-risk groups. Penetrance estimates were based on the Bayes theorem and stratified by cancer type (breast or ovarian), risk group (average, moderate, and high), and age whenever data were available. Appendix 2 provides additional details of this method (48). We also performed a meta-analysis of chemoprevention trials to more precisely estimate effectiveness and adverse effects. All chemoprevention trials reported relative risk (RR) estimates, and the logarithm of the RR (logRR) and the corresponding standard errors were calculated for each trial and used in the meta-analysis. The overall estimates of RR were obtained by using a random-effects model (53). We developed an outcomes table to determine the magnitude of potential benefits and adverse effects of testing for BRCA mutations in the general population based on best estimates from published studies and results of analyses when available. Variation associated

Quality control and quality assurance in genotypic data for genome-wide association studies

Genome‐wide scans of nucleotide variation in human subjects are providing an increasing number of replicated associations with complex disease traits. Most of the variants detected have small effects and, collectively, they account for a small fraction of the total genetic variance. Very large sample sizes are required to identify and validate findings. In this situation, even small sources of systematic or random error can cause spurious results or obscure real effects. The need for careful attention to data quality has been appreciated for some time in this field, and a number of strategies for quality control and quality assurance (QC/QA) have been developed. Here we extend these methods and describe a system of QC/QA for genotypic data in genome‐wide association studies (GWAS). This system includes some new approaches that (1) combine analysis of allelic probe intensities and called genotypes to distinguish gender misidentification from sex chromosome aberrations, (2) detect autosomal chromosome aberrations that may affect genotype calling accuracy, (3) infer DNA sample quality from relatedness and allelic intensities, (4) use duplicate concordance to infer SNP quality, (5) detect genotyping artifacts from dependence of Hardy‐Weinberg equilibrium test P‐values on allelic frequency, and (6) demonstrate sensitivity of principal components analysis to SNP selection. The methods are illustrated with examples from the “Gene Environment Association Studies” (GENEVA) program. The results suggest several recommendations for QC/QA in the design and execution of GWAS. Genet. Epidemiol. 34: 591–602, 2010.

New models of collaboration in genome-wide association studies: the Genetic Association Information Network.

The Genetic Association Information Network (GAIN) is a public-private partnership established to investigate the genetic basis of common diseases through a series of collaborative genome-wide association studies. GAIN has used new approaches for project selection, data deposition and distribution, collaborative analysis, publication and protection from premature intellectual property claims. These demonstrate a new commitment to shared scientific knowledge that should facilitate rapid advances in understanding the genetics of complex diseases.

Efficacy of Prophylactic Mastectomy in Women With Unilateral Breast Cancer: A Cancer Research Network Project

PURPOSE We investigated the efficacy of contralateral prophylactic mastectomy (CPM) in reducing contralateral breast cancer incidence and breast cancer mortality among women who have already been diagnosed with breast cancer. METHODS This retrospective cohort study comprised approximately 50,000 women who were diagnosed with unilateral breast cancer during 1979 to 1999. Using computerized data confirmed by chart review, we identified 1,072 women (1.9%) who had CPM. We obtained covariate information for these women and for a sample of 317 women who did not undergo CPM. RESULTS The median time from initial breast cancer diagnosis to the end of follow-up was 5.7 years. Contralateral breast cancer developed in 0.5% of women with CPM, metastatic disease developed in 10.5%, and subsequent breast cancer developed in 12.4%; 8.1% died from breast cancer. Contralateral breast cancer developed in 2.7% of women without CPM, and 11.7% died of breast cancer. After adjustment for initial breast cancer characteristics, treatment, and breast cancer risk factors, the hazard ratio (HR) for the occurrence of contralateral breast cancer after CPM was 0.03 (95% CI, 0.006 to 0.13). After adjustment for breast cancer characteristics and treatment, the HRs for the relationship of CPM with death from breast cancer, with death from other causes, and with all-cause mortality were 0.57 (95% CI, 0.45 to 0.72), 0.78 (95% CI, 0.57 to 1.06), and 0.60 (95% CI, 0.50 to 0.72), respectively. CONCLUSION CPM seems to protect against the development of contralateral breast cancer, and although women who underwent CPM had relatively low all-cause mortality, CPM also was associated with decreased breast cancer mortality.

Cyclin D1 Repression of Peroxisome Proliferator-Activated Receptor γ Expression and Transactivation

ABSTRACT The cyclin D1 gene is overexpressed in human breast cancers and is required for oncogene-induced tumorigenesis. Peroxisome proliferator-activated receptor γ (PPARγ) is a nuclear receptor selectively activated by ligands of the thiazolidinedione class. PPARγ induces hepatic steatosis, and liganded PPARγ promotes adipocyte differentiation. Herein, cyclin D1 inhibited ligand-induced PPARγ function, transactivation, expression, and promoter activity. PPARγ transactivation induced by the ligand BRL49653 was inhibited by cyclin D1 through a pRB- and cdk-independent mechanism, requiring a region predicted to form an helix-loop-helix (HLH) structure. The cyclin D1 HLH region was also required for repression of the PPARγ ligand-binding domain linked to a heterologous DNA binding domain. Adipocyte differentiation by PPARγ-specific ligands (BRL49653, troglitazone) was enhanced in cyclin D1−/− fibroblasts and reversed by retroviral expression of cyclin D1. Homozygous deletion of the cyclin D1 gene, enhanced expression by PPARγ ligands of PPARγ and PPARγ-responsive genes, and cyclin D1−/− mice exhibit hepatic steatosis. Finally, reduction of cyclin D1 abundance in vivo using ponasterone-inducible cyclin D1 antisense transgenic mice, increased expression of PPARγ in vivo. The inhibition of PPARγ function by cyclin D1 is a new mechanism of signal transduction cross talk between PPARγ ligands and mitogenic signals that induce cyclin D1.

Neurofibromatosis type 1: The cognitive phenotype

Visuospatial deficits have been reported in patients with neurofibromatosis type 1 (NF 1), although detailed observations regarding academic achievement are conflicting. Using neurocognitive testing and magnetic resonance imaging, we studied 12 families, each comprising one child with NF 1, an unaffected sibling of the same age range (6 to 16 years), and both biologic parents. The Full Scale IQ ranged from 70 to 130 among children with NF 1 and from 99 to 139 among unaffected siblings. A significant (p < 0.01) pairwise difference was found between each child with NF 1 and sibling on the Full Scale IQ and the Verbal IQ. On a single visuospatial test, Judgement of Line Orientation, children with NF 1 did significantly worse than siblings (p < 0.01). Children with NF 1 had significant learning disabilities in written language and reading (p < 0.05) and in neuromotor dysfunction (p < 0.005) compared with siblings. A significant correlation was found between the pairwise lowering of the Full Scale IQ and Judgment of Line Orientation scores in children with NF 1 and the number of locations in which, on magnetic resonance imaging, T2-weighted hyperintensities were seen (Full Scale IQ: p < 0.0003; Judgment of Line Orientation score: p < 0.02). We conclude that NF 1 is associated with a significantly lower Full Scale IQ, multifocal cognitive deficits (Verbal IQ, Judgment of Line Orientation score), reading disability, and neuromotor deficit. Pairwise cognitive differences correlated with the number of brain lesions on magnetic resonance imaging.

Melphalan May Be a More Potent Leukemogen than Cyclophosphamide

We have evaluated the relation between alkylating agents and leukemic disorders in 3363 1-year survivors of ovarian cancer who were treated in five randomized clinical trials and at two large medical centers. Overall, 28 patients developed acute nonlymphocytic leukemia (expected, 1.2) and 7 developed preleukemia. A 93-fold increased risk for acute nonlymphocytic leukemia was seen in 1794 women treated with chemotherapy; the incidence of leukemic disorders was 7.7/1000 women per year. Risk was highest 5 to 6 years after the first treatment and appeared to decrease thereafter. The use of radiation therapy did not affect risk. The 10-year cumulative risk (mean +/- SE) of acquiring a leukemic disorder was 8.5% +/- 1.6% after treatment with any alkylating agent, 11.2% +/- 2.6% after treatment with melphalan, and 5.4% +/- 3.2% after cyclophosphamide treatment. A dose-response relationship was apparent in 605 women receiving melphalan and suggested in 333 women receiving cyclophosphamide. Women receiving melphalan were two to three times as likely to develop leukemic disorders than were women receiving cyclophosphamide. These data indicate that choice of chemotherapeutic agent and drug dosage may influence significantly the risk for long-term adverse effects of cancer therapy.

Screening for Hereditary Hemochromatosis: A Systematic Review for the U.S. Preventive Services Task Force

The U.S. Preventive Services Task Force (USPSTF) has not previously considered screening for hereditary hemochromatosis for a recommendation as a clinical preventive service for primary care clinicians. We examined key questions to assess hemochromatosis penetrance in C282Y homozygotes (key question 1), address health outcomes of therapeutic phlebotomy (key question 2), and examine the possibility of targeted genetic screening (key question 3). Key questions for this focused systematic review were limited to addressing critical evidence gaps in order for the USPSTF to recommend screening (1, 2), and were applied using strict and consistent definitions of disease, which are described in more detail below. Background Condition Definition Hemochromatosis was originally thought to be a rare idiopathic disorder characterized by end-stage disease (cirrhosis, diabetes, and bronzed skin) but is now recognized as having a hereditary component due to an autosomal recessive inherited disorder of iron metabolism (3). In hemochromatosis, body iron accumulates and can lead to iron overload (4). In iron overload, excess iron is deposited in the liver, pancreas, heart, joints, and endocrine glands, resulting in tissue damage that can lead to disease conditions (such as cirrhosis, diabetes, heart failure, arthropathy, and impotence) (46). Iron overload can be primary (as in hereditary hemochromatosis) or secondary (for example, due to anemias with inefficient erythropoiesis or repeated blood transfusions) (7). In 1996, 2 base-pair alterations, termed C282Y and H63D, of the HFE gene on the region of HLA-A on chromosome 6 were identified in hereditary hemochromatosis (8). C282Y homozygosity is now recognized as the most common genotype in hereditary hemochromatosis (9). Estimates are that 82% to 90% of cases of hereditary hemochromatosis among white persons occur in C282Y/C282Y homozygotes (10). The other 10% to 18% of cases appear to be due to environmental factors or other genotypes. While other HFE-related and nonHFE-related genetic mutations are associated with hereditary hemochromatosis in a small number of cases (4), other genotypes do not appear to be as strongly associated with hereditary hemochromatosis (3, 9). HFE mutations are fairly common in the United States, with 1 in 10 white persons heterozygous for the HFE C282Y mutation (carriers) and 4.4 homozygotes per 1000 (4, 6). The frequency of C282Y homozygosity is much lower among Hispanic persons (0.27 in 1000), Asian Americans (<0.001 per 1000), Pacific Islanders (0.12 per 1000), and black persons (0.14 per 1000) (11). The availability of genotyping has permitted identifying persons who have the susceptible genotype but have little or no evidence of disease. Thus, individuals homozygous for the C282Y genotype can be characterized in 1 of 4 general stages: genetic predisposition without any other abnormality; iron overload without symptoms; iron overload with early symptoms; and iron overload with organ damage, especially cirrhosis (4). Clinically recognized hereditary hemochromatosis is twice as common in males and occurs predominantly in white populations (12). While the natural history is not well understood, the condition appears to have a long latent period, with wide individual variation in biochemical expression (13). This is because iron accumulation and disease expression are modified by environmental factors, such as blood loss from menstruation or donation, alcohol intake, diet, and comorbid disease (for example, viral hepatitis) (14, 15). If symptomatic organ involvement develops, it generally occurs in mid-life with nonspecific signs and symptoms (such as unexplained fatigue, joint pain, and abdominal pain) (14). Age of onset is delayed in females (16), perhaps because of blood loss through menstruation (3). The liver is the first target organ thought to be affected by iron accumulation (17) and is central to both diagnosis and prognosis (13). While a clinical diagnosis is based on serum iron studies and clinical evaluation, documented iron overload relies on 1 of 2 methods: quantitative phlebotomy with calculation of the amount of iron removed, or liver biopsy with determination of quantitative hepatic iron (18). Although liver biopsy was once essential to the diagnosis, it is currently used more as a prognostic tool (19). While hepatic iron concentration greater than 283 mol/g (reference range, 0 to 35 mol/g) is associated with cirrhosis in C282Y homozygotes (20), many patients with much higher levels do not have cirrhosis (13). Even in the absence of systemic iron overload, iron accumulates when the liver is inflamed or cirrhosed because of other causes (such as alcoholic steatohepatitis, transfusion and chronic hemolytic disorders, or chronic viral hepatitis) (21). Cirrhosis is a late-stage disease development and has been reported to shorten life expectancy (2225). Cirrhosis is also a risk factor for hepatocellular carcinoma (13) and typically occurs between the ages of 40 and 60 years (6). Cirrhosis prevention would be a major goal of screening and treatment (26). Prevalence and Burden of Disease Estimates of the general population prevalence of hemochromatosis vary because of the long preclinical period and lack of a consistent case definition. The prevalence of cases of hemochromatosis defined biochemically (elevated serum iron indices) will be higher than the prevalence of cases based on documented iron overload, with or without clinical signs and symptoms. The prevalence will be lowest for cases based on diagnosed disease (cirrhosis, diabetes) (27). Experts have recommended defining iron overload as distinct from hemochromatosis (4), and this provides an objective, although not universally accepted, standard for early disease based on documented increases in body iron stores (27). On the basis of clinically diagnosed hemochromatosis or hemochromatosis-compatible disease, 79850 hemochromatosis-associated hospitalizations (2.3 per 100000 residents) were projected in the United States over 18 years (1979 to 1997), although annual rates could not be reliably calculated (28). Of 29 million deaths from 1979 to 1992, 4858 (0.017%) were consistent with hemochromatosis as an underlying cause (12). Age-adjusted mortality rates for hemochromatosis-consistent deaths increased from 1.2 per million in 1979 to 1.8 per million in 1992. These rates were about twice as high in males as in females and in white persons as in nonwhite persons. Both of these estimates of the burden of disease suggest a disease prevalence much lower than the prevalence of associated genetic mutations, which has fueled the debate about disease penetrance. While these statistics are probably underestimates, primarily because of underdiagnosis (29), the extent of this underestimation is not clear. The prevalence of hemochromatosis-attributable morbid conditions (such as cirrhosis, diabetes, arthralgias, and fatigue or other symptoms) has been proposed as an estimate of the burden due to undiagnosed disease, particularly since diagnosis may commonly be delayed as a result of the nonspecific nature of hemochromatosis-related signs and symptoms (30). Since these signs and symptoms are also prevalent and nonspecific, however, relevant evidence must establish their prevalence due to iron overload, or their excess prevalence in association with iron overload compared with controls. In a previous study, 297 middle-aged patients with previously undetected hereditary hemochromatosis (homozygous for C282Y) had a higher prevalence of diagnosed osteoarthritis, knee symptoms, hypothyroidism, and use of antihypertensive or thyroid replacement medications than sex- and age-specific controls (31). However, general health, mental health, and 52 other questionnaire-based and clinical examinationbased measures of cardiovascular, respiratory, and liver diseases were not statistically different between case-patients and controls. In another cross-sectional comparison of 124 C282Y screening-detected adult homozygotes with 22394 wild-type/wild-type genotypic controls, common symptoms (chronic fatigue, joint symptoms, impotence, and limited general health) and signs (diabetes) were no more frequent in C282Y homozygotes than controls (32). While the relative risk for physician-diagnosed liver problems or hepatitis was increased (relative risk, 2.1 [95% CI, 1.1 to 4.0]), the proportion of C282Y homozygotes with liver problems was modest (10%). Similarly, in the Hemochromatosis and Iron Overload Screening (HEIRS) study, C282Y homozygotes had an increased odds of self-reported liver disease (odds ratio, 3.28 [CI, 1.49 to 7.22]) compared with wild-type controls. Almost one fourth, however, were not identified by screening (11). Clearly, the prevalence of hemochromatosis-attributable morbid conditions is not a simple, reliable way to estimate the disease burden associated with hemochromatosis. Rationale for Population Screening Screening for hemochromatosis or iron overload is theoretically attractive and has been widely discussed over the past 10 to 15 years, with renewed interest and a focus on hereditary hemochromatosis since the discovery of the HFE mutations (4, 3336). Although hereditary hemochromatosis appears to be ideal for population screening (7, 16, 3739) and for a new paradigm for genetics and public health (34), inadequacies in the evidence supporting genetic screening for hereditary hemochromatosis have precluded widespread support for population-based screening (4, 9, 34, 40). Aims of Focused Systematic Review This review addresses 2 major uncertainties in the evidence: How much disease is actually caused by HFE mutations? and Does therapeutic phlebotomy treatment, initiated through earlier identification of those with hereditary hemochromatosis, lead to better outcomes? We also considered evidence for high-risk (as opposed to general population) screening. Methods We focused on hereditary HFE-associ