Using Cerebral White Matter for Estimation of Nondisplaceable Binding of 5-HT_1A Receptors in Temporal Lobe Epilepsy

Abstract

The estimation of nondisplaceable binding from cerebellar white matter, rather than from whole cerebellum, was proposed for the PET tracer carbonyl-¹¹C-WAY-100635 (N-{2-[4-(2-methoxyphenyl)-1-piperazinyl]ethyl}-N-(2-pyridyl)cyclohexane-carboxamidel]) because of the heterogeneity of total ligand binding in this region. For the 5-hydroxytryptamine receptor 1A (5-HT_1A) antagonist ¹⁸F-N-{2-[4-(2-methoxyphenyl)piperazin-1-yl]ethyl}-N-2-pyridyl)trans-4-fluorocyclohexanecarboxamide (¹⁸F-FCWAY), the estimation of nondisplaceable binding from cerebellum (V_ND) may be additionally biased by spillover of ¹⁸F-fluoride activity from skull. We aimed to assess the effect of using cerebral white matter as reference region on detection of group differences in 5-HT_1A binding with PET and ¹⁸F-FCWAY.

Methods

In 22 temporal lobe epilepsy patients (TLE) and 10 healthy controls, ¹⁸F-FCWAY distribution volume in cerebral white matter (V_WM) was computed using an extrapolation method as part of a partial-volume correction (PVC) algorithm. To assess the feasibility of applying this method to clinical studies in which PVC is not performed, V_WM was also calculated by placing circular, 6-mm-diameter regions of interest (ROIs) in the centrum semiovalis on parametric images. Binding potentials were BP_F = (V_T − V_ND)/f_P and BP_F-WM = (V_T − V_WM)/f_P, where V_T is total distribution volume and f_P = ¹⁸F-FCWAY plasma free fraction. Statistical analysis was performed using t tests and linear regression.

Results

In the whole group, V_WM was 14% ± 19% lower than V_ND (P < 0.05). V_WM/f_P was significantly (P < 0.05) lower in patients than in controls. All significant (P < 0.05) reductions of 5-HT_1A receptor availability in TLE patients detected by BP_F were also detected using BP_F-WM. Significant (P < 0.05) reductions of 5-HT_1A specific binding were detected by BP_F-WM, but not BP_F, in ipsilateral inferior temporal cortex, contralateral fusiform gyrus, and contralateral amygdala. However, effect sizes were similar for BP_F-WM and BP_F. The value of V_WM calculated with the ROI approach did not significantly (P > 0.05) differ from that calculated with the extrapolation approach (0.67 ± 0.32 mL/mL and 0.72 ± 0.34 mL/mL, respectively).

Conclusion

Cerebral white matter can be used for the quantification of nondisplaceable binding of 5-HT_1A without loss of statistical power for detection of regional group differences. The ROI approach is a good compromise between computational complexity and sensitivity to spillover of activity, and it appears suitable to studies in which PVC is not performed. For ¹⁸F-FCWAY, this approach has the advantage of avoiding spillover of ¹⁸F-fluoride activity onto the reference region.

Pet studies of serotonergic 5-hydroxytryptamine receptor 1A (5-HT_1A) are of interest in several neuropsychiatric disorders. Carbonyl-¹¹C-WAY-100635 and ¹⁸F-N-{2-[4-(2-methoxyphenyl)piperazin-1-yl]ethyl}-N-2-pyridyl)trans-4-fluorocyclohexanecarboxamide (¹⁸F-FCWAY) are frequently used for quantitative measurements with PET of 5-HT_1A receptor specific binding. Quantification of the binding potential requires the estimation of nondisplaceable (free plus nonspecific) binding. Autoradiographic studies showed that, among gray matter regions, 5-HT_1A receptor concentration was lowest in the cerebellum (1,2). Therefore, this region traditionally has been used for the estimation of nondisplaceable binding for 5-HT_1A tracers.

More recently, PET studies with carbonyl-¹¹C-WAY-100635 indicated a significant concentration of 5-HT_1A in cerebellar vermis and cortex of adult subjects and suggested that cerebellar white matter could be a better reference region for the estimation of nondisplaceable binding (3,4). The potential advantage of using cerebellar white matter has been addressed in a limited number of studies (3–5).

The possibility of using a reference region other than the cerebellum is of particular interest for the 5-HT_1A tracer ¹⁸F-FCWAY (6). After intravenous injection, this tracer rapidly undergoes defluorination to ¹⁸F-fluoride, which irreversibly accumulates in the skull (7), causing substantial spillover of activity into the neighboring brain tissue. Despite the development of a correction strategy (8,9), tissue counts can be biased, especially in low-binding regions such as the cerebellum. Therefore, regions of interest (ROIs) have been drawn far from the cortical rim to avoid bias in binding potential measurement (10).

On the basis of results obtained for carbonyl-¹¹C-WAY-100635 with cerebellar white matter and because of the distance of the centrum semiovalis from the extracerebral ¹⁸F-fluoride activity, we hypothesized that cerebral white matter could be a valid alternative to the cerebellum for the estimation of nondisplaceable binding for ¹⁸F-FCWAY. The aim of this study was to assess the impact on detection of differences in 5-HT_1A receptor availability between temporal lobe epilepsy (TLE) patients and healthy controls using ¹⁸F-FCWAY binding in cerebral white matter as an estimate of nondisplaceable binding. The regional pattern of group differences obtained using the cerebellum was previously reported (10).

MATERIALS AND METHODS

Patient Selection

Data from 22 TLE patients (18 men; mean age ± SD, 37 ± 11 y) and 10 healthy volunteers (7 men; mean age, 35 ± 9 y) previously studied (10) were reexamined. Briefly, patients were referred for evaluation of medically refractory TLE. All patients were taking different combinations of antiepileptic drugs (10,11). None had experienced partial seizures for at least 2 d, or a secondarily generalized tonic clonic seizure for at least 1 mo, before PET studies. T1-weighted volumetric MR images were acquired for segmentation purposes, and T2-weighted and fluid-attenuated inversion recovery (FLAIR) images were obtained for the evaluation of mesial temporal sclerosis. The study was approved by the National Institute of Neurological Disorders and Stroke Institutional Review Board and the National Institutes of Health Radiation Safety Committee.

PET Procedure

Each patient was scanned with an Advance Tomograph (GE Healthcare). A bolus of 333 ± 74 MBq of ¹⁸F-FCWAY was injected intravenously, and dynamic scanning was performed for 120 min in 3-dimensional mode. Arterial samples were taken to quantify plasma ¹⁸F-FCWAY and ¹⁸F-fluorocyclohexanecarbox-ylic acid metabolite (¹⁸F-FC) concentrations and whole-blood activity. The ¹⁸F-FCWAY fraction unbound to plasma proteins was measured with ultracentrifugation (12).

PET Data Analysis

Radioactivity frames registered to MR images (13) were corrected for cerebral uptake of acid radioactive metabolite, intravascular radioactivity, and ¹⁸F-fluoride spillover onto the brain (8,9). The method to correct for ¹⁸F-fluoride spillover assumes that there is a known true skull time–activity curve, and each brain region receives a fractional contribution of this time–activity curve that is higher for areas that are closer to the skull (8,9). Next, a previously developed partial-volume correction (PVC) algorithm (10,14) was applied. Briefly, the PVC algorithm corrects on a frame-by-frame basis gray matter pixels for spill-out of gray matter activity and for spill-in of activity from white matter (15). No correction for partial-volume averaging between adjacent gray matter regions is performed (16,17). The procedure is based on the segmentation of MR images into binary masks (18), which are subsequently smoothed to the PET resolution; these smoothed masks are named s_GM, s_WM, and s_CSF. The corrected activity gray matter values can be calculated as follows:

$${\text{C}}_{\text{PVC}}=\frac{{\text{C}}_{\text{ORIG}}-{\text{C}}_{\text{WM}}{\text{S}}_{\text{WM}}}{{\text{S}}_{\text{GM}}},$$ Eq. 1

where C_PVC represents the corrected activity value in a gray matter pixel after PVC, C_ORIG is the original uncorrected pixel value, C_WM is the estimated white matter activity, and s_GM and s_WM are the pixel values (range, 0–1) from the smoothed masks for gray and white matter, respectively. Original and corrected pixel data were fitted to a 2-tissue-compartment model with 3 parameters using the metabolite-corrected input function to provide parametric images of ¹⁸F-FCWAY distribution volume (V).

¹⁸F-FCWAY distribution volume in cerebral white matter (V_WM) was estimated directly on parametric images. To obtain an accurate estimate of V_WM, pixel values that represent 100% white matter should be used. Such pixels have a high s_WM value (close to the maximum value of 1) and are typically found in the centrum semiovalis. Therefore, pixels with s_WM values between 0.986 and 0.995 were identified. The lower threshold of 0.986 was chosen to ensure a stable fit in all patients. Potential inaccuracies in white matter estimation introduce only marginal errors in PVC-corrected gray matter values (15,19), but the effect is expected to be more important if white matter is used as the estimate of nondisplaceable binding. An anatomic constraint was applied to avoid sampling from the basal ganglia and thalamus, in which gray matter is frequently misidentified as white matter (14). Values of these pixels were then fitted as a linear function of s_WM, and the fitted value at s_WM = 1 was used as an estimate of nondisplaceable binding.

The extrapolation method is used for studies in which PVC is performed. To assess the feasibility of using white matter for more routine clinical studies without PVC, V_WM was also calculated with an ROI procedure on parametric PET images. First, 3 contiguous transverse slices for which the centrum semiovalis displayed minimal activity were identified. Next, a 6-mm ROI was centered on the area of lowest signal in both hemispheres. This operation was performed on each identified slice, using a total of 6 ROIs. The mean pixel value was calculated in each ROI, and the 6 regional values were averaged to obtain V_WM.

Derivation of Binding Potential

Binding potentials were derived using both V_ND and V_WM, as following:

$${BP}_{\text{F}}=[(V_{\text{T}}-V_{\text{ND}})/f_{\text{P}})],$$ Eq. 2

$${BP}_{\text{F}-\text{WM}}=[(V_{\text{T}}-V_{\text{WM}})/f_{\text{P}})],$$ Eq. 3

where V_T is the total distribution volume in the target ROI, V_ND is the distribution volume in the cerebellum, and f_P is ¹⁸F-FCWAY plasma free fraction (20).

ROI Analysis

Temporal and extratemporal ROIs were drawn on MR images and applied to coregistered parametric images (10). Cerebellar ROIs were initially drawn along the outer cortical edge, then uniformly shrunk until, on visual inspection, no spillover of activity was observed. Final cerebellar ROIs were about 50% smaller than original ROIs (10). ROI measurements were computed using only gray matter pixels, as defined by the gray matter segment (18). For the cerebellum, for which gray matter segment definition is not accurate, the mean ROI value was computed from unmasked PET images coregistered to MRI volumes (10). The effect size was computed as the difference between patients and controls divided by the SD in the control group. Statistical analysis was performed using t test and linear regression, and statistical significance was set at uncorrected P < 0.05.

RESULTS

Figure 1 shows transverse images of ¹⁸F-FCWAY distribution volume in a TLE patient. In the cerebellum, the large spillover of ¹⁸F-fluoride activity masks the visualization of possible differences in specific binding between gray and white matter. At the cortical level, there is a clear difference in ¹⁸F-FCWAY binding between centrum semiovalis white matter and gray matter or mixed gray matter–white matter regions. Spillover of ¹⁸F-fluoride activity is observed in the left frontal and bilateral parietal cortex.

FIGURE 1.

Transverse MR and ¹⁸F-FCWAY distribution volume images passing through centrum semiovalis and cerebellum in TLE patient. Negligible binding of tracer in cerebellar white matter and centrum semiovalis and large spillover of ¹⁸F-fluoride activity onto cerebellum were observed. Two 6-mm-diameter, circular ROIs used for calculation of V_WM with manual approach are displayed. ¹⁸F-FCWAY distribution volume image (V/f_P) is scaled to maximum of 80 mL/mL.

Figure 2 shows representative time–activity curves from a healthy control. In white matter, activity peaked at about 5 min after injection, then progressively decreased to 57%, 24%, and 21% of peak at 30, 60, and 120 min, respectively. The activity peak in white matter was about 40% lower than that in cerebellum. White matter and cerebellar time–activity curves overlapped, starting from about 25 min after injection. As expected, in both regions the time–activity curve was markedly different from the time–activity curve of the mesial temporal cortex.

FIGURE 2.

Representative time–activity curves from healthy control for white matter (○, 1.9 cm²), cerebellum (▴, 4.2 cm²), and mesial temporal cortex (, 2.8 cm²).

Figure 3 shows the extrapolation method used to automatically estimate V_WM, as part of the PVC process. On the distribution volume images, values of voxels with smoothed white matter mask s_WM between 0.986 and 0.995 were fitted to a straight line, and the extrapolated value for s_WM = 1 was taken as an estimate of nondisplaceable binding. The volume of white matter voxels with an s_WM between 0.986 and 0.995 was 19 ± 5 mL. The quality of fit, as quantified by r² values, was good in the whole sample (r² = 0.88 ± 0.11; range, 0.59–0.99).

FIGURE 3.

Estimation of ¹⁸F-FCWAY distribution volume in white matter (V_WM) with extrapolation method. Values for V_WM are plotted vs. smoothed white matter mask (s_WM). V_WM values of voxels with s_WM between 0.986 and 0.995 (○) were fitted to straight line, and value at s_WM = 1.0 (●) was used as estimate of nondisplaceable binding.

In the whole sample, V_WM estimated with the extrapolation approach was significantly smaller than V_ND (0.72 ± 0.34 mL/mL and 0.84 ± 0.30 mL/mL; paired t test, P < 0.05). The percentage difference between V_WM and V_ND was 14% ± 19%, without significant (P > 0.05) differences between TLE patients and controls. On a subject-by-subject basis, V_WM was lower or of the same magnitude as V_ND in all but 3 subjects, in whom V_WM was 11%, 14%, and 29% higher than V_ND. A significant relationship was found between V_WM and V_ND (regression equation, V_WM = −0.10 + 0.99 · V_ND [r² = 0.76; P < 0.05]) (Fig. 4). V_WM was significantly (P < 0.05) higher in patients than in controls (0.81 ± 0.35 mL/mL and 0.55 ± 0.25 mL/mL, respectively). ¹⁸F-FCWAY f_P was significantly higher in patients than in controls (0.121 ± 0.033 vs. 0.066 ± 0.024, respectively; P < 0.05), a finding that may be attributed to antiepileptic drugs (10,11). V_WM/f_P, however, was significantly (P < 0.05) lower in patients than in controls (6.80 ± 2.56 mL/mL and 9.32 ± 3.16 mL/mL, respectively).

FIGURE 4.

Correlation between ¹⁸F-FCWAY distribution volume in white matter (V_WM) and in cerebellum (V_ND) in controls (○) and in TLE patients (●). For whole group, regression equation was V_WM = −0.10 + 0.99 · V_ND (r² = 0.76; P < 0.05).

The variability of V_WM, as quantified by the percentage coefficient of variation (%CV), was greater than that of V_ND (45.2% vs. 39.5% in controls and 43.0% vs. 31.6% in patients, respectively). Nevertheless, the variability of BP_F-WM was not significantly higher (P > 0.05) than that of BP_F, respectively. In the whole group, mean values of %CV across ROIs were 36.8% ± 5.6 and 35.5% ± 6.0, respectively, for BP_F and BP_F-WM.

All significant differences detected using BP_F (10) were also detected using BP_F-WM. Significantly (P < 0.05) lower BP_F-WM, but not BP_F, was detected in ipsilateral inferior temporal cortex, contralateral fusiform gyrus, and contralateral amygdala. However, effect sizes were similar for BP_F-WM and BP_F (Table 1).

TABLE 1.

Group Differences Detected with BP_F-WM and BP_F

Region	BPF-WM (mL/mL)			BPF (mL/mL)*
	Controls	Patients	P	Effect size	P	Effect size
I. sup. temporal	90 ± 26	77 ± 32	0.214	−0.53	0.280	−0.50
C. sup. temporal	86 ± 24	74 ± 27	0.199	−0.52	0.264	−0.49
I. mid. temporal	97 ± 28	80 ± 29	0.113	−0.60	0.154	−0.57
C. mid. temporal	93 ± 24	78 ± 25	0.096	−0.65	0.135	−0.63
I. inf. temporal	103 ± 29	80 ± 28	0.034	−0.78	0.052	−0.76
C. inf. temporal	100 ± 26	91 ± 32	0.423	−0.32	0.528	−0.29
I. fusiform	128 ± 35	82 ± 24	<0.001	−1.35	<0.001	−1.34
C. fusiform	122 ± 32	97 ± 32	0.041	−0.78	0.058	−0.76
I. parahippocampus	146 ± 38	84 ± 26	<0.001	−1.62	<0.001	−1.62
C. parahippocampus	146 ± 42	103 ± 29	0.002	−1.02	0.003	−1.02
I. hippocampus	129 ± 29	66 ± 31	<0.001	−2.20	<0.001	−2.22
C. hippocampus	126 ± 34	99 ± 30	0.025	−0.79	0.035	−0.78
I. amygdala	75 ± 17	51 ± 20	0.002	−1.39	0.005	−1.37
C. amygdala	77 ± 16	62 ± 22	0.045	−0.94	0.078	−0.89
I. insula	97 ± 28	64 ± 21	0.001	−1.18	0.001	−1.17
C. insula	97 ± 27	67 ± 20	0.001	−1.11	0.002	−1.10
I. frontal	70 ± 18	68 ± 23	0.643	−0.16	0.820	−0.11
C. frontal	67 ± 17	64 ± 23	0.648	−0.16	0.828	−0.11
I. parietal	65 ± 18	65 ± 22	0.914	0.01	0.890	0.06
C. parietal	64 ± 18	68 ± 24	0.761	0.20	0.593	0.26
I. occipital	41 ± 11	45 ± 13	0.524	0.39	0.298	0.51
R. occipital	39 ± 8	47 ± 16	0.235	1.02	0.130	1.10

^*Values obtained using V_ND as estimate of nondisplaceable binding (10).

I. = ipsilateral to focus; sup. = superior; C. = contralateral to focus; mid. = middle; inf. = inferior.

In the whole group, values of V_WM calculated with the ROI approach did not significantly differ from those obtained with the extrapolation technique (0.67 ± 0.32 mL/mL and 0.72 ± 0.34 mL/mL, respectively; P > 0.05). Group differences in specific binding were detected without any regional discrepancy in comparison to those detected with the extrapolation approach.

DISCUSSION

The cerebellum has been used in several PET studies to estimate nondisplaceable binding of 5-HT_1A. However, for both carbonyl-¹¹C-WAY-100635 (21–23) and ¹⁸F-FCWAY (8), a 2-compartment model was necessary to accurately fit cerebellar time–activity curves. This finding was attributed to a slow component of nondisplaceable binding or to the progressive accumulation of radioactive metabolites (8,21–23). This unexpected finding motivated the search for nontraditional reference regions for the estimation of non-displaceable binding.

Using PET with carbonyl-¹¹C-WAY-100635, Parsey et al. found that cerebellar white matter time–activity curves were well described by a 1-tissue-compartment model (3). Using cerebellar white matter, rather than the whole cerebellum, for the derivation of the binding potential improved the identifiability and time stability in all cortical regions. The authors concluded that for carbonyl-¹¹C-WAY-100635, cerebellar white matter could be a better reference region than the whole cerebellum (3).

For ¹⁸F-FCWAY, the estimation of nondisplaceable binding from the cerebellum has the additional problem of spillover of ¹⁸F-fluoride activity. This applies to the cortex as well, even though specific binding is high in cortical regions. A method to correct for such spillover was developed. This method, however, does not accurately take into account intersubject variations in the skull shape and thickness (8,9). Disulfiram was used to inhibit defluorination of ¹⁸F-FCWAY in humans (24). The drug reduced the accumulation of ¹⁸F-fluoride in the skull and improved the visualization of radioligand cerebral binding. However, disulfiram also affected the clearance of ¹⁸F-FCWAY and acid radioactive metabolite and jeopardized the accuracy of the compartmental analysis (24).

A simple approach to minimize the effect of ¹⁸F-fluoride spillover onto the cerebellum is to draw ROIs far from the cerebellar edge (10). However, some residual spillover from the bone or mixing of regions with different 5-HT_1A receptor concentrations cannot be excluded. Alternatively, a different reference region could be identified.

Suitability of Cerebral White Matter for Quantification of ¹⁸F-FCWAY Nondisplaceable Binding

The suitability of a region to provide an estimate of nondisplaceable binding can be inferred from kinetic analysis or preblocking studies. Adequate fitting with a 1-tissue-compartment model and negligible k₃ and k₄ values are consistent with a lack of significant concentration of specific receptors. In preblocking studies, the baseline regional distribution volume is compared with that obtained after the administration of a specific receptor cold ligand. If a region is truly devoid of specific binding, the regional value should be comparable in the 2 conditions. Carson et al. performed preblocking ¹⁸F-FCWAY PET studies in monkeys (8). The administration of unlabelled WAY-100635 in advance produced a significant reduction in total binding in the cerebellum by about 44%. This result is inconsistent with the assumption that the cerebellum is a region with only nondisplaceable binding sites for ¹⁸F-FCWAY. Unfortunately, changes of the distribution volumes in white matter were not evaluated (8). Admittedly, the lack of preblocking studies and of formal kinetic analysis, that is, analysis of goodness of fit and calculation of rate constants, represents a significant limit of this study.

Cerebral white matter would be expected to be suitable for the quantification of nondisplaceable binding. White matter is composed of bundles of myelinated axons. The myelin sheath, which is produced by glial cells, is a bimolecular layer of lipids interspersed between protein layers. No significant expression of 5-HT_1A or receptor messenger RNA was found in glial cells (25–27).

Visual inspection of both carbonyl-¹¹C-WAY-100635 (3,28) and ¹⁸F-FCWAY (8,10,24) images shows low activity in the centrum semiovalis.

Finally, starting from about 25 min after the injection of ¹⁸F-FCWAY, the cerebral white matter time–activity curve is similar to the cerebellar time–activity curve, and both time–activity curves have characteristics consistent with negligible concentration of specific binding sites, compared with the cortex.

Group Differences Using White Matter

We failed to show a substantial gain in statistical power using white matter. In TLE patients, lower 5-HT_1A receptor specific binding was detected in ipsilateral inferior temporal cortex, contralateral fusiform gyrus, and contralateral amygdala using V_WM but not V_ND (Table 1). However, by comparing P values and effect sizes, it is evident that the lack of significant group differences in these regions using BP_F (10) is likely a false-negative finding. These results are not unexpected given that regional values of specific binding are much greater than those of nondisplaceable binding, independently of the reference region. Our findings are consistent with results obtained with carbonyl-¹¹C-WAY-100635, indicating no substantial advantage in the detection of sex differences in the hippocampus using BP_F-WM, compared with using BP_F (3).

Comparison Between V_WM and V_ND

We found lower V_WM than V_ND. Although this difference was statistically significant, the magnitude of the difference (14%) was low, and several factors could contribute to this finding. Admittedly, smaller ¹⁸F-FCWAY distribution volume in white matter per se does not constitute evidence of less specific binding. Because of the low parent accumulation in these regions, differences in radioligand delivery, blood volume, radiometabolite accumulation, spillover of ¹⁸F-fluoride onto the cerebellum, and partial-volume averaging could contribute to this difference.

The lower peak of the white matter time–activity curve than of the cerebellar time–activity curve likely reflects lower flow-mediated radioligand delivery. The lower peak of the cerebellar white matter time–activity curve than of the whole cerebellum time–activity curve was reported by Parsey et al. (3) and by Hirvonen et al. (4). Starting from about 25 min after injection, the 2 time–activity curves had similar height and time dependence.

Group Differences in Nondisplaceable Binding

The absence of significant group differences in non-displaceable binding is a prerequisite for determining that measured differences of interest reflect only specific binding. Such differences can be assessed only in arterial line–based models, which are now favored less than less invasive reference tissue models (21,29–32). However, when arterial line–based models were used, significant group differences in nondisplaceable binding with 5-HT_1A tracers were reported. A higher V_ND (not corrected for f_p) was found in healthy women than in healthy men (3,33), whereas a lower uncorrected V_ND was found in depressed patients than in healthy controls (5). A lower V_ND, after correction for f_p, was previously reported in these TLE patients (10,11). If a region is devoid of specific binding and f_p drives the kinetics of the tracer, group differences in nondisplaceable binding estimates might be expected in values uncorrected for f_p, but they should be canceled out after f_p correction. These group differences could be attributed to actual differences in 5-HT_1A receptor availability, nondisplaceable binding, blood volume, or acid corrections. For ¹⁸F-FCWAY, an additional artificial cause might be the spillover of ¹⁸F-fluoride activity.

Choice of Binding Outcome

Previous PET studies with carbonyl-¹¹C-WAY-100635 found that BP_F is the most sensitive binding outcome for the accurate quantification of 5-HT_1A receptor availability (3,5,23,33). On the basis of this finding and to account for group differences in V_ND, we had chosen BP_F as primary binding outcome (10,11). For ¹⁸F-FCWAY, the magnitude of nondisplaceable binding is small, as evidenced by the low value of V_ND or V_WM, compared with V_T, in cortical and limbic areas. Thus, V_T primarily reflects specific binding. In this sample, regional group differences detected with BP_F were similar to those detected with V_T/f_p without significant loss of statistical power (data not shown). Thus, if spillover of ¹⁸F-fluoride onto the cerebellum is a concern and scientific evidence supporting the use of white matter as an estimate of nondisplaceable binding is not sufficient, the small group differences in nondisplaceable binding could be neglected and V_T/f_p could be used as primary binding outcome in future studies in TLE patients with ¹⁸F-FCWAY (34).

Variability of V_WM

The variability of V_WM was higher than that of V_ND. Comparable %CV for cerebellar white matter and whole cerebellum was detected by both Parsey et al. (3,33) and Hirvonen et al. (4). Because of the higher specific binding of cortical regions and large interregion variability, ultimately no variability is added to ¹⁸F-FCWAY binding potential measurements.

Cerebellum and TLE

In TLE patients, the use of cerebral white matter as reference region may be preferable because of cerebellar atrophy. Epilepsy itself—and some antiepileptic drugs, particularly phenytoin—may induce cerebellar structural change and hypometabolism on ¹⁸F-FDG PET (35,36). These effects may be more severe in patients with long epilepsy duration and uncontrolled seizures, who are most likely to have PET studies (35). Moreover, changes in 5-HT concentration in the cerebellum have been reported because of antiepileptic drugs and the effects of uncontrolled epilepsy itself (37).

CONCLUSION

Using cerebral white matter, rather than the cerebellum, as an estimate of ¹⁸F-FCWAY nondisplaceable binding, we found regional reductions of 5-HT_1A receptor specific binding in TLE patients, without loss of statistical power. These findings add to results obtained with carbonyl-¹¹C-WAY-100635 that support the use of white matter for the estimation of nondisplaceable binding for 5-HT_1A tracers in clinical studies and expand the emerging topic of non-traditional reference regions.