Visual Abstract
Abstract
Both cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) convert arachidonic acid to prostaglandin H2, which has proinflammatory effects. The recently developed PET radioligand 11C-PS13 has excellent in vivo selectivity for COX-1 over COX-2 in nonhuman primates. This study sought to evaluate the selectivity of 11C-PS13 binding to COX-1 in humans and assess the utility of 11C-PS13 to measure the in vivo potency of nonsteroidal antiinflammatory drugs. Methods: Baseline 11C-PS13 whole-body PET scans were obtained for 26 healthy volunteers, followed by blocked scans with ketoprofen (n = 8), celecoxib (n = 8), or aspirin (n = 8). Ketoprofen is a highly potent and selective COX-1 inhibitor, celecoxib is a preferential COX-2 inhibitor, and aspirin is a selective COX-1 inhibitor with a distinct mechanism that irreversibly inhibits substrate binding. Because blood cells, including platelets and white blood cells, also contain COX-1, 11C-PS13 uptake inhibition from blood cells was measured in vitro and ex vivo (i.e., using blood obtained during PET scanning). Results: High 11C-PS13 uptake was observed in major organs with high COX-1 density, including the spleen, lungs, kidneys, and gastrointestinal tract. Ketoprofen (1–75 mg orally) blocked uptake in these organs far more effectively than did celecoxib (100–400 mg orally). On the basis of the plasma concentration to inhibit 50% of the maximum radioligand binding in the spleen (in vivo IC50), ketoprofen (<0.24 μM) was more than 10-fold more potent than celecoxib (>2.5 μM) as a COX-1 inhibitor, consistent with the in vitro potencies of these drugs for inhibiting COX-1. Blockade of 11C-PS13 uptake from blood cells acquired during the PET scans mirrored that in organs of the body. Aspirin (972–1,950 mg orally) blocked such a small percentage of uptake that its in vivo IC50 could not be determined. Conclusion: 11C-PS13 selectively binds to COX-1 in humans and can measure the in vivo potency of nonsteroidal antiinflammatory drugs that competitively inhibit arachidonic acid binding to COX-1. These in vivo studies, which reflect the net effect of drug absorption and metabolism in all organs of the body, demonstrated that ketoprofen had unexpectedly high potency, that celecoxib substantially inhibited COX-1, and that aspirin acetylation of COX-1 did not block binding of the representative nonsteroidal inhibitor 11C-PS13.
The cyclooxygenase enzyme family is responsible for the biotransformation of arachidonic acid into various prostaglandins and thromboxanes that are major inflammatory mediators. The constitutively expressed cyclooxygenase-1 (COX-1) isoform has traditionally been thought to be responsible for maintaining the physiologic integrity of major organs such as the stomach and kidney, as well as normal platelet function. In contrast, the inducible cyclooxygenase-2 (COX-2) isoform is thought to be associated with pathologic responses to external injury or stimuli, including inflammation and pain (1). However, several studies suggest that COX-1 may play a previously underrecognized proinflammatory role in various pathologic conditions such as neurodegeneration (2,3), atherosclerosis (4), and carcinogenesis (5,6). In this respect, a selective PET radioligand might serve to image COX as a surrogate marker for the development and progression of various diseases that have inflammation as one component of their pathogenesis. Any such radioligand would also be useful for measuring the target engagement of nonsteroidal antiinflammatory drugs (NSAIDs) tested in clinical trials for those diseases.
Previous studies from our laboratory reported initial PET results using the newly developed radioligand 11C-PS13 (Supplemental Fig. 1; supplemental materials are available at http://jnm.snmjournals.org) in whole-body scans of nonhuman primates and human brain scans. PS13 is a highly potent COX-1 inhibitor (half-maximal inhibitory concentration [IC50], 1 nM) and 1,000 times more selective for COX-1 than COX-2 (7). In nonhuman primates, 11C-PS13 showed selective binding to COX-1 over COX-2 in most major organs, including the spleen, gastrointestinal tract, kidneys, and brain, as indicated by substantial blockade by COX-1 inhibitors and minimal blockade by COX-2 inhibitors (8). COX-1 is widely distributed in the brain, with the highest concentrations found in the hippocampus, occipital cortex, and pericentral cortex (9), and 11C-PS13 was found to be an excellent radioligand for measuring COX-1 in the human brain. However, the pharmacologic selectivity of 11C-PS13 for COX-1 over COX-2 has not been confirmed in humans.
The present study used whole-body imaging of healthy human volunteers to investigate the selectivity of 11C-PS13 binding to COX-1 and to assess the ability of 11C-PS13 to measure the potency of COX-1 inhibitors. The COX-1 inhibitors ketoprofen (which is preferential for COX-1) and celecoxib (which is preferential for COX-2 but also inhibits COX-1 to a lower extent (10,11)) were used. The effects of aspirin, which is 10- to 100-fold selective for COX-1 over COX-2, were also examined; notably, aspirin is the only clinically used NSAID that irreversibly inhibits both COX-1 and COX-2 by acetylation of Ser-530, which is present adjacent to the binding site for the substrate, arachidonic acid (12). Although all other clinically used NSAIDs competitively inhibit arachidonic acid binding, it was unknown whether aspirin would block the binding of a small-molecule radioligand to this site.
MATERIALS AND METHODS
To determine the selectivity of binding to COX-1, the difference in 11C-PS13 uptake in major organs and blood cells was measured for baseline or blocked conditions using pharmacologic doses of ketoprofen, celecoxib, and aspirin in 26 healthy human volunteers. To explore the ability of 11C-PS13 to measure the potency of COX-1 inhibitors, the relationship between the blockade of 11C-PS13 binding and the plasma concentrations of each blocker was examined after increasing oral doses. All participants gave written informed consent in accordance with the requirements of the National Institutes of Health Combined Neurosciences Institutional Review Board (protocol 17-M-0179; NCT03324646). A detailed description of the methods and relevant citations (9,13–17) can be found in the supplemental materials.
RESULTS
Pharmacologic Effects, Biodistribution, and Dosimetry
11C-PS13 was intravenously injected once in 10 participants at baseline and twice in 8 participants at baseline and after administration of ketoprofen (Supplemental Table 1). In another group of 8 participants, 11C-PS13 was injected 3 times: at baseline, after administration of celecoxib, and after administration of aspirin. The injected activity for 50 total scans was 693.1 ± 51.9 MBq, corresponding to a molar dose of 0.14 ± 0.07 nmol/kg. Neither the radioligand nor the blocking drugs caused any adverse effects, as assessed by participant self-report, blood pressure, pulse, temperature, respiratory rate, electrocardiography, and basic laboratory tests.
The whole-body images were notable for early distribution in the blood pool (e.g., brain, thyroid gland, heart, and spleen), accumulation in target organs (e.g., spleen, brain, lung, and kidney), and apparent metabolism in the liver, with visible excretion of activity in the gallbladder and small intestine (Supplemental Fig. 2). Radiation exposure was calculated from the organ distribution of radioactivity in the first 15 of 26 total participants (Supplemental Table 2). The mean effective dose of 11C-PS13 derived from these 15 healthy volunteers was 4.6 ± 0.6 μSv/MBq, resulting in radiation exposure of 3.4 mSv for an administered activity of 740 MBq (Supplemental Table 3). This dose is comparable to the mean dose of 21 other 11C-labeled tracers used in human research (18), thus raising no unusual concerns about radiation safety. The 3 organs receiving the highest radiation exposure (μSv/MBq) were the spleen (27.8 ± 1.1), liver (9.8 ± 0.2), and lungs (7.2 ± 0.1).
Quantitation of Organ Uptake
To determine whether the blocking drugs changed the concentration of radioligand in plasma, the average concentration of 11C-PS13 was compared at 4 time points (10, 30, 60, and 90 min) in the baseline and blocked scans. The plasma concentration of 11C-PS13 was slightly increased by ketoprofen (P = 0.036) but was unchanged by celecoxib (P = 0.484) or aspirin (P = 0.161) (Supplemental Fig. 3). Because of the change induced by ketoprofen, organ uptake was normalized in each individual by the plasma concentration of radioligand between 10 and 90 min. Although this normalized value (i.e., the ratio of organ SUV to plasma parent concentration [SUV/CP]) was used, similar measurements of potency were found using only PET data (i.e., SUV) (data not shown). A detailed description of the method used to calculate and verify SUV/CP can be found in Supplemental Figure 4. Data were not corrected for 11C-PS13 binding to plasma proteins for 2 reasons. First, the plasma free fraction was negligibly increased in blocked scans with ketoprofen (P = 0.050) and did not significantly increase in scans after blockade by celecoxib (P = 0.208) or aspirin (P = 0.575) (Supplemental Fig. 5). Second, our previous brain test–retest study suggested that the extremely low plasma free fraction of 11C-PS13, with high variability, could introduce unnecessary variance and perhaps even bias into subsequent analyses (9).
Selective Binding to COX-1 in Major Organs
As an overview of the blocking effects, the median and interquartile range (IQR) of the results were measured for each blocker, recognizing that varying doses were used for each drug. Ketoprofen, a COX-1–selective inhibitor, decreased 11C-PS13 organ uptake by the greatest amount (82% in the spleen [IQR, 59%–90%]) (Figs. 1A and 2). Celecoxib, a COX-2 preferential inhibitor, blocked much less (25% in the spleen [IQR, 2%–35%]) (Figs. 1B and 2), and aspirin had the weakest blocking effects (10% in the spleen [IQR, 0%–18%]) (Fig. 2).
Planar images of healthy volunteers after injection of 11C-PS13 under baseline conditions and after administration of blockers. (A) Under baseline conditions and after oral administration of ketoprofen (0.35 mg/kg). Upper arrow = brain; lower arrow = spleen. (B) Under baseline conditions and after oral administration of celecoxib (6.55 mg/kg). Arrow = spleen. Images were averaged from 10 to 90 min and compressed from anterior to posterior.
Percentage blockade of 11C-PS13 binding in major organs and ex vivo blood cells after pharmacologic doses of ketoprofen, celecoxib, and aspirin (n = 8 in each condition). The administered dose ranged from 1 to 75 mg for ketoprofen, 100 to 400 mg for celecoxib, and 972 to 1,950 mg for aspirin. Data are presented as median and IQR. Binding in organs from PET images was measured as tissue-to-plasma ratio from 10 to 90 min—that is, SUV/CP. For ex vivo studies, blood was withdrawn at 10, 30, 60, and 90 min. Binding of 11C-PS13 was then measured as ratio of concentration of radioligand within cells to that in plasma. We refer to this binding as the distribution volume of the blood cells.
Blood cells, including platelets, monocytes, and erythrocytes, contain COX-1 and can be used as a measure of enzyme occupancy by NSAIDs (19–21). In the blood collected at 4 time points during the scans, cells were separated from the plasma, and the uptake of radioactivity in the cells was measured and normalized to the plasma concentration of 11C-PS13. Similar to uptake in organs of the body, uptake in the ex vivo cells showed selectivity for COX-1 over COX-2, and the order of percentage blockade was ketoprofen > celecoxib > aspirin (Fig. 2). The percentage blockade measured in the ex vivo blood cells was highly consistent with the percentage blockade measured in the spleen with ketoprofen and celecoxib (Fig. 3), a finding also observed in other organs (r [Spearman correlation coefficient] = 0.762–0.982).
Consistency between percentage blockade of ex vivo blood cells and spleen by ketoprofen and celecoxib, calculated using SUV/CP.
In Vivo Potency of Ketoprofen and Celecoxib
Both ketoprofen and celecoxib substantially blocked organ uptake and could be used to calculate their in vivo IC50 relative to their plasma concentrations. Because the spleen had the highest uptake at baseline and the greatest percentage blockade, it was used for these calculations, although similar results were found for other organs. The plasma concentration of ketoprofen was significantly and positively correlated with blocking of radioligand uptake in the spleen (n = 8, r = 0.976, P < 0.001; Fig. 4A) and all other organs (r ≥ 0.762). The plasma concentration of celecoxib correlated positively with blocking in the spleen (n = 8, r = 0.767, P = 0.037; Fig. 4B), but the correlation did not reach significance in other organs. From each correlation plot, the plasma concentration of drug that gives half-maximal spleen blockade (i.e., in vivo IC50) was estimated to be less than 0.24 μM for ketoprofen and more than 2.5 μM for celecoxib, which is a greater than 10-fold difference in potency. Greater precision could not be provided because ketoprofen rarely caused less than 50% blockade and celecoxib rarely caused more than 50% blockade.
Dose–response plots of percentage blockade in spleen after oral administration of ketoprofen (A) or celecoxib (B). For ketoprofen, single oral doses ranged from 1 to 75 mg. The 3 leftmost points (i.e., those with lowest plasma concentrations of ketoprofen) were from participants who received either 1 or 5 mg orally. For celecoxib, single oral doses ranged from 100 to 400 mg, and 1 participant received the maximal dose. r = Spearman correlation coefficient.
Brain Uptake
Blockade by ketoprofen and celecoxib was examined using the subset of whole-body images that included the brain (Fig. 5). The occupancy plot determines whether specific binding shown in all regions has the same occupancy, as would be expected for a single target (e.g., COX-1) with homogeneous affinity for the radioligand (22). Uptake in brain regions was quantified as SUV/CP, as was done for whole organs.
Blockade by ketoprofen in the brain of a representative participant (A), and occupancy plots of participants after blockade by ketoprofen (colored points in 5 participants) or celecoxib (black points in 2 participants) (B). These plotted 7 participants had the greatest blockade among all participants. Each point represents a brain region, and each color represents an individual participant. The occupancy (i.e., slope) of COX-1 was about 54% for ketoprofen and 31% for celecoxib. Binding was measured as SUV/CP.
Three of the 8 participants had minimal blockade of brain uptake by ketoprofen and thus could not be used for the occupancy plot analysis; specifically, the range of y-values in these 3 participants was too small to estimate an x-intercept. For the remaining 5 participants, the occupancy plot was reasonably linear—coefficient of determination (r2) values ranged from 0.50 to 0.82—and the nondisplaceable uptake (x-intercept) was similar for 4 participants (1.06–1.96), but there was a single outlier (0.14); values are expressed in units corresponding to SUV/CP. Their occupancy showed a trend toward positive correlation with plasma ketoprofen concentration that did not reach statistical significance (Supplemental Fig. 6). Using these individually identified nondisplaceable uptake values, the average binding potential was 1.25 ± 0.7 for the first 4 participants, with an outlying value of 48 in the last participant. The regions with the highest average binding potentials were the calcarine (1.88) and lingual gyrus (1.64) of the occipital region. Interestingly, the 2 participants with the highest brain blockade by celecoxib (21% and 30%, respectively) also showed a reasonably linear occupancy plot, with average binding potentials estimated at 2.72 and 4.36.
Blockade by Aspirin
Aspirin (972–1,950 mg orally) blocked a small percentage of radioligand uptake in organs of the body. Even the highest doses of aspirin blocked only a modest percentage (<30%) of radioligand uptake in the spleen. The median blockade caused by aspirin in the spleen was only 10% (IQR, 0%–18%; Fig. 2). Like the imaging results, ex vivo studies of blood cells harvested during the PET scans showed low blockade; the median blockade of blood cell uptake of 11C-PS13 was only 2% (IQR, 0%–50%). The median plasma aspirin concentration was 27 μM (IQR, 19–37 μM), approximately 16-fold higher than the in vitro IC50 of aspirin.
To see whether aspirin was capable of blocking radioligand uptake if higher concentrations were achieved, whole blood cells collected from a healthy individual were incubated with aspirin (66 μM) or ketoprofen (1.83 μM), with each concentration being approximately 39-fold higher than the in vitro IC50 of the inhibitor. To assess the effect of the plasma itself, blood cells in plasma were compared, as were cells resuspended in phosphate-buffered saline containing 0.4% dextrose. When incubated with plasma, aspirin blocked 11C-PS13 uptake by 23% and ketoprofen by 97% (Fig. 6A). In contrast, when incubated in buffer, both aspirin and ketoprofen blocked a near-maximal amount: 90% and 93%, respectively (Fig. 6B).
In vitro experiment with incubated whole blood cells collected from a healthy participant. High doses of aspirin and ketoprofen were used; each concentration was approximately 39-fold higher than the in vitro IC50. The bar graph presents the mean and SEM from 6 divided samples with separate centrifuge and γ-counting. Percentage blockade is presented as mean percentage blockade from 6 samples of each condition. Binding of 11C-PS13 was measured as the concentration of radioligand within cells to that in plasma. We refer to this binding as the distribution volume of the blood cells (VBC).
DISCUSSION
Our results demonstrate that 11C-PS13 selectively binds to COX-1 in humans and that inhibition of its organ uptake provides a measure of the in vivo potency of NSAIDs for COX-1. 11C-PS13 uptake to COX-1 was high in major organs with abundant COX-1, such as the spleen, lungs, kidneys, gastrointestinal tract, and heart. Using percentage blockade of 11C-PS13 uptake in the spleen as the parameter of response, the in vivo potency of ketoprofen (IC50 < 0.24 μM) was more than 10-fold higher than that of celecoxib (IC50 > 2.5 μM). These in vivo potencies are consistent with the in vitro potencies of ketoprofen and celecoxib for COX-1 (10,23–25). In contrast, a single oral dose of aspirin (972–1,950 mg) blocked little radioligand uptake in organs of the body or in ex vivo blood cells, consistent with its action at a site adjacent to but not directly in the binding site for small-molecule inhibitors such as 11C-PS13 (Supplemental Fig. 1). Although the brain showed a relatively smaller percentage of blockade—probably due to the limited penetrance of blockers—the regional distribution of radioactivity was appropriate for that of COX-1 and was blocked by ketoprofen, such that the occupancy in all regions was, as expected, the same for high- and low-density regions. That is, the blockade was consistent with a single type of binding site (i.e., COX-1) and with all binding sites having the same affinity for the radioligand.
The 11C-PS13 PET images obtained under baseline and blocked conditions displayed the constitutive expression of COX-1 and were consistent with previous human postmortem studies (26) and with our in vivo studies on nonhuman primates (8). Although the major cell or tissue types with high binding of 11C-PS13 were not identified in this study, most of the organs that showed high specific binding were associated with the mononuclear phagocyte system, which includes blood monocytes and tissue macrophages in various organs such as the liver, spleen, lungs, and brain (27). Similar to observations made on nonhuman primates (8), the spleen showed the highest specific binding to COX-1, which was likely due to stagnant blood cells encompassed within the spleen sinusoids. This possibility is speculated from the remarkably similar values of percentage blockade between the spleen and ex vivo blood cells by ketoprofen and celecoxib (Fig. 3); blood cells, particularly platelets, express abundant COX-1 (19–21).
The results of the present study are consistent with the high in vitro potency of ketoprofen to inhibit COX-1 and with the selectivity of ketoprofen and celecoxib for COX-1 versus COX-2. In this PET study, ketoprofen was found to have high in vivo potency, given that doses of only 5 mg orally blocked more than 50% of organ uptake (Fig. 4A). By comparison, the typical therapeutic dose of ketoprofen is 75–300 mg (28). Regarding isozyme selectivity, the selectivity index of an NSAID for COX-1 equals the IC50 to inhibit COX-2 divided by that for COX-1, based on an in vitro assay to inhibit prostaglandin production. Although the reported values of IC50 and the selectivity index are highly variable, the consensus is that ketoprofen is essentially selective for COX-1 (selectivity index, 8–61) (10,23), whereas celecoxib is preferential only for COX-2 (selectivity index for COX-2, 1.4–30) (23–25). Thus, high doses of celecoxib are expected to block a portion of radioligand binding to COX-1.
The very low blockade of radioligand uptake by aspirin is consistent with its 2-step mechanism of action. Aspirin first weakly binds to the substrate (i.e., active) site of COX-1 and then acetylates the adjacent Ser-530, which irreversibly blocks access of bulky arachidonic acid to the substrate site (12). However, studies using COX-1 protein with amino acid substitutions suggest that this acetylation does not block access of inhibitors to the substrate binding site (12). The results of the present study directly confirm this speculation, given that the radioligand 11C-PS13, which itself is a small-molecule inhibitor, was still able to bind even after exposure to doses of aspirin that would be expected to fully acetylate COX-1 (Figs. 2 and 6A). The low-affinity binding of aspirin to the substrate site was demonstrated in vitro with high concentrations of aspirin (66 μM) in buffer rather than in plasma (Fig. 6B). Plasma proteins decrease the effectiveness of aspirin by removing the drug from plasma water, thereby making it unavailable to diffuse to the target. Under these artificial conditions of no plasma proteins, aspirin was able to displace the radioligand. Thus, at pharmacologic doses and concentrations, aspirin inhibited and irreversibly inactivated COX-1 without blocking access of inhibitors to the substrate binding site.
It should be noted that although celecoxib’s inhibition of 11C-PS13 uptake may be caused by its cross-reactivity with COX-1, it may also be due to the cross-reactivity of PS13 with COX-2. The most direct way to address this possibility would have been to use another selective inhibitor, not just preferential, for COX-2; rofecoxib, which would have been ideal for this purpose, was withdrawn from clinical use because of its high incidence of adverse cardiovascular events (29). To our knowledge, celecoxib is the most COX-2–selective clinically available NSAID in the United States. However, to examine this issue, studies were performed on monkeys using MC1 as the COX-2–selective compound and 11C-PS13 as the COX-1–selective compound (8). The results indicated that 11C-PS13 was selective for COX-1 and had no measurable binding to COX-2.
The occupancy plot of brain uptake assumes that occupancy of specific binding is the same for all regions. Importantly, this analysis showed that ketoprofen enters the brain and displaces the radioligand from COX-1 (Fig. 5). Both ketoprofen and our radioligand are highly selective for COX-1 over COX-2, and the fact that one blocked the other provides strong evidence that both selectively bind to COX-1. Interestingly, measurable occupancy was also found in the 2 participants with the highest brain blockade by celecoxib. However, the occupancy plot also showed a limitation in our ability to measure brain uptake—that is, SUV/CP data from 10 to 90 min—and CP measured with venous samples. The noise and potential bias in these measurements presumably caused the poor identifiability of nondisplaceable uptake and average binding potential, preventing us from obtaining consistent results in participants who received lower doses of the blockers. Future studies will explore whether full quantitation using continuous dynamic scans and arterial blood sampling might provide more accurate measurements.
The highly consistent blocking effect between ex vivo blood cells and the spleen suggests the potential utility of blood cells to evaluate COX-1 inhibitors in the future, even before PET images are obtained. As an example, the potency of a novel NSAID for COX-1 inhibition might be screened first with in vitro human blood cells using 11C-PS13 or a tritiated analog.
Although the brain is protected from polar radiometabolites by the blood–brain barrier, most peripheral organs, including the spleen, have no such barrier. The percentage of contamination by radiometabolites in any organ will depend on the relative composition of 3 components: specific binding to the target receptor, nonspecific binding of the parent radioligand, and radiometabolites. We chose the spleen, which has a high percentage of specific binding (Fig. 2). However, most other organs (e.g., liver) will have a greater percentage of contamination by radiometabolites, which would underestimate the measured percentage blockade by blockers. To determine whether occupancy was stable in the spleen during the scan, we analyzed the time stability of SUV/CP with a method similar to the time-stability analysis in the brain (Supplemental Fig. 4F). No evidence of radiometabolite accumulation in the spleen over time was observed.
CONCLUSION
11C-PS13 was selective for COX-1 in humans, as assessed via its distribution (which reflected that of COX-1 rather than COX-2) and by the much greater potency of ketoprofen than of celecoxib to inhibit uptake in target organs. Using in vivo uptake in organs or ex vivo uptake in blood cells, 11C-PS13 was able to measure the in vivo potencies for COX-1 of NSAIDs that act at the substrate binding site. Taken together, the results suggest that 11C-PS13 can be used to determine whether any NSAID, except for aspirin, achieves adequate concentrations to inhibit COX-1 in the target organ, whether located centrally or in the periphery.
DISCLOSURE
This study was funded by the intramural research program of the National Institute of Mental Health, NIH (projects ZIAMH002795 and ZIAMH002793; NCT03324646). No other potential conflict of interest relevant to this article was reported.
KEY POINTS
QUESTION: How selective is 11C-PS13 binding to COX-1 in humans, and can 11C-PS13 be used to measure the in vivo potency of NSAIDs?
PERTINENT FINDINGS: High 11C-PS13 uptake was observed in major organs with high COX-1 density. Ketoprofen blocked uptake in these organs far more effectively than did celecoxib, and aspirin blocked only a small percentage of uptake.
IMPLICATIONS FOR PATIENT CARE: Because COX-1 may play a proinflammatory role in the brain and periphery, a selective PET radioligand such as 11C-PS13 might be used to image COX-1 as a surrogate marker for the development and progression of diseases marked by inflammation and measure the target engagement of NSAIDs as potential treatments.
ACKNOWLEDGMENTS
We thank the NIH PET department (Chief, Peter Herscovitch) for performing the PET scanning and Ioline Henter (NIMH) for invaluable editorial assistance.
Footnotes
Published online Jul. 7, 2022.
- © 2023 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication February 23, 2022.
- Revision received June 29, 2022.