The Consortium to Establish a Registry for Alzheimer's Disease (CERAD). Part X. Neuropathology Confirmation of the Clinical Diagnosis of Alzheimer's Disease

The Consortium to Establish a Registry for Alzheimer's Disease (CERAD), a longitudinal multicenter study, was formed in 1986 to address the need for standardized instruments to evaluate patients clinically diagnosed as having Alzheimer's disease (AD). To facilitate the collection and pooling of reliable data obtained from multiple medical centers in the United States and abroad, and to promote communication among investigators, CERAD has developed clinical, neuropsychological, and neuroimaging assessment batteries ^[1,2]. CERAD also developed a standardized neuropathology protocol ^[3] to establish the neuropathologic diagnosis and to provide important correlations with clinical data and will continue to refine the protocol as appropriate to reflect new technical and scientific developments.

Although there are established clinical criteria for the diagnosis of AD, ^[4] definitive diagnosis still rests upon neuropathologic confirmation. The accuracy of the clinical diagnosis of AD is critical: allocation of public health resources, patient selection for therapeutic trials, and other clinical studies all depend upon access to reliable clinical information. To date, CERAD has enrolled more than 1,200 patients with a clinical diagnosis of AD and approximately 500 nondemented control subjects, about one-half of whom are spouses of subjects. These individuals have been evaluated using the standardized clinical and neuropsychological batteries and the data entered into the CERAD central database. At this time, autopsies have been performed on 134 of the 258 patients who have died, yielding a 52% autopsy rate. The present study describes the concordance of clinical and neuropathologic diagnoses in 106 dementia patients on whom neuropathologic study has been completed.

Methods. Patient entry and follow-up. Subjects diagnosed as having AD were recruited from participating CERAD centers as described by Morris et al ^[1]. Briefly, these patients met the CERAD criteria of (1) being age 50 years or older, (2) being able to speak and comprehend English, and (3) having a cooperative caregiver available to provide adequate history and assist with follow-up. Patients with very mild cognitive impairment were excluded, as were patients with other neurologic, medical, or psychiatric disorders having the potential to impair cognitive function. The diagnostic criteria for AD were adapted with minor modification from those of McKhann et al ^[4].

The clinical assessment battery was designed to provide experienced clinicians with the minimum information needed to make a confident diagnosis of probable AD ^[1] and included an informant-based modified Blessed Dementia Scale and the six-item Short Blessed Orientation-Memory-Concentration Test ^[5]. The Clinical Dementia Rating scale ^[6-8] was used to stage the global severity of dementia. Following initial clinical evaluation, a standardized neuropsychological assessment battery was administered by experienced psychometricians. The tests constituting this battery were chosen for their ability to detect the principal cognitive deficits of AD and to follow changes in these deficits throughout the course of the disease ^[1,9,10]. The clinical and neuropsychological assessment batteries were readministered at annual intervals.

Neuropathologic evaluation. Procedure used to establish a diagnosis of AD. Participating neuropathologists used their own routine fixation and staining methods to examine multiple regions of neocortex, entorhinal cortex, hippocampus, and subcortical structures. The frequency of neuritic plaques and neurofibrillary tangles was evaluated using the CERAD neuropathology protocol, ^[3] which involves (1) semiquantitative assessment of senile plaque and neurofibrillary tangle frequency in various brain regions, (2) determination of an age-related plaque score based on the patient's age at death and the semiquantitative measure of neuritic plaques in the neocortex, and (3) integration of the age-related plaque score with clinical information regarding the presence or absence of dementia to determine the level of certainty of the neuropathologic diagnosis of AD. Semiquantitative measures are used in the CERAD neuropathology protocol to promote standardization and to facilitate pooling of data from cases studied at multiple centers; reasonable interrater reliability for multicenter data has been observed for semiquantitative measures of plaque and tangle frequency, but significant interrater differences have been found for quantitative plaque and tangle counts ^[11].

Diagnostic criteria for Parkinson's disease (PD). PD pathology was defined as the presence of nigral degeneration (gliosis, neuronal loss, and pigmentary incontinence) and Lewy bodies at any site or the presence of significant nigral degeneration in the absence of any other disorder or disorders clearly explaining this change--eg, progressive supranuclear palsy, in which nigral degeneration is commonly observed ^[12,13]. The following working diagnostic scheme was used: if PD changes as defined above were accompanied by a clinical diagnosis of PD, a diagnosis of "definite PD" was made; in the absence of a clinical diagnosis of PD, a diagnosis of "uncertain PD" was assigned. In one case, the nigra was unavailable for assessment, but since Lewy bodies were found in multiple cortical and subcortical sites, a diagnosis of uncertain PD was permitted. (The increasing awareness of the clinical-pathologic spectrum of PD and related disorders, eg, diffuse Lewy body disease ^[14] or the Lewy body variant of AD, ^[15] the relationship of these disorders to AD, and the controversy regarding their nosology have prompted revision of the CERAD neuropathology protocol and adoption of a more descriptive approach to PD changes) ^[16].

Assessment of vascular changes. Gross evidence of cerebrovascular disease, eg, degree of atherosclerosis of vessels of the circle of Willis and major branches, was documented along with the nature, size, and general location of large infarcts (>10 mm in diameter), lacunes (cystic lesions <10 mm in diameter), small infarcts (lesions <10 mm in diameter without cystic change), and hemorrhages. A general assessment of microscopic changes was made, including arteriosclerosis and arteriolar sclerosis, microinfarcts, and periventricular or diffuse pallor of myelin. This protocol was designed to provide a broad overview of the range of vascular pathology and a framework upon which to build a more comprehensive instrument.

Results. Case material. Autopsy records on the 106 dementia subjects included in this report were derived from 20 institutions. The demographic features of these patients are summarized in table 1. Of the 106 subjects, 101 (95.3%) were white and five (4.7%) were black. All patients carried a clinical diagnosis of probable (n = 83) or possible (n = 23) AD.

Neuropathologic evaluation. The clinical diagnosis of AD was confirmed in 92 (86.8%) of the 106 patients (table 1). In an additional six patients, mild to moderate AD pathology was observed but was not believed by the neuropathologist to be the primary cause of the patient's dementia.

Cases with a primary neuropathology diagnosis of AD. Of the 92 cases with a primary neuropathologic diagnosis of AD, 19 (20.6%) had coexistent PD changes. Four of these 19 had a clinical diagnosis of PD along with the clinical diagnosis of probable AD. Nigral degeneration was observed in the 18 cases in which the nigra was available for study. Subcortical Lewy bodies, present in all 19 cases, were seen consistently in the substantia nigra and frequently in the locus ceruleus or nucleus basalis, as well. Cortical Lewy bodies, however, were found in only 11 of the 19 cases (58%)--in the entorhinal cortex, the neocortex, or both. The number of cases with cortical Lewy bodies would probably have been higher had additional regions of the neocortex, eg, anterior cingulate or insular cortex, been sampled.

Vascular lesions were noted in 26 (28.3%) of the 92 neuropathologically confirmed AD cases (table on file with the National Auxiliary Publications Service; see Note at end of article). Regardless of the nature or extent of these vascular lesions, AD was still considered by the neuropathologists to be the primary dementing illness in all cases. Twenty-three of the 26 cases had a neuropathologic diagnosis of "definite AD"; the remaining three were diagnosed as "probable AD" cases. No clear pattern emerged regarding the distribution or the nature of vascular lesions; infarcts of varying size, lacunes, and hemorrhages were observed in disparate sites. More than one type of vascular lesion (infarct, hemorrhage, lacune) was encountered in nine of the 26 cases (35%). Four of the 26 cases showed concomitant PD changes.

Cases in which AD changes were not interpreted as the primary neuropathology. Seven of the 14 patients (50%) in whom AD was not confirmed as the primary neuropathologic diagnosis carried a clinical diagnosis of "possible AD"; the remaining seven had a clinical diagnosis of probable AD. The 14 patients had been evaluated at 11 of the 20 institutions participating in the study; thus, occasional failure to confirm the clinical diagnosis of AD was not confined to one or two centers. In no instance was a potentially treatable disorder mistakenly diagnosed as AD.

The cases in which AD was not interpreted as the primary dementing illness can be divided into four general groups: PD and related disorders (table 2, cases 1 to 5), hippocampal and/or entorhinal sclerosis (table 2, cases 6 to 8), miscellaneous other disorders (table 2, cases 9 to 11), and cases with no significant pathology observed (table 2, cases 12 to 14). In six of the 14 cases, a concomitant neuropathologic diagnosis of AD (using CERAD neuropathology criteria of definite (case 7), probable (case 10), or possible AD (cases 3, 4, 8, and 11) ^[3]) was made. Case 7 had only sparse to moderate neocortical plaques; because of the patient's age, the age-related plaque score met CERAD criteria for definite AD despite the relatively mild AD pathology. While hippocampal sclerosis, entorhinal sclerosis, or both occur in some cases of AD, the neuropathologist interpreted the severe neuronal loss and gliosis as the predominant change underlying the cognitive impairment in this patient. Case 10 showed sparse neuritic plaques in the neocortex adequate for a secondary diagnosis of probable AD. The pathologist interpreted the severe neuronal loss without gliosis, neurofibrillary tangles, or plaques in the hippocampus, along with the severe gliosis in the frontal and temporal cortex and white matter, as the key changes contributing to the dementia. The remaining four cases with secondary diagnoses of possible AD showed mild AD changes with another predominant neuropathology: corticobasal ganglionic degeneration; thalamic and hypothalamic gliosis; hippocampal, amygdala, and entorhinal sclerosis; and diffuse Lewy body disease.

In summary, the neuropathologic findings showed that the 106 cases clinically diagnosed as probable AD fell into four major categories: (1) AD without evidence of other disorder, (2) AD with coexistent PD changes, (3) AD with coexistent vascular changes, and (4) cases in which the clinical diagnosis of AD was not confirmed or the AD pathology, although present to some degree, was not considered to be the primary cause of the dementia. Some overlap, however, was observed between the AD+PD and AD+vascular groups: four cases showed combined AD, PD, and vascular changes.

Discussion. Clinical diagnostic accuracy. To our knowledge, this is the first prospective clinical-pathologic study pooling data from multiple centers on subjects clinically diagnosed with AD where both the clinical and the neuropathologic features were evaluated with standardized assessment instruments. The neuropathologists confirmed the clinical diagnosis of possible or probable AD as the primary cause of dementia in 92 (87%) of our 106 patients. Among the 83 patients carrying a clinical diagnosis of probable AD, the diagnostic accuracy was 92%, and one-half of the 14 patients in whom AD was not confirmed as the primary dementing illness were clinically diagnosed as having possible AD. Previous clinical-pathologic studies have yielded a wide range of diagnostic accuracy (46 to 100%), as reported by Mendez et al ^[17]. In reviewing the literature as well as their own retrospective analysis of 650 cases diagnosed as AD by numerous practitioners, these authors concluded that diagnostic accuracy has improved over time with the increasing use of formal clinical criteria. Our results support this notion.

In the present study, the 14 subjects for whom the neuropathologists did not interpret AD as the primary cause of dementia exhibited diverse findings at autopsy. Nonetheless, many of these cases displayed hippocampal sclerosis, nigral degeneration, or frontal lobe degeneration; in nine of the 14 cases, one or more of these findings were designated as dominant neuropathologic features. Neuronal loss in these three regions, with and without coexistent changes, is recognized as a pathologic substrate of progressive dementia that may be clinically indistinguishable from AD ^[18-22]. This compendium of changes may include cases described as hippocampal sclerosis, ^[19,20] frontal lobe dementia, ^[18,22] lobar atrophy without Pick bodies (for review, see Hulette and Crain) ^[23], and mesolimbocortical dementia ^[24]. The question of whether this group of changes represents a distinct clinical-pathologic entity may be resolved as more cases are documented. Rinne et al ^[25] proposed nigral degeneration as a substrate of dementia, and indeed, dementia may be a prominent feature in PD, ^[26] diffuse Lewy body disease, ^[14,15] corticobasal ganglionic degeneration, ^[27] and Hallervorden-Spatz disease ^[28,29]. Nigral degeneration, interpreted as a primary neuropathologic finding in five of the 14 "non-AD" cases in our study, was also present to varying degrees in two other non-AD cases: a case of adult polyglucosan disease and a case of frontal lobe degeneration, both of which have been associated with dementia ^{[21,22,30,31]}. It is uncertain whether the mild gliosis described in the thalamus and hypothalamus in one case contributed significantly to the patient's cognitive deficit. Finally, the absence of any correlative pathology in three dementia cases is puzzling but not without precedent: in their series of 650 dementia cases clinically diagnosed as AD, Mendez et al ^[17] found no significant pathology in 11 cases.

Revision of the CERAD protocol. Our neuropathologic findings demonstrate the frequency with which PD or vascular changes may coexist with AD. To provide a more comprehensive evaluation of PD and vascular changes with and without concomitant AD pathology and to update the assessment of AD-related changes, eg, neuritic plaques versus diffuse plaques, we recently revised the CERAD neuropathology protocol.

Parkinson's disease and Lewy bodies. The revised CERAD neuropathology protocol ^[16] documents the frequency of Lewy bodies in the nigra and other subcortical sites; the presence or absence of neocortical Lewy bodies is also determined. Examination of the anterior cingulate cortex, a common site for cortical Lewy bodies, is recommended on all dementia cases. Furthermore, when nigral but not cortical Lewy bodies are observed on hematoxylin-eosin sections, ubiquitin immunohistochemistry on neocortical sections is recommended to enhance detection of cortical Lewy bodies. To further promote recognition of cortical Lewy bodies, the illustrated portion of the protocol depicts the morphologic spectrum of cortical Lewy bodies. The presence of spongiform change, commonly seen in certain regions along with cortical Lewy bodies, ^[32] is also documented in the revised protocol. Finally, in the absence of consensus among neuropathologists on nomenclature for diagnostic criteria for PD-related conditions, ^[33] eg, diffuse Lewy body disease, the approach taken in the revised protocol is more descriptive.

Vascular changes. To promote participation and cooperation, we did not require detailed data on vascular changes in the original CERAD neuropathology protocol. However, because of increasing interest in more precise clinical-pathologic correlation of vascular changes and dementia, we extensively revised the section on vascular pathology to include more comprehensive information on the size (expressed as a percentage of the volume of a hemisphere or nucleus, eg striatum or thalamus), age, and distribution (including vascular territory and laterality) of vascular lesions. We also added to the protocol an optional section for correlation with available neuroimaging studies, along with specific suggestions and illustrations on the handling of standard sections of cerebral white matter.

Alzheimer's disease. Investigators have affirmed the need for continued reevaluation and validation of diagnostic criteria for AD ^[34,35]. Diffuse plaques, seen in AD patients, patients with other neurodegenerative diseases, and the cognitively intact elderly, ^[36,37] are increasingly recognized as a distinct plaque subtype. Diffuse plaques not only occur in the cerebral cortex but are commonly present in subcortical locations such as the corpus striatum, ^[13] cerebellum, ^[38] diencephalon, and hypothalamus ^[39] in AD. In addition, several studies ^[40-42] have challenged previous notions that plaques and amyloid burden accumulate steadily with increasing age or duration of illness. To help resolve questions about the significance of diffuse versus neuritic plaques and age-related progression of pathology, the revised CERAD neuropathology protocol includes semiquantitative ratings of frequency and relative proportions, as well as the distribution, of these senile plaque subtypes. This information will be particularly valuable for the study of hierarchical changes when collected in control subjects and individuals with mild dementia.

Although issues of diagnosis and interpretation remain unresolved, the CERAD database will continue to provide a framework in which standardized data may be entered and stored. These raw data can be retrieved and later used to modify diagnostic algorithms and answer other questions. The protocol is not static and will continue to be improved in concert with technical and scientific developments.

Notes. For information about the CERAD assessment batteries and access to the database, contact Albert Heyman, MD, P.O. Box 3203, Duke University Medical Center, Durham, NC 27710.

Readers can obtain a table of supplementary data (1 page) from the National Auxiliary Publications Service, c/o Microfiche Publications, P.O. Box 3513, Grand Central Station, New York, NY 10163-3513. Request document no. 05140. Remit with your order (not under separate cover), in US funds only, $7.75 for photocopies or $4 for microfiche. Outside the United States and Canada, add postage of $4.50 for the first 20 pages and $1.00 for each 10 pages of material thereafter, or $1.75 for the first microfiche and $.50 for each fiche thereafter. There is a $15.00 invoicing charge on all orders filled before payment.

Acknowledgments

The following neuropathologists contributed data to this study: H. Brent Clark, MD, PhD, University of Minnesota School of Medicine, Minneapolis, MN; Barbara J. Crain, MD, PhD, Duke University, Durham, NC; Pierluigi Gambetti, MD, Case Western Reserve University, Cleveland, OH; Lawrence A. Hansen, MD, University of California at San Diego, La Jolla, CA; E. Tessa Hedley-Whyte, MD, Massachusetts General Hospital, Boston, MA; John C. Hedreen, MD, The Johns Hopkins University School of Medicine, Baltimore, MD; Ann Marie Kazee, MD, University of Rochester Medical Center, Rochester, NY; William R. Markesbery, MD, University of Kentucky Medical Center, Lexington, KY; Ann C. McKee, MD, Massachusetts General Hospital, Boston, MA; Daniel W. McKeel, Jr., MD, Washington University School of Medicine, St. Louis, MO; Carol A. Miller, MD, University of Southern California Medical Center, Los Angeles, CA; Suzanne S. Mirra, MD, Emory University and Veterans Affairs Medical Center, Atlanta, GA; David Nochlin, MD, University of Washington School of Medicine, Seattle, WA; Daniel P. Perl, MD, Mt. Sinai Medical Center and Bronx Veterans Affairs Medical Center, New York, NY; Dushyant Purohit, MD, Mt. Sinai Medical Center and Bronx Veterans Affairs Medical Center, New York, NY; Ursula T. Slager, MD, Bishop, CA; S. Mark Sumi, MD, University of Washington, Seattle, WA; Robert Terry, MD, University of California at San Diego, La Jolla, CA; Richard M. Torack, MD, Washington University School of Medicine, St. Louis, MO; John Q. Trojanowski, MD, PhD, University of Pennsylvania School of Medicine, Philadelphia, PA; and Charles R. White, MD, University of Texas-Southwestern Medical Center, Dallas, TX.