Thromboxane–prostaglandin receptor antagonist, terutroban, prevents neurovascular events after subarachnoid haemorrhage: a nanoSPECT study in rats

Rat subarachnoid haemorrhage was performed as previously described by Dusick et al. [27]. Briefly, general anaesthesia was delivered with intraperitoneal injections of ketamine (100 mg/kg) and midazolam (10 mg/kg). Autologous arterial blood was collected from the ventral tail artery, and 250 μL was percutaneously injected into the cisterna magna (day 1). A stereotactic frame was used to fix the animal head by both external auditory canals and to maintain a correct anteflexion of the head. Cisterna magna was punctured percutaneously after localization of the puncture site by manual palpation of osseous landmarks. Progression of the needle was operated with a manual depression in the connected syringe. When CSF backflow appeared, the intracisternal localization of the needle tip was confirmed. Then, blood was injected. This procedure was repeated 24 h later (day 2).

In the first step, rats were randomly assigned, by drawing lots, to one of the three experimental groups (Fig. 1): in the SAH group (n = 20), rats only received a double intracisternal injection with blood. Animals in the SAH+TBN group (n = 19) received a double intracisternal injection with blood and a daily TBN (Servier®, Suresnes, France) oral gavage (30 mg/kg per day dissolved in saline) from day 1 to day 5. TBN treatment was started immediately after the first intracisternal injection. The cerebrospinal fluid (CSF) group (n = 19) received a double intracisternal injection of 250 μL artificial CSF. Secondarily, two types of sham rats were studied: without treatment (sham group; n = 15) or with daily TBN oral gavage (30 mg/kg per day) from day 1 to day 5 (sham+TBN group; n = 5). In the sham groups, the cisterna magna was punctured without injection and any post mortem analysis was performed.

All investigators were blinded to group assignment. On days 2 and 5, body weight was recorded and rat behaviour was evaluated with the Garcia neurological scoring system [28]. The following behaviours were evaluated: spontaneous activity, symmetrical movements of limbs, forelimb outstretching, climbing the wall of a wire cage, axillary touch response, and vibrissae touch response. Each subtest was graded on a scale of 0 (worst) to 3 (best). The total score was calculated as the sum of all subtest scores.

NanoSPECT-CT technique provides in vivo, longitudinal, quantitative and functional images. Imaging approaches also offer ethical advantages by largely reducing the number of animals needed (3-R rule). Blood-brain barrier permeability was studied on day 3 by injecting 20 MBq/100 μL ^99mTc-diethylenetriaminepentacetate (DTPA), (Pentacis®, France) into the lateral tail vein. ^99mTc-DTPA is a hydrophilic agent that cannot pass through an intact blood-brain barrier. ^99mTc-DTPA was previously used to explore blood-brain barrier disruptions in clinical studies and in other rodent models of acute neurological injuries [29]. On day 4, we evaluated the uptake of ^99mTc-Anx-V128 (Atreus Pharmaceuticals®, Ottawa, Canada). ^99mTc-Anx-V128 is a novel radiolabelled agent that target the negative phosphatidyl serine exposed on cell membranes. It is used for quantification of apoptosis. Finally, we studied the kinetics of cerebral perfusion by injecting 20 MBq/150 μL ^99mTc-hexamethylpropyleneamineoxime (HMPAO) (Cerestab®, GE Healthcare, France). ^99mTc-HMPAO, a lipophilic agent that passes through the blood-brain barrier, is metabolized in neurons, and the metabolites are trapped in the brain. This agent is well-validated for exploring cerebral perfusion in animals [30] and humans [31]. Since animals remain awake during radiotracer uptake and move freely before imaging, this technique prevents the bias of cerebral blood flow disturbances secondary to anaesthetic exposure. ^99mTc-HMPAO NanoSPECT images were acquired at two time points: in the basal state (day 0) and after SAH conditions were established (day 5). Forty minutes after the radiolabelled agent injection, animals were imaged with a nanoSPECT/CTplus® camera (Mediso, Hungary). Before SPECT imaging, we obtained a 0.8-mm-slice thickness CT scan of the encephalic region. During the scan, the animals were anaesthetized with isoflurane (1.5 vol%), and body temperature was maintained. Images were reconstructed and analysed with the 3D-ROI module provided in InVivoScope® software v2.0p4 (InviCRO). After co-registration between SPECT and CT scan acquisitions, three regions of interest (ROI) were drawn, based on anatomical landmarks identified in the CT images. The brainstem ROI was delimited cranially by the spheno-occipital synchondrosis and caudally by the foramen magnum. A frontal plane, passing through the fourth ventricle, delimited the cerebellum and brainstem ROIs in the antero-posterior axis. The transverse cerebral fissure was used to distinguish the cerebellum ROI from the cerebral hemispheres. Radioactivity inside each ROI was quantified and corrected according to the tissue volume (MBq/mm³), then divided by the total effective injected dose, after correcting for radioactive decay of ^99mTc. For each ROI, the radioactivity was expressed in parts per million of the injected dose (ppmID/mm³). The uptakes of ^99mTc HMPAO on day 5 were expressed as the percentage of basal uptake (day 0).

Cerebral quantification of TP receptor agonists was performed in a subgroup of 18 rats. On day 5, rats were euthanized and harvested fresh brains were frozen with liquid nitrogen. Thromboxane-B2 (TXB-2), Prostaglandins E2 (PGE2), D2 (PGD2), F2α (PGF2α) and isoprostane 8-iso Prostaglandin A-2 (8-isoPGA2) identification and quantification were performed on a half brain homogenate with high-performance liquid chromatography (HPLC; 1290 Infinity and Zorbax column SB-C18; Agilent®, Les Ulis, France) associated with tandem mass spectrometry (MS/MS; G6460 Agilent Triple Quadripole; Agilent®, Les Ulis, France). Further details are provided in [32]. Results were expressed in picograms per milligram of protein.

Histological examination was realized in a subgroup of 19 animals. On day 5, perfusion-fixation was performed by cannulating the left ventricle. The ventricle was perfused at 100 cmH₂O pressure by 50 mL of 4% formaldehyde. Thereafter, the whole brain was removed and immersed in 4% formaldehyde overnight at 4 °C. Next, the brainstem was carefully dissected and embedded in paraffin. To examine the vascular changes in the basilar artery, axial brainstem slices (5 μm) were stained with haematoxylin and eosin. A blinded observer determined the cross-sectional areas of the basilar artery lumen with computerized image analysis (NIS-Elements 4.3, Nikon®, Champigny sur Marne, France).

The number of rats per group was assessed in a power analysis performed with online software (https://​www.​dssresearch.​com/​resources/​calculators/​sample-size-calculator-percentage/​). We expected terutroban to increase the ^99mTc-HMPAO uptake by 50% compared to the SAH group. Past studies showed that cerebral blood flow decreased after a double intracisternal injection in the rat model of SAH [26]. Based on these data, we estimated that a minimum of 14 animals for both samples would be required, for an empiric α error level of P < 0.05 and a β error level = 0.4. Therefore, we included rats to reach a minimal sample of 14 rats in each radiolabeled study group, taking into account mortality and radiolabeled injection failure. All data are presented as the mean and standard deviation (SD). Between-group comparisons was conducted with a one- or two-way ANOVA (depending on the outcome) followed by Tukey’s multiple comparisons test. Adjusted P values < 0.05 were considered significant. Statistical analyses were performed with Prism® 6.0 (GraphPad Software, Inc., La Jolla, CA). The datasets generated and/or analysed during the current study are available in the “figshare” repository, with the identifier https://​figshare.​com/​s/​f7f8f15e1d185ad7​e7f4.

The model of SAH was validated by the presence of blood clots in harvested brain of all rats in the SAH and SAH+TBN groups but not in the CSF group (Fig. 2a). Post mortem quantification of cerebral TP receptor agonist revealed the following: in the SAH group, cerebral TXB-2 and PGF2α were significantly increased compared to the CSF group (101 ± 28 versus 53 ± 14 pg per mg of protein; P < 0.001 and 126 ± 51 versus 62 ± 20 pg per mg of protein; P < 0.001 respectively). Cerebral TXB-2 and PGF-2α levels were significantly reduced in the SAH+TBN group compared to the SAH group (TXB-2 68 ± 16 versus 101 ± 28 pg per mg of protein; P = 0.02; PGF-2α 85 ± 16 versus 118 ± 50; P = 0.03; Fig. 2b).

On day 5, SAH rats treated with TBN (SAH+TBN group) displayed significant less weight loss than the non-treated SAH rats (SAH group) (Fig. 3a) showing a beneficial effect of TBN. This difference was not yet visible on day 2. TBN also improved neurological deficits as measured using the Garcia scale. On day 2 the SAH+TBN group displayed a significant higher Garcia score than the SAH group. On day 5, differences in the Garcia score remain significant between the SAH and the sham or CSF groups (Fig. 3b).

On day 3, ^99mTc-DTPA uptake was significantly increased in the SAH group compared to the sham group in the brainstem (0.8 ± 0.5 ppmID/mm³ versus 0.2 ± 0.1 ppmID/mm³, P < 0.001; Fig. 4a, c) and in the cerebellum (0.7 ± 0.4 ppmID/mm³ versus 0.2 ± 0.1 ppmID/mm³, P = 0.03). In the SAH group, an inter-ROI analysis showed that brainstem and cerebellar uptakes were significantly higher than the cerebral hemispheres uptake (P = 0.02 versus cerebellum; P = 0.01 versus brainstem; Fig. 4a, b). No difference was found between CSF and SAH groups in all studied ROIs. TBN had no significant effect on ^99mTc-DTPA uptakes of sham-operated animals. In the SAH+TBN group, a significant reduction in brainstem ^99mTc-DTPA uptake was found compared to the SAH group (0.4 ± 0.1 ppmID/mm³ versus 0.8 ± 0.5 ppmID/mm³; P = 0.03; Fig. 4c), but no significant effect was found in the cerebellum (SAH+TBN versus SAH; P = 0.46) and the cerebral hemispheres (SAH+TBN versus SAH; P = 0.7).

On day 4, ^99mTc-Anx-V128 uptake was significantly increased in the SAH group compared to the sham group in the brainstem (0.5 ± 0.2 ppmID/mm³ versus 0.2 ± 0.1 ppmID/mm³, P < 0.001), in the cerebellum (0.6 ± 0.4 ppmID/mm³ versus 0.2 ± 0.1 ppmID/mm³, P < 0.001) and in the cerebral hemispheres (0.5 ± 0.2 ppmID/mm³ versus 0.2 ± 0.1 ppmID/mm³, P < 0.001; Fig. 4d). Compared to the CSF group, ^99mTc-Anx-V128 uptake was significantly increased in the cerebellum (0.6 ± 0.4 ppmID/mm³ versus 0.3 ± 0.1 ppmID/mm³, P = 0.001). TBN had no significant effect on ^99mTc-Anx-V128 uptakes of sham-operated animals. TBN significantly reduced cerebellar ^99mTc-Anx-V128 uptake induced by SAH (0.3 ± 0.1 ppm ID/mm³ versus 0.6 ± 0.4 ppmID/mm³; P = 0.01; Fig. 4d). However, TBN did not show any significant effect on SAH-induced ^99mTc-Anx-V128 uptake in the brainstem (SAH+TBN versus SAH; P = 0.83) and in the cerebral hemispheres (SAH+TBN versus SAH; P = 0.71).

On day 5, ^99mTc-HMPAO uptakes (Figs. 5a, b) were significantly reduced in the SAH group compared to the CSF group in all studied ROIs: cerebral hemispheres (− 18 ± 20 versus 2 ± 13% of basal uptakes; P = 0.003), cerebellum (− 17 ± 18 versus 1 ± 12% of basal uptakes; P = 0.01) and brainstem (− 18 ± 17 versus 2 ± 9% of basal uptakes; P = 0.004). In the sham+TBN group, we found increased day 5 ^99mTc-HMPAO uptakes: + 9 ± 20% of basal uptakes in the cerebral hemispheres (P = 0.8 versus sham group), + 24 ± 16% of basal uptakes in the cerebellum (P = 0.05 versus sham group), and + 26 ± 12% of basal uptakes in the brainstem (P = 0.04 versus sham group). TBN also significantly prevented the SAH-induced reduction in ^99mTc-HMPAO uptake in all examined ROIs: cerebral hemispheres (3 ± 12% of basal uptakes; P = 0.002 versus SAH group), cerebellum (4 ± 21% of basal uptakes; P = 0.002 versus SAH group) and brainstem (5 ± 21% of basal uptakes; P < 0.001 versus SAH group; Fig. 5b).

Histological evaluation of the basilar artery on day 5 showed that its lumen was significantly narrowed in the SAH group compared to the CSF group (38,382 ± 6528 μm² versus 70,933 ± 14,994 μm²; P < 0.001). Terutroban significantly diminished basilar artery narrowing (56,335 ± 15,728 μm²) compared to the SAH group (P = 0.04; Fig. 6a, b).

The double autologous arterial blood intracisternal injection model had been recognized for its validity [26] and low mortality rate [53], but this model has its limitations due to the absence of vascular rupture. We have not found an increase in cerebral 8-isoPGA2 level after blood injection revealing the poor activation of the oxidative stress response in this animal model. Nonetheless, this minimally invasive rat model, by avoiding cutaneous incision and burr-hole drilling, reduces procedure-related intracranial artefacts that could be potentially detected by the highly sensitive nanoSPECT technique. Even if same volumes of blood or CSF were injected, after a strict confirmation of the intracisternal position, we do not have any data on intracranial pressure or cerebral blood flow measurement. Then, the Garcia score in this animal model lacks sensibility, and other functional tests would have been needed for a better neurobehavioural assessment of treatment effect [54]. We have found heterogeneous effects of TBN depending on the type of radiotracers and the considered ROI. For example, TBN protects BBB in the brainstem without exerting a significant effect on apoptosis in this ROI. This topographical discrepancy can be related to the site of blood injection with a diverse effect on each ROI and/or to a lack of power because of too small sample sizes. Despite our past work [34], our study lacks a histologic correlation of apoptosis with ^99mTc-Anx-V128 uptake in this specific model. ^99mTc-Anx-V128 uptake could also be enhanced by blood and subarachnoid clots as annexin V has an affinity for blood components such as activated platelets. This possibility may reduce ^99mTc-Anx-V128 specificity for apoptotic cells in SAH models. As we have found in the cerebellum ROI, TBN may play a role in apoptosis prevention. Regarding the limits of ^99mTc-Anx-V128, confirmation of this result is needed with other type of investigations dedicated to apoptosis quantification. Finally, additional studies are needed to better understand the relation between the different post-SAH events (hypoperfusion, BBB disruption, apoptosis…) and to specify the mechanistic aspects of TBN activity.