Superior Frontal Cortex Cholinergic Axon Density in Mild Cognitive Impairment and Early Alzheimer Disease

Background  Loss of cortical choline acetyltransferase (ChAT) activity contributes to end-stage Alzheimer disease (AD) dementia. In general, ChAT activity levels are stable in the neocortex in mild to moderate AD (mAD) and there is a selective up-regulation in the superior frontal cortex (SFC) in mild cognitive impairment (MCI), indicating a transient, region-specific cholinergic neuroplastic response.

Objective  To assess whether a proliferation of cholinergic axons underlies increased ChAT activity levels in the SFC in subjects with MCI.

Design  Stereologic principles were applied to assess the density of ChAT-immunoreactive fibers and axon varicosities in SFC tissue obtained postmortem from subjects with no cognitive impairment, MCI, and mAD.

Subjects  Thirty-six subjects enrolled in the Religious Orders Study, with records of annual clinical evaluation for frontal lobe specific and global cognitive functions.

Results  Compared with the group with no cognitive impairment, SFC ChAT-immunoreactive fiber and axon varicosity densities were not altered in the MCI group but were significantly reduced in the group with mAD and correlated with impaired frontal lobe and global cognitive function.

Conclusions  The lack of an increase in cholinergic axonal innervation of the SFC in MCI suggests that structural reorganization of cholinergic profiles is not the mechanism underlying the transient cholinergic plasticity reported in this region. Furthermore, the stability of cholinergic enzyme activity in mAD is likely the result of a biochemical up-regulation of ChAT protein or enzyme activity levels in the SFC, compensating for decreased regional cholinergic fibers and axon varicosities.

Deficits in cortical cholinergic enzyme activity correlate with cognitive impairment in Alzheimer disease (AD).^1-4 In early AD (mild to moderate AD [mAD]), cholinergic enzyme activity levels are preserved in all cortical regions.^5-7 Subjects with mild cognitive impairment (MCI) exhibit increased choline acetyltransferase (ChAT) activity levels in the superior frontal cortex (SFC),⁷ indicating a region-specific presynaptic cholinergic plasticity response before the onset of AD dementia. In addition, subjects with MCI have elevated levels of drebrin, a postsynaptic dendritic spine marker, in the SFC,⁸ indicating that changes occur both presynaptically and postsynaptically in this region during preclinical AD. We hypothesized that cholinergic plasticity in the SFC in subjects with MCI reflects increases in ChAT-containing fiber and axon varicosity densities. To test this hypothesis, we quantified ChAT immunoreactive (ChAT-ir) fibers and axon varicosities in SFC tissue obtained from subjects enrolled in the Religious Orders Study⁹ who died with a premortem clinical diagnosis of MCI, mAD, or no cognitive impairment (NCI). In addition, results were correlated with clinical measures of global and frontal cortex–specific cognitive function tests.

Tissue was obtained postmortem from 36 participants in the Religious Orders Study, a longitudinal clinicopathologic study of aging and AD in Catholic nuns, priests, and brothers.^7,9 Clinical evaluation was based on tests obtained within 12 months before death and consisted of a global cognitive score⁹ and assessment of frontal lobe function using a composite z score from the following 5 cognitive tests: Consortium to Establish a Registry for Alzheimer's Disease (CERAD) Word List, Immediate Recall Trials 1 through 3; Digits Backward; Digits Ordering; Symbol Digit Modalities Test, Oral Version; and Standard Progressive Matrices. Subjects were categorized as having NCI (n = 11), MCI (n = 11), or mAD (n = 14) (Table). Subjects with MCI had cognitive impairment insufficient to meet criteria for dementia.⁹ The AD group represented subjects with mild to moderate dementia based on their Mini-Mental State Examination (MMSE) scores (Table). The diagnosis of AD dementia was based on standard criteria.¹⁰ The study was approved by the human investigations committees of Rush University Medical Center, Chicago, Illinois, and the University of Pittsburgh, Pittsburgh, Pennsylvania.

Tissue blocks containing the middle third of the SFC (Brodmann area 9) were dissected as previously described.^11,12 Each case was coded and assigned neuropathologic diagnoses that included Braak staging of neurofibrillary tangle pathologic findings,¹³ CERAD,¹⁴ and NIA-Reagan (National Institute on Aging–Reagan Institute Working Group) 1997 diagnoses (Table) by a board-certified neuropathologist. Cases were excluded if tumors, large strokes, multiple lacunar infarctions, signs of encephalitis, or pathologic features of Parkinson disease were detected. Chromogen-based immunohistochemistry was performed^15-17 using a polyclonal goat anti-ChAT antibody (1:100, antihuman placental enzyme, code AB144P, lot No. 23031288; Chemicon International, Temecula, California). The specificity of this antibody for human ChAT has been reported previously.¹⁶

Densities of ChAT-ir fibers and associated varicosities were determined on 2 tissue sections chosen randomly from the processed series using a stereologic sampling-based method. Inasmuch as it was impossible to determine the total volume of Brodmann area 9 because of the limited number of sections available, density measurements were expressed in arbitrary units. Previous studies of the neocortex in subjects with AD and neurologically healthy subjects demonstrated lamina-specific variations in ChAT-ir fiber densities¹⁸ and ChAT enzyme activity.¹⁹ Accordingly, ChAT-ir fibers and axon varicosities were quantified separately in the superficial and deep laminae of the SFC, determined on adjacent matching tissue sections counterstained with cresyl violet to aid in cytoarchitecture analysis.²⁰ For each section, 6 evenly spaced regions of interest (ROIs) (×40 magnification optical fields positioned parallel to the cortical surface) were captured in the superficial gray matter (laminae II and III) and 6 corresponding ROIs in the deep gray matter (laminae V and VI). This resulted in up to 12 superficial and 12 deep ROIs per case. Under the immunolabeling conditions used in this study, the ChAT antibody penetrates throughout the full depth of the section.¹⁷ All images were captured at a z-axis level that resulted in the greatest concentration of fibers in clear focus. Each imaged ROI was superimposed with a cycloid or point-counting grid for analyses of fibers or axon varicosities, respectively.²¹ In each ROI, we counted intersections of cycloids with all ChAT-ir fibers, regardless of their size. For the axon varicosity analysis, ChAT-ir swellings associated with ChAT-ir fibers and that were in clear focus were counted in 4 randomly chosen grid boxes within the original ROI.

Results of ChAT-ir fiber and axon varicosity density analyses were compared among diagnostic groups using 1-way analysis of variance, with Tukey's method for post hoc comparisons. One-way analysis of variance and the Fisher exact and Kruskal-Wallis tests were applied to the comparison of clinical, demographic, and pathologic variables across diagnostic groups. Associations of ChAT-ir fiber and axon varicosity data with clinical, demographic, and neuropathologic variables were examined using the Spearman rank correlation or the Wilcoxon rank sum test. The level of statistical significance for all tests was set at .01 (2-sided).

There were no differences in postmortem interval, age, years of educational achievement, sex, or presence of the apolipoprotein ε4 allele among the 3 diagnostic groups (NCI, MCI, and mAD; Table). The groups differed in Braak scores (Table), and there was a trend toward a difference in NIA-Reagan and CERAD diagnoses (Table). For tests of general cognition (MMSE and Global Cognitive Score; Table) and for frontal lobe function, the NCI and MCI groups scored significantly better than the mAD group (Table).

In all subjects examined, ChAT-ir fibers and axon varicosities in the SFC were observed as a network of long processes with multiple swellings throughout the gray matter, with greater densities in more superficial (laminae II and III) compared with deep gray matter (laminae V and VI) (Figure 1). ChAT-ir fiber and axon varicosity densities in the SFC were comparable between the NCI and MCI groups but were significantly reduced in the mAD group P < .01), with the greatest differences in the SFC superficial laminae (Figure 1 and Figure 2). The results remained unchanged after correcting for cortical laminar thickness (data not shown) or when the AD group included only those subjects with an MMSE score of 20 or higher (n = 11) (data not shown). ChAT-ir fibers often contained abnormally enlarged varicosities in subjects with mAD (Figure 3).

Overall, ChAT-ir fiber and axon varicosity densities were significantly correlated (superficial gray matter, r = 0.83; deep gray matter, r = 0.78; both P < .001); similarly, ChAT-ir profile densities in superficial vs deep laminae correlated strongly (fibers, r = 0.95; varicosities, r = 0.82; both P < .001). The following correlative analyses of ChAT-ir profiles with demographic data, pathologic findings, and cognitive function are based on total (combined superficial and deep lamina) values. No association was found between ChAT-ir fiber and axon varicosity densities with any demographic variable. Both ChAT-ir profiles correlated directly with scores on the MMSE (fibers, r = 0.57, P < .001; varicosities, r = 0.53, P < .001) and Global Cognitive Score (fibers, r = 0.52, P < .001; varicosities, r = 0.50, P = .001) and with the composite z score for frontal lobe function (fibers, r = 0.54, P < .001; varicosities, r = 0.54, P < .001). There was a trend toward an inverse correlation of ChAT-ir fiber and varicosity densities with frequencies of neuritic amyloid plaques (fibers, r = −0.26, P = .06; varicosities, r = −0.33, P = .03) and diffuse amyloid plaques (fibers, r = −0.13, P = .22; varicosities, r = −0.35, P = .02) in the SFC.

The present study demonstrates that cholinergic projections to the SFC are not altered in MCI but are markedly reduced in early stages of AD. The distribution of cholinergic axonal profiles in the SFC, with greater fiber density in superficial laminae across all clinical diagnostic groups, supports previous morphologic^18,22 and laminar microchemical analyses of ChAT activity in the SFC.¹⁹ Ultrastructural electron microscopic studies revealed that ChAT-ir fibers and varicosities are thin unmyelinated axons and boutons containing synaptic vesicles²³ that make presynaptic perikaryal contacts.²² Because as many as 67% of cortical cholinergic axon varicosities make synaptic contacts,²⁴ the reduction of ChAT-ir fibers and varicosities observed in the SFC in subjects with mAD in the present study likely reflects loss of cholinergic synaptic specializations.

Stable densities of SFC ChAT-ir fibers and axon varicosities in MCI and their loss in mAD was not expected in light of our previous biochemical findings of increased ChAT activity in this region in subjects with MCI compared with subjects with NCI or mAD from the same cohort.⁷ Together, these observations suggest that in MCI, plasticity of the cholinergic system occurs through biochemical up-regulation of ChAT enzyme activity rather than an increase (sprouting) in fiber projections to the SFC. In mAD, SFC ChAT-ir axonal projections are reduced (present study), whereas biochemically measured ChAT enzyme activity is comparable to levels observed in NCI,⁷ indicating that a biochemical up-regulation also occurs in mAD. This agrees with the suggestion that during AD progression, synapse loss precedes loss of ChAT activity in the SFC.⁶ Increased ChAT activity⁷ and drebrin protein levels⁸ in MCI and up-regulation of the high-affinity choline transporter in AD²⁵ indicate that multiple compensatory mechanisms are recruited to maintain cholinergic neural transmission in the frontal cortex in MCI and mAD. Similar compensatory mechanisms may occur in the hippocampus, which is also characterized by increased ChAT activity levels in MCI.⁷

The preservation of ChAT immunoreactivity,¹⁷ despite changes in high-affinity trkA¹¹ and low-affinity p75^NTR neurotrophin receptors¹² in cholinergic basal forebrain neurons suggests that the increase in ChAT enzyme activity levels in MCI⁷ and its stability despite loss of cholinergic axons and varicosities in mAD results from altered metabolic status of cholinergic basal forebrain neurons²⁶ that project to the SFC. Alternatively, increased ChAT activity levels could result from impaired axonal transport reflected by enlarged ChAT-ir axon varicosities in the SFC in subjects with mAD. Similar abnormal ChAT-ir axon varicosities were described in previous studies of AD,^27,28 and their functionality remains to be determined.

There were several methodologic limitations in this study. Although our sampling approach was based on unbiased stereologic principles, the entire SFC was unavailable for evaluation. Therefore, a systematic uniformly random sampling of the entire SFC was impossible. This also prevented estimates of SFC volume and precluded conversion of density estimates into estimates of total numbers of fibers and axon varicosities. There is no stereologic method for uniform sampling of cortical structures for laminar analyses. Any biases associated with our sampling approach are limited, based on the assumption that fiber innervation and the cytoarchitecture of the SFC are homogeneous if consistently sampled perpendicular to the pial surface, as in the present study. Potential biases were further minimized by maintaining strict sampling and analytical procedures, enabling fair comparisons among the 3 clinical diagnostic groups.

Stability of the SFC cholinergic infrastructure and increased ChAT activity levels in MCI could explain why cholinesterase inhibitors are not as effective in these patients as one would expect.²⁹ Reduced densities of SFC ChAT-ir fibers and varicosities in mAD may hinder the ability of the cholinergic system to up-regulate ChAT activity sufficiently in more advanced disease stages. Failure of this compensatory mechanism may contribute to clinical deterioration during the transition from MCI to mAD. The goal of future therapies should be to sustain neocortical cholinergic afferents by supporting the viability and functionality of cholinergic basal forebrain neurons³⁰ in an attempt to prevent or delay the onset of clinical dementia or to impede further decline in individuals with the earliest signs of cognitive dysfunction.

Correspondence: Steven T. DeKosky, MD, Department of Neurology, University of Pittsburgh School of Medicine, 341 Fifth Ave, Ste 811, Pittsburgh, PA 15213 (DeKoskyST@upmc.edu).

Accepted for Publication: January 12, 2007.

Author Contributions:Study concept and design: Ikonomovic, Abrahamson, Mufson, and DeKosky. Acquisition of data: Ikonomovic, Abrahamson, and Isanski. Analysis and interpretation of data: Ikonomovic, Abrahamson, Isanski, Wuu, Mufson, and DeKosky. Drafting of the manuscript: Ikonomovic, Abrahamson, Mufson, and DeKosky. Critical revision of the manuscript for important intellectual content: Ikonomovic, Abrahamson, Isanski, Wuu, Mufson, and DeKosky. Statistical analysis: Abrahamson and Wuu. Obtained funding: Ikonomovic, Mufson, and DeKosky. Administrative, technical, and material support: Ikonomovic, Mufson, and DeKosky. Study supervision: Ikonomovic, Abrahamson, and DeKosky.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grant P01 AG14449 from the National Institute on Aging, National Institutes of Health.

Additional Contributions: David A. Bennett, MD, Elizabeth J. Cochran, MD, Julie Schneider, MD, and Traci Colvin, MPH, oversaw data collection and neuropathologic evaluations, and James T. Becker, PhD, assisted with cognitive domain data analyses.