To the Editor: In their review of the evaluation of dementia(Aug. 1 issue),1 Geldmacher and Whitehouse expressed their "opinion. . . that apolipoprotein E testing will not be ready for clinicaluse as either a predictive or a diagnostic test until betterpopulation-based estimates of risk and appropriate genetic counselingare available." There is virtual unanimity regarding the lackof support for apolipoprotein E genotyping in the predictionof asymptomatic persons — whether or not they have relativeswith Alzheimer's disease.2 No apolipoprotein E genotype predictsaccurately whether or when Alzheimer's disease will develop.However, the rationale for using apolipoprotein E genotypingfor diagnosis is quite different.2,3
Four separate studies have found that the positive predictivevalue of carrying an apolipoprotein E 4 allele is 96 to 100percent, despite different strategies of ascertainment3,4,5(and Welsh-Bohmer KA, et al.: unpublished data). More autopsy-confirmedseries are needed to establish the confidence limits, but anydiagnostic test that can accurately predict the eventual neuropathologicdiagnosis with a positive predictive value of more than 95 percentcannot be ignored.
Given that diagnostic resources are no longer considered unlimited,can we justify not using a highly specific, relatively inexpensive,diagnostic adjunct that has useful positive predictive valuefor more than half the patients? The current routine evaluationof dementia, especially the use of low-yield imaging studies,is designed to find uncommon causes. Now, with genetic information,it may be time to modify the paradigm. A standard diagnosticevaluation consisting of a structured history taking, reliableand complete neurologic examination, blood chemistry evaluation,thyroid-function tests, and the like, along with apolipoproteinE genotyping would seem to be appropriate initially. Computedtomography and magnetic resonance imaging could be consideredon the basis of the history and physical examination, ratherthan scheduled routinely for every patient.
Allen D. Roses, M.D. Ann M. Saunders, Ph.D. Duke UniversityMedical Center Durham, NC 27710
Editor's Note: Duke University Medical Center has licensed apatent to Athena Neurosciences for diagnostic applications ofapolipoprotein E genotyping. Drs. Roses and Saunders are consultantsto Athena Neurosciences, which offers apolipoprotein E genotypingonly to patients certified by their physicians to be cognitivelyimpaired.
References
Geldmacher DS, Whitehouse PJ. Evaluation of dementia. N Engl J Med 1996;335:330-336. [Free Full Text]
National Institute on Aging/Alzheimer's Association Working Group. Apolipoprotein E genotyping in Alzheimer's disease. Lancet 1996;347:1091-1095. [Medline]
Saunders AM, Hulette O, Welsh-Bohmer KA, et al. Specificity, sensitivity, and predictive value of apolipoprotein-E genotyping for sporadic Alzheimer's disease. Lancet 1996;348:90-93. [CrossRef][Medline]
Kakulas BA, Wilton SD, Fabian VA, Jones TM. Apolipoprotein-E genotyping in diagnosis of Alzheimer's disease. Lancet 1996;348:483-483. [Medline]
Smith AD, Jobst KA, Johnston C, Joachim C, Nagy Z. Apolipoprotein-E genotyping in diagnosis of Alzheimer's disease. Lancet 1996;348:483-484.
To the Editor: The excellent, succinct review of dementia byDrs. Geldmacher and Whitehouse omitted one important conditionassociated with dementia — alcoholism. Some cognitivedysfunction can be measured in 50 to 70 percent of alcoholics,and 10 percent of alcoholics have severe cognitive impairmentrequiring institutional care.1 Alcohol-related dementia wasidentified in 24 percent of cognitively impaired residents ofa long-term care facility, ranking second only to Alzheimer'sdisease as a cause of dementia.2 The cause of dementia in alcoholicsis multifactorial.3 Wernicke's encephalopathy, which is causedby thiamine deficiency, is a relatively common disorder in alcoholicsand results in both selective memory loss (Korsakoff's syndrome)and global cognitive changes that may be indistinguishable fromthose due to Alzheimer's disease.4 Autopsy studies indicatethat the diagnosis of acute and chronic Wernicke's encephalopathyis frequently overlooked.3 Alcoholics are subject to a varietyof nutritional deficiencies, traumatic brain injury, hepatocerebraldegeneration, alcohol neurotoxicity, and Marchiafava–Bignamisyndrome; alone or in combination, these may contribute to dementia.A history of alcoholism should be carefully elicited from patientswith cognitive dysfunction. Abstinence from alcohol and nutritionaltherapy may lead in some patients to stabilization or even improvementof cognitive function.
Michael E. Charness, M.D. Veterans Affairs Medical Center WestRoxbury, MA 02132
References
Eckardt MJ, Martin PR. Clinical assessment of cognition in alcoholism. Alcohol Clin Exp Res 1986;10:123-127. [Medline]
Carlen PL, McAndrews MP, Weiss RT, et al. Alcohol-related dementia in the institutionalized elderly. Alcohol Clin Exp Res 1994;18:1330-1334. [CrossRef][Medline]
Charness ME, Simon RP, Greenberg DA. Ethanol and the nervous system. N Engl J Med 1989;321:442-454. [Medline]
Torvik A, Lindboe CF, Rogde S. Brain lesions in alcoholics: a neuropathological study with clinical correlations. J Neurol Sci 1982;56:233-248. [CrossRef][Medline]
To the Editor: We were surprised that functional imaging ofthe brain with single-photon-emission computed tomography (SPECT)was not discussed in the review of the evaluation of dementia.SPECT has been used for over a decade and typically shows decreasedperfusion in the bilateral posterior temporoparietal cortexin Alzheimer's disease, frontal hypoperfusion in Pick's disease,multiple asymmetric perfusion defects in vascular dementia,multifocal abnormalities in AIDS dementia complex, and a normalpattern in pseudo-dementia (depression).1 Although not entirelyspecific, a finding of bilateral posterior temporoparietal hypoperfusionin a patient with cognitive deficits is highly suggestive ofAlzheimer's disease. In mild Alzheimer's disease (patients witha score of 20 to 24 on the Mini–Mental State Examination),SPECT still has a sensitivity of more than 80 percent. Conversely,a normal pattern of perfusion in patients with memory loss hasa high negative predictive value and, in the appropriate clinicalsettings, favors the diagnosis of depressive disorder ratherthan early Alzheimer's disease.2 On the basis of the evidencefrom several well-designed clinical studies and clinicopathologicalseries, SPECT is considered an established technique to supportthe clinical diagnosis of Alzheimer's disease.3
Michel Rubinstein, M.D. Roger Denays, M.D. Andréa Collier,M.D. Association Hospitalière Etterbeek-Ixelles B-1050Brussels, Belgium
References
Holman BL, Devous MD. Functional brain SPECT: the emergence of a powerful clinical method. J Nucl Med 1992;33:1888-1904. [Free Full Text]
Holman BL, Johnson KA, Gerada B, Carvalho PA, Satlin A. The scintigraphic appearance of Alzheimer's disease: a prospective study using technetium-99m-HMPAO SPECT. J Nucl Med 1992;33:181-185. [Erratum, J Nucl Med 1992;33:484.] [Free Full Text]
Assessment of brain SPECT: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 1996;46:278-285. [Free Full Text]
To the Editor: In their excellent review on the evaluation ofdementia, Geldmacher and Whitehouse suggest that the diagnosisof normal-pressure hydrocephalus might be confirmed by radioisotope-diffusionstudies. Although this test may confirm the presence of communicatinghydrocephalus, it will not assist clinicians in managing casesof presumed normal-pressure hydrocephalus. The value of isotopecisternography for predicting the postsurgical clinical outcomeis no higher than that of combined clinical and CT data.1 Sinceisotope cisternography is invasive, expensive, and time-consuming,performing this test in patients with presumed normal-pressurehydrocephalus should be discouraged.
Jan Vanneste, M.D. Sint Lucas Andreas Ziekenhuis 1061 AF Amsterdam,the Netherlands
References
Vanneste J, Augustijn P, Davies GAG, Dirven C, Tan WF. Normal-pressure hydrocephalus: is cisternography still useful in selecting patients for a shunt? Arch Neurol 1992;49:366-370. [Abstract]
To the Editor: Geldmacher and Whitehouse state that the annualincidence of Creutzfeldt–Jakob disease in the United Statesis 1 per 167,000 population. This is the incidence for a subgroupof the population that is 70 to 74 years of age. The study towhich the authors refer, an epidemiologic analysis of nationalmortality data by Holman et al.,1 reports an "average annualage-adjusted mortality rate of 0.9 deaths per million persons,a rate consistent with published estimates of the crude incidenceworldwide of 1 case per million persons." The study by Masterset al.2to which Holman et al. refer states, "In the UnitedStates, the average annual mortality rate is used by epidemiologiststo estimate annual disease incidence due to the rapidly fatalcourse of the disease for most patients with CJD [Creutzfeldt–Jakobdisease]."
Elizabeth W. Lazzara, M.D. College of Physicians and Surgeonsof Columbia University New York, NY 10032
References
Holman RC, Khan AS, Kent J, Strine TW, Schonberger LB. Epidemiology of Creutzfeldt-Jakob disease in the United States, 1979-1990: analysis of national mortality data. Neuroepidemiology 1995;14:174-181. [CrossRef][Medline]
Masters CL, Harris JO, Gajdusek DC, Gibbs CJ Jr, Bernoulli C, Asher DM. Creutzfeldt-Jakob disease: patterns of worldwide occurrence and the significance of familial and sporadic clustering. Ann Neurol 1979;5:177-188. [CrossRef][Medline]
The authors reply:
To the Editor: Charness correctly notes alcoholism as an importantpotential contributor to dementia. We agree that it deservesemphasis, and we presumed, perhaps unwisely, that practitionersroutinely assess alcohol use by their patients.
Rubinstein et al. identify the strong points of SPECT in prototypicalcases, but its value in atypical, and therefore diagnosticallychallenging, cases of dementia is less clear. We agree withVanneste that although some clinicians continue to use isotopecisternography, its cost effectiveness is questionable.
Lazzara correctly notes that we misstate the incidence of Creutzfeldt–Jakobdisease. The overall incidence in the United States is 0.9 permillion. The figure of 1 per 167,000 cited in our article, however,is probably appropriate for the older groups that are frequentlyassessed for dementia.
In response to Roses and Saunders, we appreciate the new informationconcerning apolipoprotein E genotype testing in dementia andrestate our view that in clinical practice asymptomatic personsshould not be tested. They would receive far less informationthan persons with symptoms requiring diagnosis.1 The predictivevalues in selected autopsy populations do not necessarily reflectthose in clinical populations. Knowing the apolipoprotein Estatus does supply more information about the likelihood ofcertain diagnoses, but the usefulness of the information inthe routine evaluation of patients with dementia is uncertain.
We agree with Roses and Saunders that imaging studies mightbe conducted more selectively, but they do not specify whereapolipoprotein E fits in their decision tree. It remains uncertainwhether apolipoprotein E testing will really alter our use ofimaging and other diagnostic tests and save money.
Not addressed by Roses and Saunders is another major issue raisedby the literature2 as to how people use complex diagnostic andgenetic-risk data. Current efforts to teach physicians, patients,and their families how to interpret the results of apolipoproteinE genotyping are inadequate. If they were more adequate, wethink that fewer persons would opt to be tested.
Clinical-research use of apolipoprotein E, building on the excitingbasic-science discoveries, should be encouraged, but at presentclinical use should be highly discretionary. In the future,apolipoprotein E genotyping may prove useful in decisions aboutbrain biopsy and about shunting in cases of possible normal-pressurehydrocephalus, and in other situations,3 but it does requirefurther study. Fortunately, many agree with us, and such studiesare under way.4
Peter J. Whitehouse, M.D., Ph.D. David S. Geldmacher, M.D. UniversityHospitals of Cleveland Cleveland, OH 44106
References
Breitner JC. APOE genotyping and Alzheimer's disease. Lancet 1996;347:1184-1185. [Medline]
National Institute on Aging/Alzheimer's Association Working Group. Apolipoprotein E genotyping in Alzheimer's disease. Lancet 1996;347:1091-1095.
Andrews LB, Fullarton JE, Holtzman NA, Motulsky AG, eds. Assessing genetic risks: implications for health and social policy. Washington, D.C.: National Academy Press, 1994.
American College of Medical Genetics/American Society of Human Genetics Working Group on APOE and Alzheimer Disease. Statement on use of apolipoprotein E testing for Alzheimer disease. JAMA 1995;274:1627-1629. [Abstract]
Correction
to
Geldmacher and Whitehouse, 
4 allele is 96 to 100 percent, despite different strategies of ascertainment3,4,5 (and Welsh-Bohmer KA, et al.: unpublished data). More autopsy-confirmed series are needed to establish the confidence limits, but any diagnostic test that can accurately predict the eventual neuropathologic diagnosis with a positive predictive value of more than 95 percent cannot be ignored. 
Previous
