¹²³I-5-IA-85380 SPECT Measurement of Nicotinic Acetylcholine Receptors in Human Brain by the Constant Infusion Paradigm: Feasibility and Reproducibility

Subjects

Ten healthy subjects including 3 men (32–50 y) and 7 women (20–60 y) provided written informed consent provided written informed consent. This study was approved by the Yale University School of Medicine Human Investigation Committee, Radiation Safety Committee, the West Haven Veteran Affairs Connecticut Healthcare System Human Subjects Subcommittee, and the Food and Drug Administration. Eligibility for the study was determined by a physical examination; laboratory blood tests, including complete blood counts, liver and thyroid function; urine drug screen; and an electrocardiogram. Study participants had no history or presence of a serious medical, neurologic, or psychiatric illness and had no implanted metallic devices. Female participants of child-bearing potential had negative pregnancy tests. Each subject participated in two ¹²³I-5-IA SPECT studies separated by 10.7 ± 6.8 d (range, 5–28 d).

Radiopharmaceutical Preparation and Plasma Measurements

¹²³I-5-IA was prepared as described previously (19) to give a product with radiochemical purity of 95.8% ± 2.2% and 97.2% ± 1.6% and an overall yield of 44.0% ± 20.0% and 45.0% ± 18.5% for the test and retest studies, respectively. Specific activity was >185 MBq/nmol. The plasma parent, metabolite fractions, and free fraction (f₁) were measured in venous blood samples (19).

Clearance of ¹²³I-5-IA in L/h was calculated as R₀/Css, where R₀ is the infusion rate of the radiotracer in kBq/h and Css is the total (free and protein bound) tracer concentration in kBq/mL in plasma at equilibrium (20).

Data Acquisition

Subjects were pretreated with stable supersaturated potassium iodide (800 mg). Two antecubital venous catheters were placed and ¹²³I-5-IA was administered by a bolus (125.8 ± 14.8 MBq) and continuous infusion (18.1 ± 1.5 MBq/h) at a constant rate using a computer-controlled pump (Gemini PC-1; IMED pump) for a bolus-to-infusion ratio of 6 h in subject 1 and 7 h in the remaining 9 subjects (Table 1). Before scanning, 5 external fiducial markers containing 0.04–0.19 MBq of ¹²³I were placed on the head along the canthomeatal line.

The pharmacologic safety of ¹²³I-5-IA administration was assessed by standard laboratory blood tests and urinalysis within 2 mo before the test scan and 24 h after each scan. Vital signs, including electrocardiogram, blood pressure, heart rate, and respiration rate, were measured 30–60 min before, within 5 min, and 30–60 min after ¹²³I-5-IA administration.

Simultaneous transmission and emission scans were acquired with a line source containing 740 MBq ⁵⁷Co on a Picker PRISM 3000XP, 3-head camera (Philips). SPECT acquisitions (30 min) and venous blood sampling were performed every 60 min starting at 2 h after the bolus and continuing throughout the infusion (14 h for subjects 1 and 2 and 10 h for subjects 3–9). The sensitivity of the PRISM 3000XP equipped with a low-energy, ultra-high-resolution fanbeam collimator (photopeak window, 159 keV ± 10%; matrix, 128 × 128) is uniform across the field of view. To control for day-to-day variation in camera sensitivity, a ⁵⁷Co-distributed source was measured with each experiment. A sealed aliquot of this solution (18.5 kBq) was also measured using the γ-counter (1282 Compugamma; Wallac Inc.) used for measuring the plasma activity for cross-calibration of SPECT counts and plasma counts. This system provides a transaxial resolution and an axial resolution of 14.3- and 18.1-mm full width at half maximum, respectively, measured with ¹²³I line sources and water in a cylindric phantom.

MRI

To identify brain regions on the ¹²³I-5-IA SPECT scan, sagittal MR images were obtained on a 1.5-T Signa camera (GE Healthcare) with a spoiled GRASS (gradient recall acquisition in the steady state) sequence with echo time = 5 ms, repetition time = 24 ms, number of excitations = 1, matrix = 256 × 192, and field of view = 34 cm.

Image Analysis

SPECT emission images were filtered using a 3-dimensional (3D) Butterworth filter (order = 10, cutoff frequency = 0.24 cycle/pixel) and reconstructed using a filtered backprojection algorithm with a ramp filter on a 128 × 128 matrix to obtain 50 slices with a pixel size of 2.06 × 2.06 × 3.56 mm in the x-, y-, and z-axes. Nonuniform attenuation correction was performed (15). The MR image was coregistered to the SPECT image using the coregistration function in SPM99 (Statistical Parametric Mapping, version 99). A coregistered MR image was used to guide the placement of standard 2-dimensional ROI templates. A 3D volume of interest (VOI) was generated for each region and transferred to the coregistered SPECT image to determine regional radioactive densities (cpm/pixel) (Fig. 1). Test–retest analyses were performed on the mean of all cortical brain areas, including frontal, parietal, anterior cingulate, temporoinsular, and occipital cortices (cortex) and also on the mean of the caudate and putamen (striatum).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Two-dimensional standardized regions of interest are placed on 3 representative transverse images of ¹²³I-5-IA uptake. FC = frontal cortex; PC = parietal cortex; mFC = medial frontal cortex; TIC = temporoinsular cortex; AC = anterior cingulate; OC = occipital cortex; Cd = caudate; Th = thalamus; Pt = putamen; CB = cerebellum.

Outcome Measures

Regional SPECT activity measurements (cpm/pixel) were converted to absolute units of activity (kBq/mL) as described previously (10). The percentage change per hour was calculated by dividing the slope for the activity values (6–8 h, the earliest time steady state was apparent) by the mean activity (6–8 h) (21). The test–retest reproducibility of the following was evaluated: plasma (clearance, f₁, total parent, and free parent) and regional brain activity (kBq/mL, uncorrected and corrected for infusion rate); %ID/mL (percentage injected dose/mL brain tissue); V_T′ (ratio of total uptake to total plasma parent concentration, calculated as [regional brain activity]/total plasma parent activity between 6 and 8 h) and V_T (ratio of total uptake to free plasma parent concentration, calculated as [regional brain activity]/free plasma parent. Regional brain activity (kBq/mL) was divided by the infusion rate (kBq/h) of the radiotracer to normalize differences in brain activity caused by different infusion rates for test–retest comparisons.

The equilibrium volume of distribution of a compartment is defined as the ratio of the tracer concentration in the compartment to the free tracer in plasma at equilibrium (Vi = Ci/f₁Ca). V_T is the tissue equilibrium volume of distribution equal to the sum of V₂ (nondisplaceable) and V₃ (receptor bound) at equilibrium of a tracer dose. Bound/Free = maximum density of binding sites/affinity of radiotracer = binding potential or (B/F = B_max/K_d = BP). V_T is measured directly by measuring the ratio of brain activity (average values between 6 and 8 h) divided by the plasma free activity (kBq/mL brain/kBq/mL blood) (17,22).

Statistical Analysis

Within-subject variability between test and retest conditions was calculated as the absolute value of the difference of the test and retest measure divided by the mean of the test and retest and expressed as a percentage. Repeated-measures ANOVA was used to evaluate within- and between-subject differences for each plasma and brain measure for the 10 pairs of scans on the basis of time. The reliability of the outcome measures was determined relative to between-subject variance by calculation of the intraclass correlation coefficient (ICC or ρ) as described (23). Interrater reliability was determined for 4 raters who analyzed 6 scans by the ICC.

Pharmacologic Effects

All subjects completed each scan session without symptoms from the injection. There were no changes in electrocardiogram, blood pressure, heart rate, and respiration rate or in blood or urinalysis measures relative to baseline measurements in any of the study subjects after ¹²³I-5-IA administration.

Feasibility of Bolus-Plus-Constant-Infusion Paradigm

Time–activity curves for total and free plasma ¹²³I-5-IA activity stabilized after 3 h with rates of change (1.9%/h ± 4.0%/h) between 6 and 8 h of infusion. Time–activity curves for brain ¹²³I-5-IA activity stabilized after 3 h in striatum, cortex, and cerebellum and 5 h in thalamus. Rates of change were low for cerebellum (0.2%/h ± 1.9%/h), cortex (0.6%/h ± 2.2%/h), striatum (1.2%/h ± 3.0%/h), and thalamus (2.5%/h ± 2.7%/h) and supported equilibrium modeling between 6 and 8 h of infusion.

Variability and Reliability of Clearance, f₁, and Plasma Parent

Clearance of plasma parent demonstrated good test–retest variability (12.3%) and reasonable reliability (ICC = 0.71, Table 1; Fig. 2). Variability was good for total parent and the f₁ (12.5% and 10.9%, respectively) and poor (24.3%) for free parent. Within-subject variability was not significantly different for any measure. Reliability was better for total parent (ICC = 0.81) than for free parent (ICC = 0.68) because of the poor reliability of the f₁ measurement (ICC = 0.35).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Total plasma parent (•, ○) and free parent (▪, □) ¹²³I-5-IA activity are illustrated for test (filled symbols) and retest (open symbols) of representative subject. ¹²³I-5-IA was administered using bolus-plus-constant-infusion paradigm (bolus-to-infusion ratio = 6.94 and 6.96 for test and retest studies, respectively).

Variability and Reliability of Regional Brain Activity and Outcome Measures

Mean regional activity values (kBq/mL) ranged from 5.3 to 13.5 kBq/mL and mean %ID/mL ranged from 0.0014 to 0.0038 %ID/mL for low-density (cortex) and high-density (thalamus) regions, respectively (Fig. 3; Table 2). The test–retest variability was 11.1%–16.4% and the reliability (ICC) ranged from 0.83 to 0.90 for regional brain activity (kBq/mL). The variability improved (range, 8.2%–9.9%) when the regional brain activity (kBq/mL) was corrected for the infusion rate, whereas the reliability (ICC) did not change. The %ID/mL (<13.3% mean test–retest variability) demonstrated good variability (range, 9.5%–13.3%) and excellent reliability (ICC = 0.93–0.96) because of the correction for injected dose. There were no statistically significant differences in any measure between the test and retest studies.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

¹²³I-5-IA activity in thalamus (•, ○), striatum (▴, ▵), and cortex (♦, ⋄) is illustrated for test (filled symbols) and retest (open symbols) of representative subject.

The mean V_T′ and V_T values calculated between 6 and 8 h of infusion ranged from 18.2 to 47.1 and 49.2 to 139.4 for low-density (cortex) and high-density (thalamus) brain regions, respectively (Table 3). The test–retest variability (range, 7.0%–8.9%) and the reliability (ICC = 0.30–0.64) for V_T′ were good and better than the variability (range, 12.9–14.6) and reliability (ICC = 0.28–0.60) observed for V_T. The greater variability for V_T relative to V_T′ is attributed to higher variability of the f₁. There was no significant difference in V_T′ and V_T between test and retest studies.

Interrater Reliability of Region-of-Interest Measurements

Interrater differences in image analysis may also alter reliability. The accuracy of the analysis for regional brain ¹²³I-5-IA uptake measurements between 4 different raters for the test and retest ¹²³I-5-IA images from 3 subjects (subjects 2, 6, and 7) was evaluated. Calculation of the ICCs indicated that measures of each region of interest were highly reliable, with ICC values > 0.95 for thalamus, striatum, cortex, and cerebellum. These findings indicate excellent reproducibility between raters for the image-processing steps, including MR coregistration, reorientation to the anterior commissure–posterior commissure line, and also region-of-interest analyses.

Equilibrium Imaging of ¹²³I-5-IA SPECT

This study evaluated the feasibility and reproducibility of the bolus-plus-constant-infusion (B/I) paradigm for equilibrium imaging and associated outcome measures using ¹²³I-5-IA SPECT. In the equilibrium model, a steady state of radiotracer in blood is established and maintained by constant infusion so that steady state between free radiotracer in blood and brain and between the specifically bound compartment and free radiotracer in brain is established (17,22,24,25). The equilibrium paradigm offers several advantages over the more commonly used kinetic compartmental model, including the following: (a) equilibrium measures are not influenced by differences in blood flow that may differ between clinical populations; (b) the calculation of a volume of distribution equivalent does not require derivation of transfer rate constants that are difficult to identify because of the greater variability inherent in quantitation of nonlinear kinetics; (c) equilibrium outcome measures are insensitive to the ratio of peripheral clearance and brain washout because the terminal half-life is zero at equilibrium; (d) no arterial blood sampling is required; and (e) computation procedures are more straightforward, simpler, and readily amenable to within-subject receptor occupancy studies (17,25). This paradigm has been implemented successfully for other SPECT radiotracers—¹²³I-iomazenil (17,26,27), ¹²³I-iodobenzamide (28), and ¹²³I-epidepride (20)—and PET radiotracers—¹⁸F-cyclofoxy (22), ¹⁸F-altanserin (29,30), ¹⁸F-deuteroaltanserin (21,30,31), ¹⁸F-fallypride (32), and ¹¹C-raclopride (33).

To evaluate the B/I paradigm for ¹²³I-5-IA binding to β₂-nAChRs, 4 parameters were assessed: (a) time required for radiotracer binding to reach stable levels throughout all brain regions, including low-density (cortex) and high-density (thalamus) areas; (b) percentage change per hour of regional brain uptake across the infusion time frame; (c) accuracy of measurement of activity levels in low-density (cortex) regions during equilibrium; and (d) reproducibility of plasma and brain activity and the outcome measures (V_T′ and V_T) during equilibrium. Stable ¹²³I-5-IA levels in plasma and brain were established by approximately 5 h of infusion in all subjects. Equilibrium defined as <5% change per hour in ¹²³I-5-IA activity was established earlier (<3 h) in lower density cortical brain areas and later (∼5 h) in the high-density thalamus. The time frame of 6–8 h was chosen to evaluate equilibrium measures because both plasma and regional brain activities appeared to have achieved steady state, as demonstrated by a mean change of <2.5%/h, and because this interval is reasonable for clinical studies. Brain regions with lower receptor densities (e.g., cerebellum and cortex) all demonstrated <1%/h differences, whereas the high-density thalamus demonstrated a change of 2.5%/h between 6 and 8 h of infusion. This change per hour was less than our a priori criterion of <5%/h and is similar to equilibrium studies with other radiotracers (17,20,21,27,31). This finding differs from an earlier study that stated stable levels of ¹²³I-5-IA uptake did not occur until 12–14 h (10). This conclusion was based on a smaller number of subjects (n = 6) for whom different B/I ratios (range, 4–11.4 h) were used. The extreme range in B/I ratios was chosen based on simulation of the kinetic bolus data and represents the optimal B/I ratio for equilibrium imaging of high-density (B/I = 11.4 h) and low-density (B/I = 4.0 h) brain regions. The wide-ranging B/I ratios resulted in highly variable times that equilibrium was achieved and, in some cases, it was not achieved. In the present study, a constant B/I ratio (7.0 h) was used for most subjects (9/10) and equilibrium was routinely established by 5–6 h across the 10 subjects.

The time (6–8 h) to reach stable levels of binding throughout the brain may present the caveat that counts in low-density brain areas (cortex) may be too low to be measured with high reproducibility (10). Our analyses demonstrated very stable activity measures throughout the cortex, with variability and reliability in the same range as other PET and SPECT radiotracers (20,31,34), suggesting that counts are adequate to reliably measure cortical nAChR using ¹²³I-5-IA SPECT. This conclusion differs from the earlier study (10) that claimed that cortical counts were too low for reliable measurements. The different conclusions may be due to the time that cortical activity was evaluated (e.g., 6–8 h in the present study and 12–14 h in the earlier study) or also may be due to the application of a scatter correction in the earlier study that resulted in low counts in low-density cortical brain areas. Though, theoretically, scatter correction increases the accuracy of SPECT measurements, this correction may introduce error in the precision of the measurement of low-density regions by decreasing “good” counts and increasing variability. However, because scatter correction enhances the theoretical accuracy, it needs to be evaluated in a test–retest study.

Reproducibility of Plasma and Brain Outcome Measures

The test–retest variability and reliability of plasma activity and regional brain activity and outcome measures were evaluated for ¹²³I-5-IA SPECT. The plasma total parent, f₁, and clearance of ¹²³I-5-IA all demonstrated good variability (<12.5%), whereas the variability for the free parent was not as good (24.3%). The reliability of the total parent was good (ICC = 0.81), whereas the reliability of the f₁ was poor (ICC = 0.35) because of high within-subject variability associated with this measurement that has been observed for other radiotracers (20,31,35). Moreover, the poor reliability of the f₁ measurement likely contributed to the lower reliability of free parent measurements (ICC = 0.68). Improved methods for measurement of free parent would significantly decrease the variability associated with V_T.

Regional brain activity values (kBq/mL) demonstrated good test–retest variability (<16.4%) that improved after correction for infusion rate (<9.9%). Overall reliability for all brain activity measures was good (ICC > 0.82). The %ID/mL also demonstrated good test–retest variability (<13.3%) and excellent reliability (ICC > 0.93), likely because of the correction for differences in injected dose. The test–retest variability of V_T′ was excellent (<8.9%); however, the reliability was not as good (ICC = 0.30–0.64). The test–retest variability for V_T was also good (<14.6%), and the reliability was less notable (ICC = 0.28–0.60). The lower variability and reliability observed for V_T is in keeping with the increased variability that occurs due to the assessment of the percentage of plasma protein binding. Thus, V_T′ demonstrated the best reproducibility for assessment of ¹²³I-5-IA binding to β₂-nAChRs in brain. V_T′ and V_T are proportional to B_max/K_d under equilibrium conditions and the assumption that plasma protein binding and nondisplaceable uptake are uniform between individuals. If group differences in plasma protein binding exist, V_T is the appropriate outcome measure and will require larger sample sizes to override the higher variability and poorer reliability.

Overall, the test–retest variability of the outcome measures studied was very good (<16%, with the exception of the free parent at 24%), whereas the reliability was not as impressive. Variability is a measure of the absolute difference between the test and retest scans. In contrast, reliability assessed by the ICC compares within-subject differences and between-subject differences and is more robust. This phenomenon of good variability and poor reliability has been noted previously (34) and occurs when the within-subject variability is similar to the between-subject variability. Given the narrow age range and relatively homogeneous sample studied herein, this is not surprising. Importantly, the variability observed for plasma and brain measures is within the same range as other PET and SPECT radiotracers (31,34,35).

V_T′ and V_T do not correct for nondisplaceable uptake. Preclinical studies in nonhuman primates have indicated that there is no gray matter background region for ¹²³I-5-IA (1,4,5,13). The cerebellum, which is often used as a background region, demonstrated at least 35% displacement of ¹²³I-5-IA uptake by nicotine or the nicotinic agonist cytisine in nonhuman primates (4). This finding has recently been confirmed in human subjects, where up to 70% of ¹²³I-5-IA cerebellar uptake was displaced by the nicotine delivered from smoking 2 consecutive cigarettes (E. Mitsis and J.K. Staley, unpublished data). White matter may be used to measure nondisplaceable uptake, but the related PET radiotracer ¹⁸F-6-FA demonstrated specific binding in white matter tracts but not the corpus callosum (36). This question will be addressed with ¹²³I-5-IA SPECT in human subjects by measuring the absolute nondisplaceable uptake (V₂) after the administration of saturating doses of nicotine. In summary, in the absence of a background region, V_T′ and V_T may be used as outcome measures under the assumptions that nondisplaceable uptake is uniform between individuals.