Clinical Review

A practical guide to interpretation and clinical application of personal genomic screening

BMJ 2009; 339 doi: http://dx.doi.org/10.1136/bmj.b4253 (Published 29 October 2009) Cite this as: BMJ 2009;339:b4253
  1. Emily Edelman, project director1,
  2. Charis Eng, Sondra J and Stephen R Hardis chair of cancer genomic medicine, professor and chair2
  1. 1National Coalition for Health Professional Education in Genetics, Lutherville, MD 21093, USA
  2. 2Cleveland Clinic Genomic Medicine Institute, Cleveland, OH 44195, USA
  1. Correspondence to: C Eng, Genomic Medicine Institute, Cleveland Clinic, 9500 Euclid Avenue, NE-50, Cleveland, OH 44195, USA engc{at}ccf.org

    Summary points

    • Personal genomic screening is a rapidly evolving area of direct to consumer genetic testing

    • Such screening uses data from genome-wide association studies to predict a person’s disease risk and tests using many genetic markers at once

    • Companies may offer genomic testing for common diseases and traits as well as rare mendelian conditions

    • Screening results may include predictions about physical or personality traits, pharmacogenetic profiles, and ancestry testing

    • Doctors should use caution when a patient presents with such screening results because their accuracy and usefulness are unknown

    • Instead, risk of disease should be based on validated clinical methods, such as family history and clinical and biochemical measures

    Researchers are identifying genetic associations for common diseases at a rapid pace. With enormous amounts of genome-wide association data available, several for-profit companies worldwide have developed testing services to provide clients with predictions about their risks for a broad spectrum of medical and non-medical conditions and traits—from ancestry grouping, to earwax type, to prostate cancer. Many of these companies provide services direct to the consumer, without input or authorisation from a doctor or other healthcare provider.

    Sources and selection criteria

    Because of the various types of genetic associations analysed and the rapid evolution of tests offered by personal genomic screening companies, data on evidence based assessments of these services are limited. We searched PubMed for combinations of terms such as “direct-to-consumer”, “personal-genomic testing”, “personal-genomic profile”, “clinical utility”, “clinical validity”, and “evidence” to identify studies assessing the current state of evidence.

    We searched the Genetics and Public Policy Center at Johns Hopkins University and the Human Genetics Commission’s 2007 report More Genes Direct to identify direct to consumer genetic testing companies that provide personal genomic scanning for multiple conditions. We reviewed the websites of many of the companies.

    Doctors may be asked for advice on medical matters that arise from such screening tests.1 As yet, we have insufficient evidence to provide recommendations for action based on most of these results. This article offers a practical guide to help doctors review, interpret, and communicate to patients information about the results of personal genomic screening tests.

    What is personal genomic screening?

    Personal genomic screening is a relatively new type of genetic testing, which is based on multiple statistical comparisons, or genome-wide associations. In contrast, validated mendelian genetic testing focuses on one or a few genes, which are chosen because of the medical and family history and are strongly associated with disease. If a gene mutation is identified, the person is at high risk of developing the associated disease. Validated genetic testing is offered by healthcare providers to people at high risk of mendelian genetic disease or as part of cost effective public health measures (for example, newborn screening). Diseases that can be tested for in this way include Huntington’s disease, familial adenomatous polyposis, and Tay-Sachs disease.

    Personal genomic screening, or personal genomic testing or profiling, uses data derived from genome-wide association studies to predict a person’s chance of developing many different conditions throughout his or her lifetime. Most genome-wide association studies are case-control studies that look at hundreds to thousands of people to identify genetic markers that are statistically more common in people with disease than in controls. These genetic markers are called single nucleotide polymorphisms (SNPs). They are common in the general population and may or may not have a known functional consequence. Genome-wide association studies typically provide odds ratios of the likelihood that a person with a particular genotype will develop certain diseases.

    For which common diseases and traits have genetic markers been identified?

    Genome-wide association studies have identified SNPs associated with many common diseases such as type 1 and type 2 diabetes, cardiovascular disease, prostate cancer, breast cancer, colon cancer, asthma, Crohn’s disease, and rheumatoid arthritis.2 Furthermore, genome-wide association studies have identified associations for non-medical traits or characteristics such as eye colour, skin pigmentation, height, and intelligence. Arguably, the most promising application of genome-wide association research is that of pharmacogenomics, the ability to predict a person’s drug metabolism or response to a specific drug. Personal genomic screening analyses a person’s genotype for many of these common conditions, but the specific disease profile will vary between companies.3 4 5 Personal genomic screening may also assess specific point mutations for selected mendelian genetic conditions, such as hereditary breast and ovarian cancer syndrome, cystic fibrosis, and sickle cell anaemia. Some companies, without input or interpretation from a healthcare provider, now offer full genome sequencing, which they say will report on rare mendelian conditions and common diseases.6 Some also offer testing on ancestry or racial origin, and the genetic similarity between one individual and another.

    How is a personalised genome screening test ordered?

    At least 38 web based companies worldwide offer direct to consumer genetic testing.7 8 Many of these companies offer personal genomic screening for multiple conditions at once. Interested clients may create a profile, pay online with a personal credit card, and receive a sample collection kit in the mail within days (box 1). Other companies offer the option of having the test ordered by, and the results sent to, a doctor. Some companies employ doctors to coordinate test referrals and results. The biological specimen needed depends on the company and may be saliva, a buccal swab, or blood.

    Box 1 Overview of personal genomic screening results

    • Results may be reported to the doctor or directly to the patient (or both)

    • Results may be updated as data from more genome-wide association studies are included in the company’s reports. For this reason, new conditions can appear on a client’s results page or their risk for a particular condition may increase or decrease

    • Results typically include a relative risk or odds ratio compared with the general population and an absolute, or lifetime, risk for disease

    • The results often include the name of the specific single nucleotide polymorphism(s) (SNP(s)) tested and information about the associated gene or functional consequence of the SNP, if known.

    • Access for patients:

      1. Direct to consumer testing companies provide the results directly to the patient, via a web portal or written material (or both)

      2. The web portals are often interactive and allow clients to delve into their genomic results, create personal profiles, participate in research, and e-chat with other clients and sometimes health or scientific experts about their results

    • Access for doctors:

      1. Companies that provide doctors with access to the results do so via an interactive web portal, similar to that accessed by the client, or by a written report

      2. These reports summarise the patient’s relative risk and absolute risk. If a patient is found to have a high risk of a particular disease, information on disease surveillance or prevention measures may be provided

    • A few companies have genetic counsellors or other medical professionals available to work with doctors

    How accurate are genome screening results?

    The accuracy of any type of genetic test can be described by its analytical validity and clinical validity. Analytical validity is the accuracy and reliability of the genotype testing technology itself7 9—in this case, if a person is genotyped for a particular SNP, how reproducible would the result be? Although the analytical validity of personal genomic screening is assumed to be high, it has not been uniformly tested.10 In addition, although many personal genomic screening companies adhere to the recommended standards for laboratory quality, some do not, and this may result in poor analytical validity.

    Clinical validity evaluates how well the test results correlate with disease or the ability of the test to detect or predict the associated disorder (phenotype).9 Clinical validity includes an assessment of sensitivity, specificity, and positive predictive value for each SNP-disease association.

    Many, if not most, genetic tests for mendelian disorders have high clinical validity. In contrast, very few (if any) personal genomic tests for common traits are clinically valid. For example, if a result of personal genomic screening states that a client’s genotype at a certain SNP is associated with a 1.21 relative risk for prostate cancer, and the general population risk of prostate cancer is 16%, then this person’s absolute lifetime risk is 19.36%.3 11 A prospective longitudinal cohort or observational study would be the ideal method to test the validity of this predicted 1.21-fold increase and 19.36% lifetime risk. Such a study could calculate the sensitivity (probability that a person who will develop prostate cancer had an increased risk on the basis of SNP genotyping), specificity (probability that a person who will not develop prostate cancer had an average or decreased risk on the basis of SNP genotyping), and positive predictive value (probability that a person with an increased risk of prostate cancer on the basis of SNP genotyping will actually develop prostate cancer). These traditional applications of sensitivity, specificity, and positive predictive value are further complicated by the ongoing debate in the medical community over what constitutes “high” risk for particular people for a specific disease.

    Importantly, risk predictions based on SNP data derived from retrospective genome-wide association studies may not correlate with those based on prospective studies. In addition, SNP-disease associations derived from studies conducted in one population may not apply to other populations of different ethnic groups. Companies may also use different SNPs for risk prediction, and different genotype frequencies, odds ratios, and average lifetime risks, which can adversely affect clinical validity. This may result in different companies assigning contradictory risks for a condition to the same person. Unfortunately, assessment of clinical validity is not available for most SNP-disease associations currently tested for, because performing prospective trials is costly and time intensive.

    The challenge of assessing clinical validity for SNP-disease associations is further complicated by the use of multiple SNPs to assess risk for a particular disease. In general, SNP associated risks are not additive. Rather than using just one SNP to predict prostate cancer, for example, companies are combining relative risks from many SNPs to predict the risk of prostate cancer. These overall relative risks are calculated using algorithms, which again have not been validated through the necessary studies, so the clinical validity of these results is unknown.

    What should a doctor do with the results of personal genomic screening?

    Doctors are faced with several challenges if a patient presents with direct to consumer personal genomic screening results: deciding which, if any, results are clinically valid and medically actionable for the patient; evaluating the most appropriate medical intervention for the patient, if results are actionable; and providing genetic counselling and interpretation of the results. Doctors should be prepared to discuss the following matters with the patient:

    • Current status of evidence regarding clinical validity and usefulness, which is generally limited.

    • The risk of a false negative for a disease in families with a high risk for a mendelian condition that was not discovered on personal genomic screening.

    • The risk of a false negative when mendelian or monogenic disease testing is carried out but only one or a few mutations are tested (unless sequencing has been performed).

    • Results of common disease risk predications based on multiple SNPs do not account for all genetic contribution to disease and do not consider the environmental or acquired risk factors for disease.

    • Risk prediction based on SNP genotype may not be applicable to people of certain ethnic origins.

    Determining action and intervention

    Insufficient evidence is available for most SNP-disease associations detected by personal genomic screening to warrant medical management being altered on the basis of the results. The clinical utility of a test is the measure of how useful it is clinically—the likelihood of the test results guiding management that will significantly improve health related outcomes, or being useful in medical, personal, or public health decision making.9 Most personal genomic screening results offer limited practical information about what kinds of behavioural changes or medical interventions are appropriate on the basis of an increased (or decreased) risk.1 12 13 Outcomes of behavioural changes must be studied before personal genomic screening results can be incorporated into routine clinical care.1 14

    If a patient presents with results of personal genomic screening that include abnormal results of validated tests for mendelian disease—such as cystic fibrosis or haemochromatosis—the patient should be given genetic counselling and personalised medical management. The result may need to be confirmed by re-testing in a registered clinical genetic diagnostic laboratory.

    An international effort is being made to set standards for evaluation of analytical validity, clinical validity, and clinical utility in genetic testing, but these standards are not yet established.9 Most work has focused on mendelian genetic tests ordered as part of standard of care practice, such as factor V Leiden testing. As regulations shift to deal with genomic testing as well, practical guidelines for assessing the validity and utility of personal genomic screening should become available.15

    Should a doctor recommend personal genomic screening?

    Evidence that doctors should recommend that patients undergo personal genomic screening is lacking. Insufficient data are available on the clinical validity and utility of most SNP-disease associations. Validated clinical risk assessment measures that combine blood chemistry profiles, health behaviours, and family history remain the “gold standard” for predicting disease. A recent prospective cohort study analysing an SNP strongly associated with cardiovascular disease in 22 129 white women found that adding data on this SNP to family history and C reactive protein values did not improve risk classification.16 Another large prospective study found that, although taking account of 11 SNPs associated with the risk of type 2 diabetes slightly improved the ability to predict the development of this disease, clinical and family history risk factors were the predominant indicators of future disease.17

    Thus, there is no evidence that personal genomic screening is medically indicated. If a patient has a personal or family history suggestive of an inherited genetic syndrome, validated genetic testing in the setting of genetic counselling is recommended.18 19 20 21 22

    Potential benefits and harms of personal genomic screening

    We have no direct evidence that personal genomic screening harms the patient, just as we have no direct evidence that it benefits the patient. Potential benefits include increased awareness of disease risk and health status, increased adherence to public health prevention measures, and increased personal responsibility for health.23

    Some potential risks of genetic and genomic testing include increased stress and anxiety, strain on family relationships, risk to privacy, and potential for genetic discrimination. There is also a risk of both false positive results and false negative results. A false negative result has obvious medical implications—the person may not be aware that he or she is still at increased risk of disease. A false positive result may increase personal anxiety, expose a healthy person to unnecessary intervention and treatment, and result in wasted healthcare expenses.1 23 In the case of genetic testing, a false positive result in one family member can affect the emotional wellbeing and medical treatment of relatives.23 Even standard screening tests such as surveillance for cervical cancer or colon cancer produce false positives, which can result in expensive and invasive follow-up. Most tests on a personal genomic screen do not satisfy the recommended criteria for screening tests, such as a high predictive value for the disease in question, effective treatment for the preclinical stage of disease, and evidence that treatment improves patient outcomes (box 2).24

    Box 2 Criteria for establishing and evaluating a screening programme24

    When considering the validity and usefulness of personal genomic analysis to screen for common disease, it is helpful to refer to the general criteria for screening protocols:

    • The disease causes a serious risk for the individual and a burden on society

    • An effective treatment is available for the disease in question

    • The natural course of the disease is understood

    • Screening can detect a presymptomatic or early symptomatic stage of disease, and treatment is useful in the early stage

    • The screening test has high specificity, high sensitivity, and carries a low risk

    • People who test positive on the screening test have access to treatment

    • Screening and follow-up is cost effective when balanced with the burden of disease on the individual and on society

    Doctors and other healthcare providers need to differentiate personal genomic screening from genome-wide association studies performed for research purposes because patients may seek their advice on whether they should participate in such research studies. Patients should be encouraged to participate in well designed genome-wide research studies because these and related prospective observational studies should, in the end, inform patient care.

    Privacy and confidentiality

    Laws protecting patients’ privacy vary between countries. Patients should be advised to read the consent form for any personal genomic screening test carefully and consider who will have access to their data and medical information. Although it is in the company’s best interest to protect its clients’ privacy, client data may be used for outside research. Even if de-identified aggregate, or summary, genotype data are presented from a research analysis or database, it may be possible to identify the presence of an individual in the study population.25

    Conclusion

    Although public interest in direct to consumer personal genomic screening is growing, this type of genomic testing has unknown clinical validity and little clinical utility at this time. Doctors should use traditional risk assessment measures until further research has clarified the medical, financial, ethical, and social implications of personal genomic screening.

    Genomic information will almost certainly be clinically useful, but only after focused research has been carried out. Research should link genomic data with clinical outcome; investigate gene-gene interactions and gene-environment interactions; study the link between pharmacogenomics and clinical outcomes; look at how genomic information relates to family history in risk assessments; investigate how to inform evidence based genomic counselling; and aim to develop educational strategies for the public and healthcare providers. This last aim is important because the explosion in genetic and genomic knowledge in the past decade has left the public and healthcare providers behind.

    Questions for future research

    • What is the analytical validity, clinical validity, and clinic usefulness of personal genomic testing based on single nucleotide polymorphism (SNP) genotyping?

    • In what situation (medical, subset of patients, ancestry, environment, etc) would SNP genotyping improve risk prediction and help target interventions?

    • What information does a consumer need to make an informed decision about personal genomic testing?

    • What information do healthcare providers need to educate patients about the value of personal genomic testing results?

    • What are the healthcare information technology needs and standards for storing and accessing genomic data in a confidential manner?

    • What is the role of government in determining policy on access to and regulation of genetic and genomic testing?

    Some ongoing research studies

    • 1000 Genomes Project: an international consortium that aims to sequence more than 1000 genomes. All genomic data are accessible to the public

    • Coriell Personal Medicine Collaboration (US): this multidisciplinary project aims to enrol 100 000 participants to study the usefulness of using genomic data for health management

    • ClinSeq Project (US): the National Human Genome Research Institute is enrolling participants to undergo genomic sequencing to study the technological, medical, and genetic counselling problems of using genomic sequencing in a clinical setting

    • Scripps Genomic Health Initiative (US): this project will study 10 000 participants to see the effect of receiving personal genomic screening data on decisions about health behaviour over 20 years

    • Multiplex Initiative (US): this study evaluates the attitude of the public and the individual decision making process involved in ordering genomic testing, receiving the results, and making subsequent health behaviour decisions

    • Physician Perspectives on Genomic Testing (US): Cleveland Clinic is surveying doctors to evaluate how much they know about personal genomic testing and its usefulness

    • Risk Assessment Comparison: Family History versus Genomic Screening (US): Cleveland Clinic is comparing risk assessments based on family health history and personal genome testing

    Tips for non-specialists

    • Patients who present with personal genomic screening results should be advised that the data to support medical intervention or altered health behaviours for most single nucleotide polymorphism-disease associations are limited

    • Consider referring patients who are interested in personal genomic screening to clinicians with specialised genetic training for further genetic counselling

    Additional educational resources

    Resources for healthcare professionals
    • Evaluation of Genomic Applications in Practice and Prevention (www.egappreviews.org)—A working group focused on developing a standard process for evaluating genetic and genomic testing

    • National Coalition for Health Education in Genetics (www.nchpeg.org)—Organisation dedicated to promoting genetic education for practising healthcare providers. It aims to establish core competencies in genetics for all healthcare professionals

    Resources for patients
    • Trust it or trash it? (www.trustortrash.org)—A free resource for patients to evaluate the content, quality, and usability of genetic information

    • National Human Genome Research Institute (www.genome.gov/19016903)—Genetics and genomics for patients and the public

    A patient’s perspective

    Susan is a 32 year old woman of Chinese ancestry who is in good health. She is a proactive patient who is interested in a healthy lifestyle and disease prevention. While watching the news one morning, she heard about a new kind of genetic testing that would clarify her risk for more than 20 illnesses. She purchased a personal genomic scan and recently received her results.

    She was relieved to learn that her risk of Crohn’s disease, type 2 diabetes, and heart attack was low. However, she was concerned to find that her risk of breast cancer was higher than in the general population (10%) (fig 1). She visited her doctor to discuss increased surveillance for breast cancer.

    Figure1

    Susan’s personal genomic screening result for breast cancer

    The doctor took a detailed personal and family history to evaluate her risk of several diseases, including breast cancer. Susan had no family history of cancer. When assessing Susan’s personal history, family history, and personal genomic screening results, he considered the following questions:

    • Is this a reputable laboratory?

    • What is the accuracy of this risk estimate for breast cancer in my patient?

    • What is the efficacy for earlier or more aggressive breast cancer screening for my patient on the basis of this result?

    • Will increased screening be paid for by health insurance or the NHS on the basis of this personal genomic screening result?

    • Is this risk estimate applicable to my patient, who is of Chinese ethnicity, when most genome-wide association studies have been performed on white people?

    Susan’s doctor determined that her personal genomic screening results provided insufficient data to recommend increased or earlier breast cancer surveillance. He recommended standard of care breast cancer screening and provided Susan with general education about personal genomic screening.

    Notes

    Cite this as: BMJ 2009;339:b4253

    Footnotes

    • Contributors: EE performed the literature search and wrote the first draft of the manuscript. CE provided the scientific and conceptual oversight and expert input on genetics and genomic medicine throughout the drafting process. CE revised the manuscript draft and ensured that key references were included. CE is guarantor.

    • Funding: CE is the Sondra J Hardis and Stephen R Hardis endowed chair of cancer genomic medicine at the Cleveland Clinic, is a recipient of the Doris Duke distinguished clinical scientist award, and is an American Cancer Society clinical research professor.

    • Competing interests: None declared

    • Provenance and peer review: Commissioned; externally peer reviewed.

    • Patient consent not required (patient anonymised, dead, or hypothetical).

    References