Jun Jie Tan's List: Genomics and Bioinformatics

• Next month, about 5,500 first-year students will receive testing kits in the mail and be asked to submit DNA swabs to test three genes. The genes include those related to the ability to break down lactose, metabolize alcohol and absorb folates.
• But the Center for Genetics and Society, a Berkeley public interest organization, and the Council for Responsible Genetics, which is based in Cambridge, Mass., say the project disregards the potential harmful use of the information.
• "What we hoped to do is expose our students to the science behind personalized medicine and engage them in the discussion about what the future may hold," Schlissel said. "We could have sent all the students a magazine article or book to read, but we thought it would be far more engaging to actually involve in them in a genetic test."

1 more annotation...

• has discovered how one particular gene regulates epithelial cells cells that normally form in sheets and are polarized to enable the transport of molecules in a single direction. It's this loss of polarity that is thought to play an important role in breast tumor development.
• By using mouse models, Muller discovered that the cells do not form neat structures when the gene malfunctions. "In fact, the first mouse model had a skin defect and was completely incapable of forming sheets of epithelial cells. This gene is frequently lost in breast cancer, significant proof that this gene might play an important role," he said.

• Scientists report that breast cancer risk assessment models, which predict a woman's chance of developing breast cancer, do not perform better when they include common inherited genetic variants recently linked to the disease. Therefore, recommendations for breast cancer screening or treatments will remain unchanged for most women. The study, led by investigators from the National Cancer Institute (NCI), part of the National Institutes of Health, appears in the March 18, 2010, New England Journal of Medicine.
• "In the past three years, genome-wide association studies have identified multiple common genetic variants associated with breast cancer. The extent to which adding these variants to existing models could improve clinical recommendations had not been tested in a large population of women prior to this study," said Sholom Wacholder, Ph.D., senior investigator in NCI's Division of Cancer Epidemiology and Genetics (DCEG). "When we included these newly discovered genetic factors, we found some improvement in the performance of risk models for breast cancer, but it was not enough improvement to matter for the great majority of women."
• Findings from genome-wide association studies (GWAS) to date have pinpointed several locations in the human genome, called single-nucleotide polymorphisms (SNPs), where genetic variation is associated with cancer risk. SNPs are the most common type of variation, affecting just a single building block of DNA. SNPs are used in GWAS to identify chromosome regions that are associated with disease. Studies to characterize the biologic effects of the variants associated with breast cancer are now being conducted to help clarify their role in breast cancer risk.
• the investigators examined the predictive accuracy of the Gail model -- the most commonly used breast cancer risk model -- for this group of women. The Gail model uses information on a woman's own personal medical and reproductive history, as well as the history of breast cancer among her first-degree relatives (mother, sisters, and children) to estimate her risk of developing invasive breast cancer within the next five years, or over her lifetime. The investigators then tested the accuracy of a SNP model and found that it was as good as the Gail model alone. An inclusive model, using both SNPs and Gail factors, performed only slightly better than either model alone.

2 more annotations...

• We used information on traditional risk factors and 10 common genetic variants associated with breast cancer in 5590 case subjects and 5998 control subjects, 50 to 79 years of age, from four U.S. cohort studies and one case–control study from Poland to fit models of the absolute risk of breast cancer. With the use of receiver-operating-characteristic curve analysis, we calculated the area under the curve (AUC) as a measure of discrimination. By definition, random classification of case and control subjects provides an AUC of 50%; perfect classification provides an AUC of 100%. We calculated the fraction of case subjects in quintiles of estimated absolute risk after the addition of genetic variants to the traditional risk model.
• The inclusion of newly discovered genetic factors modestly improved the performance of risk models for breast cancer. The level of predicted breast-cancer risk among most women changed little after the addition of currently available genetic information.
• The Breast Cancer Risk Assessment Tool,^4,5 commonly referred to as the Gail model, summarizes information about a woman's reproductive history, breast cancer in close relatives, and previous breast biopsies to estimate the probability of the development of breast cancer in subsequent years; the estimated breast-cancer risk based on the Gail model has been used for counseling, informing decisions about the use of tamoxifen,⁶ and determining sample size in randomized prevention trials.⁷ In this study, we empirically evaluated the contribution of a set of 10 newly established common genetic variants as an alternative to and as a supplement to the components of the Gail model in 5590 case subjects and 5998 control subjects, 50 to 79 years of age, included in the National Cancer Institute's Cancer Genetic Markers of Susceptibility multistage genomewide association study of breast cancer.
• In our study involving 5590 case subjects with breast cancer and 5998 control subjects, the addition of information on 10 genetic variants to a standard clinical breast-cancer risk model predicted the risk of breast cancer only slightly better than the clinical model alone. In the inclusive model, the ROC curve was above 60% of the possible AUC. That is, about 60% of the time, a randomly selected patient with breast cancer had a higher estimated risk than the risk for a randomly selected woman in whom breast cancer did not develop during the follow-up period. By contrast, a single dichotomous risk factor detected in 60% of case subjects and 40% of control subjects (odds ratio, 2.25; AUC, 60%) would discriminate about as well as our model by the AUC criterion. We saw similarly modest improvement using measures based on the change in estimated risk. Although our data suggest that the SNP-only model predicted risk slightly better than the Gail model, the nongenetic clinical variables are available at essentially no cost, whereas the costs of obtaining genetic information are likely to be substantial.
• Although the Gail model and SNP-inclusive models may help to identify groups of women who have an increased risk of breast cancer for trials of interventions, none of the models in our set of data accurately predicted the development of breast cancer. Our results indicate that the recent identification of common genetic variants does not herald the arrival of personalized prevention of breast cancer in most women. Even with the addition of these common variants, breast-cancer risk models are not yet able to identify women at reduced or elevated risk in a clinically useful way.

3 more annotations...

• But when Amy Berrington de Gonzalez, D.Phil., and colleagues at the Johns Hopkins Bloomberg School of Public Health in Baltimore, looked at breast cancer mortality statistics in this group of women following five annual mammograms starting at various ages, they found a disturbing trend: far more cases of breast cancer developed than were expected.
• It's also important to note that other researchers have questioned whether all women -- not only those with hereditary breast cancer -- are putting themselves at risk with yearly mammograms. For example, research published last fall showed that breast cancer rates soared after regular mammography was started in four Norwegian countries ( http://www.naturalnews.com/024901.html). In addition, Samuel S. Epstein M.D., Professor Emeritus of Environmental Medicine at the University of Illinois at Chicago School of Public Health, and his colleagues conducted a review of 47 scientific articles about mammography. Their article, "Dangers and Unreliability of Mammography: Breast Examination is a Safe, Effective, and Practical Alternative", published in the International Journal of Health Services (2001;31(3):605-15) concluded that mammogram screening carries many dangers, including induction and promotion of breast cancer, falsely positive and negative diagnosis of breast cancer, and over-diagnosis.

• Formed in 2008, the consortium brings together leading cancer researchers from around the world, working together to catalogue the genetic changes of the 50 most common cancers - 500 genomes from each cancer type - and make the results freely available on the internet. 
  
• "Given the tremendous potential for relatively low-cost genomic sequencing to reveal clinically useful information, we anticipate that in the not so distant future, partial or full cancer genomes will routinely be sequenced as part of the clinical evaluation of cancer patients," say the authors in the paper.
• "For example, you might find that the aberrations in a subtype of colon cancer are the same as the aberrations in a subtype of melanoma. In that case, the treatment that works in the colon cancer may be appropriate for the melanoma. So you'd go ahead and test it." 
  
   "The problem we have is the complexity of cancer. No two tumours are the same, even within the same type of cancer. They may look the same under the microscope, but their molecular aberrations vary greatly."
• "The consortium's internet-based databanks will help us treat specific cancers with specific treatments. Not only that, the information will help us understand why some treatments work and others do not, and then design better drugs to target faulty elements or mechanisms." 
  
• "If you approach the problem with new technologies, you can quickly match the drug with molecular aberrations in specific cancers, and narrow the trial phase down to a few months."
• "We've hypothesised about that in the past, but having the evidence to prove the difference is exciting. Right from the outset we know everything there is to know about one person's tumour at the genomic, transcriptomic and epigenomic levels. We might not understand it, but we've got the data." 
  

4 more annotations...

• A case in point is breast cancer. In this issue of the Journal, Wacholder et al.³ report that when 10 recently identified SNPs that have been associated with breast cancer were added to the Gail model, a commonly used tool to assess the risk of breast cancer, discrimination and prediction in estimating risk were only moderately improved. The Gail model is based on self-reported risk factors (age at menarche, number of breast biopsies, age at birth of the first child, and family history of breast cancer). The authors conclude that the clinical usefulness of common genetic variants for identifying women at increased risk for breast cancer is, at present, still negligible. Intriguingly, however, Wacholder et al. also found that a model with the 10 SNPs alone performed just as well as the Gail model in predicting the development of breast cancer. The addition of the SNP data to the Gail model almost doubled its discriminatory power, which was expressed as the area under the receiver-operating-characteristic curve (AUC); the AUC was calculated to be 61.8% in the final combined model. The authors correctly conclude that such an AUC is still far too low for application in clinical settings, but in admitting so, they also implicitly dismiss the pure Gail model for this purpose.
• Their disappointment seems a little premature, however. The 10 SNPs analyzed by Wacholder et al. occur frequently in women, but each confers only slightly elevated risks of breast cancer (with odds ratios of 1.05 to 1.25). A positive family history is one of the major risk factors for breast cancer; on average, it almost doubles the risk.⁴ Rare high-risk variants, such as those in the breast-cancer genes BRCA1 and BRCA2, account for about 25% of this familial risk. In contrast, the 10 common breast-cancer SNPs account for less than 5% of the familial risk.⁴ Depending on the relative risk per allele, several hundred to more than 1000 of such SNPs are required to account for the doubling of the relative risk in a sister of a woman with breast cancer.² Clearly, the 10 SNPs assessed by Wacholder et al. are no more than the tip of the iceberg. A more pressing question is why, after the completion of several genomewide association studies of breast cancer, only a dozen risk alleles have been identified.
• A limitation of the currently identified breast-cancer SNPs is that we have very few clues as to how they function to increase the risk of breast cancer. It is plausible that once a cellular network within which a SNP operates has been unraveled, the resulting biologic defect will reveal a much stronger association with breast cancer than the SNP that led to the detection of the pathway. For example, the risk allele of a SNP in intron 2 of fibroblast growth factor receptor 2 (FGFR2) (1 of the 10 SNPs studied by Wacholder et al.) seems to be associated with relatively high levels of FGFR2 messenger RNA.⁵ Yet, even in homozygous carriers of the risk allele, there is a wide range of expression levels of FGFR2, implicating other factors in its regulation. Hence, if high expression levels of FGFR2 alone contribute to an increased risk of breast cancer, then determining these levels (or the activity of the pathway in which FGFR2 operates) might be a much more accurate way to determine risk than genotyping the intronic SNP.
• While the dissection of the genetic architecture of breast cancer continues, we should address the limitations of current risk-prediction models. If the reference population is composed of women who abstain from alcohol, have a normal weight, have had more than five children, had menarche at a late age (>15 years of age), and have a history of breast-feeding for a total of more than 10 years, the major portion of breast-cancer risk among women today would be accounted for by the lack of these factors alone.^6,7 However, a reference group of women with these characteristics, although dominant until a century ago, has almost disappeared in Western populations. The challenge is therefore to predict whether breast cancer will develop in women who are already at increased risk because of their modern lifestyle.
• Present-day genomewide association studies are still severely limited for the detection of gene–gene and gene–environment interactions, which are likely to be profoundly important. Investigating these interactions is not going to be easy, but one way forward might be through the so-called intermediate phenotypes that are strongly associated with the cause of breast cancer and that have a substantial genetic component themselves. Interesting candidates are breast density on mammographic screening and chromosomal radiosensitivity. An unraveling of the biologic and genetic factors of these and other markers associated with breast cancer (such as atypical hyperplasia, increased bone mineral density, and involution of lobular tissue) might identify cellular pathways and their regulatory factors that are also causally involved in determining the risk of breast cancer.
• For women seeking advice on their personal risk of breast cancer, it is obviously too early to incorporate SNP testing into a counseling procedure, although such tests are already advertised for this purpose on the Internet.

4 more annotations...

• Boston (DbTechNo) - Results of a new study show that genetic tests can not predict who is most likely to develop breast cancer in their lifetime.
• Researchers compared the traditional method of asking women a series of health questions to gain an understanding of their past history, to genetic tests to check for the likelyhood of breast cancer.
• A total of 11,588 women age 50 to 79 took part in the study and more than 5000 of them had previously been diagnosed with breast cancer.

1 more annotation...

• Women with breast cancer before age 55 who carry an inherited mutation in the breast cancer susceptibility genes BRCA1 or BRCA2 are four times more likely to develop cancer in the breast opposite, or contralateral, to their initial tumor as compared to breast cancer patients without these genetic defects. These findings, by Fred Hutchinson Cancer Research Center breast cancer epidemiologist Kathleen Malone, Ph.D., and colleagues, were published online April 5 in the Journal of Clinical Oncology.
• Compared to non-carriers, breast cancer patients with a BRCA1 mutation had a 4.5-fold increased risk and those with a BRCA2 mutation had a 3.4-fold increased risk of a subsequent contralateral breast cancer, the researchers found. Carriers of either mutation who were diagnosed with breast cancer before age 55 faced an 18 percent cumulative probability of developing cancer in the opposite breast within 10 years as compared to a 5 percent cumulative probability among women who were mutation-free.
• In addition, the study revealed that among those who harbored a BRCA1 mutation, the younger they were at the time of initial diagnosis the higher was their risk of developing contralateral breast cancer. For example, mutation carriers diagnosed initially in their early to mid 30s had a 31 percent cumulative probability of developing contralateral breast cancer within 10 years as compared to a 7 percent probability among non-carriers.
• While only about 5 percent of breast cancer patients across all age groups carry a BRCA mutation, the younger a woman is at the time of her first breast cancer diagnosis, the more likely she is to have such a mutation.

2 more annotations...

• My analysis revealed that 10,079 SNPs were contained within the regions where my brothers and I share diplotypes. Of these 10,079 SNPs, only 86 of them had any genotyping errors!
• This means that the genotyping calling was 99.15% accurate for these regions. Moreover, of the errors recorded, 79 of them occurred when there was a genotype call for some of the siblings and a null call (–) for others. Only 7 errors occurred where there was inconsistency in the genotype assigned to the siblings. The results are summarized in the table to the left.
• My conclusion: the genotyping error rate is very low, less than 1% for the Illumina platform used by 23andMe. Even taking null calls into account, this number is still below 1%. My siblings and I shared 99.15% genotype identity in a region where we all share both parental haplotypes. I am very pleased with the accuracy.

1 more annotation...

• The best-known epigenetic marker is DNA methylation. The initial finding of global hypomethylation of DNA in human tumors⁵ was soon followed by the identification of hypermethylated tumor-suppressor genes,^{6,7,8,9,10,11} and then, more recently, the discovery of inactivation of microRNA (miRNA) genes by DNA methylation.^12,13 These and other demonstrations of how epigenetic changes can modify gene expression have led to human epigenome projects¹⁴ and epigenetic therapies.¹⁵ Moreover, we now know that DNA methylation occurs in a complex chromatin network and is influenced by the modifications in histone structure that are commonly disrupted in cancer cells.^16,17,18,19
• DNA methylation has critical roles in the control of gene activity and the architecture of the nucleus of the cell. In humans, DNA methylation occurs in cytosines that precede guanines; these are called dinucleotide CpGs.^26,27 CpG sites are not randomly distributed in the genome; instead, there are CpG-rich regions known as CpG islands, which span the 5' end of the regulatory region of many genes. These islands are usually not methylated in normal cells.^26,27 The methylation of particular subgroups of promoter CpG islands can, however, be detected in normal tissues.
• The low level of DNA methylation in tumors as compared with the level of DNA methylation in their normal-tissue counterparts was one of the first epigenetic alterations to be found in human cancer.⁵ The loss of methylation is mainly due to hypomethylation of repetitive DNA sequences and demethylation of coding regions and introns — regions of DNA that allow alternative versions of the messenger RNA (mRNA) that is transcribed from a gene.³⁸ A recent large-scale study of DNA methylation with the use of genomic microarrays has detected extensive hypomethylated genomic regions in gene-poor areas.²⁴ During the development of a neoplasm, the degree of hypomethylation of genomic DNA increases as the lesion progresses from a benign proliferation of cells to an invasive cancer³⁹ (Figure 1).
• Hypermethylation of the CpG islands in the promoter regions of tumor-suppressor genes is a major event in the origin of many cancers. The initial reports of hypermethylation of the CpG islands in the promoter region of the retinoblastoma tumor-suppressor gene (Rb)^6,7 were followed by the findings that hypermethylation of the CpG island was a mechanism of inactivation of the tumor-suppressor genes VHL (associated with von Hippel–Lindau disease), p16^INK4a,^8,9,10,11 hMLH1 (a homologue of MutL Escherichia coli),²⁶ and BRCA1 (breast-cancer susceptibility gene 1).^26,51

2 more annotations...

• Physicians have long recognized that pinpointing specific causes of disease in individual patients enables therapies that are the most likely to confer benefit with the fewest adverse effects. We also recognize the potential for disease prevention through identification of specific risk factors and mitigation of their effects. For a century, we have known that many of these risk factors are genetic. In the past 20 years, the genomic revolution has translated this knowledge into a new understanding of disease: mutations that cause more than 2000 mendelian diseases have been identified, which has led to the rewriting of textbooks of pathophysiology of every organ system and the identification of rational targets for therapeutic intervention. Genes also play a major role in risk for virtually every common disease, affording the possibility of identifying persons who have a specific inherited predisposition.
• In this issue of the Journal, Lupski and colleagues report on their study that shows the power of this new technology.⁴ They used whole-genome sequencing to make a specific diagnosis in a family in which four siblings were affected by Charcot–Marie–Tooth disease, a peripheral polyneuropathy. Mutations in 31 known genes and additional unidentified loci can produce Charcot–Marie–Tooth disease. The investigators produced nearly 90 billion base pairs of genomic sequence in one affected subject (sufficient to ensure that both alleles at nearly every base pair have been sampled repeatedly) and identified variations from the reference sequence. As expected, they found a large number of common and novel variants. When they examined genes known to be mutated in patients with Charcot–Marie–Tooth disease, they found two compelling mutations in SH3TC2 (the SH3 domain and tetratricopeptide repeats 2 gene), which causes autosomal recessive Charcot–Marie–Tooth disease. They also found complete cosegregation of these mutations with disease status in the family, providing convincing evidence that these SH3TC2 mutations are the cause of Charcot–Marie–Tooth disease in this family.
• The sequence production for this project cost less than $50,000. More traditional approaches could have obtained the same answers; nonetheless, the study provides a striking proof of principle. Moreover, there is every reason to believe that the cost of sequencing will continue to plummet. Owing to innovation and intense competition, the cost of sequence production 2 years from now will almost certainly be at most one tenth of the current cost of using current technologies. Moreover, there are widespread efforts to advance new technologies to achieve further drastic drops in cost.⁵

1 more annotation...

• Nowhere is this transformation more apparent than in oncology. Cancer is a complex disease. Our current taxonomy of cancers, which is based mostly on histopathology, includes more than 200 distinct entities arising from diverse types of cells. In addition, tumors have somatic mutations and epigenetic changes, many of which are specific to individual neoplasms; these molecular abnormalities influence the expression of genes that control a tumor's growth, invasiveness, metastatic potential, and responsiveness or resistance to chemotherapy. The genetic complexity of cancer probably explains the clinical diversity of histologically similar tumors, but it has been difficult to study this diversity with traditional methods, which are best suited to investigating one gene at a time. The advent of DNA microarray technology, however, permits the quantitative measurement of complex, multigene expression patterns in cancer.
• Microarrays are wonderful tools for discovery, allowing researchers to obtain unbiased surveys of gene expression in tissue samples, but some have questioned their direct clinical application for individualized diagnosis and treatment planning. Microarrays have an appreciable failure rate and occasionally show significant interreplicate and interbatch variability in measurement. A second concern is that in a single sample, microarrays measure thousands of variables, most of which are irrelevant to the clinical end point under investigation. Complex statistical and computational tools are thus required to extract informative patterns from raw microarray data. Current technology also requires snap-frozen tissue for microarray-based gene-expression profiling. It is usually possible for established tumor banks to provide small frozen specimens for the initial discovery of clinically useful gene-expression profiles, but validation studies are often limited by the availability of tissue, since tumor specimens are generally fixed in formalin rather than frozen. These limitations pose considerable obstacles to the routine use of microarrays in the clinical laboratory, where tests must be highly reliable and easy to interpret.
• Oncology is spearheading the movement to an era of personalized diagnosis and treatment planning with the development and implementation of increasingly accurate molecular diagnostic tools. Although clinical diagnostic tests have traditionally taken a backseat to therapeutic agents in cancer medicine, change is at hand. In principle, it should be possible to create molecular diagnostic tools that can predict the response of all human tumors to single agents or combination chemotherapy, thereby allowing for precise, individualized matching of molecular diagnosis with treatment. Moreover, early studies in cancer genomics have focused on microarray-based RNA-expression profiling, teaching us how to grapple with complex biologic data sets in the genomic era. Lessons learned from this initial experience are already informing our use of newer techniques for obtaining systemwide molecular views of disease.

1 more annotation...

• If genetics has been misunderstood, genomics is even more mysterious — what, exactly, is the difference? Genetics is the study of single genes and their effects. "Genomics,"⁴ a term coined only 15 years ago, is the study not just of single genes, but of the functions and interactions of all the genes in the genome. Genomics has a broader and more ambitious reach than does genetics. The science of genomics rests on direct experimental access to the entire genome and applies to common conditions, such as breast cancer⁵ and colorectal cancer,⁶ human immunodeficiency virus (HIV) infection,⁷ tuberculosis,⁸ Parkinson's disease,⁹ and Alzheimer's disease.¹⁰ These common disorders are also all due to the interactions of multiple genes and environmental factors. They are thus known as multifactorial disorders. Genetic variations in these disorders may have a protective or a pathologic role in the expression of diseases.
• Mutations can also decrease the risk of a disease. One example of this is a 32-bp deletion (a frame shift) in a chemokine receptor gene, CCR5. Persons who are homozygous for this deletion prove almost completely resistant to infection with HIV type 1, and those who are heterozygous for the deletion have slower progression from infection to AIDS. These effects arise because CCR5 is an important part of the mechanism by which HIV enters the cell.²⁵
• The study of genomics will most likely make its greatest contribution to health by revealing mechanisms of common, complex diseases, such as hypertension, diabetes, and asthma. So far, most genes involved in common diseases have been identified by virtue of their high penetrance — that is, the mutations lead to disease in a fairly large proportion of people who have them. Examples include mutations in BRCA1 and BRCA2 (increasing the risk of breast and ovarian cancer),²⁶ HNPCC (increasing the risk of hereditary nonpolyposis colorectal cancer),⁶ MODY 1, MODY 2, and MODY 3 (increasing the risk of diabetes),²⁷ and the gene for -synuclein (causing Parkinson's disease).⁹ One can think of these as near-mendelian subgroups of disease within a larger group of affected persons. If a person has such a mutation, the likelihood of disease is great. However, each of these highly penetrant mutations associated with common disease has a prevalence in the general population of only one in several hundred to several thousand people.
• One characteristic of the human genome with medical and social relevance is that, on average, two unrelated persons share over 99.9 percent of their DNA sequences.³ However, given the more than 3 billion base pairs that constitute the human genome, this also means that the DNA sequences of two unrelated humans vary at millions of bases. Since a person's genotype represents the blending of parental genotypes, we are each thus heterozygous at about 3 million bases. Many efforts are currently under way, in both the academic and commercial sectors, to catalogue these variants, commonly referred to as "single-nucleotide polymorphisms" (SNPs), and to correlate these specific genotypic variations with specific phenotypic variations relevant to health.
• Some SNP–phenotype correlations occur as a direct result of the influence of the SNP on health. More commonly, however, the SNP is merely a marker of biologic diversity that happens to correlate with health because of its proximity to the genetic factor that is actually the cause. In this sense, the term "proximity" is only a rough measure of physical closeness; instead, it connotes that, as genetic material has passed through 5000 generations from our common African ancestral pool, recombination between the SNP and the actual genetic factor has occurred only rarely. In genetic terms, the SNP and the actual genetic factor are said to be in linkage disequilibrium (Figure 4).
• An extension of the current efforts to catalogue SNPs and correlate them to phenotype are efforts to map and use haplotypes. Whereas a SNP represents a single-nucleotide variant, a haplotype represents a considerably longer sequence of nucleotides (averaging about 25,000), as well as any variants, that tend to be inherited together. SNPs and haplotypes will be the key to the association studies (i.e., studies of affected persons and control subjects) necessary to identify the genetic factors in complex, common diseases, just as family studies have been important to the identification of the genes involved in monogenic conditions.^35,36 Also, at least until whole-genome sequencing of individual patients becomes feasible clinically, the identification of SNPs and haplotypes will prove instrumental in efforts to use genomic medicine to individualize health care.

4 more annotations...

• But such premature attempts at popularizing genetic testing seem to neglect key aspects of the established multifaceted evaluation of genetic tests for clinical applications. First, there is the question of a test's analytic validity, "its ability to accurately and reliably measure the genotype of interest."¹ Although appropriate monitoring and oversight of the analytic validity of genetic tests remain largely unaddressed,² most researchers report that the analytic validity of these platforms is very high. It is likely that sample-handling errors are a greater threat to the validity of results than are genotypic misclassification errors. Yet even very small error rates per SNP, magnified across the genome, can result in hundreds of misclassified variants for any individual patient. Without transparent quality-control monitoring and proficiency testing, the real-world performance of these platforms is uncertain.
• Second, one must consider clinical validity, or the ability of the test to detect or predict the associated disorder.¹ Components of clinical validity include the test's sensitivity, specificity, and positive and negative predictive value. This is the area in which the data are in the greatest flux, and even the ardent proponents of genomic susceptibility testing would agree that for most diseases, we are still at the early stages of identifying the full list of susceptibility-associated variants. Most of the diseases listed by the direct-to-consumer testing companies (e.g., diabetes, various cancers, and heart disease) are so-called complex diseases thought to be caused by multiple gene variants, interactions among these variants, and interactions between variants and environmental factors. Thus, a full accounting of disease susceptibility awaits the identification of these multiple variants and their interactions in well-designed studies. What we have now is recognition of a limited number of variants associated with relative risks of diseases on the order of 1.5 or lower. Risk factors with this level of relative risk clearly do a poor job of distinguishing people who will develop these diseases from those who will not.^3,4
• Finally, there is the issue of the test's clinical utility, or the balance of its associated risks and benefits if it were to be introduced into clinical practice.¹ Measures of utility address the question at the heart of the clinical application of a test: If a patient is found to be at risk for a disease, what can be done about it? This is the arena in which there are virtually no data available on the health impact of genomewide analysis. There are very few observational studies and almost no clinical trials that demonstrate the risks and benefits associated with screening for individual gene variants — let alone testing for many hundreds of thousands of variants. Thus, any claim to clinical utility currently rests on the assumption that interventions that have proven successful in the general population will behave the same way in a genetically at-risk population. Many of these interventions — such as smoking cessation, weight loss, increased physical activity, and control of blood pressure — are likely to be broadly beneficial in relation to many diseases, regardless of a person's genetic susceptibility to a specific disease.
• It may be argued that knowledge of increased susceptibility to a disease, such as type 2 diabetes, for which protective lifestyle interventions exist, will motivate patients to follow relevant recommendations. Yet as intuitively appealing as this contention may be, evidence to support it, particularly in the case of low-penetrance alleles, is scanty. The flip side, of course, is that patients who test negative may be falsely reassured and thus less motivated to comply with preventive recommendations. In the absence of evidence of efficacy, this rationale for susceptibility testing should be regarded with skepticism.
• So what advice should a physician offer patients? For the patient who appears with a genome map and printouts of risk estimates in hand, a general statement about the poor sensitivity and positive predictive value of such results is appropriate, but a detailed consumer report may be beyond most physicians' skill sets. For the patient asking whether these services provide information that is useful for disease avoidance, the prudent answer is "Not now — ask again in a few years." More information is needed on the clinical utility of this information in the light of existing disease-specific opportunities for prevention or early detection and the potential value that genomic profiles can add to that of simpler tools, such as the family health history. Finally, given the risk of commercial exploitation, if patients are determined to proceed, perhaps because they are simply curious, are genetic hobbyists, or are "early adopters" of new technology, it would make sense to encourage them to enroll in formal scientific studies.

3 more annotations...