Current issues of ACP Journal Club are published in Annals of Internal Medicine


Genotype testing improved the clinical diagnosis of Alzheimer disease

ACP J Club. 1998 Sep-Oct;129:48. doi:10.7326/ACPJC-1998-129-2-048

Source Citation

Mayeux R, Saunders AM, Shea S, et al. Utility of the apolipoprotein E genotype in the diagnosis of Alzheimer's disease. N Engl J Med. 1998 Feb 19;338:506-11.



Is apolipoprotein E (APOE) genotype testing useful for diagnosing Alzheimer disease (AD) in patients with dementia?


Blinded comparison of the diagnostic accuracy of APOE genotypes and clinical findings with pathologic findings on autopsy.


26 AD centers in the United States.


2188 of 3177 patients (mean age at diagnosis 72 y, 51% women, 97% white) referred for evaluation of dementia.

Description of tests and diagnostic standard

Clinical diagnoses were made according to the criteria from the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association Working Group; the Diagnostic and Statistical Manual of Mental Disorders, third edition and third edition revised; or Cummings and Benson (Butterworth-Heinemann, 1992). APOE genotype diagnoses (i.e., presence of an APOE [egr ]4 allele) were determined from blood or other tissues using amplification by polymerase chain reaction. Neuropathologic diagnoses were made according to standardized criteria from the Consortium to Establish a Registry for Alzheimer's Disease.

Main outcome measures

Sensitivity, specificity, and area under receiver-operating characteristic (ROC) curves.

Main results

1770 patients had pathologically confirmed AD. Test characteristics for clinical diagnosis and APOE [egr ]4 are listed in the Table. When the results of clinical diagnosis were stratified for individual genotypes, the sensitivity and specificity did not change. The addition of APOE [egr ]4 testing for patients who met clinical criteria for AD increased the specificity of clinical diagnosis to 84% but decreased the sensitivity to 61%. The difference in the areas under the ROC curve for clinical diagnosis with and without the APOE genotype was 4% (99% CI 2% to 6%, P < 0.001).


In patients with dementia, clinical diagnosis of AD had high sensitivity and low specificity. In clinically diagnosed AD, the sequential addition of APOE testing increased the specificity of clinical diagnosis but reduced the sensitivity.

Sources of funding: National Institutes of Health; Charles S. Robertson Memorial Gift; Blanchette Hooker Rockefeller Gift.

For correspondence: Dr. R. Mayeux, Columbia University College of Physicians and Surgeons, Gertrude H. Sergievsky Center, 630 West 168th Street, New York, NY 10032, USA. FAX 212-305-2518.

Table. Test characteristics for clinical diagnosis and apolipoprotein E (APOE) testing for Alzheimer disease*

Testing Sensitivity (95% CI) Specificity (CI) +LR -LR ROC curve
Clinical 93% (92 to 94) 55% (50 to 59) 2.07 0.13 0.84
APOE [egr ]4 65% (62 to 67) 68% (64 to 73) 2.03 0.51 0.80

*+LR = likelihood ratio for presence of disease if the test is positive; -LR = likelihood ratio if the test is negative; LRs calculated from data in article. ROC = receiver-operating characteristic.


The well-established association of APOE [egr ]4 and AD is an important advance. The study by Mayeux and colleagues of patients from U.S. AD centers reports a false-positive rate of only 6% of 1142 persons with clinical AD and the [egr ]4 allele. However, 35% of 1643 patients with pathologically confirmed AD without [egr ]4 were presumably misclassified. The cost of improving specificity was 567 false-negative AD diagnoses compared with 124 reclassified true-negative diagnoses, which would have been false-positive diagnoses without the test.

Two questions: First, which costs more—missing a large number of patients with AD or avoiding about 25% as many persons with false-positive results? In practice, missing so many persons who have AD is a problem. In research, avoiding false-positive results to improve precision might be more important. Second, is this population similar to those seen by clinicians? Studies (1-3) have shown that patients in research or referral centers are younger, better educated, more severely demented, and have a higher frequency of [egr ]4 allele (45% in research vs 36% in non-autopsy-confirmed community cases) (3). The population of this study was unusual—79% of patients met clinical criteria for AD, and 80% met pathologic criteria. Most clinicians probably do not see such a population, and if they do, it is likely that a decision analytic test-treatment threshold approach (4) would conclude that with a base-rate of about 80%, "do not test, get on with treatment."

It is unlikely, or at least premature to conclude, that the APOE allele has a place in routine clinical practice. It does, however, have value for research.

Eric B. Larson, MD, MPH
University of WashingtonSeattle, Washington, USA


1. Kokmen E, Ozsarfati Y, Beard CM, O'Brien PC, Rocca WA. Impact of referral bias on clinical and epidemiological studies of Alzheimer's disease. J Clin Epidemiol 1996;49:79-83.

2. Barnhart RL, van Belle G, Edland Sd, et al. Geographically overlapping Alzheimer's disease registries: comparisons and implications. J Geriatr Psychiatry Neurol. 1995;8:203-8.

3. Tsuang D, Kukull W, Sheppard L, et al. Impact of sample selection on APOE epsilon 4 allele frequency: a comparison of two Alzheimer's disease samples. J Am Geriatr Soc. 1996;44:704-7.

4. Sackett DL, Haynes RB, Guyatt GH, Tugwell P, eds. Clinical Epidemiology: A Basic Science for Clinical Medicine. Boston: Little, Brown; 1991:145-8.