Apolipoprotein E genotype: utility in clinical practice in Alzheimer's disease.

Abstract

In this issue, Dr. Hyman has presented a reasoned view of the epidemiology and clinical literature supporting the association of apolipoprotein E (APOE) alleles and the risk of Alzheimer's disease (AD). However, tests are used in clinical practice for individuals, not populations. Odds ratios are not particularly helpful to either the patient or concerned relatives in the early stages of dementia.1–3 The critical question directly relevant to the patient is whether APOE (or any other test) is useful as a diagnostic test early in the work-up. A secondary clinical concern is the ramifications for other relatives who are asymptomatic. This ties into the analysis of APOE in prediction of asymptomatic individuals. The literature is unanimous that there is little practical clinical use of APOE genotyping for prediction of risk of normal individuals.1–5 Whether the person is a relative of a patient or not, prediction based on APOE genotype is not accurate, providing only relative, non-quantitative risk estimates at the present time. Many reported studies calculate risk using “APOE-ε4 positive” individuals, not distinguishing between each APOE genotype. The age of onset and risk distributions are different for ε4/ε4, ε3/ε4. and ε2/ε4, so much of the combined data are even more approximate.1–7 However, without repeating this recent and extensive literature, it is quite accurate to simply state the lack of applicability for individual risk. Even with an ε4/ε4 genotype at age 50 years, it is not possible to predict whether an individual will develop AD at age 55 years or more than 90 years. The same could be said for each APOE genotype, except the relative risks are less at each age. There are, however, very few ε4/ε4 individuals (2 to 3% of the population at age 50 years) reported who are cognitively intact over the age of 90 years, despite extrapolations from epidemiology studies with very small populations older than age 80 years.8–10 In Dr. Hyman's own reported series of patients who were 90+ years old, there were no (0/28) normal ε4/ε4 homozygous individuals, and the ε4 allele frequency was reduced to 0.036 from 0.15.11 All of the remaining normal ε4 carriers (less than 25% of the normal proportion of ε4 carriers) had the ε3/ε4 genotype. [see also p 0000] Dr. Hyman states that “the critical issue is, among individuals with dementia, what is the chance that someone who is APOE ε4 positive has AD rather than another type of dementia.” He then returns to clinical series and epidemiology studies to discuss the “chance that someone with AD has inherited” either one or two copies of ε4. Both the clinical and epidemiological series have an uncertain and unknown diagnostic accuracy for AD and cannot be assumed to represent AD because there is a “4 to 35% error rate” in diagnosis.4 My viewpoint is that the “critical issue” is the accuracy of information that can be given to the patient early in the clinical diagnostic work-up. Patients are encountered in clinical practice one by one, and each requires accurate information early in their course of disease, not at autopsy. The important data for clinical applicability is the positive predictive value of a particular APOE genotype for a symptomatic patient early enough in the disease to allow the patient to participate in decisions concerning his or her own care and planning. Does APOE genotyping help establish a diagnosis of AD disease? If so, for what proportion of diagnostic evaluations is the test helpful?1, 2 There are now three studies that were not available to Dr. Hyman at the time he wrote his article.12–14 All three studied groups of subjects were ascertained prospectively and had autopsy studies, the current gold standard for accurate diagnosis. Each asked the same questions: What was the specificity of APOE genotyping for identifying AD patients, what was the positive predictive value for an individual with a designated APOE genotype, and for what proportion of individuals was APOE information valuable (sensitivity)? The first series of 67 patients was diagnosed clinically as probable AD in our Center, followed during life, and had subsequent autopsy confirmation.12 They were indistinguishable from the patient group from our Center with probable AD who had no autopsies. The diagnostic accuracy of AD was 85% (57/67), similar to many previous series in the literature. The remaining 15% of patients did not have AD, but had another dementia resembling AD. None of these 10 patients had an ε4 allele. The ε4 allele was carried by 75% (43/57) of AD patients, with or without associated Lewy bodies. Thus, for 43 of 67 patients, the finding of an ε4 allele during the early diagnostic work-up would have predicted AD with 100% accuracy. The non-ε4 group contains 14 AD patients and 10 non-AD patients. Thus the test would have been useful for positive diagnosis of AD in 64% of the patients, and of no diagnostic value in 36%. A second independent series examined the APOE using paraffin-embedded tissue blocks from subjects identified from 20 Consortium to Establish a Registry for Alzheimer's Disease (CERAD) centers who participated prospectively in clinical and subsequent neuropathological studies.13 This series consisted of patients identified early in their course of dementia and included age-matched normal individuals who participated as controls. The CERAD criteria were developed from these patients. In this series, the final diagnosis of AD was confirmed in 117/134 subjects (87%), of which 89 carried an ε4 allele. Of the 17 subjects in the non-AD group, only one ε2/ε4 individual did not have CERAD criteria for probable or definite AD. This subject was the only non-AD patient in the autopsy series with a primary CERAD neuropathological diagnosis of “possible AD.” In this series, the specificity (16/17) is 94%, but the positive predictive value is 99% (89/90). There will be more data available from the CERAD study as several participating centers have yet to submit tissue blocks from their autopsied patients (Welsh-Bohmer et al., manuscript submitted). The third series represents the experience of the Department of Neuropathology at the Royal Perth Hospital (RPH) in Western Australia.15 RPH provides the only Neuropathology Department for a very large area, including foreign countries in the SE Asia and the Indian Ocean. Most of the series were older people referred from nursing facilities or to the Medical Examiner. They were selected for the study because their clinical diagnosis was either “dementia” or “AD.” Thus, this is a very heterogeneous group of multiethnic individuals with a paucity of clinical information, in some ways similar to the heterogeneous requests for autopsies from general physicians received at Alzheimer's Disease Centers in the US. In this older series, 81% (54/66) of patients had AD, but only 48% (25/54) of AD patients carried the ε4 allele. None of the 12 patients with a non-AD diagnoses carried an ε4 allele. The positive predictive value was 100%, so the test would have been useful in about half the patients. Similar studies are being performed at multiple institutions, some with very exacting diagnostic criteria like the OPTIMA study in Oxford, and others with less well characterized clinical data. A large ADC Consortium study with thousands of patients is also now in progress. Thus many more autopsy studies of prospectively ascertained subjects will allow the predictive value to be measured with clinically useful confidence limits. There is now very strong support for the use of APOE genotyping as a diagnostic adjunct. After ruling out reversible causes of dementia, APOE genotyping is the first diagnostic test that can differentiate a large proportion of clinical AD patients (64% Duke, 66% CERAD, 38% RPH) with high predictive value. The positive predictive value of particular autosomal dominant AD mutations is similarly high, yet age of onset predictions for individuals are still only estimates. These mutations are rare, representing much less than 1 % of AD patients.16 The real “critical question” is whether an APOE genotype can help in the early diagnostic work-up of cognitively impaired patients, not what proportion of normal people at any age carry a particular APOE allele. Positive answers are beginning to come into the literature. APOE genotyping data will also have a major impact on the standard or “typical” dementia clinical work-ups, especially when cost/benefit analyses are performed on all currently used diagnostic tests. Clinical trials can now be designed to allow more rational interpretations: with the ε4 positive group containing AD patients, and the ε4 negative group containing both AD and non-AD patients. Treatments that are effective for AD would be expected to be observed in both groups, perhaps better in the ε4 positive group since the ε4 negative group contains most of the non-AD patients. A therapy that is effective in the ε4 negative group may well have its maximum benefit in non-AD patients. For the general practitioners without direct access to expensive imaging tests, but with the ability to perform reliable neurological exams, APOE genotyping can have a major effect on the cost of a diagnostic work-up. In addition, the majority of patients and families will benefit from earlier AD diagnosis than is now possible. In the United States, imaging tests are frequently used for routine screening, sometimes in lieu of a routine and reliable neurological examination. In the United Kingdom, where imaging tests are less commonly used for evaluation of dementia, there does not appear to be a significant problem with frequently missed space-occupying lesions that present solely with cognitive impairment, i.e., no headache, seizures, other history or neurological signs. Health service research studies that evaluate the cost and benefit of each test used as part of a dementia diagnostic work-up are also now in progress. There is something to be said for a thorough history and neurological examination. Space-occupying lesions and normal pressure hydrocephalus usually have other historical data and may have informative neurological exams.17 Only 2% of those very rare, anecdotal missed cases would have the e4/e4 genotype. Screening with APOE genotyping for AD is far more cost-effective than screening with multiple imaging studies. A patent has been issued to Duke University in the United States and several foreign countries for the application of APOE genotyping by any method to the diagnosis of AD. Several subsequent patent applications from other individuals and institutions have not been allowed. Athena Neurosciences acquired the exclusive license to the Duke patent to develop diagnostic applications of APOE genotyping. At the present time, Athena Neurosciences limits APOE genotyping to patients with cognitive impairment, certified by the signature of the referring physician who submits the request. Dr. Roses is a Consultant for AD diagnostics for Athena Neurosciences. Scientists at Duke University, including Dr. Roses, the Joseph Bryan Scholars Fund, and other Not-For-Profit entities receive, or will receive, royalties from Athena Neurosciences. STATEMENT OF OWNERSHIP MANAGEMENT AND CIRCULATION (Required by 39 U.S.C. 3685) . Publication Title: JOURNAL OF THE AMERICAN GERIATRICS SOCIETY; 2. Publication no.: 0002-8614 . Filing Date: 10-01-96; 4. Issue Frequency: monthly; 5. No. of issues published annually: 12; 6. Annual subscription price: $124.00. 7. Complete mailing address of known office of publication: 351 West Camden Street, Baltimore, MD 21201-2436. . Complete mailing address of the headquarters or general business offices of the publisher: 351 West Camden Street, Baltimore, MD 21201-2436. . Full names and complete mailing address of publisher, editor, and managing editor: Publisher: Williams & Wilkins, 351 West Camden Street, Baltimore, MD 21201-2436; Editor: Dr. William Applegate, M.D., MPH, Editor, The University of TN, Memphis School of Medicine, 66 N. Pauline, Suite #232, Memphis, TN 38105. Managing Editor: Elizabeth Webb, Editorial Coordinator, The University of TN, Memphis School of Medicine, 66 N. Pauline, Suite #232, Memphis, TN 38105. . Owner: American Geriatrics Society, 770 Lexington Ave. Suite 300, New York, NY 10021 . Known bondholders, mortgagees, and other security holders owning or holding 1 percent or more of total amount of bonds, mortgages, or other securities: None. . Purpose, function, and nonprofit status: Has not changed during preceding 12 months. . Publication Name: JOURNAL OF THE AMERICAN GERIATRICS SOCIETY . Issue Date for circulation data: July, 1996. Vol. 44, #7. . Extent and nature of circulation: Average number of copies each issue during preceding 12 months: (a) Total No. copies (Net Press Run), 10191. (b) Paid and/or requested circulation; (1) Sales through dealers and carriers, street vendors and counter sales (not mailed), 385; (2) Paid or Requested mail subscriptions (include Advertisers Proof Copies/Exchange Copies), 8726. (c) Total paid and/or requested circulation (sum of 15b(1) and 15b(2)), 9111. (d) Free distribution by mail (samples, complimentary, and other free copies), 252. (e) Free distribution outside the mail (carriers or other means), none. (f) Total free distribution (sum of 15d and 15e), 252. (g) Total distribution (Sum of 15c and 15f) 9363. (h) Copies not distributed: (1) Office use, leftovers, spoiled, 828; (2) Return from news agents, none. (i) Total (sum of 15g, 15h(1), and 15h(2)), 10191. Percent Paid and/or Requested Circulation (15c/15g × 100) 97.73%. Actual no. copies of single issue published nearest to filing date: (a) Total no. copies (Net Press Run), 9840. (b) Paid and/or requested circulation; (1) Sales through dealers and carriers, street vendors and counter sales (not mailed), 475; (2) Paid or Requested mail subscriptions (include Advertisers Proof Copies/Exchange Copies), 8415. (c) Total paid and/or requested circulation (sum of 15b(1) and 15b(2)), 8890. (d) Free distribution by mail (samples, complimentary, and other free copies), 231. (e) Free distribution outside the mail (carriers or other means), none. (f) Total free distribution (sum of 15d and 15e), 231. (g) Total distribution (Sum of 15c and 15f), 9121. (h) Copies not distributed: (1) Office use, leftovers, spoiled, 719; (2) Return from news agents, none. (i) Total (sum of 15g, 15h(1), and 15h(2)), 9840. Percent Paid and/or Requested Circulation (15c/15g × 100), 97.47%. . This Statement of Ownership will be printed in the December issue of this publication. . I certify that the statements made by me above are correct and complete. Alma J. Wills Publisher