PET measurement of cyclooxygenase-2 using a novel radioligand: upregulation in primate neuroinflammation and first-in-human study

Abstract

Background: Cyclooxygenase-2 (COX-2), which is rapidly upregulated by inflammation, is a key enzyme catalyzing the rate-limiting step in the synthesis of several inflammatory prostanoids. Successful positron emission tomography (PET) radioligand imaging of COX-2 in vivo could be a potentially powerful tool for assessing inflammatory response in the brain and periphery. To date, however, the development of PET radioligands for COX-2 has had limited success.

Methods: The novel PET tracer [¹¹C]MC1 was used to examine COX-2 expression [1] in the brains of four rhesus macaques at baseline and after injection of the inflammogen lipopolysaccharide (LPS) into the right putamen, and [2] in the joints of two human participants with rheumatoid arthritis and two healthy individuals. In the primate study, two monkeys had one LPS injection, and two monkeys had a second injection 33 and 44 days, respectively, after the first LPS injection. As a comparator, COX-1 expression was measured using [¹¹C]PS13.

Results: COX-2 binding, expressed as the ratio of specific to nondisplaceable uptake (BP_ND) of [¹¹C]MC1, increased on day 1 post-LPS injection; no such increase in COX-1 expression, measured using [¹¹C]PS13, was observed. The day after the second LPS injection, a brain lesion (~ 0.5 cm in diameter) with high COX-2 density and high BP_ND (1.8) was observed. Postmortem brain analysis at the gene transcript or protein level confirmed in vivo PET results. An incidental finding in an unrelated monkey found a line of COX-2 positivity along an incision in skull muscle, demonstrating that [¹¹C]MC1 can localize inflammation peripheral to the brain. In patients with rheumatoid arthritis, [¹¹C]MC1 successfully imaged upregulated COX-2 in the arthritic hand and shoulder and apparently in the brain. Uptake was blocked by celecoxib, a COX-2 preferential inhibitor.

Conclusions: Taken together, these results indicate that [¹¹C]MC1 can image and quantify COX-2 upregulation in both monkey brain after LPS-induced neuroinflammation and in human peripheral tissue with inflammation.

Trial registration: ClinicalTrials.gov NCT03912428. Registered April 11, 2019.

Keywords: Cyclooxygenase 1; Cyclooxygenase 2; Inflammation; Lipopolysaccharide; Positron emission tomography; Rheumatoid arthritis.

Fig. 1

Edema and/or hemorrhage after the first and second injections of lipopolysaccharide (LPS). a, b After the first injection of LPS in monkey 2, the T2 magnetic resonance images (MRI) (a) showed edema surrounding the injection site in the right putamen and a leftward shift of the midline. The [¹¹C]PBR28 scans (b) obtained on post-LPS day 3 showed increased TSPO binding in the right putamen. The orthogonal cross-hairs mark the injection site in the right putamen. c, d MRI and COX-2 scans in monkey 3. The T1-weighted MRI (c) was obtained after the first LPS injection but prior to the second one and showed an intracerebral hemorrhage, which is localized with cross-hairs. This hemorrhage was adjacent to, but overlapped with, the injection site, marked with a red arrow. The COX-2 scan (d) in monkey 3 showed high uptake overlying the pre-existing hemorrrhage and not the injection site. e, f MRI and COX-2 scans in monkey 4 showed results similar to those in monkey 3. The T1-weighted MRI (e) obtained between the two LPS injections showed an intracerebral hemorrhage that was offset from the injection site. The COX-2 scan (f) showed high uptake overlying the pre-existing hemorrhage and not the injection site

Fig. 2

Parametric images of distribution volume (V_T) showing [¹¹C]MC1 uptake before (top row) and on day 1 (middle row) after the first lipopolysaccharide (LPS) injection in monkey 2, followed by pharmacological blockade by MC1 (0.3 mg/kg, i.v.) on the same day (bottom row). LPS increased [¹¹C]MC1 uptake globally, which was blocked by cold parent. Orthogonal cross-hairs show the injection site, right putamen. The blockade by MC1 was incomplete—i.e., 78% according to the Lassen plot (Fig. 4). Thus, higher doses of MC1 would have caused even greater blockade

Fig. 3

Whole brain time-activity curves of [¹¹C]MC1 uptake before (a) and on day 1 after the first lipopolysaccharide (LPS) injection (b) in monkey 2, under baseline and blocked conditions by MC1 (0.3 mg/kg, i.v.) on the same day. LPS increased whole brain [¹¹C]MC1 uptake, which was blocked by cold parent

Fig. 4

Lassen plot to determine cyclooxygenase-2 (COX-2) occupancy and non-displaceable uptake (V_ND) of [¹¹C]MC1 in brain after a single lipopolysaccharide (LPS) injection in monkey 2. Each point represents a brain region from a single animal scanned twice in one day: at baseline and after enzyme blockade by MC1 (0.3 mg/kg i.v. administered five minutes before the second radioligand injection). The slope of the straight-line fit provides the occupancy (78%), meaning that MC1 occupied 78% of all COX-2 molecules. The x-intercept provides the V_ND (4.3 mL cm⁻³)

Fig. 5

Time-course comparison of [¹¹C]MC1 standard uptake value (SUV) images obtained on day 1, day 3, and day 8 after the first lipopolysaccharide (LPS) injection in monkey 2. Orthogonal cross-hairs show the injection site, right putamen. [¹¹C]MC1 uptake was greatest on day 1, decreased on day 3, and was lowest on day 8

Fig. 6

Quantitative fluorescent in situ hybridization (FISH) of the gene transcript of PTGS2 (the gene that encodes COX-2) after a single lipopolysaccharide (LPS) injection in monkey 1. In contrast to healthy control, marked PTGS2 transcript was detected in prefrontal cortex (PFC) and putamen, both ipsi- and contralaterally to the injection site in putamen. PTGS2 transcript was primarily colocalized (~ 75%) in neurons in cortex and putamen

Fig. 7

Parametric distribution volume (V_T) images of [¹¹C]MC1 uptake in monkey 3 before lipopolysaccharide (LPS) injection (top row), on day 1 after the second (LPS) injection (middle row), and after blockade by cold MC1 (1 mg/kg, i.v., bottom row). Orthogonal cross-hairs show the injection site, right putamen. [¹¹C]MC1 uptake was markedly increased near the injection area in the right putamen after LPS injection. Cold MC1 blocked radioligand binding to cyclooxygenase-2 (COX-2) at the injection site to a level lower than that in the remainder of brain. Because this dose of MC1 achieved almost complete blockade (i.e., 75%), the residual uptake closely reflected nondisplaceable distribution volume (V_ND). Thus, V_ND in the area of the lesion was reduced below that of normal brain

Fig. 8

Postmortem histologic findings after the second lipopolysaccharide (LPS) injection in monkey 3. From a to d Cellular staining (hematoxylin and eosin (H&E)) of contralateral (a, c) and ipsilateral (b, d) putamen. The red square in a or b indicates the sampling area where each larger image (c or d) was taken. The ipsilateral putamen (d) showed a marked increase in the number of infiltrating leukocytes. The contralateral putamen (c) showed little or no change. Scale bar = 100 μm. From e to j Confocal photographs demonstrate the expression of cyclooxygenase-2 (COX-2) in neutrophils. The top panel (e, f, g) shows the expression of COX-2 (e) and neutrophil elastase (f). The merged fluorescent image (g) shows strong colocalization of COX-2 and neutrophil elastase. The contralateral hemisphere (h, i, j) shows negligible COX-2 (h) and neutrophil (i) staining. COX-2 (red), neutrophil elastase (green), nuclear protein (blue). Scale bar = 10 μm. From k to p Confocal photographs demonstrate the expression of COX-2 in neurons. The top panel (k, l, m) shows the expression of COX-2 (k) and neuronal nuclear protein (NeuN) (l) near the partially necrotic core. The merged fluorescent image (m) shows occasional co-localization of COX-2 in NeuN-positive cells. The contralateral hemisphere (n, o, p) showed only NeuN-positive cells. COX-2 (red), NeuN (green), nuclear protein (blue). Scale bar = 20 μm

Fig. 9

Human positron emission tomography (PET) images of cyclooxygenase-2 (COX-2) and translocator protein (TSPO) in patient 1 with rheumatoid arthritis and a healthy control. Increased [¹¹C]MC1 uptake in the bilateral hand joints reflected increased COX-2 binding in Patient 1 compared to the healthy control. TSPO binding represented by [¹¹C]ER176 uptake in the same individuals showed consistent distribution with COX-2 binding in the bilateral hand joints of patient 1. The increased [¹¹C]MC1 uptake in patient 1 with rheumatoid arthritis was blocked by celecoxib

Fig. 10.

Time-activity curves of [¹¹C]MC1 uptake in representative body parts of patient 1 with rheumatoid arthritis under baseline and blocked conditions. Celecoxib (400 mg) was used as the blocking agent