¹²³I-5-IA-85380 SPECT Imaging of Nicotinic Receptors in Alzheimer Disease and Mild Cognitive Impairment

Human Subjects

The study sample consisted of 12 subjects with probable AD (age, 75.0 ± 9.0 y [mean ± SD]), 10 subjects with MCI (73.8 ± 6.0 y), and 10 age-matched HC subjects (73.8 ± 8.3 y). All subjects were of European American descent and were nonsmokers. On the basis of responses to a smoking history questionnaire, 18 of the 32 subjects were never-smokers (<100 lifetime cigarettes). Six subjects with AD, 5 subjects with MCI, and 3 subjects with HC were former smokers but had not smoked for ≥20 y. Urine cotinine levels were not determined. All female subjects were postmenopausal. The demographic and cognitive characteristics of the study participants are shown in Table 1.

Screening Procedures and Neuropsychologic Assessment

Screening procedures included a medical history, physical and neurologic examinations, electrocardiography, serum chemistries, thyroid function studies, complete blood count, urinalysis, and urine toxicology screen. Subjects were also assessed with the Mini-Mental State Examination (MMSE) (28), NYU Paragraph (Guild Memory Test word paragraph) (29), Clinical Dementia Rating Scale (CDR) (30), and Cornell Scale for Depression in Dementia (CSDD) (31). Subjects were excluded for significant medical, psychiatric, or neurologic conditions that could contribute to cognitive impairment. No subject was taking psychotropic medications or drugs known to affect the brain's cholinergic system within 4 wk of screening. In particular, none of the subjects with AD was taking a cholinesterase inhibitor or memantine. Subjects were also excluded if they had a modified Hachinski Ischemic Score (32) of greater than 4 or a CSDD score of greater than 12.

Patients with AD were required to meet standard diagnostic criteria for probable AD (33) and to have a CDR score of ≥0.5 and an MMSE score of ≤26. Subjects with MCI were required to meet standard research diagnostic criteria for amnestic MCI (13), including performance on the NYU Paragraph that fell 1.5 SDs below the mean of age-matched and education-adjusted normative data. They were also required to have a CDR score of 0.5 and an MMSE score of ≥24. HC subjects were required to have memory scores within the reference range for age, a CDR score of 0, and an MMSE score of >26. All subjects (or their responsible next of kin) signed informed consent forms approved by the Yale University School of Medicine Human Investigation Committee.

An array of neuropsychologic tests was chosen to assess a broad range of cognitive functions, including attention, memory, executive function, processing speed, language, and verbal and semantic fluencies. On the day of the SPECT scan, the subjects underwent a comprehensive neuropsychologic evaluation that included the tests shown in Tables 1 and 2. The reading subtest of the Wide-Range Achievement Test, third edition, and the Stroop Test could not be administered to one AD subject because of reading impairment.

¹²³I-5-IA SPECT Studies

Subjects were imaged with SPECT and ¹²³I-5-IA administered as a bolus plus a constant infusion as described previously (21). In brief, subjects received a potassium iodide (0.8 g)–saturated solution before radiotracer administration. ¹²³I-5-IA was synthesized as previously described (34) and administered as a bolus (159 ± 11 MBq) and a continuous infusion (22 ± 0 MBq/h) at a bolus-to-infusion ratio of 7.0 h for 8 h. The total administered dose was 363 ± 26 MBq. When this paradigm is used, ¹²³I-5-IA reaches equilibrium binding in the brain by 6–8 h (21).

Three consecutive 30-min emission scans and one 15-min simultaneous transmission and emission protocol scan were acquired between 6 and 8 h with a Picker PRISM 3000XP (n = 29) or 3000XP2 (n = 3) camera equipped with a low-energy, ultra-high-resolution fanbeam collimator (photopeak window, 159 keV ± 10%; matrix, 128 × 128). The difference in sensitivity between the 2 cameras has been found to be less than 2.5% (4 MBq of the ¹²³I source at the center of rotation of each camera). To control for day-to-day variations in camera sensitivity, we measured a ⁵⁷Co-distributed source with each experiment. The axial resolution (full width at half maximum) was 12.2 mm, measured with a ¹²³I line source in water in a cylindric phantom. Plasma samples were collected before radiotracer injection and at the beginning and end of emission scans to quantify the total parent concentration and the free fraction (percentage) of the parent tracer in plasma (f_P) (34).

SPECT emission images were reconstructed and analyzed as previously described, including nonuniform attenuation correction (21,23). Sagittal MR images were obtained with a 1.5-T GE Healthcare Signa camera with spoiled gradient recalled echo sequences (echo time, 5 ms; repetition time, 24 ms; number of excitations, 1; matrix, 256 × 192; field of view, 34 cm). The coregistered and reoriented MR image was used to guide the placement of standardized 2-dimensional region-of-interest templates (Fig. 1) (21). No attempt was made to correct for partial-volume effects. The mean from 2 raters who conducted image analyses is reported. Both raters were unaware of subject identity and diagnostic group.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

Placement of 2-dimensional standardized regions of interest on 3 representative transverse images of ¹²³I-5-IA uptake. AC = anterior cingulate; CB = cerebellum; FC = frontal cortex; OC = occipital cortex; PC = parietal cortex; Str = striatum; TC = temporal cortex; Th = thalamus.

Statistical Analyses

On the basis of previous postmortem studies (9,35,36), the a priori hypothesis was that α₄β₂-nAChR availability would be decreased broadly in neocortical brain regions—particularly temporal and frontal but also parietal and occipital cortices—in subjects with AD compared with HC subjects, with receptor availability in subjects with MCI being between those in subjects with AD and HC subjects. Nicotinic receptor–binding studies in the postmortem AD brain have generally shown less consistent findings in subcortical areas, such as the striatum and thalamus, with no changes in the cerebellum (9,35,36). We also conducted an exploratory analysis to determine whether regional α₄β₂-nAChR availability was correlated with neuropsychologic measures within each subject group.

We previously validated 2 equilibrium outcome measures for ¹²³I-5-IA (21): V_T (V_T′ in our earlier publications (21–23,37)) is the volume of distribution, which is calculated as the ratio of the regional brain concentration to the total plasma parent concentration; V_T/f_P (V_T in our earlier publications (21–23,37)) is the tissue equilibrium volume of distribution, which is equal to the sum of V₂ (nondisplaceable) and V₃ (receptor bound) and which is calculated as the ratio of the regional brain concentration to the free plasma parent concentration (total plasma parent concentration × f_P). We analyzed V_T as the primary outcome measure in the present study because of its greater test–retest reliability. However, we also performed a confirmatory analysis of V_T/f_P, which has the advantage of not assuming intersubject uniformity of f_P.

The effect of diagnosis (HC, MCI, and AD) on regional α₄β₂-nAChR availability (V_T and V_T/f_P, average values at 6–8 h) in 8 brain regions was determined by multivariate analysis of covariance (MANCOVA), controlling for age and sex. Age was included in the model because α₄β₂-nAChR availability has been shown to decline significantly with age in the same 8 brain regions (23). The Pearson r was used to examine the relationship between regional α₄β₂-nAChR availability (V_T) and neuropsychologic measures. Given the exploratory nature of this analysis, no correction for multiple comparisons was applied. However, the effect of Bonferroni adjustment was considered. Group differences in neuropsychologic measures were analyzed by analysis of covariance (ANCOVA), with age, sex, and education entered into the model as covariates. Group differences in demographic and cognitive characteristics were analyzed by ANOVA or the χ² test. All statistical analyses were performed with SPSS (SPSS Inc.) and included 2-tailed tests of significance.

Subject Characteristics

Table 1 summarizes the characteristics of each diagnostic group with regard to demographics and cognitive function. HC subjects, subjects with MCI, and subjects with AD did not differ with respect to age (F_2,29 = 0.23; P = 0.79), sex (χ² = 2.13; P = 0.34), education (F_2,29 = 2.36; P = 0.11), or Wide-Range Achievement Test, third edition, reading subtest standard score (F_2,28 = 1.40; P = 0.26) as an estimate of premorbid intelligence. In contrast, in measures of global function and cognitive performance, the 3 groups differed significantly, as shown in Table 1. Impairments in subjects with AD spanned a range of mild to moderate severity, as evidenced by MMSE (20.4 ± 5.9) and CDR (1.00 ± 0.52) scores. However, the majority of impairments were of mild (CDR score, 1; n = 6) or very mild (CDR score, 0.5; n = 4) severity. Although no subject had clinically important depressive symptoms, there was a significant effect of group on the CSDD score (F_2,29 = 4.49; P = 0.020), in that patients with AD scored nominally higher (2.3 ± 2.9) than HC subjects (0.0 ± 0.0).

Table 2 shows the results obtained from the larger neuropsychologic battery, organized to reflect the various cognitive domains assessed. Group comparisons of neuropsychologic measures were done with raw scores and ANCOVA analyses, controlling for age, sex, and education. All tests in the battery demonstrated significant effects of diagnostic group, with arithmetic differences generally following the pattern of HC > MCI > AD; however, the statistical significance of post hoc pairwise comparisons varied considerably, as shown in Table 2. The results of these analyses broadly supported the previous construct of amnestic MCI (13) as involving objective memory impairment but normal general cognitive function. All 4 memory measures revealed significant differences between subjects with MCI and HC subjects, but only 1 of 8 nonmemory tests (category fluency) revealed such a difference.

Effect of AD and MCI on Regional α₄β₂-nAChR Availability

The primary outcome measure (V_T) assumed that f_P does not vary as a function of diagnostic group (HC, MCI, and AD). Indeed, ANOVA analyses revealed no group differences in f_P (HC, 33.3% ± 2.5%; MCI, 32.2% ± 5.4%; AD, 33.3% ± 4.2%; F_2,29 = 0.12; P = 0.89), total plasma parent concentration (HC, 0.38 ± 0.10 kBq/mL; MCI, 0.42 ± 0.10 kBq/mL; AD, 0.44 ± 0.13 kBq/mL; F = 0.44; P = 0.65), or free plasma parent concentration (F = 0.47; P = 0.63).

Mean values for regional α₄β₂-nAChR availability (V_T) are shown in Table 3 and graphically depicted in Figure 2. MANCOVA results indicated a significant overall effect of diagnostic group on mean V_T (Wilks Λ test: F_16,40 = 2.25; P = 0.019). However, counter to the previously mentioned hypothesis, univariate analyses revealed no significant regional differences among the subject groups for any of the 8 regions, including the 4 neocortical regions. Neither age (F_8,20 = 2.16; P = 0.078) nor sex (F_8,20 = 0.62; P = 0.76) was significant in the MANCOVA model. In a confirmatory analysis of the alternate outcome measure (V_T/f_P), MANCOVA results again indicated a significant overall effect of diagnostic group on mean α₄β₂-nAChR availability (Wilks Λ test: F_16,40 = 2.22; P = 0.021). However, univariate analyses again revealed no significant regional differences among the subject groups for any of the 8 regions (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Regional nAChR availability (V_T) in 8 brain regions as measured with ¹²³I-5-IA and SPECT in older HC subjects (n = 10), subjects with amnestic MCI (n = 10), and subjects with probable AD (n = 12). MANCOVA results indicated significant overall effect of diagnostic group on mean nAChR availability (Wilks Λ test: F_16,40 = 2.25; P = 0.019). However, univariate analyses revealed no significant differences among subject groups for any region.

Relationship Between Regional α₄β₂-nAChR Availability and Neuropsychologic Measures

An exploratory analysis of the relationship between α₄β₂-nAChR availability (V_T) in 6 brain regions (eliminating the cerebellum and the striatum) and performance on 10 neuropsychologic measures for each of the diagnostic groups is shown in Table 4. Significant (P < 0.05) correlations were consistently in the predicted direction (greater α₄β₂-nAChR availability correlating with better task performance). After Bonferroni adjustment (α = 0.05/162, or 0.00031), only the correlation between the Trail Making Test, Part B, and the anterior cingulate in the HC group (r = 0.91) remained significant (P = 0.00029). The HC group accounted for the majority of significant correlations between neuropsychologic measures and regional α₄β₂-nAChR availability (V_T). Because of limitations in spatial resolution with SPECT, we did not attempt to measure ¹²³I-5-IA availability in medial temporal lobe structures, including the hippocampal formation and the entorhinal cortex, for correlations with measures of episodic memory.

Comparison with In Vivo Imaging Studies

Our results are thus in agreement with those of Ellis et al. (26), who observed no significant reduction in α₄β₂-nAChR binding, as measured with 2-¹⁸F-fluoro-A-85380 (2-¹⁸F-FA) and PET, in subjects with early AD compared with age-matched HC subjects. However, our results stand in contrast to those of O'Brien et al. (24), Sabri et al. (25), and Terrière et al. (27), who all found significant reductions in α₄β₂-nAChR in patients with AD (24,25) or MCI (25,27), as measured with the ligand ¹²³I-5-IA (24,27) or 2-¹⁸F-FA (25). Using ¹²³I-5-IA, O'Brien et al. found significant reductions in the activity ratios of the region of interest to cerebellum in the frontal, striatal, right medial temporal, and pontine regions in 16 patients with AD compared with 16 control subjects. Sabri et al. reported significant reductions in the 2-¹⁸F-FA binding potential (using the corpus callosum as a reference region) in the hippocampus, caudate, frontal cortex, temporal cortex, posterior cingulate, anterior cingulate, and parietal cortex in 17 patients with moderate to severe AD and 6 subjects with amnestic MCI compared with 10 older HC subjects. That study contained very little subject characterization, including demographics, cognitive or functional status, smoking status, or concomitant medications, thus limiting comparison with other studies. Terrière et al. studied 9 subjects with amnestic MCI and 10 age-matched HC subjects by using ¹²³I-5-IA and statistical parametric mapping analysis with global normalization; the subjects with MCI evidenced discrete reductions in uptake in the medial temporal cortex.

The divergent results among these studies may be related to several factors, including subject differences and outcome measures. Severity level may account for some differences because positive results were obtained in the studies of O'Brien et al. (subjects with somewhat more advanced AD; MMSE score, 17.2 ± 5.7) (24) and Sabri et al. (subjects with moderate to severe AD) (25), whereas negative results were observed in subjects with milder AD in the study of Ellis et al. (MMSE score, 22.5 ± 2.5) (26) and the present study (MMSE score, 20.4 ± 5.9). To be sure, these severity differences were not great and do not explain the discrepant results even in studies of MCI. The finding by Terrière et al. (27) of reduced α₄β₂-nAChR binding restricted to the medial temporal cortex may be particularly susceptible to partial-volume effects, given that medial temporal lobe structures are also most atrophied in amnestic MCI (37). However, this feature is a limitation of all studies, none of which have corrected for the effects of atrophy. All studies that provided information about the subjects appropriately excluded current smokers. Only O'Brien et al. (24) included subjects taking cholinesterase inhibitors (7/16) in the AD group; however, these authors reported substantially the same results when these subjects were excluded from the analysis. The 2 studies in which gray matter reference regions—either the cerebellum (24) or the whole brain (27)—were used were limited by substantial nicotine-specific binding in these regions (20,21).

Older PET studies with ¹¹C-nicotine had revealed reductions in nAChR-binding sites in subjects with moderate AD (14,15). Significantly lower levels of ¹¹C-nicotine binding were observed in the frontal cortex, temporal cortex, and hippocampus in patients than in these regions in control subjects (14,15). These ¹¹C-nicotine studies were limited by high nonspecific binding, rapid metabolism and washout from the brain, and a strong dependence on cerebral blood flow. Some of these studies involved the use of a dual-tracer model, with the administration of ¹⁵O-H₂O and ¹¹C-nicotine, to generate a flow-compensated parameter of ¹¹C-nicotine binding (14,15).

Comparison with Postmortem Studies

Our results also disagree with those of the majority of postmortem binding studies of AD, which have consistently revealed that the nAChR subtype is reduced (reviewed in (9)). However, these studies were done largely with autopsy specimens from patients in the terminal stages of the disease (2,6). Studies that have specifically included patients with mild AD (7,10) or MCI (11) are extremely rare and have tended to suggest that reductions in nAChR binding may be restricted to the moderate or even the severe stages of the disease. Perry et al. (7) showed that ³H-epibatidine binding in the temporal cortex was inversely correlated with clinical severity (CDR rating). However, these investigators also observed no discernible decline in nAChR binding in individuals with CDR scores of 0–2, but only in the severe stages of the disease (7). Pimlott et al. (10) reported no significant differences in ¹²⁵I-5-IA binding in several brain regions, including the caudate, putamen, and insular cortex, between control subjects and patients with mild AD (n = 4); however, the neocortical and medial temporal regions were not examined. Interestingly, the only study of nAChR binding in subjects with MCI found no difference in ³H-epibatidine binding between subjects with no dementia and subjects with MCI (11).

More broadly, several postmortem investigations have suggested the preservation during the early stages of AD of a range of cholinergic markers, including neocortical acetylcholinesterase activity (3), choline acetyltransferase activity (3,4), and cholinergic neurons in the nucleus basalis of Meynert (12). However, the results of one study have suggested that the early neuropathologic stages of AD may be associated with a loss of neocortical choline acetyltransferase activity (39). Conversely, other studies have reported reductions in the number of neurons in the nucleus basalis containing immunoreactivity for TrkA (40) and the low-affinity p75 neurotrophin receptor (41) in MCI and mild AD. These collective clinicopathologic observations—in conjunction with the results of the present in vivo imaging study—support the interpretation that there may be some phenotypic changes but minimal degeneration of cholinergic basal forebrain neurons in MCI and mild AD. Alternatively, to the extent that neurodegeneration occurs, it may be compensated for by the upregulation of cholinergic markers (4), including nAChRs.

Neuropsychologic Analyses

Our failure to observe reductions in receptor availability in subjects with AD and MCI is more striking in light of the dramatic and expected cognitive differences among our subject groups. For all neuropsychologic and clinical measures obtained (with the exception of the Trail Making Test, Part A), the performance of the MCI group was intermediate between those of the AD group and the HC group. This finding is consistent with the concept of MCI as a transitional stage between normal aging and the onset of AD. The results of these analyses also broadly supported the previous construct of amnestic MCI (13) as involving objective memory impairment but normal general cognitive function. All 4 memory measures revealed significant differences between subjects with MCI and HC subjects, but only 1 of 8 nonmemory tests (Category Fluency) revealed such a difference.

When the relationship between α₄β₂-nAChR availability and several neuropsychologic variables was explored with (uncorrected) correlational analyses, several plausible correlations emerged—mostly in the HC group. The HC group showed the majority of significant correlations between neuropsychologic measures and regional α₄β₂-nAChR availability (V_T), as performance on both the Trail Making Test, Part B, and the Digit Symbol Test correlated with V_T in 4 brain regions (anterior cingulate, frontal cortex, temporal cortex, and thalamus). However, after Bonferroni adjustment, only the correlation between the Trail Making Test, Part B, and V_T in the anterior cingulate in the HC group remained significant. Although exploratory, this finding is interesting in light of recent evidence that genetic variations in α₄β₂-nAChR affect performance on attentional tasks (42) as well as the activation of an attentional network that includes the anterior cingulate (43).

Two other research teams have examined specific neuropsychologic correlations with in vivo nAChR binding in subjects with AD or HC subjects. Kadir et al. (16) found that mean cortical ¹¹C-nicotine binding was correlated with performance on the Digit Symbol Test and the Trail Making Test, Part A—which they grouped as “attention” tests—but not with measures of episodic memory or visuospatial ability in 27 patients with mild AD. Unlike these investigators, we did not find correlations between nAChR availability and the Digit Symbol Test or the Trail Making Test, Part A, in the AD group. Indeed, the only significant correlations in this group involved Category Fluency. However, our results bear some similarity to those of Kadir et al. in that the significant correlations in the HC group included the Trail Making Test, Part A, the Digit Symbol Test, and the Trail Making Test, Part B (which also has strong attentional demands but is thought to require more executive control and set shifting). In their study with 2-¹⁸F-FA, Ellis et al. (26) used an extensive neuropsychologic test battery and found that nAChR binding in the frontal cortex in patients with mild AD correlated significantly with performance on a simple reaction time task. After correcting for multiple comparisons, they observed no significant correlations for the HC group. We did not measure simple reaction time for comparison with this finding.