- Email: [email protected]
APOE ε4 allele in Alzheimer's and non-Alzheimer's dementias
THE LANCET
Characteristics of Alzheimer’s disease patients with and without ApoE4 allele SIR—Although, as a marker of Alzheimer’s disease (AD), apolipoprotein E4 (ApoE4) has been found to have the ability to discriminate in 90% of cases in case-control studies, in general population studies it is less specific. Examining subtypes of subclinical cognitive disorder, we reported1 the existence of two groups at high risk of AD: one characterised by deterioration in language and attentional functions with relative conservation over time of memory, and the second by impairment in primary and secondary memory without deterioration in language. Subsequent analysis of phenotypes led to the discovery that people with the ApoE ε4 allele were found only in the second group. We have subsequently undertaken a preliminary comparison of incident senile dementia cases in our longitudinal study with and without an ε4 allele. We have reported1 a cross-sectional study of 833 normal elderly people and follow-up over 2 years of 397 subjects identified as having subclinical cognitive impairment. The 2year incidence of dementia in this cohort (18%) is very high. 12 cases of AD diagnosed by DSM-III-based neurological examination have been examined in detail. Of these, four are ApoE 3/4, seven ApoE 3/3, and one ApoE 2/3. Cognitive ability in the year before diagnosis of AD was examined by performance on a computerised examination permitting the estimation of response latencies on four types of task: attention, memory, language, and visuospatial analysis. From the 159 variables derived from this examination eight cognitive subscores may be measured.1 Test scores were converted to standardised scores and the figure shows results for ApoE4 and non-ApoE4 cases of AD on cognitive tests where the scores obtained by each individual in the smaller ApoE4-related group lie outside the 95% confidence intervals for the non-ApoE4 group. The results suggest a tendency for non-ApoE4 cases to have poorer performance (longer response time) than ApoE4 cases on tasks of attention and on word and syntax comprehension, and lower scores on verbal fluency. On the other hand they were found to perform better on tasks of immediate and delayed recall. No difference was found in visuospatial or implicit memory tasks. Regional cerebral blood flow was examined with singlephoton emission computed tomography (SPECT). Data were analysed both on the basis of visual inspection and regions-ofinterest (ROI) quantification. Mean counts of the cerebellar cortex (75% max isolevel) were taken for normalisation of brain activity. SPECT results revealed the expected bilateral temporoparietal flow reductions in ApoE4 cases in all but one subject who had an asymmetrical reduction. Of the five non2·0
ApoE4
1·5
Non-ApoE4
1·0
Normal elderly
0·5 0 –0·5 –1·0 –1·5 –2·0 Visual attention RT Dual task RT
Word comp RT
Verbal fluency Syntax comp RT
Figure: Task performance results RT=response times, comp=comprehension.
960
Delayed recall Immediate recall
ApoE4 cases for whom SPECT data are available, three show bilateral frontal reduction, one unilateral frontal reduction, and the other diffuse unilateral atrophy (this patient is also thought to have vascular disease). From our study it would appear that in a cohort of subjects at risk of developing AD the 2-year incident cases are predominantly non-ApoE4-related. Non-ApoE4 cases of AD show some tendency, at least in this early stage of the disease, to manifest frontal lobe rather than temporoparietal pathology. Is non-ApoE4 AD a separate disease entity? Are cholinesterase inhibitors more efficient in ApoE4 AD because temporal rather than frontal areas are involved? The following shortcomings in our observations must be taken into account: (1) Weiner et al2 have observed frontal lobe reduction in 24% of cases of AD so our results could be due to chance; (2) in our series the diagnosis of AD has not been confirmed by necropsy; and (3) cases of non-ApoE4 AD with temporoparietal pathology (or conversely cases of ApoE4 AD with frontal pathology) may have died before clinical examination. *Karen Ritchie, Pierre-Olivier Kotzki, Jacques Touchon, Jean-Paul Cristol INSERM Cognitive Ageing Research Group, Centre Val D’Aurelle, Parc Euromedecine, 34094 Montpellier, France
1 2
Ritchie K, Leibovici D, Ledésert B, Touchon J. A typology of subclinical senescent cognitive disorder. Br J Psychiatry 1996; 168: 470–76. Weiner MF, Wighton-Benn WH, Risser R, Svetlik D, Tintner R. Xenon-133 SPECT-determined regional cerebral blood flow in Alzheimer’s disease: what is typical? J Neuropsychiatry 1993; 5: 415–18.
APOE ε4 allele in Alzheimer’s and non-Alzheimer’s dementias SIR—Accumulating evidence shows that the apolipoprotein E (APOE) ε4 allele is strongly associated with late-onset familial and sporadic Alzheimer’s disease (AD) as a major susceptibility gene or risk factor that increases the incidence and lowers the age of onset of AD in a dose-dependent manner.1 However, the diagnostic value of APOE genotyping remains enigmatic. Saunders and colleagues (July 13, p 90)2 report that the presence of the APOE ε4 allele in demented patients is highly specific and predictive for AD. To assist this issue further, we present data on APOE testing from demented patients with necropsy-confirmed diagnoses. We examined 72 cases with AD (mean age 77 years) and 20 cases with non-AD dementias (mean age 72) in a nearly consecutive necropsy series from 1985 to 1994 at the University of Pennsylvania. The non-AD dementias included four patients with Parkinson’s disease and dementia without evidence of AD pathology (PDD), three with dementia lacking distinct histology (DLDH), three with progressive supranuclear palsy (PSP), five with Lewy-body dementia (LBD) and mild co-existing AD pathology, one with amyotrophic lateral sclerosis with dementia, one with Pick’s disease, one with adult polyglucosan-body disease, one with dementia due to hydrocephalus, and one with WernickeKorsakoff syndrome. The neuropathological diagnosis of AD was established by the criteria recommended by the US National Institute on Aging.3 APOE genotyping was done from either frozen or paraffin-embedded brain tissue without knowledge of the diagnosis, as described earlier.4 In our necropsy-confirmed series, the sensitivity of the APOE ε4 allele was 52·8% (38/72), the specificity was 60·0% (12/20), the positive predictive value 82·6% (38/46), and the negative predictive value 26·1% (12/46) (table). Indeed, the seven non-AD cases carrying the APOE 4/3 genotype included
Vol 348 • October 5, 1996
THE LANCET
4/4 3/4 2/4 3/3 2/3 2/2
0·6
Number of cases AD (n=72)
Other dementias (n=20)
7 31 0 26 8 0
1 7 0 10 2 0
Table: Results of APOE genotyping in necropsy-confirmed cases with AD and other dementias
PDD (n=3), LBD (n=2), PSP (n=1), and DLDH (n=1). The APOE 4/4 genotype also was found in one patient with LBD. Our results indicate that the presence of the APOE ε4 allele is not predictive of AD alone since the allele also occurs in other dementias. Despite a limited sampled size for the nonAD dementia group in the present study, these data are in agreement with reports5 suggesting that there are limitations in the use of APOE genotyping as a sole diagnostic test for AD. Shigeru Itabashi, *Hiroyuki Arai, Susumu Higuchi, Hidetada Sasaki, John Q Trojanowski *Department of Geriatric Medicine, Tohoku University School of Medicine, Sendai 980, Japan; Department of Psychiatry, National Institute of Alcoholism, Kurihama National Hospital, Kanagawa; and Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, USA
1
2
3 4
5
Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 1993; 261: 921–23. Saunders AM, Hulette C, Welsh-Bohmer KA, et al. Specificity, sensitivity, and predictive value of apolipoprotein-E genotyping for sporadic Alzheimer’s disease. Lancet 1996; 348: 90–93. Khachaturian ZS. Diagnosis of Alzheimer’s disease. Arch Neurol 1985; 42: 1097–105. Arai H, Higichi S, Muramatsu T, et al. Apolipoprotein E gene in diffuse Lewy body disease with or without co-existing Alzheimer’s disease. Lancet 1994; 344: 1307. National Institute on Aging/Alzheimer’s Association Working Group. Apolipoprotein E genotyping in Alzheimer’s disease. Lancet 1996; 347: 1091–95.
Impact on management of new diagnostic tests in Alzheimer’s disease SIR—According to MEDLINE more than 1400 papers have been written over the past 5 years on the “diagnosis” of “Alzheimer’s disease” (AD). Single-photon emission computed tomography (SPECT), magnetic resonance imaging (MRI), sensitivity of pupil dilation, apolipoprotein E (APOE) genotyping, and cerebrospinal-fluid (CSF) assays for specific proteins are among the latest advances that are strongly advocated as promising diagnostic tests. However, there appears to be a discrepancy between enthusiasm about the potential value of these new tests and conservation of current diagnostic methods in everyday clinical practice. Is the medical profession inert and unduly reluctant to use these innovations, or are there other explanations? In clinical practice diagnostic tests in patients with (suspected) dementia should be selected on the basis of their potential to change management. Discussing with patients or caregivers a diagnosis with such devastating consequences as in AD requires high levels of diagnostic certainty. Under the assumption that caregivers explicitly state they want to know the diagnosis, estimates for the minimum level of certainty for discussing a diagnosis of AD varied from 0·75 to 0·90 among eight of our colleagues. These estimates for a “counsellingthreshold probability”, analogous to “treatment-threshold” for curable disease,1 are within the range of accuracy that can be obtained by careful application of current diagnostic criteria mainly relying on clinical examination and use of ancillary investigations only to exclude other causes of
Vol 348 • October 5, 1996
124
Relative frequency
APOE genotype
0·5 102 0·4 0·3 0·2 0·1
11 0
0 Technical capacity
Diagnostic accuracy
Diagnostic Therapeutic impact impact
0 Patient outcome
Figure: Histogram showing frequency distribution of 237 studies (1991–95) into SPECT, MRI, CSF, APOE genotyping, or pupil dilation in AD across levels of test evaluation3 Numbers above bars indicate absolute number of studies.
dementia.2 Thus, at present, there is little to gain from any new diagnostic technology for AD. With the framework proposed by Kent and Larson for assessing efficacy of new diagnostic methods,3 we did a 199195 MEDLINE search for original studies published in English on the diagnostic use of SPECT, MRI, CSF, APOE, and pupil dilation in dementia. Case-series without controls were classified as studies of technical capacity, as were studies only including healthy controls and those with fewer than ten patients because the resulting confidence intervals are too wide to be useful. Reporting of measures such as sensitivity, specificity, or receiver-operator-characteristics was typical for studies of diagnostic accuracy. For classification as a study of diagnostic impact a judgment on a test’s accuracy compared with existing alternatives was required. If the focus of a study was related to changes in treatment, including counselling decisions, a study qualified as one with therapeutic impact. Assessment of the effects on the functional status of patients or caregiver were required for classification at the highest level of the framework as study of patients’ outcomes.3 Studies with equivocal characteristics were allowed the benefit of the doubt and were classified at a higher level. The distribution of diagnostic studies in AD is extremely skewed according to this analysis (figure). There were no studies at levels of clinical efficacy relating to therapeutic (or counselling) impact or to patients’ and caregivers’ outcomes. New diagnostic tests were hardly ever measured by their potential to increase diagnostic confidence in relation to information available from the clinical examination or from other ancillary investigations. The extreme focus on lower levels of test evaluation may explain the repetitive cultivation of high hopes in this field. However, testimonial views on the clinical utility of a new diagnostic method are simply inappropriate if they are based exclusively on research into its technical capacity or diagnostic accuracy. Methods are needed to increase the value of diagnostic studies for clinical medicine by measuring impact of testing at the level of diagnostic confidence in physicians, and ultimately, at the level of patients’ (or caregivers’) health status, satisfaction, or burden. Many resources for research could be saved if studies of new diagnostic technologies for AD included these assessments in earlier stages of test evaluation. *W A van Gool, H van Crevel Department of Neurology, Academic Medical Center, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, Netherlands
1 2 3
Pauker SG, Kassirer JP. Therapeutic decision making: a cost-benefit analysis. N Engl J Med 1975; 293: 229–34. Corey-Bloom J, Thal LJ, Galasko D, et al. Diagnosis and evaluation of dementia. Neurology 1995; 45: 211–18. Kent DL, Larson EB. Disease, level of impact, and quality of research methods: three dimensions of clinical efficacy assessment applied to magnetic resonance imaging. Invest Radiol 1992; 27: 245–54.
961
Characteristics of Alzheimer’s disease patients with and without ApoE4 allele SIR—Although, as a marker of Alzheimer’s disease (AD), apolipoprotein E4 (ApoE4) has been found to have the ability to discriminate in 90% of cases in case-control studies, in general population studies it is less specific. Examining subtypes of subclinical cognitive disorder, we reported1 the existence of two groups at high risk of AD: one characterised by deterioration in language and attentional functions with relative conservation over time of memory, and the second by impairment in primary and secondary memory without deterioration in language. Subsequent analysis of phenotypes led to the discovery that people with the ApoE ε4 allele were found only in the second group. We have subsequently undertaken a preliminary comparison of incident senile dementia cases in our longitudinal study with and without an ε4 allele. We have reported1 a cross-sectional study of 833 normal elderly people and follow-up over 2 years of 397 subjects identified as having subclinical cognitive impairment. The 2year incidence of dementia in this cohort (18%) is very high. 12 cases of AD diagnosed by DSM-III-based neurological examination have been examined in detail. Of these, four are ApoE 3/4, seven ApoE 3/3, and one ApoE 2/3. Cognitive ability in the year before diagnosis of AD was examined by performance on a computerised examination permitting the estimation of response latencies on four types of task: attention, memory, language, and visuospatial analysis. From the 159 variables derived from this examination eight cognitive subscores may be measured.1 Test scores were converted to standardised scores and the figure shows results for ApoE4 and non-ApoE4 cases of AD on cognitive tests where the scores obtained by each individual in the smaller ApoE4-related group lie outside the 95% confidence intervals for the non-ApoE4 group. The results suggest a tendency for non-ApoE4 cases to have poorer performance (longer response time) than ApoE4 cases on tasks of attention and on word and syntax comprehension, and lower scores on verbal fluency. On the other hand they were found to perform better on tasks of immediate and delayed recall. No difference was found in visuospatial or implicit memory tasks. Regional cerebral blood flow was examined with singlephoton emission computed tomography (SPECT). Data were analysed both on the basis of visual inspection and regions-ofinterest (ROI) quantification. Mean counts of the cerebellar cortex (75% max isolevel) were taken for normalisation of brain activity. SPECT results revealed the expected bilateral temporoparietal flow reductions in ApoE4 cases in all but one subject who had an asymmetrical reduction. Of the five non2·0
ApoE4
1·5
Non-ApoE4
1·0
Normal elderly
0·5 0 –0·5 –1·0 –1·5 –2·0 Visual attention RT Dual task RT
Word comp RT
Verbal fluency Syntax comp RT
Figure: Task performance results RT=response times, comp=comprehension.
960
Delayed recall Immediate recall
ApoE4 cases for whom SPECT data are available, three show bilateral frontal reduction, one unilateral frontal reduction, and the other diffuse unilateral atrophy (this patient is also thought to have vascular disease). From our study it would appear that in a cohort of subjects at risk of developing AD the 2-year incident cases are predominantly non-ApoE4-related. Non-ApoE4 cases of AD show some tendency, at least in this early stage of the disease, to manifest frontal lobe rather than temporoparietal pathology. Is non-ApoE4 AD a separate disease entity? Are cholinesterase inhibitors more efficient in ApoE4 AD because temporal rather than frontal areas are involved? The following shortcomings in our observations must be taken into account: (1) Weiner et al2 have observed frontal lobe reduction in 24% of cases of AD so our results could be due to chance; (2) in our series the diagnosis of AD has not been confirmed by necropsy; and (3) cases of non-ApoE4 AD with temporoparietal pathology (or conversely cases of ApoE4 AD with frontal pathology) may have died before clinical examination. *Karen Ritchie, Pierre-Olivier Kotzki, Jacques Touchon, Jean-Paul Cristol INSERM Cognitive Ageing Research Group, Centre Val D’Aurelle, Parc Euromedecine, 34094 Montpellier, France
1 2
Ritchie K, Leibovici D, Ledésert B, Touchon J. A typology of subclinical senescent cognitive disorder. Br J Psychiatry 1996; 168: 470–76. Weiner MF, Wighton-Benn WH, Risser R, Svetlik D, Tintner R. Xenon-133 SPECT-determined regional cerebral blood flow in Alzheimer’s disease: what is typical? J Neuropsychiatry 1993; 5: 415–18.
APOE ε4 allele in Alzheimer’s and non-Alzheimer’s dementias SIR—Accumulating evidence shows that the apolipoprotein E (APOE) ε4 allele is strongly associated with late-onset familial and sporadic Alzheimer’s disease (AD) as a major susceptibility gene or risk factor that increases the incidence and lowers the age of onset of AD in a dose-dependent manner.1 However, the diagnostic value of APOE genotyping remains enigmatic. Saunders and colleagues (July 13, p 90)2 report that the presence of the APOE ε4 allele in demented patients is highly specific and predictive for AD. To assist this issue further, we present data on APOE testing from demented patients with necropsy-confirmed diagnoses. We examined 72 cases with AD (mean age 77 years) and 20 cases with non-AD dementias (mean age 72) in a nearly consecutive necropsy series from 1985 to 1994 at the University of Pennsylvania. The non-AD dementias included four patients with Parkinson’s disease and dementia without evidence of AD pathology (PDD), three with dementia lacking distinct histology (DLDH), three with progressive supranuclear palsy (PSP), five with Lewy-body dementia (LBD) and mild co-existing AD pathology, one with amyotrophic lateral sclerosis with dementia, one with Pick’s disease, one with adult polyglucosan-body disease, one with dementia due to hydrocephalus, and one with WernickeKorsakoff syndrome. The neuropathological diagnosis of AD was established by the criteria recommended by the US National Institute on Aging.3 APOE genotyping was done from either frozen or paraffin-embedded brain tissue without knowledge of the diagnosis, as described earlier.4 In our necropsy-confirmed series, the sensitivity of the APOE ε4 allele was 52·8% (38/72), the specificity was 60·0% (12/20), the positive predictive value 82·6% (38/46), and the negative predictive value 26·1% (12/46) (table). Indeed, the seven non-AD cases carrying the APOE 4/3 genotype included
Vol 348 • October 5, 1996
THE LANCET
4/4 3/4 2/4 3/3 2/3 2/2
0·6
Number of cases AD (n=72)
Other dementias (n=20)
7 31 0 26 8 0
1 7 0 10 2 0
Table: Results of APOE genotyping in necropsy-confirmed cases with AD and other dementias
PDD (n=3), LBD (n=2), PSP (n=1), and DLDH (n=1). The APOE 4/4 genotype also was found in one patient with LBD. Our results indicate that the presence of the APOE ε4 allele is not predictive of AD alone since the allele also occurs in other dementias. Despite a limited sampled size for the nonAD dementia group in the present study, these data are in agreement with reports5 suggesting that there are limitations in the use of APOE genotyping as a sole diagnostic test for AD. Shigeru Itabashi, *Hiroyuki Arai, Susumu Higuchi, Hidetada Sasaki, John Q Trojanowski *Department of Geriatric Medicine, Tohoku University School of Medicine, Sendai 980, Japan; Department of Psychiatry, National Institute of Alcoholism, Kurihama National Hospital, Kanagawa; and Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, USA
1
2
3 4
5
Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 1993; 261: 921–23. Saunders AM, Hulette C, Welsh-Bohmer KA, et al. Specificity, sensitivity, and predictive value of apolipoprotein-E genotyping for sporadic Alzheimer’s disease. Lancet 1996; 348: 90–93. Khachaturian ZS. Diagnosis of Alzheimer’s disease. Arch Neurol 1985; 42: 1097–105. Arai H, Higichi S, Muramatsu T, et al. Apolipoprotein E gene in diffuse Lewy body disease with or without co-existing Alzheimer’s disease. Lancet 1994; 344: 1307. National Institute on Aging/Alzheimer’s Association Working Group. Apolipoprotein E genotyping in Alzheimer’s disease. Lancet 1996; 347: 1091–95.
Impact on management of new diagnostic tests in Alzheimer’s disease SIR—According to MEDLINE more than 1400 papers have been written over the past 5 years on the “diagnosis” of “Alzheimer’s disease” (AD). Single-photon emission computed tomography (SPECT), magnetic resonance imaging (MRI), sensitivity of pupil dilation, apolipoprotein E (APOE) genotyping, and cerebrospinal-fluid (CSF) assays for specific proteins are among the latest advances that are strongly advocated as promising diagnostic tests. However, there appears to be a discrepancy between enthusiasm about the potential value of these new tests and conservation of current diagnostic methods in everyday clinical practice. Is the medical profession inert and unduly reluctant to use these innovations, or are there other explanations? In clinical practice diagnostic tests in patients with (suspected) dementia should be selected on the basis of their potential to change management. Discussing with patients or caregivers a diagnosis with such devastating consequences as in AD requires high levels of diagnostic certainty. Under the assumption that caregivers explicitly state they want to know the diagnosis, estimates for the minimum level of certainty for discussing a diagnosis of AD varied from 0·75 to 0·90 among eight of our colleagues. These estimates for a “counsellingthreshold probability”, analogous to “treatment-threshold” for curable disease,1 are within the range of accuracy that can be obtained by careful application of current diagnostic criteria mainly relying on clinical examination and use of ancillary investigations only to exclude other causes of
Vol 348 • October 5, 1996
124
Relative frequency
APOE genotype
0·5 102 0·4 0·3 0·2 0·1
11 0
0 Technical capacity
Diagnostic accuracy
Diagnostic Therapeutic impact impact
0 Patient outcome
Figure: Histogram showing frequency distribution of 237 studies (1991–95) into SPECT, MRI, CSF, APOE genotyping, or pupil dilation in AD across levels of test evaluation3 Numbers above bars indicate absolute number of studies.
dementia.2 Thus, at present, there is little to gain from any new diagnostic technology for AD. With the framework proposed by Kent and Larson for assessing efficacy of new diagnostic methods,3 we did a 199195 MEDLINE search for original studies published in English on the diagnostic use of SPECT, MRI, CSF, APOE, and pupil dilation in dementia. Case-series without controls were classified as studies of technical capacity, as were studies only including healthy controls and those with fewer than ten patients because the resulting confidence intervals are too wide to be useful. Reporting of measures such as sensitivity, specificity, or receiver-operator-characteristics was typical for studies of diagnostic accuracy. For classification as a study of diagnostic impact a judgment on a test’s accuracy compared with existing alternatives was required. If the focus of a study was related to changes in treatment, including counselling decisions, a study qualified as one with therapeutic impact. Assessment of the effects on the functional status of patients or caregiver were required for classification at the highest level of the framework as study of patients’ outcomes.3 Studies with equivocal characteristics were allowed the benefit of the doubt and were classified at a higher level. The distribution of diagnostic studies in AD is extremely skewed according to this analysis (figure). There were no studies at levels of clinical efficacy relating to therapeutic (or counselling) impact or to patients’ and caregivers’ outcomes. New diagnostic tests were hardly ever measured by their potential to increase diagnostic confidence in relation to information available from the clinical examination or from other ancillary investigations. The extreme focus on lower levels of test evaluation may explain the repetitive cultivation of high hopes in this field. However, testimonial views on the clinical utility of a new diagnostic method are simply inappropriate if they are based exclusively on research into its technical capacity or diagnostic accuracy. Methods are needed to increase the value of diagnostic studies for clinical medicine by measuring impact of testing at the level of diagnostic confidence in physicians, and ultimately, at the level of patients’ (or caregivers’) health status, satisfaction, or burden. Many resources for research could be saved if studies of new diagnostic technologies for AD included these assessments in earlier stages of test evaluation. *W A van Gool, H van Crevel Department of Neurology, Academic Medical Center, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, Netherlands
1 2 3
Pauker SG, Kassirer JP. Therapeutic decision making: a cost-benefit analysis. N Engl J Med 1975; 293: 229–34. Corey-Bloom J, Thal LJ, Galasko D, et al. Diagnosis and evaluation of dementia. Neurology 1995; 45: 211–18. Kent DL, Larson EB. Disease, level of impact, and quality of research methods: three dimensions of clinical efficacy assessment applied to magnetic resonance imaging. Invest Radiol 1992; 27: 245–54.
961