- Email: [email protected]
Memory and mental status correlates of modified Braak staging
Neurobiology of Aging 20 (1999) 573–579
Memory and mental status correlates of modified Braak staging夞, 夞夞 E. Grobera,*, D. Dicksonb, M.J. Sliwinskia, H. Buschkea, M. Katza, H. Crystala, R.B. Liptona,c a
Department of Neurology, Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, NY, USA b Department of Pathology, Mayo Clinic, Jacksonville, FL, USA c Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, NY, USA Received 27 October 1998; received in revised form 20 July 1999; accepted 18 August 1999
Abstract We assessed the relationships of performance on memory and mental status tests and neuropathologic stage of Alzheimer’s disease as defined by Braak and Braak in 29 patients from a prospective clinicopathologic series. We predicted that memory changes would occur at an earlier Braak stage than mental status changes. Staging was accomplished by matching the topographic distribution of neurofibrillary lesions detected with tau immunocytochemistry to the best fitting diagram published by Braak and Braak. Higher Braak stages were associated with decrements in performance on both memory and mental status tests. As predicted, memory performance declined from stages II to III and mental status did not decline until stages III to IV. The association between memory and Braak stage was unchanged after adjusting for neocortical senile plaques, whereas adjustments for Braak stage eliminated the association between cognitive functioning and amyloid burden. We conclude that Braak staging provides a useful summary of Alzheimer’s disease neuropathology, which is associated with both memory and mental status performance. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Memory; Mental status; Braak stage; Alzheimer’s disease
1. Introduction Neuropathology is the gold standard for diagnosing Alzheimer’s disease (AD). A well-established system for staging the spread of neurofibrillary degeneration in AD was developed by Braak and Braak [6]. Their recent study on the distribution of Alzheimer’s neuropathology by age supports the value of their approach but does not provide clinical information regarding individual cases [11]. Nonetheless, the results of clinicopathological studies establish Braak staging as a useful approach for characterizing AD neuropathology [2,8]. In this study, we evaluated the relationship of Braak stage to performance on memory and mental status tests in
夞 This research was supported by the Einstein Aging Study, National Institute on Aging Grant AG03949. 夞夞 This paper was presented in part at the 49th Annual Meeting of the American Academy of Neurology, Boston, MA, 1997. * Corresponding author. Tel.: ⫹1-914-963-5601; fax: ⫹1-914-9635602. E-mail address: [email protected]
prospectively studied subjects. We predicted that advances in neuropathologic stage would be associated with declines in cognitive tests administered close to the time of death. We also sought to clarify the behavioral correlates at early Braak stages. Because memory decline is one of the earliest neuropsychological manifestations of preclinical and early AD [4,5,15,16,18,20,21,27,31,32], we hypothesized that decrements in memory performance would occur at early stages of pathologic AD. Because mental status tests have good sensitivity for detecting more advanced dementias [30], we predicted that mental status would not decline until later-stage disease. We also sought to determine the extent to which the relationship between cognitive status and Braak stage was modified by amyloid burden [19,23,24,29].
2. Materials and methods 2.1. Subjects We selected subjects for inclusion in this study from the clinical-pathologic database of the Einstein Aging Study
0197-4580/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 1 9 7 - 4 5 8 0 ( 9 9 ) 0 0 0 6 3 - 9
574
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Table 1 Characteristics of the subjects by Braak stagea Braak stage N Age at death Gender Education (years) Post interval (years) BIMC score Memory-z score SUM SP a
0 3 91.6 (1.7) 1 F; 2 M 11.3 2.7 (1.4) 4.3 (1.5) 1.4 (.4) 0.0
1 3 88.2 (.69) 3F 14.0 2.0 (1.0) 1.3 (1.5) 1.2 (.4) 85.0 (100)
II
III
IV–VI
11 87.4 (4.7) 7 F; 4 M 10.7 2.1 (1.0) 3.8 (2.4) 1.0 (.7) 70.4 (69.7)
8 86.7 (9.6) 5 F; 3 M 11.6 1.2 (.6) 7.1 (9.3) .17 (.54) 160.9 (74.4)
4 84.2 (4.1) 3 F; 1 M 9.0 2.8 (1.3) 24.0 (6.2) ⫺.64 (1.01) 234.2 (31.4)
Standard deviations are in parenthesis.
(EAS). Eligible subjects had mental status and memory tests in life and a complete brain autopsy. The target sample included individuals without pathologic evidence of dementia and individuals with pathologic AD without evidence of other forms of dementia. Subjects had an average age of 87.2 (range: 66 –97), 19 were female, and they averaged 11.1 years of education (range: 5–20). Brains with evidence of significant vascular disease, Lewy body disease, or other non-AD pathology were excluded. The median time between cognitive testing and death was 20 months. Clinical diagnoses were assigned using DSM III-R criteria for dementia [1] and NINCDS-ADRDA criteria for AD [22]. Twenty subjects were considered normal at the time of their death, 5 were diagnosed as having AD, and 2 as having vascular dementia. There were 2 subjects who did not meet DSM III-R criteria for dementia; one had memory and cognitive impairment but no functional decline and the other had cognitive impairment. Characteristics of the subjects classified by Braak stage are summarized in Table 1. 2.2. Memory tests Memory was assessed using one of two verbal learning tests: the Selective Reminding Test (SRT) [9] in 11 subjects and the Free and Cued Selective Reminding Test (FCSRT) [10,13] in 18 subjects. FCSR differs from SR by including a study procedure that controls cognitive processing and a reminding procedure that allows for cued recall [17]. Despite these procedural differences, the correlation between the two tests in a sample of 72 nondemented EAS participants was very high (r ⫽ 0.71). Therefore, we computed a single standardized memory score, using the FCSR or the SR test. 2.3. Selective reminding procedure Subjects read a list of 12 unrelated words presented one at a time at 5-s intervals. Immediately after reading the words, subjects recalled as many of them as possible in any order. Before each of the remaining 5 test trials, the subject was reminded only of those words that were not recalled on the immediately preceding trial. Scores ranged from 0 to 72. Test procedures are described in detail elsewhere [17,21].
2.4. FCSR procedure Briefly, the test begins with a study procedure that required subjects to search a card containing 4 pictured objects (e.g., grapes) and point to and name aloud each picture when its category cue (fruit) was given verbally. After all 4 pictures were identified, immediate cued recall of those pictures was tested. Once immediate cued recall for a set of pictures was correct, the next set of 4 was presented, continuing until all 16 items were studied. The study procedure was followed by 3 test trials, each consisting of free recall followed by cued recall of these items not retrieved by free recall. Free recall scores ranged from 0 to 48. This test has been described in prior publications [13,17]. 2.5. Mental status Mental status was assessed by the Blessed Information Memory and Concentration (BIMC) test [3], which is comprised of test items that evaluate orientation, attention, concentration, memory, and retrieval of factual information. Errors are scored and range from 0 to 33. 2.6. Neuropathology procedures The cases were staged using a modification of the Braak and Braak [6] staging system, which has been shown to have good interrater reliability [25,26]. Thin paraffin sections (10 m thick) of the inferior temporal lobe including the hippocampus, entorhinal cortex, occipitotemporal gyrus, and inferior temporal gyrus, were immunostained with a mouse monoclonal antibody to tau protein (PHF-1) using standard laboratory procedures and an avidin-biotin peroxidase kit (Vector Labs, Burlingame, CA) with diaminobenzidine as the chromogen. The sections were counterstained with hemotoxylin. Staging was accomplished by matching the topographic distribution of neurofibrillary lesions to the best fitting diagram published by Braak and Braak [6]. The staging scheme could be roughly operationalized as follows with subsequent stages showing changes of the earlier stages: I. Neurofibrillary degeneration (NFD) confined to transentorhinal cortex (layer IV); II. NFD in entorhinal
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
575
Fig. 1. Representative thioflavin-S stained sections of midfrontal cortex of study cases with no senile plaques (a), many non-neuritic plaques (arrows) (b), or many neuritic plaques (arrows) (c). The latter case also had neocortical neurofibrillary tangles (open arrowhead). The signal in cases with no plaques is autofluorescent lipofuscin pigment in neurons and glia. Layer II of the entorhinal cortex of the same cases immunostained with PHF-1 and counterstained with hematoxylin reveals no NFT in a case with a Braak stage of I (d; inset shows neuritic pathology in the transentorhinal region of this case), sparse neurofibrillary tangles in a case with a Braak stage II (e) and numerous neurofibrillary tangles in a case with a Braak stage of V (f).
cortex (layer II); III. NFD in hippocampus (CA1 and subiculum); IV. NFD throughout entorhinal cortex; V. NFD in inferior temporal neocortex (mild to moderate lesion density); and VI. NFD in inferior temporal neocortex (moderate to marked lesion density). Quantification of senile plaques was accomplished by thioflavine S staining. The cortical regions were sampled before dissection to assure consistency of sampling and to obtain perpendicular sections of gyri. The areas that were sampled were the midfrontal cortex (Brodmann’s area 9 or 46), superior temporal gyrus (area 38), inferior parietal lobule (area 37), and calcarine cortex (area 17). After embedding in paraffin and sectioning at 7-m thickness, the sections were stained with thioflavine S staining and examined with an Olympus BH-2 fluorescent microscope equipped with a 100-watt mercury lamp and a “true blue” band pass filter set (490-nm). Multiple fields were counted along the side of the gyrus, avoiding the crest of the gyrus and the depth of the sulcus. Representative fields with the highest and lowest counts were recorded and the average of the high and low fields were recorded. In the event that the SP counts exceeded 50, which is more than twice the number of SP needed to satisfy the Khachaturian criteria [19], the counts were truncated to 50. SP were counted without bias to the type of amyloid (i.e., diffuse or compact) and without regard to the presence of neuritic elements. In other studies, these SP counts were found to correlate highly with the amyloid burden determined by image analysis with amyloid antibody methods [28]. The sum of the SP counts for all cortical areas (sumSP) was our measure of amyloid burden. We sought to determine the extent to which amyloid
burden modified the relationship between cognitive status and Braak stage. Fig. 1 illustrates findings with tau immunostaining of the entorhinal cortex layer II neurons in representative cases with Braak Stages of I (d), II (e), and V (f). A range of cortical pathology with thioflavine S is also illustrated with some cases having no neocortical plaques (a), cases with many diffuse or non-neuritic plaques (b), and cases with many neuritic plaques and some neocortical neurofibrillary tangles. 2.7. Analysis One-way analysis of variance (ANOVA) was used to examine the relationship between Braak staging and memory as well as BIMC. Because of an interest in early stages and because of sample size limitations, we collapsed subjects into the following groups by Braak stage: 0, I, II, III, IV–VI. All significance tests were conducted at the 0.05 level of significance. After detecting significant main effects, 95% confidence intervals for differences between adjacent Braak stages were examined. A square-root transform was applied to the raw BIMC scores because they were positively skewed.
3. Results Analysis of variance confirmed that higher Braak stages were associated with declining performance both on the BIMC test (F[4,24] ⫽ 8.78, p ⬍ 0.001, Mse ⫽ 1.00) and
576
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Fig. 2. BIMC errors as a function of Braak stage. Open squares indicate the mean score for each stage. Vertical bars show the 95% confidence interval for the means.
the memory test (F[4,24] ⫽ 6.90, p ⬍ .001, Mse ⫽ 0.46). The mean BIMC and memory performance for adjacent Braak stages (i.e., 0 vs. I, I vs. II, II vs. III, and III vs. IV–VI) are displayed graphically in Figs. 2 and 3. Fig. 2 shows that mean BIMC performance remained relatively constant from Braak stage 0 to stage III, and only at stages IV–VI was there an increased number of errors on the BIMC. Fig. 3 shows that for the memory score, the drop in performance occurred at stage III. Figs. 4 and 5 present the differences in average BIMC and
memory performance between adjacent Braak stages and the 95% confidence intervals for these differences. In Fig. 4, the 95% confidence intervals for the differences in BIMC overlap 0.0 until the Braak stage III vs. IV–VI comparison. In contrast, the confidence intervals for the stage II vs. III comparison does not include 0.0, indicating detectable differences in memory performance between Braak stage II and III. Table 2 shows the Spearman correlation coefficients for BIMC errors, the standardized memory score, Braak score, and the sum of senile plaques (sumSP). The memory score
Fig. 3. Memory performance as a function of Braak stage. Open squares indicate the mean score for each stage. Vertical bars show the 95% confidence interval for the means.
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
577
Fig. 4. Differences in average BIMC errors between adjacent Braak stages and the 95% confidence intervals for the differences.
is negatively correlated with Braak stage and senile plaque count indicating that as pathology increases memory declines. The BIMC score is directly correlated with pathology measures because the BIMC increases as disease progresses. To determine whether the relationship between Braak score and the cognitive measures was due to the association with senile plaque count, partial correlations that adjusted for plaque count were performed. After adjusting for plaque count, the association between memory and Braak stage was still significant (r ⫽ 0.58, p ⬍ 0.001) though the association between mental status and Braak
stage was reduced to marginal levels (r ⫽ 0.34, p ⬍ 0.08). In contrast, partial correlations that adjusted for Braak stage eliminated the association between both plaques and memory (r ⫽ 0.004, NS) as well as plaques and mental status (r ⫽ ⫺0.003, NS).
4. Comment This study demonstrates that Braak stage is associated with performance on both memory and mental status tests
Fig. 5. Differences in average memory performance between adjacent Braak stages and the 95% confidence intervals for the differences.
578
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Table 2 Spearman correlation coefficients
BIMC MEMORY BRAAK SUM SP
BIMCa
MEMORY
BRAAK
SUM SP
—
⫺.54** —
.46** ⫺.72* —
.32 ⫺.51* .71* —
* p ⬍ .0001; ** p ⬍ .01. a BIMC, Blessed Information Memory and Concentration test score; MEMORY, z-score from Selective Reminding or Free and Cued Selective Reminding memory tests; BRAAK, stage of AD neuropathology; SUM SP, sum of non neuritic senile plaques in cortex, hippocampus, and Basal forebrain sections.
supporting the clinical relevance of the neuropathologic staging model. Because the staging system reflects the sequential unfolding of AD neuropathology, we predicted that, at early Braak stages, memory decline might occur without mental status decline. Both memory and mental status test performance were stable at pathology stages I and II. Either this is a clinically silent period with pathologic progression without definite behavioral change or this is a period of behavioral decline that the current instruments cannot measure reliably. A goal for future research is to identify neuropsychological or biological markers for the earliest stages of disease. Memory scores declined from Braak stage II to stage III. Mental status decline was not observed until after Braak stage III. The finding that memory decline occurred at stage II to III when subjects were considered clinically nondemented supports our view that systematic memory impairment occurs when early stage AD is present pathologically but is not yet clinically diagnosable. Recent studies have demonstrated memory impairment in nondemented individuals who are in the preclinical stages of AD and subsequently manifest the typical dementia syndrome [4,5,15,16, 18,21,31]. In two Einstein Aging studies, we found that memory impairment was a powerful predictor of future dementia five years before it was clinically diagnosable using free recall from FCSR in one sample [15] and SR in a separate sample [18]. Subjects with impaired free recall on FCSR at baseline developed dementia (RR ⫽ 75.2, 95% CI ⫽ 9.9 –567) over 5 years of follow-up at dramatically higher rates than subjects with intact free recall after adjusting for age, gender, and education [15]. Impaired free recall on FCSR and an informant report of memory decline may be useful in identifying persons with mild cognitive impairment (MCI) [27]. Patients with MCI are at increased risk for developing dementia, ranging from 1% to 25% per year [27]. Comparisons between the relative value of tangle versus plaque development as an index of clinical severity are beyond the scope of this paper because staging of SP was not done. Our interest was in the association of each pathologic variable independent of the other with memory impairment in preclinical and very early AD. Amyloid burden,
indexed by the sum of SP in all the major lobes of the neocortex, did not modify the association between memory performance and Braak stage though the correlation between the memory and amyloid burden was eliminated when adjusted for Braak stage. This pattern of associations suggests that memory impairment is related to tangle development in preclinical and early AD independently of amyloid burden in the neocortex. Staging of SP may have revealed a different pattern of associations between the cognitive measures, Braak stage, and amyloid burden. A recent autopsy series of clinically nondemented elderly suggests that plaques and tangles develop independently of the other in healthy aging but that in preclinical AD, there is an interaction between them [29]. The two pathological substrates for the clinically nondemented cases identified in the Price and Morris series corresponds roughly to the two substrates we have characterized previously having minimal or no cerebral amyloid deposition (i.e., normal aging) and the other having moderate to marked amyloid deposition (i.e., pathological aging) [12]. We agree that the latter subgroup of clinically nondemented subjects may have preclinical AD and that in the present series, many subjects who have stage II or III pathology are likely to have preclinical AD. Other clinicopathological studies support the clinical relevance of the neurofibrillary staging model [2,8]. Braak stage and mental status were significantly correlated (r ⫽ 0.68) in 26 brains from individuals who were repeatedly assessed with the BIMC. The clinical protocols of stage I or II cases did not generally include intellectual impairment, whereas stage III or IV cases typically showed mild to modest impairment of cognition and personality changes. The presence of dementia was noted in the clinical protocols of all stage V and VI cases. Based on these findings, the authors suggest that stages I and II correspond to a clinically silent period or preclinical phase of AD and that the morphological changes and clinical symptoms in stages III and IV represent incipient AD. We recognize several limitations in our study. First, 19 of the 22 subjects in Braak stages I to III were considered clinically nondemented at the time of their last evaluation. Had they lived longer they might not have satisfied established clinical and neuropathological criteria for AD. Second, because the average time between cognitive testing and death was 20 months, the extent of pathology at the time of death most likely overestimates the extent of pathology at the last clinical evaluation. Using time-to-death as a covariate in the analyses yielded very similar results to those reported here. Third, the difference in sensitivity of the cognitive tests to early stage disease may reflect differences in task difficulty. Although the BIMC is generally regarded as an omnibus mental status test of several cognitive domains, inspection of the BIMC questions suggest that nearly all depend on various kinds of memory. Because FCSR and SR are multitrial memory tests, they assess new learning
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
ability as well as memory, which may be more sensitive to cognitive impairment in early AD.
References [1] American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 3rd edition, revised. Washington, DC: American Psychiatric Association, 1980. [2] Bancher C, Braak H, Fischer P, Jellinger KA. Neuropathological staging of Alzheimer lesions and intellectual status in Alzheimer’s and Parkinson’s disease. Neurosci Lett 1993;162:179 – 82. [3] Blessed G, Tomlinson BE, Roth M. The association between quantitative measure of dementia and of senile change in the cerebral gray matter of elderly subjects. Br J Psychiatry 1968;114:797– 811. [4] Bondi MW, Salmon DP, Galasko D, Thomas RG, Thal LJ. Neuropsychological function and apolipoprotein E genotype in the preclinical detection of Alzheimer’s disease. Psychol Aging, in press, 1999. [5] Bowen J, Teri L, Kukull W, McCormick W, McCurry S, Larson E. Progression to dementia in patients with isolated memory loss. Lancet 1997;349:763–5. [6] Braak H, Braak E. Neuropathological staging of Alzheimer-related changes. Acta Neuropath 1991;82:239 –59. [7] Braak H, Braak E. Frequency of stages of Alzheimer-related lesions in different age categories. Neurobiol Aging 1997;18:351–7. [8] Braak H, Duyckaerts C, Braak E, Piette F. Neuropathological staging of Alzheimer-related changes correlates with psychometrically assessed intellectual status. In: Corian B, Iqbal K, Nicolini M, Winblad B, Wisniewski H, Zatta PF, editors. Alzheimer’s disease: advances in clinical and basic research. Third International Conference on Alzheimer’s Disease and Related Disorders. Chichester: Wiley, 1993. p. 131–7. [9] Buschke H. Selective reminding for analysis of memory and learning. J Verbal Learning Verbal Behav 1973;12:543–50. [10] Buschke H. Cued recall in amnesia. J Clin Exp Neuropsychol 1984; 6:433– 40. [11] Dickson DW. The value of cross sectional neuroanatomical studies as a conceptual framework for prospective clinicopathological studies. Neurobiol Aging 1997;18:382– 6. [12] Dickson D, Crystal H, Mattiace LA, Masur D, Blau A, Davies P, Yen SH, Aronson M. Identification of normal and pathological aging in prospectively studied nondemented elderly humans. Neurobiol Aging 1991;13:179 – 89. [13] Grober E, Buschke H. Genuine memory deficits in dementia. Dev Neuropsychol 1987;3:13–36. [14] Grober E, Buschke H, Crystal H, Bang S, Dresner R. Screening for dementia by memory testing. Neurology 1988;38:900 –3. [15] Grober E, Lipton R, Hall C, Crystal H. Memory impairment on Free and Cued Selective Reminding predicts dementia. Neurology 1999;in press. [16] Grober E, Kawas C. Learning and retention in preclinical and early Alzheimer’s Disease. Psychol Aging 1997;12:183– 8.
579
[17] Grober E, Merling A, Heimlich T, Lipton RB. Free and cued selective reminding and selective reminding in the elderly. J Clin Exp Neuropsychol 1997;19:643–54. [18] Hall CB, Lipton RB, Sliwinski M, Stewart WF. A change point model for estimating the onset of cognitive decline in preclinical Alzheimer’s disease. Statistics Med 1999, in press. [19] Khachaturian ZS. Diagnosis of Alzheimer’s disease. Arch Neurol 1985;42:1097–105. [20] Linn RT, Wolf PA, Bachman DL, Knoefel JE, Cogg JL, Belanger AJ, Kaplan EG, D’Agostino RB. The preclinical phase of probable Alzheimer’s disease. Arch Neurol 1995;52:485–90. [21] Masur DM, Sliwinski M, Lipton RB, Blau AD, Crystal HA. Neuropsychological prediction of dementia and the absence of dementia in healthy elderly persons. Neurology 1994;44:1427–32. [22] McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan E. Clinical diagnosis of Alzheimer’s disease: report of the NINCDSADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 1984;34:939 – 44. [23] Mirra SS, Heyman A, McKeel D, Sumi SM, Crain BJ, Brownlee LM, Vogal FS, Hughes JP, van Belle G, Berg L. The consortium to establish a registry for Alzheimer’s Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer’s Disease. Neurology 1991;41:479 – 86. [24] Morris JC, Storandt M, McKeel DW, Rubin EH, Price JL, Grant EA, Berg L. Cerebral amyloid deposition and diffuse plaques in “normal” aging: Evidence for presymptomatic and very mild Alzheimer’s disease. Neurology 1996;46:707–19. [25] Nagy Z, Yilmazer-Hanke DM, Braak H, Braak E, Schultz C, Hanke J. Assessment of the pathological stages of Alzheimer’s disease in thin paraffin sections: a comparative study. Dementia Geriatr Cognitive Dis 1998;9:140 – 4. [26] Nagy Z, Vatter-Bittner B, Braak H, Braak E, Yilmazer DM, Schultz C, Hanke J. Staging of Alzheimer-type pathology: an interraterintrarater study. Dementia Geriatr Cognitive Dis 1997;8:248 –51. [27] Peterson RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. Mild Cognitive Impairment: Clinical Characterization and Outcome. Arch Neurol 1999;56:303– 8. [28] Poulos B, Crystal H, Katz M, Kuslansky G, Grober E, Lipton R, Dickson DW. Amyloid burden determined by image analysis correlates with conventional histochemical methods and clinical measures of cognitive decline. J Neuropathol Exp Neurol 1997;56:615. [29] Price JL, Morris JC. Tangles and plaques in nondemented aging and “preclinical” Alzheimer’s disease. Ann Neurol 1999;45:358 – 68. [30] Ritchie K, Fuhrer R. A comparative study of the performance of screening tests for senile dementia using receiver operating characteristics analysis. J Clin Epidemiol 1992;45:627–37. [31] Tierney MC, Szalai JP, Snow G, et al. Prediction of probable Alzheimer’s disease in memory-impaired patients: a prospective longitudinal study. Neurology 1996;46:661–5. [32] Welsh K, Butters N, Hughes J, et al. Detection of abnormal memory decline in mild cases of Alzheimer’s Disease using CERAD neuropsychological measures. Arch Neurol 1991;48:278 – 81.
Memory and mental status correlates of modified Braak staging夞, 夞夞 E. Grobera,*, D. Dicksonb, M.J. Sliwinskia, H. Buschkea, M. Katza, H. Crystala, R.B. Liptona,c a
Department of Neurology, Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, NY, USA b Department of Pathology, Mayo Clinic, Jacksonville, FL, USA c Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, NY, USA Received 27 October 1998; received in revised form 20 July 1999; accepted 18 August 1999
Abstract We assessed the relationships of performance on memory and mental status tests and neuropathologic stage of Alzheimer’s disease as defined by Braak and Braak in 29 patients from a prospective clinicopathologic series. We predicted that memory changes would occur at an earlier Braak stage than mental status changes. Staging was accomplished by matching the topographic distribution of neurofibrillary lesions detected with tau immunocytochemistry to the best fitting diagram published by Braak and Braak. Higher Braak stages were associated with decrements in performance on both memory and mental status tests. As predicted, memory performance declined from stages II to III and mental status did not decline until stages III to IV. The association between memory and Braak stage was unchanged after adjusting for neocortical senile plaques, whereas adjustments for Braak stage eliminated the association between cognitive functioning and amyloid burden. We conclude that Braak staging provides a useful summary of Alzheimer’s disease neuropathology, which is associated with both memory and mental status performance. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Memory; Mental status; Braak stage; Alzheimer’s disease
1. Introduction Neuropathology is the gold standard for diagnosing Alzheimer’s disease (AD). A well-established system for staging the spread of neurofibrillary degeneration in AD was developed by Braak and Braak [6]. Their recent study on the distribution of Alzheimer’s neuropathology by age supports the value of their approach but does not provide clinical information regarding individual cases [11]. Nonetheless, the results of clinicopathological studies establish Braak staging as a useful approach for characterizing AD neuropathology [2,8]. In this study, we evaluated the relationship of Braak stage to performance on memory and mental status tests in
夞 This research was supported by the Einstein Aging Study, National Institute on Aging Grant AG03949. 夞夞 This paper was presented in part at the 49th Annual Meeting of the American Academy of Neurology, Boston, MA, 1997. * Corresponding author. Tel.: ⫹1-914-963-5601; fax: ⫹1-914-9635602. E-mail address: [email protected]
prospectively studied subjects. We predicted that advances in neuropathologic stage would be associated with declines in cognitive tests administered close to the time of death. We also sought to clarify the behavioral correlates at early Braak stages. Because memory decline is one of the earliest neuropsychological manifestations of preclinical and early AD [4,5,15,16,18,20,21,27,31,32], we hypothesized that decrements in memory performance would occur at early stages of pathologic AD. Because mental status tests have good sensitivity for detecting more advanced dementias [30], we predicted that mental status would not decline until later-stage disease. We also sought to determine the extent to which the relationship between cognitive status and Braak stage was modified by amyloid burden [19,23,24,29].
2. Materials and methods 2.1. Subjects We selected subjects for inclusion in this study from the clinical-pathologic database of the Einstein Aging Study
0197-4580/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 1 9 7 - 4 5 8 0 ( 9 9 ) 0 0 0 6 3 - 9
574
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Table 1 Characteristics of the subjects by Braak stagea Braak stage N Age at death Gender Education (years) Post interval (years) BIMC score Memory-z score SUM SP a
0 3 91.6 (1.7) 1 F; 2 M 11.3 2.7 (1.4) 4.3 (1.5) 1.4 (.4) 0.0
1 3 88.2 (.69) 3F 14.0 2.0 (1.0) 1.3 (1.5) 1.2 (.4) 85.0 (100)
II
III
IV–VI
11 87.4 (4.7) 7 F; 4 M 10.7 2.1 (1.0) 3.8 (2.4) 1.0 (.7) 70.4 (69.7)
8 86.7 (9.6) 5 F; 3 M 11.6 1.2 (.6) 7.1 (9.3) .17 (.54) 160.9 (74.4)
4 84.2 (4.1) 3 F; 1 M 9.0 2.8 (1.3) 24.0 (6.2) ⫺.64 (1.01) 234.2 (31.4)
Standard deviations are in parenthesis.
(EAS). Eligible subjects had mental status and memory tests in life and a complete brain autopsy. The target sample included individuals without pathologic evidence of dementia and individuals with pathologic AD without evidence of other forms of dementia. Subjects had an average age of 87.2 (range: 66 –97), 19 were female, and they averaged 11.1 years of education (range: 5–20). Brains with evidence of significant vascular disease, Lewy body disease, or other non-AD pathology were excluded. The median time between cognitive testing and death was 20 months. Clinical diagnoses were assigned using DSM III-R criteria for dementia [1] and NINCDS-ADRDA criteria for AD [22]. Twenty subjects were considered normal at the time of their death, 5 were diagnosed as having AD, and 2 as having vascular dementia. There were 2 subjects who did not meet DSM III-R criteria for dementia; one had memory and cognitive impairment but no functional decline and the other had cognitive impairment. Characteristics of the subjects classified by Braak stage are summarized in Table 1. 2.2. Memory tests Memory was assessed using one of two verbal learning tests: the Selective Reminding Test (SRT) [9] in 11 subjects and the Free and Cued Selective Reminding Test (FCSRT) [10,13] in 18 subjects. FCSR differs from SR by including a study procedure that controls cognitive processing and a reminding procedure that allows for cued recall [17]. Despite these procedural differences, the correlation between the two tests in a sample of 72 nondemented EAS participants was very high (r ⫽ 0.71). Therefore, we computed a single standardized memory score, using the FCSR or the SR test. 2.3. Selective reminding procedure Subjects read a list of 12 unrelated words presented one at a time at 5-s intervals. Immediately after reading the words, subjects recalled as many of them as possible in any order. Before each of the remaining 5 test trials, the subject was reminded only of those words that were not recalled on the immediately preceding trial. Scores ranged from 0 to 72. Test procedures are described in detail elsewhere [17,21].
2.4. FCSR procedure Briefly, the test begins with a study procedure that required subjects to search a card containing 4 pictured objects (e.g., grapes) and point to and name aloud each picture when its category cue (fruit) was given verbally. After all 4 pictures were identified, immediate cued recall of those pictures was tested. Once immediate cued recall for a set of pictures was correct, the next set of 4 was presented, continuing until all 16 items were studied. The study procedure was followed by 3 test trials, each consisting of free recall followed by cued recall of these items not retrieved by free recall. Free recall scores ranged from 0 to 48. This test has been described in prior publications [13,17]. 2.5. Mental status Mental status was assessed by the Blessed Information Memory and Concentration (BIMC) test [3], which is comprised of test items that evaluate orientation, attention, concentration, memory, and retrieval of factual information. Errors are scored and range from 0 to 33. 2.6. Neuropathology procedures The cases were staged using a modification of the Braak and Braak [6] staging system, which has been shown to have good interrater reliability [25,26]. Thin paraffin sections (10 m thick) of the inferior temporal lobe including the hippocampus, entorhinal cortex, occipitotemporal gyrus, and inferior temporal gyrus, were immunostained with a mouse monoclonal antibody to tau protein (PHF-1) using standard laboratory procedures and an avidin-biotin peroxidase kit (Vector Labs, Burlingame, CA) with diaminobenzidine as the chromogen. The sections were counterstained with hemotoxylin. Staging was accomplished by matching the topographic distribution of neurofibrillary lesions to the best fitting diagram published by Braak and Braak [6]. The staging scheme could be roughly operationalized as follows with subsequent stages showing changes of the earlier stages: I. Neurofibrillary degeneration (NFD) confined to transentorhinal cortex (layer IV); II. NFD in entorhinal
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
575
Fig. 1. Representative thioflavin-S stained sections of midfrontal cortex of study cases with no senile plaques (a), many non-neuritic plaques (arrows) (b), or many neuritic plaques (arrows) (c). The latter case also had neocortical neurofibrillary tangles (open arrowhead). The signal in cases with no plaques is autofluorescent lipofuscin pigment in neurons and glia. Layer II of the entorhinal cortex of the same cases immunostained with PHF-1 and counterstained with hematoxylin reveals no NFT in a case with a Braak stage of I (d; inset shows neuritic pathology in the transentorhinal region of this case), sparse neurofibrillary tangles in a case with a Braak stage II (e) and numerous neurofibrillary tangles in a case with a Braak stage of V (f).
cortex (layer II); III. NFD in hippocampus (CA1 and subiculum); IV. NFD throughout entorhinal cortex; V. NFD in inferior temporal neocortex (mild to moderate lesion density); and VI. NFD in inferior temporal neocortex (moderate to marked lesion density). Quantification of senile plaques was accomplished by thioflavine S staining. The cortical regions were sampled before dissection to assure consistency of sampling and to obtain perpendicular sections of gyri. The areas that were sampled were the midfrontal cortex (Brodmann’s area 9 or 46), superior temporal gyrus (area 38), inferior parietal lobule (area 37), and calcarine cortex (area 17). After embedding in paraffin and sectioning at 7-m thickness, the sections were stained with thioflavine S staining and examined with an Olympus BH-2 fluorescent microscope equipped with a 100-watt mercury lamp and a “true blue” band pass filter set (490-nm). Multiple fields were counted along the side of the gyrus, avoiding the crest of the gyrus and the depth of the sulcus. Representative fields with the highest and lowest counts were recorded and the average of the high and low fields were recorded. In the event that the SP counts exceeded 50, which is more than twice the number of SP needed to satisfy the Khachaturian criteria [19], the counts were truncated to 50. SP were counted without bias to the type of amyloid (i.e., diffuse or compact) and without regard to the presence of neuritic elements. In other studies, these SP counts were found to correlate highly with the amyloid burden determined by image analysis with amyloid antibody methods [28]. The sum of the SP counts for all cortical areas (sumSP) was our measure of amyloid burden. We sought to determine the extent to which amyloid
burden modified the relationship between cognitive status and Braak stage. Fig. 1 illustrates findings with tau immunostaining of the entorhinal cortex layer II neurons in representative cases with Braak Stages of I (d), II (e), and V (f). A range of cortical pathology with thioflavine S is also illustrated with some cases having no neocortical plaques (a), cases with many diffuse or non-neuritic plaques (b), and cases with many neuritic plaques and some neocortical neurofibrillary tangles. 2.7. Analysis One-way analysis of variance (ANOVA) was used to examine the relationship between Braak staging and memory as well as BIMC. Because of an interest in early stages and because of sample size limitations, we collapsed subjects into the following groups by Braak stage: 0, I, II, III, IV–VI. All significance tests were conducted at the 0.05 level of significance. After detecting significant main effects, 95% confidence intervals for differences between adjacent Braak stages were examined. A square-root transform was applied to the raw BIMC scores because they were positively skewed.
3. Results Analysis of variance confirmed that higher Braak stages were associated with declining performance both on the BIMC test (F[4,24] ⫽ 8.78, p ⬍ 0.001, Mse ⫽ 1.00) and
576
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Fig. 2. BIMC errors as a function of Braak stage. Open squares indicate the mean score for each stage. Vertical bars show the 95% confidence interval for the means.
the memory test (F[4,24] ⫽ 6.90, p ⬍ .001, Mse ⫽ 0.46). The mean BIMC and memory performance for adjacent Braak stages (i.e., 0 vs. I, I vs. II, II vs. III, and III vs. IV–VI) are displayed graphically in Figs. 2 and 3. Fig. 2 shows that mean BIMC performance remained relatively constant from Braak stage 0 to stage III, and only at stages IV–VI was there an increased number of errors on the BIMC. Fig. 3 shows that for the memory score, the drop in performance occurred at stage III. Figs. 4 and 5 present the differences in average BIMC and
memory performance between adjacent Braak stages and the 95% confidence intervals for these differences. In Fig. 4, the 95% confidence intervals for the differences in BIMC overlap 0.0 until the Braak stage III vs. IV–VI comparison. In contrast, the confidence intervals for the stage II vs. III comparison does not include 0.0, indicating detectable differences in memory performance between Braak stage II and III. Table 2 shows the Spearman correlation coefficients for BIMC errors, the standardized memory score, Braak score, and the sum of senile plaques (sumSP). The memory score
Fig. 3. Memory performance as a function of Braak stage. Open squares indicate the mean score for each stage. Vertical bars show the 95% confidence interval for the means.
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
577
Fig. 4. Differences in average BIMC errors between adjacent Braak stages and the 95% confidence intervals for the differences.
is negatively correlated with Braak stage and senile plaque count indicating that as pathology increases memory declines. The BIMC score is directly correlated with pathology measures because the BIMC increases as disease progresses. To determine whether the relationship between Braak score and the cognitive measures was due to the association with senile plaque count, partial correlations that adjusted for plaque count were performed. After adjusting for plaque count, the association between memory and Braak stage was still significant (r ⫽ 0.58, p ⬍ 0.001) though the association between mental status and Braak
stage was reduced to marginal levels (r ⫽ 0.34, p ⬍ 0.08). In contrast, partial correlations that adjusted for Braak stage eliminated the association between both plaques and memory (r ⫽ 0.004, NS) as well as plaques and mental status (r ⫽ ⫺0.003, NS).
4. Comment This study demonstrates that Braak stage is associated with performance on both memory and mental status tests
Fig. 5. Differences in average memory performance between adjacent Braak stages and the 95% confidence intervals for the differences.
578
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
Table 2 Spearman correlation coefficients
BIMC MEMORY BRAAK SUM SP
BIMCa
MEMORY
BRAAK
SUM SP
—
⫺.54** —
.46** ⫺.72* —
.32 ⫺.51* .71* —
* p ⬍ .0001; ** p ⬍ .01. a BIMC, Blessed Information Memory and Concentration test score; MEMORY, z-score from Selective Reminding or Free and Cued Selective Reminding memory tests; BRAAK, stage of AD neuropathology; SUM SP, sum of non neuritic senile plaques in cortex, hippocampus, and Basal forebrain sections.
supporting the clinical relevance of the neuropathologic staging model. Because the staging system reflects the sequential unfolding of AD neuropathology, we predicted that, at early Braak stages, memory decline might occur without mental status decline. Both memory and mental status test performance were stable at pathology stages I and II. Either this is a clinically silent period with pathologic progression without definite behavioral change or this is a period of behavioral decline that the current instruments cannot measure reliably. A goal for future research is to identify neuropsychological or biological markers for the earliest stages of disease. Memory scores declined from Braak stage II to stage III. Mental status decline was not observed until after Braak stage III. The finding that memory decline occurred at stage II to III when subjects were considered clinically nondemented supports our view that systematic memory impairment occurs when early stage AD is present pathologically but is not yet clinically diagnosable. Recent studies have demonstrated memory impairment in nondemented individuals who are in the preclinical stages of AD and subsequently manifest the typical dementia syndrome [4,5,15,16, 18,21,31]. In two Einstein Aging studies, we found that memory impairment was a powerful predictor of future dementia five years before it was clinically diagnosable using free recall from FCSR in one sample [15] and SR in a separate sample [18]. Subjects with impaired free recall on FCSR at baseline developed dementia (RR ⫽ 75.2, 95% CI ⫽ 9.9 –567) over 5 years of follow-up at dramatically higher rates than subjects with intact free recall after adjusting for age, gender, and education [15]. Impaired free recall on FCSR and an informant report of memory decline may be useful in identifying persons with mild cognitive impairment (MCI) [27]. Patients with MCI are at increased risk for developing dementia, ranging from 1% to 25% per year [27]. Comparisons between the relative value of tangle versus plaque development as an index of clinical severity are beyond the scope of this paper because staging of SP was not done. Our interest was in the association of each pathologic variable independent of the other with memory impairment in preclinical and very early AD. Amyloid burden,
indexed by the sum of SP in all the major lobes of the neocortex, did not modify the association between memory performance and Braak stage though the correlation between the memory and amyloid burden was eliminated when adjusted for Braak stage. This pattern of associations suggests that memory impairment is related to tangle development in preclinical and early AD independently of amyloid burden in the neocortex. Staging of SP may have revealed a different pattern of associations between the cognitive measures, Braak stage, and amyloid burden. A recent autopsy series of clinically nondemented elderly suggests that plaques and tangles develop independently of the other in healthy aging but that in preclinical AD, there is an interaction between them [29]. The two pathological substrates for the clinically nondemented cases identified in the Price and Morris series corresponds roughly to the two substrates we have characterized previously having minimal or no cerebral amyloid deposition (i.e., normal aging) and the other having moderate to marked amyloid deposition (i.e., pathological aging) [12]. We agree that the latter subgroup of clinically nondemented subjects may have preclinical AD and that in the present series, many subjects who have stage II or III pathology are likely to have preclinical AD. Other clinicopathological studies support the clinical relevance of the neurofibrillary staging model [2,8]. Braak stage and mental status were significantly correlated (r ⫽ 0.68) in 26 brains from individuals who were repeatedly assessed with the BIMC. The clinical protocols of stage I or II cases did not generally include intellectual impairment, whereas stage III or IV cases typically showed mild to modest impairment of cognition and personality changes. The presence of dementia was noted in the clinical protocols of all stage V and VI cases. Based on these findings, the authors suggest that stages I and II correspond to a clinically silent period or preclinical phase of AD and that the morphological changes and clinical symptoms in stages III and IV represent incipient AD. We recognize several limitations in our study. First, 19 of the 22 subjects in Braak stages I to III were considered clinically nondemented at the time of their last evaluation. Had they lived longer they might not have satisfied established clinical and neuropathological criteria for AD. Second, because the average time between cognitive testing and death was 20 months, the extent of pathology at the time of death most likely overestimates the extent of pathology at the last clinical evaluation. Using time-to-death as a covariate in the analyses yielded very similar results to those reported here. Third, the difference in sensitivity of the cognitive tests to early stage disease may reflect differences in task difficulty. Although the BIMC is generally regarded as an omnibus mental status test of several cognitive domains, inspection of the BIMC questions suggest that nearly all depend on various kinds of memory. Because FCSR and SR are multitrial memory tests, they assess new learning
E. Grober et al. / Neurobiology of Aging 20 (1999) 573–579
ability as well as memory, which may be more sensitive to cognitive impairment in early AD.
References [1] American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 3rd edition, revised. Washington, DC: American Psychiatric Association, 1980. [2] Bancher C, Braak H, Fischer P, Jellinger KA. Neuropathological staging of Alzheimer lesions and intellectual status in Alzheimer’s and Parkinson’s disease. Neurosci Lett 1993;162:179 – 82. [3] Blessed G, Tomlinson BE, Roth M. The association between quantitative measure of dementia and of senile change in the cerebral gray matter of elderly subjects. Br J Psychiatry 1968;114:797– 811. [4] Bondi MW, Salmon DP, Galasko D, Thomas RG, Thal LJ. Neuropsychological function and apolipoprotein E genotype in the preclinical detection of Alzheimer’s disease. Psychol Aging, in press, 1999. [5] Bowen J, Teri L, Kukull W, McCormick W, McCurry S, Larson E. Progression to dementia in patients with isolated memory loss. Lancet 1997;349:763–5. [6] Braak H, Braak E. Neuropathological staging of Alzheimer-related changes. Acta Neuropath 1991;82:239 –59. [7] Braak H, Braak E. Frequency of stages of Alzheimer-related lesions in different age categories. Neurobiol Aging 1997;18:351–7. [8] Braak H, Duyckaerts C, Braak E, Piette F. Neuropathological staging of Alzheimer-related changes correlates with psychometrically assessed intellectual status. In: Corian B, Iqbal K, Nicolini M, Winblad B, Wisniewski H, Zatta PF, editors. Alzheimer’s disease: advances in clinical and basic research. Third International Conference on Alzheimer’s Disease and Related Disorders. Chichester: Wiley, 1993. p. 131–7. [9] Buschke H. Selective reminding for analysis of memory and learning. J Verbal Learning Verbal Behav 1973;12:543–50. [10] Buschke H. Cued recall in amnesia. J Clin Exp Neuropsychol 1984; 6:433– 40. [11] Dickson DW. The value of cross sectional neuroanatomical studies as a conceptual framework for prospective clinicopathological studies. Neurobiol Aging 1997;18:382– 6. [12] Dickson D, Crystal H, Mattiace LA, Masur D, Blau A, Davies P, Yen SH, Aronson M. Identification of normal and pathological aging in prospectively studied nondemented elderly humans. Neurobiol Aging 1991;13:179 – 89. [13] Grober E, Buschke H. Genuine memory deficits in dementia. Dev Neuropsychol 1987;3:13–36. [14] Grober E, Buschke H, Crystal H, Bang S, Dresner R. Screening for dementia by memory testing. Neurology 1988;38:900 –3. [15] Grober E, Lipton R, Hall C, Crystal H. Memory impairment on Free and Cued Selective Reminding predicts dementia. Neurology 1999;in press. [16] Grober E, Kawas C. Learning and retention in preclinical and early Alzheimer’s Disease. Psychol Aging 1997;12:183– 8.
579
[17] Grober E, Merling A, Heimlich T, Lipton RB. Free and cued selective reminding and selective reminding in the elderly. J Clin Exp Neuropsychol 1997;19:643–54. [18] Hall CB, Lipton RB, Sliwinski M, Stewart WF. A change point model for estimating the onset of cognitive decline in preclinical Alzheimer’s disease. Statistics Med 1999, in press. [19] Khachaturian ZS. Diagnosis of Alzheimer’s disease. Arch Neurol 1985;42:1097–105. [20] Linn RT, Wolf PA, Bachman DL, Knoefel JE, Cogg JL, Belanger AJ, Kaplan EG, D’Agostino RB. The preclinical phase of probable Alzheimer’s disease. Arch Neurol 1995;52:485–90. [21] Masur DM, Sliwinski M, Lipton RB, Blau AD, Crystal HA. Neuropsychological prediction of dementia and the absence of dementia in healthy elderly persons. Neurology 1994;44:1427–32. [22] McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan E. Clinical diagnosis of Alzheimer’s disease: report of the NINCDSADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 1984;34:939 – 44. [23] Mirra SS, Heyman A, McKeel D, Sumi SM, Crain BJ, Brownlee LM, Vogal FS, Hughes JP, van Belle G, Berg L. The consortium to establish a registry for Alzheimer’s Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer’s Disease. Neurology 1991;41:479 – 86. [24] Morris JC, Storandt M, McKeel DW, Rubin EH, Price JL, Grant EA, Berg L. Cerebral amyloid deposition and diffuse plaques in “normal” aging: Evidence for presymptomatic and very mild Alzheimer’s disease. Neurology 1996;46:707–19. [25] Nagy Z, Yilmazer-Hanke DM, Braak H, Braak E, Schultz C, Hanke J. Assessment of the pathological stages of Alzheimer’s disease in thin paraffin sections: a comparative study. Dementia Geriatr Cognitive Dis 1998;9:140 – 4. [26] Nagy Z, Vatter-Bittner B, Braak H, Braak E, Yilmazer DM, Schultz C, Hanke J. Staging of Alzheimer-type pathology: an interraterintrarater study. Dementia Geriatr Cognitive Dis 1997;8:248 –51. [27] Peterson RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. Mild Cognitive Impairment: Clinical Characterization and Outcome. Arch Neurol 1999;56:303– 8. [28] Poulos B, Crystal H, Katz M, Kuslansky G, Grober E, Lipton R, Dickson DW. Amyloid burden determined by image analysis correlates with conventional histochemical methods and clinical measures of cognitive decline. J Neuropathol Exp Neurol 1997;56:615. [29] Price JL, Morris JC. Tangles and plaques in nondemented aging and “preclinical” Alzheimer’s disease. Ann Neurol 1999;45:358 – 68. [30] Ritchie K, Fuhrer R. A comparative study of the performance of screening tests for senile dementia using receiver operating characteristics analysis. J Clin Epidemiol 1992;45:627–37. [31] Tierney MC, Szalai JP, Snow G, et al. Prediction of probable Alzheimer’s disease in memory-impaired patients: a prospective longitudinal study. Neurology 1996;46:661–5. [32] Welsh K, Butters N, Hughes J, et al. Detection of abnormal memory decline in mild cases of Alzheimer’s Disease using CERAD neuropsychological measures. Arch Neurol 1991;48:278 – 81.