Evaluation of Brain Nuclear Medicine Imaging Tracers in a Murine Model of Sepsis-Associated Encephalopathy

Sepsis-associated encephalopathy (SAE) is a devastating complication of severe acute systemic inflammation. It causes both acute and long-lasting neurological dysfunction and contributes to the mortality of patients with sepsis [1]. Current clinical approaches are mainly based on the earliest possible diagnosis and treatment of the systemic inflammation, but our knowledge of the pathophysiological processes overwhelming the brain at this early stage of sepsis is far from complete. Understanding these processes could lead to the development of disease-specific diagnostic and therapeutic approaches that could potentially protect the brain from systemic inflammation and improve mortality.

Much of our current knowledge of SAE has been gathered from animal studies [2]. One of the most important animal models is the lipopolysaccharide (LPS)-induced murine systemic inflammation model. Following the systemic administration of LPS, the mouse brain exhibits a variety of acute and long-lasting alterations including the elevation of inflammatory cytokines [3‐7], microglia activation [8, 9], neuron damage [3], altered neurotransmission [10], oxidative stress [3, 11], blood-brain barrier changes [3, 12] vascular adhesion [13], or invasion of immune cells [14]. Similarities have been found between this mouse model and human SAE [12, 15‐18], making it also a model of murine SAE. A favorable approach to investigating the brain during systemic inflammation is multimodal nuclear medicine imaging [19, 20]. This approach could provide a means to investigate the little-known spatiotemporal distribution and correlations of multiple parameters related to pathophysiology. Brain region-specific connections between the pathophysiologic processes also provide important implications for neuroinflammation in general.

Even if a radiopharmaceutical is highly specific to a certain target, its biodistribution may not be dependent on a single biological process. In turn, many different pathophysiological factors can influence uptake by the specified target (e.g., an increase in 2-deoxy-2-[¹⁸F]fluoro-d-glucose ([¹⁸F]FDG) uptake could be caused by a wide variety of processes) [21]. Parameters measured in healthy brain or during neuroinflammation could be determined by quite different disease-specific processes.

The aim of this study was to assess whether quantitative multimodal in vivo imaging with a set of widely used radiotracers (Table 1) could be used to investigate a set of brain alterations and their region-specific connections associated to the early phase of neuroinflammation induced by systemic LPS injection in mice.

We investigated the following: brain perfusion with [^99mTc]HMPAO single photon emission computed tomography (SPECT), brain glucose metabolism with [¹⁸F]FDG positron emission tomography (PET), neuron damage with the central benzodiazepine receptor ligand [¹²⁵I]iomazenil SPECT, and microglia activation with the 18 kDa translocator protein (TSPO, or, peripheral benzodiazepine receptor, PBR) ligand [¹²⁵I]CLINME SPECT. We described microglia activation with immunohistochemistry (IHC) and oxidation state by a fluorometric ex vivo glutathione assay. These methods have been validated for the respective alterations in multiple models (see references in Table 1).

Tissue hypoperfusion is one of the hallmarks of sepsis syndrome and the brain is not an exception. In humans, decreased perfusion and impaired vascular autoregulation have been reported by multiple authors [17, 35‐37]; however, this mechanism seems to be controversial [1]. Our dual SPECT measurement showed reduced [^99mTc]HMPAO uptake in the brain of LPS-treated animals. Similar distributions were observed both in the control group and the LPS-treated group but the measured uptake quantities were significantly reduced in the latter (Fig. 2a–c). The decreased perfusion might lead to metabolic imbalance and subsequent early and late phase adaptation of glucose transport and utilization by the brain’s most metabolically active cells, astroglia and neurons.

Cerebral metabolic alterations have been previously suggested in SAE [38]. A decrease in cerebral glucose metabolism measured with [¹⁸F]FDG-PET after 24 h following LPS injection in rats has previously been reported [39]. In contrast, we have observed an early increase in [¹⁸F]FDG uptake 5 h following the induction of systemic inflammation in mice (Fig. 4a–c). Significantly enhanced [¹⁸F]FDG uptake values were observed in the cerebrum, cortex, and cerebellum (p < 0.05). Our measurements were carried out on anesthetized mice to avoid introducing additional variability resulting from an awake uptake phase [40]. The opposite alterations in perfusion and [¹⁸F]FDG uptake could be explained by two mechanisms: neurovascular decoupling or the metabolic activity of microglia and infiltrating immune cells. Decoupling during inflammation has been reported in both human [41] and animal studies [42] but it would not fully explain the rise in [¹⁸F]FDG uptake we measured. Both SAE and the LPS model leads to an increased microglial activity and the infiltration of peripheral immune cells in the brain. These cells also express glucose transporters and can contribute to [¹⁸F]FDG PET signal during neuroinflammation [43] making them the most likely cause of the increased [¹⁸F]FDG uptake we observed.

In order to be able to image two isotopes with SPECT in the same animal at the same time, we used [¹²⁵I]iodine. Mouse imaging with [¹²⁵I]iodine is a well-established quantitative possibility even with minuscule injected activities such as 0.2 MBq per animal [44‐47]. For [¹²⁵I]iodine containing radiopharmaceuticals, we used potassium perchlorate to competitively inhibit iodine uptake of different peripheral tissues via the sodium iodine symporter (NIS) [48, 49].

Neuronal damage and cell death has been previously described both in human SAE and animal models of sepsis [2]. Neuron loss could be the mechanism leading to long-term cognitive impairment observed in critically ill patients [50]. Radiolabeled iomazenil and flumazenil are widely regarded as nuclear medicine tracers indicating neuronal integrity and neuron loss [51‐53]. Surprisingly, our measurements showed that [¹²⁵I]iomazenil, a partial inverse agonist of the central benzodiazepine receptor, has an increased uptake in the brains of LPS-treated mice. (Fig. 3a–c). In a previous study, Parente A. et al. investigated the possibility of experimental neuroinflammation influencing the cerebral pharmacokinetics of [¹¹C]flumazenil [54]. They observed no significant differences in radiotracer uptake between control and herpes simplex encephalitis rats. Contrarily, our results suggest that brain [¹²⁵I]iomazenil uptake (a SPECT analogue of [¹¹C]flumazenil) can be directly influenced by neuroinflammation during the early phase of systemic inflammation. Several putative mechanisms could contribute to the increased uptake. GABA_A receptors are present on microglia [55], astrocytes [56‐58], and infiltrating immune cells [59, 60]. Furthermore [¹²⁵I]iomazenil can also bind to the peripheral benzodiazepine receptor (TSPO) with micromolar affinity which has an increased glial expression during neuroinflammation [61]. [¹²⁵I]iomazenil as an ester type molecule can be easily degraded by tissue esterase [62]. The additionally injected neostigmine (cholinesterase enzyme blocker in order to enhance plasma stability of [¹²⁵I]iomazenil) could have increased the availability of [¹²⁵I]iomazenil in the brain making low affinity TSPO binding more likely. Since all of these non-neuronal mechanisms that arise during neuroinflammation can play a role in the measured signal, [¹²⁵I]iomazenil is an unreliable marker of neuronal damage in the LPS model and also possibly other models of sepsis. On the other hand, these results raise important questions regarding the GABA_A system during neuroinflammation and a potential role for [¹²⁵I]iomazenil as an immune system-related radiotracer of neuroinflammation.

Various studies have confirmed the presumed role of TSPO as a marker of neuroinflammation [63, 64] based on its up-regulated expression on microglial cells, astrocytes, and increased ligand binding after neural damage [65] but its exact functional role is unknown [66]. In our experiments, we applied [¹²⁵I]CLINME for TSPO imaging. In the LPS-treated group, significantly enhanced (p = 0.05) [¹²⁵I]CLINME uptake values were measured in the cerebrum, and a marked, but statistically not significant enhancement in the other brain regions of the treated group (Fig. 5a–c). The lack of significant results is most likely due to the low signal-to-noise ratio of our measurements resulting from the combination of low injected activity and small regions of interest. Due to the larger size of the cerebrum VOI, the noise has a lesser impact on the activity measured there. Elevated TSPO expression in LPS-induced systemic inflammation has also been observed in non-human primates [67] and human subjects [68].

The results of the correlation studies (Table 2) outline that the brain region-specific pairwise correlation of [¹²⁵I]iomazenil, [^99mTc]HMPAO, and [¹⁸F]FDG uptake values is different between the control and LPS-treated group. The brain region dependence of correlation coefficients is much lower in the LPS-treated animals than the controls. In healthy animals, [¹⁸F]FDG, [¹²⁵I]iomazenil, and [^99mTc]HMPAO uptake mostly depends on cerebral glucose metabolism, GABA_A receptor density, and cerebral perfusion, respectively. In the LPS-treated animals, the highly positive correlation between [¹⁸F]FDG and [¹²⁵I]iomazenil uptake in all investigated brain regions suggest that inflammatory processes could indeed influence both of these values as discussed earlier. Further supporting this hypothesis, microglia activation was also significantly elevated regardless of brain region (based on IHC and [¹²⁵I]CLINME SPECT results). The highly negative correlations between [^99mTc]HMPAO and [¹⁸F]FDG or [¹²⁵I]iomazenil also fit into this idea if we assume that cerebral hypoperfusion could indicate the severity of inflammation and thus correlate with the metabolic activity and activation state of microglia and infiltrating immune cells that positively contribute to [¹⁸F]FDG and [¹²⁵I]iomazenil signal.

P2Y12 and CD45 double-labeling immunohistochemical (IHC) studies proved the activation of microglia in all the examined brain regions of the LPS-treated animals (Fig. 6). The metabotropic purinergic receptor P2Y12 is expressed by resting and activated microglia which can be used to distinguish them from other CNS cells or myeloid lineage cells (e.g., recruited leukocytes) [69, 70]. Although its expression levels were shown to highly depend on the activation and polarization states of microglia [49, 71], here it was used only to identify them and assess their morphology. CD45 is a cell surface glycoprotein expressed in all nucleated hematopoietic cells [72]. It has been shown that CD45 expression is up-regulated in activated microglia in different diseases and models [73‐76]. By assessing the morphology and CD45 immunoreactivity of microglia, we were able to distinguish between activated and resting cells with a high degree of certainty.

In conclusion, we have described the brain region-specific uptake of a set of widely used radiotracers ([^99mTc]HMPAO, [¹²⁵I]iomazenil, [¹⁸F]FDG) during the early phase of LPS-induced murine systemic inflammation. Our results suggest that inflammatory processes can directly contribute to the uptake of [¹²⁵I]iomazenil and [¹⁸F]FDG masking the neuroinflammation-induced neuron damage and hypometabolism of neural tissue, respectively. Furthermore, we have showed that [^99mTc]HMPAO and [^125I]CLINME can be used to detect cerebral hypoperfusion and microglia activation, respectively, as early as 4 h following the i.v. injection of LPS. Further investigation of the metabolic activity of different brain cells and the status of the GABA receptor system of infiltrating immune cells would be necessary to determine the exact source of the measured signal differences during the early phase of systemic inflammation.