Neuropathologic Outcome of Mild Cognitive Impairment Following Progression to Clinical Dementia FREE

The concept of mild cognitive impairment (MCI) has become a focus for study of the evolution of Alzheimer disease (AD).^{1 - 3} Early identification of patients destined to develop AD will allow for earlier intervention in an attempt to slow or halt the progression of disease. Clinically, subtypes of MCI have been recognized, with the amnestic subtype (aMCI) having an elevated risk of progressing to clinical AD.^{1 - 4} It is unclear whether the clinical diagnosis of aMCI predicts progression to pathologic AD. Pathologic confirmation of AD in these cases is important to validate these clinical observations and to justify our use of treatments for AD in aMCI.^{5 - 8}

Several small studies^{9 - 14} focusing on postmortem neuropathologic evaluation of MCI cases have concluded that many of these patients exhibit typical AD changes, including both neurofibrillary tangles and neuritic plaques. Not all cases of MCI show these changes, suggesting that pathologic heterogeneity exists.^{9 - 14} Cases of MCI with neuropathologic findings that range from a complete absence of significant AD pathologic abnormalities to sufficient pathologic abnormalities to warrant a firm neuropathologic diagnosis of AD have been reported. Other cases with predominant neuropathologic findings consistent with the diagnoses of vascular dementia, dementia with Lewy bodies, and argyrophilic grain disease (AGD) have been reported.^{9 - 14} The presence of neurofibrillary tangles and neuritic plaques in cognitively normal individuals further complicates the neuropathologic analysis, making the assessment of AD pathologic abnormalities quantitative rather than absolute.^{10 - 11 ,14 - 21} Detailed neuropathologic analyses need to be performed to validate the conclusion that aMCI represents early AD pathologically. Large-scale neuropathologic studies of aMCI are lacking owing to limited brain tissue availability.

An alternative strategy addressing the issue of clinicopathologic correlation in aMCI includes neuropathologic analysis of patients with aMCI who are prospectively followed up and have converted to dementia. These patients would be expected to show full development of the neuropathologic features underlying the cognitive impairment. This would facilitate the quantitative assessment of AD pathologic abnormalities as well as allow for higher sensitivity in detecting possible contributing pathologic abnormalities. Currently, such studies are lacking in the literature. Our study was undertaken to clarify unanswered questions and concerns related to the pathologic outcome of subjects progressing from aMCI to dementia.

We identified all of the subjects enrolled in the Mayo Alzheimer Disease Patient Registry community-based cohort who were diagnosed with aMCI from 1993 to 2001, were prospectively followed up, converted to dementia, and underwent neuropathologic examination.^{4 ,22} Nonamnestic forms of MCI were excluded from the present analysis. Neurologic testing, neuropsychologic examinations, neuroimaging, and laboratory analyses were performed at entry into the Mayo Alzheimer Disease Patient Registry and then annually or biennially during the follow-up period. Clinical diagnosis was determined by a consensus committee comprising neurologists, neuropsychologists, a neuropsychiatrist, nurse specialists, and a geriatrician following review of all of the available data. Inclusion criteria for the present analysis included a prospectively determined diagnosis of aMCI on at least 1 consensus evaluation,³ subsequent progression to clinical dementia, and neuropathologic evaluation. These studies have been approved by the Mayo Institutional Review Board, Rochester, Minn.

Clinical criteria for the diagnosis of aMCI included a memory problem, intact general cognitive functions and activities of daily living, evidence of cognitive dysfunction with predominant memory involvement on formal testing, and absence of dementia.³ Specific neuropsychologic cutoff scores were not used; rather, the subjects were diagnosed with aMCI if their memory performance was impaired out of proportion to their other cognitive domains as described previously.⁴ Cases were subclassified into single-domain or multiple-domain aMCI according to published criteria.^{1 - 2} Cognitive impairment in attention and executive function, language skills, or visuospatial skills in addition to memory impairment was used to classify subjects as having multiple-domain aMCI. We used Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition²³ or Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition²⁴ criteria for the diagnosis of dementia, and we used the National Institute on Neurologic and Communicative Disorders and Stroke/Alzheimer Disease and Related Disorders Association criteria²⁵ for the diagnosis of AD. Cognitive screening measures included the Mini-Mental State Examination,²⁶ Kokmen Short Test of Mental Status,²⁷ Clinical Dementia Rating scale,²⁸ Global Deterioration Scale,²⁹ and Mattis Dementia Rating Scale.³⁰ Apolipoprotein E (APOE) genotyping was performed on blood samples according to established protocols.³¹

The brains were processed in accordance with the recommendations of the Consortium to Establish a Registry for Alzheimer Disease (CERAD).³² Brain tissue from the left hemisphere was fixed in 10% to 15% buffered formalin for 7 to 10 days. Brain areas that were sampled included superior and middle frontal gyri (plane just anterior to temporal tip), superior and middle temporal gyri (plane of mammillary body), inferior parietal lobule (plane 1 cm behind posterior pole of splenium), calcarine (primary visual) cortex, anterior cingulate (plane of anterior commissure), hippocampus with adjacent inferior temporal cortex (level of the lateral geniculate body), amygdala and entorhinal cortex (level of the mammillary bodies), nucleus basalis, basal ganglia, cerebellum, thalamus with subthalamic nucleus, midbrain (with substantia nigra), pons (with locus coeruleus), and cerebellum as well as representative sections of any lesions noted grossly. Following routine processing in paraffin and cutting, sections were stained with hematoxylin-eosin. Selected sections were stained with modified Bielschowsky, Luxol fast blue and periodic acid–Schiff, and thioflavin-S and immunostained for β-amyloid (clone 6F/3D; Novocastra, Newcastle upon Tyne, England), phosphorylated tau (clone AT8; Endogen, Woburn, Mass), 4-repeat tau (ET3, recognizing exon 10 of the tau protein specifically³³ ; a gift from Peter Davies, PhD, Albert Einstein College of Medicine, Bronx, NY), α-synuclein (clone LB509; Zymed Laboratories, San Francisco, Calif), ubiquitin (polyclonal; Dako, Glostrup, Denmark), neurofilament protein (clone 2F11; Dako), α-β-crystallin (polyclonal; Chemicon, Temecula, Calif), and glial fibrillary acid protein (polyclonal; Dako).

Neuropathologic evaluation included classification according to criteria by Khachaturian,³⁴ CERAD,³² National Institute on Aging and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer's Disease,³⁵ and Braak staging of neurofibrillary degeneration.³⁶ Clinical information was available to the neuropathologist to assist in the final determination of CERAD assignment.³² Size, location, and histologic age of large and small vessel infarcts were recorded. Acute or subacute infarcts were not considered clinically significant with respect to chronic antemortem neurologic features. Microvascular disease, microinfarcts, amyloid angiopathy, and intracranial atherosclerosis were assessed using semiquantitative grading scales according to the National Alzheimer Coordinating Center protocol.³⁷ The presence or absence of vascular contributions to cognitive decline in these subjects was then dichotomized as being either present or absent according to consensus agreement between 2 examining neuropathologists (J.E.P. and D.W.D.). Lewy body pathology was analyzed on α-synuclein–immunostained sections and was categorized as brainstem, limbic, or neocortical.³⁸ Argyrophilic grain disease was visualized with both Gallyas stain and the ET-3 monoclonal antibody, and it was included as a pathologic diagnosis only if there was significant involvement of the medial temporal lobe, evidence for tau-positive coiled bodies in white matter, and “ballooned” neurons in the amygdala.^{33 ,39 - 40}

Following review by the 2 neuropathologists, a consensus diagnosis was rendered, taking into account the dominant neuropathologic features and the relative contributions of the neuropathologic findings in regard to the subjects' memory impairment and diagnosis of dementia. Final consensus neuropathologic diagnosis (AD vs non-AD) was used to group subjects for within-group and comparative analyses.

Rank sum test, Fisher exact test, and χ² tests were used to compare both quantitative and categorical aspects of the clinical and neuropathologic findings. P<.05 was considered statistically significant.

Thirty-five subjects fulfilled entry criteria. One patient meeting the criteria described earlier was excluded from the analysis, as the autopsy findings were rendered uninterpretable by the coincident development of a high-grade infiltrative astrocytoma. The demographic features of the study cohort are shown in Table 1. No significant differences were found between neuropathologically confirmed AD and non-AD groups for any of the demographic variables studied.

All of the 34 subjects had memory impairment at the time of initial diagnosis of MCI. Twenty-four subjects initially had isolated memory impairment and were classified as having aMCI.^{1 - 3} The remaining 10 subjects had deficits in other cognitive domains in addition to memory impairment and were classified as having multiple-domain aMCI.

As required by the entry criteria for this analysis, all of the subjects became demented over the course of clinical follow-up. Of the 24 subjects with aMCI, 19 received a final clinical diagnosis of AD; 9 of the 10 subjects with multiple-domain aMCI also had final clinical diagnoses of AD. Two subjects were given the clinical diagnosis of vascular dementia, and 1 was thought to have progressed to dementia with Lewy bodies. In 3 cases, the dementia was considered difficult to classify. Sixteen of the 34 subjects were thought to have additional contributing clinical diagnoses (Table 2).

The final consensus neuropathologic diagnosis of AD was assigned to 24 (71%) of 34 total subjects, 18 (75%) of 24 subjects who initially had aMCI, 6 (60%) of 10 subjects who had multiple-domain aMCI, and 21 (75%) of 28 subjects with the final clinical diagnosis of AD. The group of subjects with non-AD pathologic diagnoses included 3 subjects with Lewy body disease, 2 with hippocampal sclerosis, 2 with nonspecific tauopathy (1 of whom also had Binswanger disease), and 1 each with AGD, frontotemporal lobar degeneration with hippocampal sclerosis, and progressive supranuclear palsy (Table 3).

The neuropathologic analysis is shown in Table 4 and Table 5. There was variable expression of AD pathologic abnormalities among cases given non-AD pathologic diagnoses. Twenty-eight (82%) of the 34 subjects had 2 or more pathologic processes that were thought to contribute to the dementia. Twelve subjects (35%) had significant vascular disease (Table 6), 18 (53%) had AGD, and 9 (26%) had Lewy body disease.

No significant differences in cognitive measures or clinical scales were seen between cases with neuropathologically proven AD and those with other neuropathologic diagnoses. Between-group comparisons at onset of MCI, crossover to dementia, and final evaluation prior to death for all of the clinical variables are shown in Table 7.

The majority of patients initially diagnosed with aMCI who went on to develop dementia progressed both clinically and pathologically to AD. However, not all of the cases progressing from aMCI to dementia developed the clinical or pathologic features of AD. Multiple pathologic abnormalities were also present within some cases. Our study highlights the heterogeneous pathologic outcomes of aMCI following progression to dementia in a community-based cohort. The presence of cognitive impairments in addition to memory problems did not seem to predict who would have AD pathologically. These data raise the question of potential pathologic heterogeneity of subjects recruited into diagnostic and therapeutic trials in MCI.^{5 - 6}

Current evaluation of AD pathologic abnormalities in postmortem samples can be performed using criteria by Khachaturian,³⁴ CERAD,³² National Institute on Aging and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer's Disease,³⁵ or Braak staging.³⁶ These criteria have been developed with the goals of improved diagnostic accuracy and the development of universal standards. Using all of the 4 neuropathologic criteria, our data clearly demonstrate a high variability in clinicopathologic correlation rates depending on the neuropathologic criteria used (88% using CERAD criteria to 68% using National Institute on Aging and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer's Disease criteria). The use of different rating scales for assessment of AD pathologic abnormalities limits comparisons between studies. The provision of data using all of the current AD pathologic rating scales allows these data to be interpreted within the context of other reports in the literature.

These data support previous studies^{9 - 14} suggesting that pathologic abnormalities in the medial temporal lobe are the neuropathologic substrate for aMCI. The medial temporal lobe pathologic abnormality in the majority of cases was Alzheimer-type neurofibrillary degeneration. While neurofibrillary degeneration confined to limbic regions (Braak stages III-IV) can be seen in cognitively normal individuals, it is often associated with cognitive and functional impairment sufficient to meet Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition²⁴ criteria for dementia in others.^{9 - 11 ,14 - 21 ,36} Several cases in this series did not have significant neurofibrillary degeneration in limbic regions; rather, other pathologic processes including hippocampal sclerosis, AGD, and frontotemporal lobar degeneration were present. Overall, the same pathologic features seen in fully developed dementia after progression from aMCI are seen in a milder form in cases coming to autopsy with the clinical diagnosis of aMCI.⁴¹ Studies^{9 - 14} from other centers addressing clinicopathologic correlations in MCI have also shown similar results taking the variations in clinical and neuropathologic criteria used into account. Just as the clinical expression of MCI appears to represent an intermediate state between normal cognition and dementia reflecting medial temporal lobe dysfunction, the neuropathologic expression of MCI appears to be correlated with medial temporal pathologic abnormalities.

The pathologic heterogeneity observed in our study was similar to heterogeneity seen in cases that underwent autopsy with a final clinical diagnosis of aMCI.⁴¹ Several of the aMCI cases had evidence of hippocampal sclerosis or AGD in addition to Alzheimer-type pathologic abnormalities, suggesting that anatomy of neurodegeneration is more predictive of the clinical syndrome than the specific pathologic process. That which unites these various processes is their predilection for the medial temporal lobe.

The lack of a quantitative grading scale for vascular pathologic abnormalities in cognitive decline complicates a standardized interpretation of its contribution to dementia in this study; however, it is clear from the extent of vascular pathologic abnormalities and the distribution of the lesions (Table 6) that vascular disease plays an important role in the development of dementia following conversion from aMCI.⁹ Although a significant degree of variability among cases was seen for both large and small vessel infarcts, moderate to severe microvascular disease was almost universal. Only a single case had mild microvascular changes among those considered to have a vascular contribution to dementia, suggesting that microvascular disease may play a more important role in cognitive decline than overt cerebral infarction.⁴²

Argyrophilic grain disease increased in frequency from normal subjects (31%) to subjects with aMCI (47%) to subjects with dementia (53%), similar to the trends seen for neurofibrillary pathologic abnormalities and amyloidosis described earlier.^{20 ,41} Although the specificity of AGD to cognitive impairment has yet to be established, recent work⁴³ has suggested that it can be the sole substrate of dementia in cases with sufficient pathologic abnormalities. The role of contributing pathologic abnormalities to the diagnosis of dementia is important and should not be overlooked. Analysis of other autopsy series of community-based cohorts have shown similar results, suggesting that pure AD may be relatively uncommon.^{44 - 48}

No demographic or cognitive features predicted the final neuropathologic diagnoses of this cohort of subjects with dementia who transitioned from aMCI. The lack of association of apolipoprotein E status with final neuropathologic outcome may be secondary to the advanced age and low apolipoprotein ε4 allele frequency in this series. Additionally, the duration of aMCI or dementia did not correlate with either the degree of AD pathologic abnormalities or the final neuropathologic diagnosis. Moreover, none of the cognitive assessment measures used in this study predicted final pathologic findings. Further research using more detailed analyses of cognitive measures and tests from the full neuropsychometric battery may be useful in predicting incipient AD, allowing further refinement in the diagnostic criteria for aMCI.

The weaknesses of this study lie not only in its small sample size but also in a potential selection bias. Although prospectively recruited as part of a community-based study, the subjects were of relatively advanced age and may not be representative of aMCI at younger ages. Moreover, as in all autopsy studies, the analysis is limited by inherent selection bias as to who comes to postmortem evaluation. The average duration of MCI was only 3 years in the study cohort, suggesting a conversion rate in excess of that reported in community-based studies of MCI.^{1 - 2 ,4 ,6 ,49 - 50} It is possible that the rapid rate of disease progression and the underlying neuropathologic characteristics seen in this study are representative of only a subset of those patients diagnosed with aMCI. Further postmortem neuropathologic studies on patients progressing through aMCI to clinical dementia will be needed to validate these findings.

Correspondence: Ronald C. Petersen, PhD, MD, Mayo Clinic College of Medicine, 200 First St SW, Rochester, MN 55905 (peter8@mayo.edu).

Accepted for Publication: September 8, 2005.

Author Contributions:Study concept and design: Jicha, Parisi, Knopman, and Petersen. Acquisition of data: Jicha, Parisi, Dickson, Johnson, Ivnik, Tangalos, Boeve, Knopman, and Petersen. Analysis and interpretation of data: Jicha, Parisi, Dickson, Cha, Boeve, Knopman, Braak, and Petersen. Drafting of the manuscript: Jicha, Parisi, Johnson, Cha, and Petersen. Critical revision of the manuscript for important intellectual content: Jicha, Parisi, Dickson, Ivnik, Tangalos, Boeve, Knopman, Braak, and Petersen. Statistical analysis: Cha and Petersen. Obtained funding: Petersen. Administrative, technical, and material support: Jicha, Parisi, Dickson, Johnson, Ivnik, Tangalos, Boeve, Braak, and Petersen. Study supervision: Parisi, Boeve, Knopman, and Petersen.

Funding/Support: This study was supported by grants U01 AG06786 and P50 AG16574 from the National Institute on Aging and by the Robert H. and Clarice Smith and Abigail Van Buren Alzheimer's Disease Research Program.

Acknowledgment: We gratefully acknowledge Peter Davies, PhD, for supplying the ET-3 monoclonal antibody used in this study.