Introduction

Sepsis is a systemic response to infection with severe organ dysfunction caused by an unbalanced host immune response, in an attempt to eliminate invasive microorganisms [1]. After the infection, pathogens and their compounds, pathogen-associated molecular patterns (PAMPs), are identified by antigen-presenting cells via pattern recognition receptors (PRRs). The interaction between PAMPs and PRRs promotes the activation of pro-inflammatory pathways releasing cytokines, chemokines, and acute-phase proteins such as tumor necrosis factor alpha (TNF-α), interleukin (IL)-1β, IL-6, and pentraxin-3 (PTX-3) PMID [2,3,4]. Additionally, endogenous constituents from damaged tissue or those actively released from cells during inflammatory processes can bind and activate PRRs. These endogenous products, named damage-associated molecular patterns (DAMPs), are sterile inductors of the immune system, and the most studied are heat shock protein (HSP), high mobility group box-1 protein (HMGB-1), S-100 proteins, advanced glycation end products (AGEs), and mitochondrial sub products [5, 6]. In clinical studies, patients diagnosed with severe sepsis and septic shock presented an overlapping network of PAMPs and DAMPs with elevated levels of HMGB-1, HSP, and the S100 calcium-binding protein B (S100B) family in their bloodstream [7, 8]. This exacerbated host immune response increases the blood-brain barrier (BBB) permeability, facilitating the infiltration of immune cells from the bloodstream into the brain, which together with the brain immune response causes cell damage. Additionally, evidence suggests that sepsis survivors present long-term neurological sequelae with decline in cognitive function [9]. The pathway by which sepsis triggers cognitive dysfunction probably includes systemic metabolic disorders, increased host immune response, oxidative and nitrosative stress, and BBB disruption, followed by immune cells infiltrating the brain and severe microglial activation [9]. Thus, based on the high incidence of sepsis in the world and post-sepsis cognitive impairment, this systematic review aims to (i) identify mechanisms by which sepsis induces long-term neurological sequelae, particularly substantial decline in cognitive function in sepsis survivor patients and in pre-clinical sepsis models; (ii) provide evidence of biomarkers involved in brain neuroinflammation that can predict cognitive impairment in sepsis patients and in pre-clinical sepsis models; and (iii) draw attention to adjuvant treatment as a new avenue to prevent cognitive impairment post-sepsis.

Methods

We accomplished this systematic review as stated in a prospective protocol using PRISMA statement guidelines [10]. The review protocol is registered at PROSPERO (registration number: CRD42017071755; http://www.crd.york.ac.uk/prospero).

Literature Search Strategy

A systematic review of pre-clinical and clinical studies was conducted to evaluate mechanisms by which sepsis induces long-term neurological sequelae. The studies were identified by searching the PubMed/MEDLINE (National Library of Medicine), PsycINFO, and EMBASE (Ovid) databases for peer-reviewed journals that were published until January 2018. To identify additional relevant citations, we conducted forward searches in LILACS (Latin American and Caribbean Health Sciences Literature), IBECS (Bibliographical Index in Spanish in Health Sciences), and Web of Science. The abovementioned databases were searched with the following combinations of keywords: (“sepsis” OR “septic shock” OR “septicemia” OR “lipopolysaccharide” OR “LPS” OR “cecal ligation and puncture” OR “cecal ligation and perforation” OR “CLP”) AND (“cognitive impairment” OR “encephalopathy” OR “delirium” OR “dementia” OR “psychiatric disorder” OR “sickness behavior” OR “neurocognitive impairment” OR “Alzheimer’s disease” OR “schizophrenia” OR “mental disorder” OR “depressive disorder” OR “memory” OR “functional deficits” OR “functional impairment” OR “stress disorder” OR “post-traumatic stress disorder”).

Review of Interventions for Health, Patient, Intervention, Comparators, Outcome Measures, and Study Design (PICO)

We posed the questions “Do sepsis survivor patients have increased risk of neuropsychiatric manifestations?” and “What is the mechanism by which sepsis induces long-term neurological sequelae, particularly substantial cognitive function decline in survivor patients and in pre-clinical sepsis models?”

Eligibility Criteria

We included the original peer-reviewed articles with no language restriction and with pre-clinical and clinical studies to study the mechanisms by which sepsis induces long-term neurological sequelae and cognitive impairment. We omitted review articles, in vitro studies, and studies that included patients with previous disease as a risk factor for sepsis.

Screening

A total of 3555 articles were included in the primary screening. Reference management software (EndNote X7 for Windows from Thomson Reuters, 2013) was used for screening purposes. After the omission of 562 duplicates, a total of 2993 articles were selected for the study. The retrieved studies were first screened on the basis of their title and abstract and 2819 articles were further omitted on the basis of the exclusion criteria (reviews, in vitro studies, previous disease as a risk factor to acquire sepsis). The full-text articles of the remaining 174 articles were obtained and thoroughly evaluated for a second screening. At the end of the second screening, 130 articles were ultimately included after 44 articles were discarded on the basis of the exclusion criteria (Fig. 1).

Fig. 1
figure 1

Flowchart study design

Full size image

Article Selection

Primarily, two authors screened the titles and abstracts for eligibility (PS and ACSA). Any controversies regarding the studies were resolved through unison checking. Upon agreement from the two authors, valid references on the basis of the selection criteria were selected for final inclusion, and full-text PDFs were obtained and analyzed for their data. The third and fourth authors, TB and VVG, settled issues whenever a consensus could not be reached between the first two authors.

Data Extraction

The data were extracted from the comprehensively reviewed journal articles in a methodical manner. The extracted variables included in our review are as follows: sample size (n), sepsis model, inflammatory biomarkers, intervention, and behavioral task in pre-clinical studies. For clinical studies, we extracted the following: the study design, sample size (n), sepsis assessment, inflammatory biomarkers, intervention, and how the inflammatory marker profile was associated with cognitive impairment in sepsis survivor patients.

Results and Discussion

Pre-Clinical Sepsis Studies

Sepsis Pre-Clinical Models

There are different models to induce sepsis, including the cecum ligation and puncture or perforation (CLP), colon ascendens stent peritonitis (CASP), lipopolysaccharide (LPS) induced, bacterial infusion, bacterial sepsis, and fibrin experimental peritonitis models; however, a translational model to mimic the clinical symptoms is crucial [11]. The CLP is considered the gold standard model to study sepsis [12]. The CLP model includes ligation of the cecum distal to the ileocecal valve and puncture of the cecum to permit leakage of fecal substances into the peritoneum, triggering peritonitis that ultimately causes sepsis [13]. The CASP model was first described by Zantl and colleagues in 1998 [14]. In this model, a stent is inserted into the ascending colon by puncture and immobilized with a suture to the colonic wall. Stent insertion allows transmigration of colonic flora from the gut into the peritoneal cavity. The LPS sepsis model mimics an infection triggered by Gram-negative bacteria. LPS is the major component of the bacterial outer membrane localized in the outer layer of the membrane in non-capsulated bacterial strains. LPS may be administered intraperitoneally (i.p.) or intravenously (i.v.) in rodents, and it can increase pro-inflammatory cytokines, such as TNF-α, IL-1β, and IL-6 in the plasma and in the peritoneal cavity, which attain peak levels between 1.5 and 4 h and decline after 8 h of LPS administration. However, the mortality rates are comparable to those in the CLP sepsis model [15]. In the bacteria infusion model, the rodents receive an i.p. infusion of Escherichia coli at a concentration of 6.5 × 108 colony forming units (CFU) over 12 h. In this model of peritonitis, the rodents reproduce several clinical features observed in human sepsis [16]. In the bacterial sepsis model, rodent feces are macerated in 0.5 mL sterile saline to produce a 5 mg/mL (w:v) suspension. The suspension is centrifuged, and the supernatant is recovered and injected via i.p. into the rodent [17]. In the fibrin experimental peritonitis model, 0.5% bovine fibrin clots containing 2 × 108E. coli are implanted into the rodent peritoneal cavity. Mortality was reduced to 0% in 24 h; however, on day 10 after fibrin implantation, the mortality rate was between 90 and 100% [18].

Cognitive Impairment in Pre-Clinical Sepsis Model

In pre-clinical sepsis models, cognitive impairment and neuropsychiatric-like behavior have been identified from early hours after sepsis until several months after recovery.

The most common forms of cognitive impairment found in different sepsis models were impairment of aversive memory, learning, locomotor and exploratory activities, short-term and long-term memories, depressive-like behavior, anxiety-like behavior, anhedonia-like behavior, and fear memory. A study by Bozza et al. demonstrated that rodents inoculated with intraperitoneal feces presented with avoidance memory impairment 24 h after inoculation [19]. The CLP sepsis model also presented with memory impairment evaluated by novel object recognition (novel recognition memory) 24 h after the surgery [20]. In an endotoxin sepsis model, at 24 h after LPS-challenge, sepsis animals presented cognitive impairment evaluated by habituation to T-maze (memory and spatial learning), rota rod (motor coordination and balance), and activity cage tests [21]. A high dose of LPS (60 mg/kg) caused hypothermia as well as impaired spontaneous locomotor activity at 24 h after injection in BALB/c mice [22]. Four days after LPS-induced sepsis, the rodents presented depressive-like behavior assessed by the sucrose preference test [23]. Rodents showed similar behavior on the same test even with CLP-induced sepsis [23]. The Morris water maze task (spatial memory) was evaluated from the 4th to the 7th day after CLP. The rodents presented spatial and working memory (Y-maze task) impairment on day 7 after CLP surgery [24, 25]. Rodents subjected to CLP presented cognitive impairment evaluated by novel object recognition task at 9 days after surgery [26]. At 10 days after CLP surgery, rodents presented memory impairment evaluated by step-down inhibitory avoidance (aversive memory), habituation to open field (locomotor and exploratory activities), the continuous multiple-trials step-down inhibitory avoidance task (learning), the novel object recognition task (short-term and long-term memories), the forced swimming task (depressive-like behavior), elevated plus-make (anxiety-like behavior), sweet consumption (anhedonia-like behavior), and decreased contextual freezing time in a fear conditioning test (fear memory) [27,28,29,30,31]. CLP sepsis presented impairment of numbers of crossings and rearings in the open field task on day 15 after surgery [32]. After 28 days of LPS administration, C57BL/6 mice presented cognitive impairment evaluated by the novel object recognition, elevated plus maze, and tail-suspension tasks [33]. One month after LPS administration, C57BL/6 mice presented a reduction in sucrose preference, which is a measure of anhedonia [34]. On day 30 after CLP surgery, Wistar rats presented cognitive impairment evaluated by step-down inhibitory avoidance, the continuous multiple-trials step-down inhibitory avoidance task, and habituation to open field [35,36,37,38]. However, in another study, researchers observed no differences in the recognition memory indicator between CLP and sham groups, which demonstrate a rescue of short-term memory after 30 days of sepsis induction [26]. Sepsis survivor mice did not show impairment in contextual fear conditioning or trace fear conditioning at 50 days after CLP surgery; however, they demonstrated impairment in extinction of conditioned fear [39]. On day 60 after CLP surgery, Wistar rats did not show impairment of aversive, habituation, and novel object recognition memories, nor did the rodents present depressive-like behavior [37, 40, 41]. Patients may also develop neuropsychiatric manifestations, such as anxiety, depression, or post-traumatic stress disorder (PTSD), which can have an intense effect on their lives and reduce their probability of returning to work [42]. Thus, pre-clinical models of sepsis help better understand long-term outcomes following sepsis and possible new therapeutic approaches to prevent cognitive impairment triggered by sepsis.

Pathophysiology of Cognitive Impairment in Pre-Clinical Sepsis Model

Sepsis is a severe clinical condition associated with high host immune response to infection [43]. After infection, PAMPs are recognized by Toll-like receptors (TLR), and their interaction markedly upregulates the transcription of genes involved in pro-inflammatory responses [5]. Thus, this activation leads the nuclear translocation of the transcription factor nuclear factor-κB (NF-κB) to produce and deliver pro-inflammatory mediators. A number of pre-clinical studies have shown an auto amplification of pro-inflammatory cytokines, such as TNF-α, IL-6, and IL-1β, in the first few hours after sepsis induction [44], followed by BBB disruption [45]. The disruption of BBB and elevated levels of matrix metalloproteinases (MMP)-9 and MMP-2 activities were observed in the micro vessels of the cortex and hippocampus of rodents subjected to CLP surgery [45]. Consistent with the immune response, DAMPs, which are latter endogenous constituents produced or delivered by damaged tissue, bind to different receptors and exacerbate the host immune response. HMGB-1 may act on brain micro vasculature endothelial cells to disrupt BBB integrity, thus facilitating the entry of neurotoxic substances into the brain [46]. Elevated serum levels of HMGB-1 were noted from 4 to 12 weeks after CLP [46, 47], and among septic patients, serum HMGB-1 levels were significantly lower in survivors than in non-survivors patients [48]. The anti-HMGB-1 monoclonal antibody improved memory impairment and brain pathology in the CLP sepsis model [46]. There are other receptors involved in the sepsis immune activation, such as c-type lectin receptors (CLR), nucleotide binding oligomerization domain (NOD)-like receptors (NLR), receptors for advanced glycation end-products (RAGE), RIG-I-like receptors (RLR), and intra-cytosolic DNA sensors [49]. RAGE expression increased in the hippocampus and the pre-frontal cortex at 30 days after CLP sepsis, when rats were showing cognitive impairment [50]. Serum level of sRAGE amplified with the development of disseminated intravascular coagulation and the severity of sepsis in patients [51]. HMGB-1, AGEs, and S-100 proteins also bind to RAGE, leading to its activation and subsequent neuroinflammation by targeting NF-κB and increasing the expression of early growth response protein-1 (ERG-1) [52, 53]. In another study, amyloid-beta peptide interacted with RAGE-bearing cells in the vessel wall that resulted in transport of amyloid-beta peptide across the BBB and increased pro-inflammatory cytokines and endothelin-1 (ET-1) expression. Inhibition of RAGE-ligand interaction blocked accumulation of amyloid-beta peptide in brain parenchyma in a genetically manipulated mouse [54]. At 30 days after CLP surgery in rodents, there is an increase in hippocampal and pre-frontal cortex levels of amyloid-beta peptide and a decrease in synaptophysin levels associated with simultaneous cognitive impairment. Together, these results imply in the pre-clinical sepsis model that HMGB1-RAGE-signaling activation may lead to long-term cognitive impairment observed during post-sepsis [38]. The inflammasome gene profile was modulated in septic patients with an increase of NLR family CARD domain containing-4 (NLRC-4) and NLR family pyrin domain containing-3 (NLRP-3) and a decrease of NOD-1 and NLRP-1 expression in septic patients compared to healthy controls; the expression levels of the pro-inflammatory cytokines IL-1β and IL-18 were higher in septic patients with a greater magnitude in non-survivors [55]. NLRP-3 is a multiprotein complex formed by the adaptor protein apoptosis-associated speck-like protein containing CARD (ASC) and pro-caspase-1 that regulates the activation of caspase-1, which proteolytically maturates IL-1β and IL-18. Mice with genetic deficiency of NLRP-3 presented inhibition in inflammatory responses and enhanced survival rates after CLP surgery [56]. Another study showed that NLRP-3 deleted mice subjected to CLP surgery had increased survival rates and decreased autophagy and enhanced phagocytosis [56]. LPS-induced mice presented long-term depressive-like behavior and recognition memory deficit. Additionally, NLRP-3, ASC, and caspase-1 expressions and IL-1β, IL-18, and TNF-α levels increased followed by microglial activation in an LPS-induced sepsis model. These effects were blocked by a selective irreversible inhibitor of caspase-1 (Ac-Tyr-Val-Ala-Asp-chloromethylketone) [57]. In another study, mice subjected to CLP surgery presented an increase of Iba-1, IL-1β, and NLRP-3 expression and apoptosis in the hippocampus followed by spatial memory impairment evaluated by the Morris water maze. Inhibition of microglia decreased pro-inflammatory markers and prevented the memory impairment [58]. Microglial gene expression showed an increase of anti-microbial genes and the S-100A family of genes for at least 2 weeks after CLP sepsis surgery; however, the genes did not express cytokines that were observed in the entire brain. CLP-induced sepsis resulted in long-term neuroinflammation sustained due to interactions among various cell types, including resident microglia and peripheral myeloid cells [39]. A pre-clinical systematic review evaluated the effect of peripheral inflammatory activation on microglia. A total of 51 studies were identified with different doses of LPS (0.33 to 200 mg/kg) and live or heat-killed pathogens as a peripheral infectious stimulus. After LPS administration, microglial activation was noted 6 h after challenge, which persisted for at least 3 days. Live E. coli triggered microglial activation after 2 days and heat-killed bacteria after 2 weeks. Microglial activation was associated with TLR-2, TLR-4, TNF-α, and IL-1β expression [59]. Consistent with pre-clinical studies, three cases of post-mortem sepsis, when patients’ right frontal pole was removed and studied at autopsy, the expression of the astrocyte marker glial fibrillary acidic protein (GFAP), cluster of differentiation 68 (CD68), and CD45 microglial markers were all increased in the brain [60]. Thus, in addition to inflammatory markers, HMGB-1/RAGE, NLRP-3, and activation of microglia play roles in the pathophysiology of post-sepsis cognitive impairment.

Intervention to Prevent Cognitive Impairment in Pre-Clinical Sepsis Model

Adjunctive Dexamethasone Therapy

This study evaluated the effect of dexamethasone on mortality, circulating corticosterone and adrenocorticotropin hormone (ACTH) levels, body and adrenal gland weight, anhedonia-like behavior, and aversive memory in CLP sepsis survivor rats. Wistar rats received dexamethasone as an adjuvant treatment for 7 days. Ten days after CLP sepsis, the rats were evaluated for aversive memory, sweet food consumption, and body and adrenal gland weight. Sepsis caused anhedonia-like behavior, memory impairment, increased adrenal gland weight, and increased plasma levels of corticosterone and ACTH. Dexamethasone treatment normalized the adrenal gland weight and plasma levels of corticosterone and ACTH. Additionally, dexamethasone decreased mortality and anhedonia-like behavior and prevented aversive memory impairment [30]. In this study, on days 10 and 30 after CLP surgery, the Wistar rats were subjected to training for an inhibitory avoidance task. Immediately after the training session, the animals received a single injection of saline, epinephrine, naloxone, dexamethasone, or glucose, and 24 h later the animals were subjected to the test. The CLP sepsis survivor rats that received adjuvant treatment with the different aforementioned drugs presented a difference between the training and test sessions, showing retention of aversive memory [40].

Antidepressant Drugs (Fluoxetine and Imipramine)

Antidepressant drugs demonstrate an increase in neurogenesis in the adult rodent hippocampus, and there is also some evidence that antidepressant drug treatment increases peripheral brain-derived neurotrophic factor (BDNF) levels in patients [61, 62]. After 28 days of LPS administration and fluoxetine treatment, C57BL/6 mice were subjected to novel object recognition, elevated plus maze, and tail-suspension tasks. The LPS mice presented cognitive impairment in all the tasks; however, fluoxetine treatment prevented behavioral changes. After the behavioral tasks, fluoxetine treatment was discontinued for 7 days to evaluate biochemical markers. Fluoxetine decreased the Iba-1 microglia marker in the hippocampus, early growth response protein 1 (EGR1) immunoreactivity in the CA1, and bromodeoxyuridine (Brdu) immunoreactive cells in the dentate gyrus [33]. Wistar rats subjected to CLP presented with decreased consumption of sucrose, showing anhedonia-like behavior. Additionally, there were decreases in hippocampus weight and BDNF levels and increases in adrenal gland weight and plasma levels of corticosterone and ACTH. Imipramine treatment prevented depressive-like behavior, decreased corticosterone and ACTH plasma levels, increased BDNF levels, and normalized hippocampal and adrenal gland weight [30]. This study suggested that depressive-like behavior and hypothalamic–pituitary–adrenal axis (HPA) axis changes induced by sepsis may be prevented with antidepressant treatment. In another study from the same research group, it was demonstrated that depressive-like behavior in CLP sepsis survivor rats was reversed after imipramine administration [63]. When LPS-treated C57BL/6 mice were assessed after 1 month, they presented immobility in the tail suspension task and showed a decrease in sucrose preference. Fluoxetine administered 90 min before the behavioral tasks decrease the immobility in the tail suspension task in post-septic animals. However, the authors did not evaluate the effect of fluoxetine on the sucrose preference task [34]. Thus, the antidepressant-like fluoxetine and imipramine indicated a beneficial effect post-sepsis in neuropsychiatric manifestations.

Erythropoietin Treatment

Sprague-Dawley rats were subjected to CLP surgery. The rodents were treated with exogenous recombinant human erythropoietin at a dose of 5 units per day infused consecutively for 7 days into the left lateral ventricle. Seven days after CLP surgery, the animals were subjected to open field exploration, the inhibitory avoidance training and test, and the Morris water maze for spatial learning and memory functions. These tasks indicated sepsis-induced emotional and cognitive deficits; however, recombinant human erythropoietin adjuvant treatment prevented impairment of aversive and spatial memories. AKT/mTOR pathway-mediated neuronal protective effects of erythropoietin were observed in the CLP sepsis group [64]. In another study, Wistar rats were subjected to CLP and treated with a single dose of recombinant human erythropoietin and killed at 6 and 24 h after CLP surgery. Treatment with erythropoietin decreased lipid peroxidation, catalase (CAT), superoxide dismutase (SOD), and creatine kinase activity in the hippocampus of rats subjected to CLP. To study the behavior, erythropoietin was administered once a day for 4 days after CLP surgery, and aversive memory was evaluated on day 10. Mortality was decreased only during erythropoietin adjuvant treatment. After the treatment, the mortality rate was equal between the CLP saline group and CLP erythropoietin group; however, erythropoietin prevented cognitive impairment, as evidenced by improvement in aversive memory in a step-down inhibitory avoidance task [65].

Heparins (Dalteparine, Enoxaparine, or Nadroparine) Treatment

Sepsis is commonly complicated by coagulopathy by disseminated intravascular coagulation [66]. Heparin and low-molecular weight heparin decreased mortality and end-organ failure following experimental sepsis [22, 66]. In this study, i.p. LPS challenge produced sepsis in BALB/c mice. The mice were treated with nadroparine, orenoxaparine, or dalteparine. Nadroparine pretreated 2 h before LPS challenge, but not synchronous injection, inhibited the hypothermic response. Nevertheless, pretreatment with equal doses of enoxaparine or dalteparine had no significance on the hypothermia. The high dose of LPS (60 mg/kg) increased the hypothermia and inhibited spontaneous locomotor activity 24 h after treatment. Synchronous nadroparine treatment reduced the hypothermia and eliminated the reduction in spontaneous locomotor activity [22].

HMG-CoA Reductase Inhibitor (Atorvastatin, Lovastatin, or Simvastatin)

Statin is a hydroxymethylglutaryl coenzyme A (HMG-CoA) reductase inhibitor that can control hypercholesterolemia. In an experimental model, simvastatin prevented LPS-induced septic shock in rats [67]. Additionally, experimental sepsis was induced in Swiss Webster mice by i.p. injection of fecal material (feces from naïve Swiss Webster, 5 mg/mL). Atorvastatin or simvastatin did not prevent mortality in septic mice; still, survivors presented lower clinical scores. The atorvastatin or simvastatin treatments decreased pro-inflammatory cytokines, brain lipid peroxidation myeloperoxidase levels, and microglial activation in septic mice. Intravital examination of the brain vessels showed a decrease of functional capillary density and an increase of leukocyte adhesion that were prevented by both atorvastatin and simvastatin. At 15 days after sepsis, mice survivors presented cognitive dysfunction related to hippocampal and aversive amygdala-dependent memories. Statin treatments prevented cognitive impairment assessed by step-down inhibitory avoidance and Morris water maze tasks [68]. In another study from the same research group, the authors showed that sepsis survival with simvastatin treatment was improved in Swiss Webster mice but not in C57BL/6 mice, compared to controls, whereas statins reduced sepsis severity in both mice types at 24 h after induction. Lovastatin or simvastatin retained avoidance memory compared to the control group [19].

Inhibition of Indoleamine 2,3-Dioxygenase Pathway

Kynurenine pathway activation has been reported in several neurological diseases as an effect of host immune response [69,71,71]. The enzyme indoleamine 2,3-dioxygenase (IDO) is the most important connection between the immune system and the kynurenine pathway [72]. The kynurenine pathway is the main route for tryptophan metabolism in mammals. In its first step, tryptophan is converted to kynurenine in a reaction catalyzed by IDO and tryptophan 2,3-dioxygenase (TDO). Then, IDO converts tryptophan into kynurenine that is metabolized into other catabolites through the activity of enzymes within the kynurenine pathway [73]. In this study, sepsis induced hippocampus-dependent cognitive impairment, evidenced by decreased contextual freezing time in a fear conditioning test along with an increase in the hippocampal microglial marker, Iba-1, TNF-α, IL-1β, IL-6, kynurenine, the ratio of kynurenine/tryptophan, and IDO activity and a decreased tryptophan level in mice subjected to the CLP model. A single peripheral administration of l-kynurenine, a metabolite of the amino acid l-tryptophan that is produced by many cells in response to immune activation, induced a deficit in the cognitive impairment in the control group. Nevertheless, mice treated with 1-methyl-d, l-tryptophan (IDO inhibitor) did not experience these changes [31, 74]. In another study, polymicrobial sepsis increased the activity of mitochondrial complexes I, II-III, and IV at 24 h after CLP. However, IDO-1/2 inhibition normalized the activity of these complexes in the hippocampus. Additionally, Wistar rats presented impairment of habituation and aversive memories 10 days after CLP, while the adjuvant treatment with the IDO-1/2 inhibitor prevented these alterations [75]. The results suggest that IDO-dependent neurotoxic kynurenine metabolism was a cause of sepsis-induced cognitive impairment, and IDO inhibitors might be a new avenue as adjuvant treatment for sepsis-associated encephalopathy.

Inhibition of Histone Deacetylase (Sodium Butyrate, Suberoylanilide Hydroxamic Acid, Trichostatin A, or Valproic Acid)

The authors investigated the effect of class I histone deacetylase (HDAC) inhibitor (valproic acid) and suggested that it can prevent cognitive deficits in a sepsis mouse model. The C57BL/6 mice received valproic acid once daily for 14 uninterrupted days commencing either immediately or 2 weeks after CLP surgery. No difference in mortality rate was observed between valproic acid and saline-treated sepsis groups, when valproic acid was administered immediately after CLP for 14 days. However, treatment with valproic acid increased BDNF concentration and IL-1β levels and decreased the activity of caspase-3 with simultaneous increases in acetyl-H3K9 and acetyl-H4K12 levels, compared to the saline-treated sepsis group. Valproic acid prevented cognitive impairment in spatial learning memory, as seen in Morris water maze and Y-maze tasks [76]. In this study, Wistar rats subjected to CLP presented aversive memory impairment. The animals presented an increase of HDAC activity in the hippocampus and cortex 24 h after CLP and in the pre-frontal cortex and hippocampus 10 days after CLP. The adjuvant treatment with sodium butyrate, a class I HDAC inhibitor, prevented memory impairment. Additionally, sodium butyrate presented a late inhibitory effect on HDAC activity in the pre-frontal cortex and in the hippocampus after 10 days of CLP surgery, with no influence of HDAC expression at 24 h after CLP surgery [77]. A study by Fang et al. investigated whether a septic brain was epigenetically modulated by HDACs, using the CLP model in Sprague-Dawley rats. The rats were treated with trichostatin A (TSA) or suberoylanilide hydroxamic acid (SAHA) inhibitors of classes I and II of the HDAC family, respectively, for 7 days after CLP surgery. The HDACs’ inhibition improved spatial learning and memory dysfunction on the Morris water maze task in septic rats. Hippocampal acetylated histone 3 (AcH3), acetylated histone 4 (AcH4), cytoplasmic HDAC4, and B cell lymphoma-extra-large (Bcl-XL) were inhibited in the brain of septic animals. Hippocampal bcl-2-like protein 4 (Bax) and nuclear HDAC4 expressions were increased in CLP group; however, the treatment with HDAC inhibitors preserved the changes of Bcl-XL and Bax [78].

Inhibition of Matrix Metalloproteinases

Matrix metalloproteinases are important for tissue formation, neuronal cell renovation, and BBB homeostasis. During inflammation, MMPs may digest tight junctions and basement membrane proteins, contributing to an increase in the BBB permeability, thus affecting brain homeostasis [79]. Thirty-five hours after CLP induction, Wistar rats presented aversive memory impairment assessed by the inhibitory avoidance task. An intracerebroventricular (i.c.v.) administration of MMP-2 and MMP-9 inhibitors in a single dose after sepsis induction prevented cognitive impairment and BBB disruption [45]. In another study from the same research group, Wistar rats presented aversive memory impairment at 30 days after CLP surgery. Adjuvant treatment with MMP-2/9 inhibitors prevented cognitive impairment and decreased the amyloid-β deposition in the rat brain [38].

Inhibition of Pro-Inflammatory Cytokines

Pro-inflammatory cytokines are essential for a strong host inflammatory response. Nevertheless, early sepsis mortality is caused by an acute and harmful pro-inflammatory response, while the second sepsis phase is associated with acute immunosuppression, which make the patients susceptible to long-term risk for life-threatening secondary infections [80]. The IL-1β receptor antagonist (IL-1ra) prevented the BBB disruption in the pre-frontal cortex, hippocampus, and striatum; decreased the levels of IL-1β, IL-6, and TNF-α in the pre-frontal cortex and striatum at 24 h; and prevented cognitive impairment assessed by habituation to an open field and step-down inhibitory avoidance tasks at 10 days in Wistar rats subjected to CLP [81]. IL-1ra administration also ameliorated long-term potentiation (LTP) in the hippocampus in a CLP mouse model [82]. In another study, wild-type mice presented memory impairment in the novel object recognition task at 10 days after sepsis induced by CLP. However, these deficits were not observed in tumor necrosis factor receptor-1 (TNFR1) knockout mice, and the absence of TNFR1 in mice subjected to CLP surgery triggered a higher BDNF expression in the hippocampus [83].

Mechanistic Target of Rapamycin Inhibitor

Rapamycin is a mechanistic target of rapamycin (mTOR) inhibitor. The mTOR-C1 signaling stimulates cell growth by inducing and inhibiting anabolic and catabolic processes is responsible for cell cycle development, and it is sensitive to rapamycin. mTOR-C2 signaling is insensitive to acute rapamycin treatment; however, chronic rapamycin contact can disrupt its structure [84]. The purpose of this research strategy was to study the neuroprotective effect of rapamycin in Kunming mice subjected to CLP. Fourteen days after CLP surgery, mice were subjected to the Morris water maze, and then the hippocampus was dissected to evaluate mTOR expression. Rapamycin prevented cognitive impairment in the CLP group but did not affect the total mTOR targets. Phosphorylated mTOR targets decreased (p-mTOR-Ser2448, p-p70S6k-Thr389, and p-AKT-S473), autophagy indicators increased (LC3-II, Atg5, Atg7), and P62 decreased in the hippocampus of the rapamycin-treated CLP mice [85]. Sepsis by CLP model enhanced the phosphorylation of Akt, mTOR, and p70S6K along with hippocampal neuronal loss, abnormal neuronal morphology, and impaired long term cognitive performance, suggesting that sepsis-induced hippocampal neurodegeneration triggers Akt/mTOR signaling pathway. However, rapamycin rescued cognitive deficits in acute phase, 14 days after CLP surgery, with no influence on chronic phase cognitive impairment, 60 days after CLP surgery, or long-term neuronal loss in hippocampal CA1 region [86].

Nonconventional Antibiotic Treatment (Minocycline or Tigecycline)

Minocycline is a tetracycline derivative that can cross the BBB and present anti-inflammatory activity with neuroprotective characteristics that limit inflammation and oxidative stress [87]. Although tigecycline is architecturally similar to minocycline, the modifications to the molecule resulted in a prolonged spectrum of its activity and reduced susceptibility to the development of resistance, compared with other tetracycline antibiotics [88]. Sprague-Dawley rats were subjected to traumatic brain injury and CLP procedure to induce sepsis. Immediately following injury, the animals received minocycline, tigecycline, or saline. Mortality in the animals subjected to combined traumatic brain injury and CLP was reversed by both minocycline and tigecycline administration. Minocycline, but not tigecycline, decreased the extent of cortical tissue damage, TNF-α expression in the pericontusional cortex, and microglial activation. Both antibiotics had effects on recovery of cognitive deficits observed following combined traumatic brain injury and CLP surgery [89]. Sepsis also decreased hippocampal Neuregulin-1 (NRG-1) concentrations, which was reversed by minocycline adjuvant treatment. Minocycline also reduced microglia activation and prevented cognitive impairment in C57BL/6 mice subjected to CLP [31] and LTP in the hippocampus of C57BL/6 mice with sepsis [82]. In another study, sepsis increased oxidative damage, pro-inflammatory cytokines, BBB permeability, and cognitive impairment, while these alterations were prevented by minocycline treatment with simultaneous improvement in long-term cognitive impairment evaluated by the inhibitory avoidance task and habituation to open field [90].

Inhibition of Leukocyte Influx into Central Nervous System

In this study, female C57BL/6 (B6; H-2 Kb) mice were treated with anti-NK1.1 monoclonal antibody to deplete natural killer (NK) cells prior to LPS or CLP sepsis induction. In the LPS sepsis model, after disruption of the BBB, conventional CD11b(+) CD27(+) NK cells migrated into the brain. Additionally, depletion of NK cells previous to LPS treatment decreased neutrophil recruitment, IL-1β, IL-6, and TNF-α levels in the brain. NK cell depletion reduced depression-like behavior in LPS-treated mice, as indicated by reduced sucrose preference and changes in serotonin metabolism-associated enzymes and proteins, including tryptophan hydroxylase 2 (TPH2), monoamineoxidase (MAO-A), and serotonin transporter in the mice’s brains. The NK depleted CLP group presented a decrease of TNF-α and IL-1β levels in the brain. Depressive-like behavior was prevented by NK depletion treatment prior to CLP surgery, showing a result similar to that found in the LPS-induced sepsis model [23]. Wistar rats subjected to CLP surgery presented increased levels of IL-1β, IL-6, TNF-α, thiobarbituric acid reactive species (TBARS), nitrosative stress, and BBB dysfunction. After CLP surgery, the Wistar rats received anti-CD40 as a treatment or isotype immunoglobulin (Ig) IgG as a control. The inhibition of CD40-CD40 ligand activation by anti-CD40 (CD40 molecule, TNF receptor superfamily member-5) did not influence the mortality rate but decreased CD40–CD40L levels, cytokines, oxidative damage, and BBB dysfunction. Additionally, anti-CD40 prevented aversive and non-aversive long-term memory impairment 10 days after CLP surgery in sepsis survivor rats compared to the CLP/IgG group [90].

Phytotherapeutic Compounds

  • Physostigma venenosum: Physostigmine is an AChE inhibitor in the beginning extracted from P. venenosum (Calabar bean) and Hippomane mancinella (Manchineel tree) [91]. The LTP in the excitatory synapses of the hippocampal neurons was affected in septic rats compared to controls, suggesting that synaptic plasticity is affected by sepsis and that hippocampal neurons are involved in septic delirium. Physostigmine, the cholinesterase inhibitor improved the LTP suggesting that cholinergic neurotransmission is linked to septic encephalopathy [92].

  • Huperzia serrata: Huperzine-A is an acetylcholinesterase inhibitor and N-methyl-d-aspartate (NMDA) receptor antagonist extracted from Huperzia serrata. Wistar rats were treated with huperzine-A and then subjected to LPS-induced sepsis. Then memory was evaluated at 3, 12, and 24 h after LPS administration by Morris water maze task. Huperzine-A treatment prevented impairment of spatial memory induced by LPS. Huperzine-A also improved cholinergic function by augmenting hippocampal levels of (ChAT), muscarinic acetylcholine receptor-1, and acetylcholine (ACh). In addition, TNF-α and IL-1β protein and gene expression decreased in the hippocampus at 3, 12, and 24 after LPS-induced sepsis in rats treated with huperzine-A [93].

  • Panax ginseng: Ginsenosides are a class of steroid glycosides, and triterpene saponins, found Panax ginseng. Ginsenosides modulated expressions and functions of receptors such as tyrosine kinase receptors, serotonin receptors, NMDA receptors, and nicotinic acetylcholine receptors [94]. C57BL/6 mice were subjected to CLP and treated with Ginsenoside-Rg1 1 h before the CLP. The ginsenoside-Rg1 improved the survival rate and rescued from the learning and memory deficits as noted on Morris Water maze. This adjuvant treatment also decreased microglial marker Iba-1, reduced the expression of TNF-α, IL-1β, and IL-6; activation of caspase 3; expression of microtubule-associated protein 1A/1B-light chain 3 (LC3) and nucleoporin p62 (p62) in hippocampus [24].

  • Resveratrol: Resveratrol is a polyphenol that has antioxidant activity and decreases the activation of SIRT-1 [95]. In another study, C57BL/6 mice were subjected to CLP surgery and presented an increase of Iba-1, IL-1β, NLRP-3 expression, and apoptosis in the hippocampus followed by spatial memory impairment evaluated by Morris water maze [58]. However when the rodents were treated with resveratrol, there is a decrease in microglial markers and pro-inflammatory markers with simultaneous reduction in in memory impairment. Resveratrol protected against sepsis-associated encephalopathy by inhibiting the NLRP-3/IL-1β axis in microglia [58].

Radical Scavenging

  • Alpha-lipoic acid: This study evaluated the effect of α-lipoic acid (ALA; 200 mg/kg) an antioxidant compound on brain dysfunction in Wistar rats subjected to CLP. Animals were divided into sham + saline, sham + ALA, CLP + saline, and CLP + ALA groups. Twelve, 24 h, and 10 days after surgery, the hippocampus, prefrontal cortex, and cortex were assayed for TNF-α and IL-1β, BBB permeability, nitrite/nitrate concentration, MPO activity, TBARS formation, protein carbonyls, SOD and CAT activity, and neurotrophins levels. Treatment with ALA decreased TNF-α and IL-1β levels, MPO activity, nitrite/nitrate concentration, and lipid peroxidation with simultaneous increase in CAT activity. ALA also enhanced NGF levels in hippocampus and cortex and prevented cognitive as measured in novel object recognition test after sepsis [96].

  • N-acetylcysteine and/or deferoxamine: On days 10 and 30 after CLP surgery, Wistar rats were subjected to inhibitory avoidance task, habituation to an open field, and continuous multiple-trials step-down inhibitory avoidance task. The sepsis group presented memory impairment that was prevented by use of N-acetylcysteine plus deferoxamine (NAC/DFX) adjuvant treatment [27]. In another study from the same research group, CLP sepsis inhibited mitochondrial electron transport chain complexes I and II; however, treatment with NAC/DFX, taurine, or RC-3095 (gastrin-releasing peptide receptor antagonist) prevented these changes on complexes I and II in the rat brain. CLP sepsis also increased creatine kinase (CK) activity in hippocampus, cerebral cortex, cerebellum, and striatum that were prevented by the NAC/DFX or taurine treatments [97,100,99].

  • Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase inhibitor (apocynin): For this interventional study, C57BL/6 mice were subjected to CLP and treated with a NADPH oxidase inhibitor knows as apocynin. The time consumed in the center of the arena was reduced on day 13 but not on day 29 in the CLP/vehicle group compared with the control, whereas apocynin treatment prevented the decrease in the CLP group. On day 14, apocynin treatment also prevented the decrease in the freezing time of CLP mice group. Sepsis triggered cognitive impairment, which was followed by selective phenotype loss of parvalbumin interneurons and increased gp91 (phox), 4-hydroxynonenal, MDA (malondialdehyde), IL-1β, and IL-6 expressions; however, apocynin treatment reduced these inflammatory and oxidative markers [100]. In another study, CLP sepsis caused hypotension, hyperlactatemia, renal and hepatic dysfunction, along with an increase in the levels of IL-6, IL-1β, macrophage inflammatory protein (MIP), and late-cognitive deficits. Apocynin decreased the levels of H2O2 and reduced the oxidative stress [101]. Sepsis was induced in wild-type and gp91 (phox) knockout mice by CLP. The absence of NOX2 in gp91 (phox−/−) mice prevented glial cell activation. Alternatively, experimental sepsis was induced in C57BL/6 mice by an i.p. injection of the fecal slurry for behavioral studies to avoid surgery and the mice were treated with apocynin. Pharmacological inhibition of NOX2 with apocynin prevented hippocampal oxidative stress and development of long-term cognitive impairment assessed by step-down inhibitory avoidance task and Morris water maze [102]. Wild-type C57/BL6 mice and mice deficient for the inducible nitric oxide synthase gene (NOS2−/−) were subjected to sepsis model by LPS administration. LPS increased NOS2 expression in wild-type mice compared to NOS2−/−mice. Wild-type mice showed more behavioral impairment in eight-arm radial maze and open field tasks on LPS treatment compared to NOS2−/−mice, suggesting that LPS-induced NOS2 linked NO production. LPS-treated wild-type mice had increased brain mRNA levels of TNF-α, IL-1β, and RANTES [103].

  • Hydrogen gas: Hydrogen gas (H2) is an antioxidant that decrease toxic reactive oxygen species (ROS) such as hydroxyl radical (•OH); however, hydrogen-rich saline (HRS) is more appropriate for clinical application [104, 105]. In this study, the survival rate was superior among Wistar rats submitted to CLP that had received HRS compared with those that had received no treatment. CLP group presented cognitive impairment evaluated by Morris water maze, cell damage categorized by histopathologic changes and oxidative damage in the hippocampus. These changes were attenuated by HRS dose-dependent treatment [106]. ICR mice underwent CLP or sham operation and were treated with 2% H2 for 60 min. The H2 treatment reduced the levels of pro-inflammatory cytokines and oxidative products and increased activities of antioxidant enzymes in serum and hippocampus. Further, the H2 treatment stimulated the expression and transposition of nuclear factor erythroid 2-related factor 2 (NRF2) and the expression of cytoplasmic heme oxygenase-1 (HO-1). In addition, H2 prevented cognitive impairment in sepsis group evaluated by Y-maze and fear conditioning test [25].

  • Disufenton sodium (NXY-059): NXY-059 is a disulfonyl derivative of the neuroprotective spin trap phenylbutynitrone (PBN) and its hydrolysis/oxidation product MNT are very powerful scavengers of free radicals. In this study, NXY-059 showed no improvement on mortality rate of Swiss Webster mice subjected to CLP sepsis. On cognitive evaluation, the animals treated with NXY-059 improved when compared to controls, by a reduction in the numbers of crossings and rearings in the open field test [32].

Miscellaneous (Agonists, Antagonists, and Inhibitors of Different Receptors)

  • Acetylcholinesterase inhibitor (Rivastigmine): Wistar rats were subjected to CLP sepsis and 3 days after surgery the animals received rivastigmine as adjuvant treatment or saline for 7 days. Ten days after surgery, rats were submitted to habituation to an open-field memory test. CLP group presented habituation memory impairment; however, CLP rats treated with rivastigmine presented a reduction in the number of crossings and rearings on the open field habituation between test and training session, indicating memory acquisition [107].

  • β2 adrenergic receptor agonist (Salmeterol): Mice received pre-treatment with salmeterol 1 h before LPS sepsis induction reduced the expression of hippocampal pro-inflammatory cytokines TNF-α, IL-1β, and IL-6. Subsequently, the cognitive impairment induced by systemic LPS stimulation was attenuated by salmeterol treatment [108].

  • CB1and CB2 (cannabidiol) receptor agonist: The pharmacological effects of cannabinoid type I (CB1) and cannabinoid type II (CB2) are mediated through G protein coupled receptors [109]. Cannabidiol improved cognition in multiple pre-clinical models such as schizophrenia, Alzheimer’s disease, brain ischemia, cerebral malaria, hepatic encephalopathy, bacterial meningitis, and sepsis [97,100,99, 110, 111]. CLP Wistar rats treated with cannabidiol at different doses reduced the mortality, MDA levels, carbonyl levels, and prevented cognitive impairment as evidenced by aversive memory improvement on inhibitory avoidance task compared to CLP plus vehicle group in both acute and chronic phases of sepsis [97,100,99].

  • Cholinesterase inhibitor (eserine) and selective CB2receptor agonist (JWH-133): In this study, the authors evaluated the effect of a cholinesterase inhibitor (eserine), a selective CB2 receptor agonist (JWH-133), on LPS-induced sepsis. Wistar rats received LPS injection and 30 min after the injection, the animals were treated with eserine, JWH-133, or eserine plus JWH-133. At 24 h after LPS administration and adjuvant treatment, the animals were subjected to the habituation to T maze, rota rod, and activity cage tests. The adjuvant treatments improved the cognitive skills, locomotor and exploratory activity, and motor co-ordination that were impaired with LPS when compared to LPS plus vehicle. Eserine, JWH-133, or eserine plus JWH-133 also prevented an increase of IL-6, vascular cell adhesion molecule-1 (VCAM-1), and oxidative-nitrosative stress in terms of MDA and iNOS gene expression [21].

  • Acetylation of cyclophilin D (CypD): C57BL/6J mice were subjected to CLP surgery and assigned in six groups: sham group, CLP group, CypD siRNA transfection (CypD-si) group, CypD control siRNA transfection (CypD-c) group, Sirtuin (SIRT) 3 overexpression vector pcDNA3.1 (SIRT3-p) group, and SIRT3 empty vector pcDNA3.1 (SIRT3-v) group. The CypD-si and CypD-c groups were transfected with CypD siRNA and CypD control siRNA. In addition, the SIRT3-p and SIRT3-v groups received SIRT3 pcDNA3.1 and vector pcDNA3.1, respectively. CypD and acetylation of CypD levels increased in the hippocampus of mice subjected to CLP surgery. In addition, increasing SIRT3 and decreasing CypD prevented cognitive impairment, apoptosis, and protected the integrity of mitochondrial membrane. Activated SIRT3-mediated deacetylation of CypD reduced the learning and memory impairment evaluated by Morris water maze in CLP sepsis [112].

  • d-Arg-2′, 6′-dimethyltyrosine-Lys-Phe-NH2 (SS-31 peptide): The aim of this research strategy was to evaluate the effects of the mitochondria-targeted peptide SS-31 on mitochondrial function and cognition in CLP sepsis model. C57BL/6 mice were treated with peptide SS-31 (5 mg/kg) administrated after surgery and later on once daily for six uninterrupted days. Seven days after CLP surgery, surviving mice were subjected to open field and fear conditioning tests and the hippocampus was collected for biochemical analysis. SS-31 treatment improved survival rate, ameliorated the behaviour performance on open field test, and increased the freezing time in 24 h context test. SS-31 treatment prevented a decrease of mitochondrial complexes I and III enzyme activities, prevented an increase of ROS, and a reduction of ATP content in the hippocampus of CLP-induced mice. In addition, SS-31 protected the integrity of mitochondrial membrane, prevented apoptosis and neuronal damage, decreased the levels of IL-1β, and NLRP-3 expression in the hippocampus of CLP-induced mice [113].

  • d-Serine (The NMDA receptor co-agonist d-serine): Sepsis was induced by CLP or by a single intraperitoneal injection of LPS, 8 mg/kg in C57BL/6J mice. Sepsis reduced the protein and mRNA levels of NMDA receptor subunits GluN2A, GluN2B, and GluN1 but not synaptophysin levels or the hippocampal neuronal number in both CLP and LPS mice in the first week. d-serine, co-agonist of NMDA receptors, limited the LPS induced damage, including the cognitive impairment, NMDA receptor subunits loss, neuro-inflammation, oxidative stress, and the hippocampal decrease of p-CREB. As sepsis-induced NMDA receptor loss is interfering with hippocampal changes, NMDA receptors are target platform for future interventions [114].

  • Electroacupuncture: Sprague-Dawley rats were pre-treated with different waveforms of electroacupuncture (Baihui and bilateral Tsusanli acupoints). After electroacupuncture pre-treatment, the animals were subjected to CLP surgery. The survival rates increased in the CLP rats pre-treated with continuous wave, dilatational wave, and intermittent wave compared to CLP group. All waveforms prevented memory impairment evaluated by Morris water maze task in septic survivor rats. Electroacupuncture pre-treatment decreased the production of TNF-α, IL-6, MDA, and increased the activity of SOD and CAT in serum and hippocampus at 48 h after CLP surgery. Electroacupuncture pre-treatment also decreased the expression of TLR-4, NF-κB, and Iba-1 in the hippocampus of CLP sepsis rats [115].

  • Neuregulin (NRG)-1: NRG-1 is a ligand for the receptor tyrosine-protein kinase (erbB)-3 and erbB4 members of the epidermal growth factor receptor (EGF) family receptors. The NRG-1 is produced from neurons to stimulate the formation and maintenance of radial glial cells. CLP sepsis-induced anxiety-like behavior and hippocampal-dependent cognitive impairment, as demonstrated by significantly augmented distance spent in the open field task and reduced freezing time to context in the fear conditioning task [31]. The NRG1-β1 adjuvant treatment prevented the sepsis-induced cognitive impairment; however, the treatment with the EGFR inhibitor, AG1478, nullified the NRG1-β1 effect on behaviour task.

  • Glutamatergic neurotransmission inhibitor (Riluzole): Riluzole is a glutamate release inhibitor drug. Wistar rats were subjected to CLP surgery and received riluzole, 30 min after the surgery, and every 12 h as continuing treatment. The outcome of riluzole on the survival rate, body weight and temperature, leukocyte amount, neurological investigation scores, and brain edema were evaluated at 6 and 48 h after CLP surgery. CLP rats presented survival rates of 89, 50, and 28% at 6, 24, and 48 h, respectively. In CLP rats treated with riluzole, the survival rate improved to 94, 72, and 50% at the same time points. Riluzole also decreased MDA, and increased glutathione (GSH) levels in the rat brain and improved weight loss, body temperature, brain edema, and BBB permeability. In addition, Bederson’s neurological examination scores [116] decreased in the CLP rats treated with riluzole [117].

  • [Gly14]-Humanin (HNG): HNG is a derivative of humanin (HN). The HN is a liberated bioactive peptide that decreases cell toxicity by inhibiting c-JunNH2-terminal kinase [118]. In this study, ICR mice were subjected to CLP surgery and treated with HNG peptide. The HNG treatment reduced IL-1β, IL-6, and TNF-α levels; GFAP-positive astrocytes and Iba-1-positive microglia biomarkers that were elevated 16 h after CLP sepsis. On day 21 after sepsis surgery, mice were subjected to Y-maze task and they presented impairment in working memory. HNG treatment prevented cognitive impairment in working memory and also improved basal forebrain cholinergic neuronal loss and reduced synaptic plasticity caused by sepsis [119].

  • Guanosine: Guanosine is a purine nucleoside thought to have neuroprotective properties [120]. Wistar rats subjected to CLP were treated with i.p. guanosine injection. Twelve and 24 h after CLP surgery, TBARS and protein carbonyls formation were evaluated. Guanosine treatment decreased TBARS and carbonyl levels in the brain. On day 10, another group of rats were subjected to habituation to an open-field apparatus, inhibitory avoidance ask, object recognition task, and forced swimming task. Guanosine treatment prevented memory impairment and depressive-like behavior [121].

  • High mobility group box 1 (HMGB-1) inhibitor: Polymicrobial sepsis was induced by CLP in BALB/c mice. Sepsis survivor mice presented an increase of HMGB-1 levels in the serum for at least 12 weeks after CLP, along with learning and memory impairment. The anti-HMGB-1 monoclonal antibody was provided once a day for 3 days. Animals were subjected to SHIRPA, open-field task, black-and-white alley test, and navigational test. Administration of anti-HMGB-1 antibody improved memory impairment and brain pathology. To test their hypothesis, recombinant HMGB-1 was administrated to naïve mice and memory was evaluated. Interestingly, administration of recombinant HMGB-1 to naïve mice caused memory impairment [46].

  • Intermittent fasting diet: Wistar rats on intermittent fasting diet were deprived of food for 24 h every other day for 30 days. On day 31, the rats had access to food for 24 h and received LPS intravenously for sepsis induction. Intermittent fasting diet improved cognitive deficits by decreasing the pro-inflammatory cytokine expression, and enhancing neurotrophic support. LPS administration exhibited impairment in cognitive performance in the Barnes maze and inhibitory avoidance tasks, without changes in locomotor activity, that were improved in intermittent fasting diet rats [122].

  • Immunoglobulin therapy: Wistar rats were subjected to CLP surgery and received IgG or immunoglobulins enriched with IgA and IgM (IgGAM) at 5 min after the CLP surgery. On days 10, 30, and 60, the animals were subjected to open field, elevated plus maze, and forced swimming tasks. In IgG and IgGM groups, the mortality decrease to 30 and 20%, respectively. On day 10, the rats presented depressive-like behavior that was prevented by both treatments. However, on day 30 and on day 60 after CLP surgery, the rodents did not present depressive-like behavior [123].

  • Metformin: It is a drug used to treat type-2 diabetes and it exerts anti-inflammatory and anti-oxidant effect [124, 125]. Metformin treatment increased the survival rate, protected BBB integrity, attenuated neuronal apoptosis, brain edema, oxidative damage, and pro-inflammatory cytokine levels, and improved cognitive function along with an increase in Akt phosphorylation. However, LY294002, a phosphatidylinositol-3-kinase (PI3K) inhibitor reverted the metformin’s neuroprotective effect theorizing that metformin’s neuroprotective effect might be from activation of PI3K/Akt signaling pathway [126].

  • N-acetyl-5-methoxy tryptamine (melatonin): Melatonin is a hormone that is produced by the pineal gland in animals and is an important physiological sleep regulator in humans that presented anti-inflammatory and antioxidant properties in pre-clinical models [127,133,129]. Melatonin treatment immediately after surgery improved survival rate with no behavioral change and reduced plasma IL-1β levels, whereas melatonin treatment 7 days after surgery improved cognitive assessments by reverting hippocampal BDNF and GDNF levels, suggesting that melatonin could be a novel therapeutic solution for sepsis associated encephalopathy [130].

  • N-Methyl-d-aspartate (NMDA) receptor antagonist (MK-801): Sepsis was induced by CLP surgery in Wistar rats. Animals were treated with a single dose of MK-801 and 10 days after the surgery, memories were evaluated by different tasks. MK-801 adjuvant treatment prevented impairment in aversive memory and short and long-term memories as evaluated by inhibitory avoidance task and novel object recognition task respectively [29].

  • Nicotinic acetylcholine receptor agonist (nicotine): Wistar rats were subjected to CLP sepsis and received an adjuvant treatment with nicotine. The animals were treated with nicotine or vehicle every day per 1 week before and/or 1 week after sepsis surgery. At 30 min after the last administration of nicotine, the rats were subjected to the open field, elevated plus-maze, and step-down inhibitory avoidance tasks. The constant nicotine treatment did not change the survival rate in the sepsis group. Moreover, while sepsis group showed no significant changes on locomotor activity, the treatment with nicotine during 1 week after CLP decreased the locomotion of sepsis-surviving rats in the open field. Both nicotine treatments (prior and/or after CLP surgery) enhanced the sepsis-induced anxiety-like behavior. Nicotine also was able to recover short-term and long-term inhibitory avoidance memory impairments, detected in sepsis survivors, only when administered during two successive weeks (prior and after CLP surgery) [131].

  • RAGE antagonist: Rat polyclonal anti-RAGE (RAGEab) (100 μg/kg saline) was administered bilaterally into the hippocampus at days 15, 17, and 19 after CLP. Control animals received 100 μg/kg of isotype IgG. Serum proinflammatory markers (TNFα, IL-1β, and IL-6), levels of RAGE, RAGE ligands (S100B, Nε-[carboxymethyl]lysine, HSP70, and HMGB1), brain levels of TLR4, GFAP, neuronal NOS, Aβ, and p-tauSer202 all increased during post-CLP period. Intracerebral administration of RAGE antibody post-CLP reversed these changes and also attenuated the cognitive deficits that resulted from sepsis. The data suggest that RAGE induces neuronal damage that might be alleviated with anti-RAGE treatments [132] .

  • Serine/threonine kinase glycogen synthase kinase 3β (GSK3β) inhibitor: TDZD-8 is a thiadiazolidine derivative that acts as a non-ATP competitive inhibitor of the serine/threonine kinase glycogen synthase kinase 3β (GSK3β) [133]

  • Glucagon-like peptide (GLP-1): Liraglutide is an equivalent of human GLP-1 and acts as a GLP-1 receptor agonist with insulinotropic activity [134]. Swiss mice were subjected to sepsis by CLP and on day 30 post-surgery, the animals presented memory impairment evaluated by novel object recognition and step-down inhibitory avoidance task, but they demonstrated normal performance when re-evaluated on day 45 after CLP surgery. Cognitive impairment in post-septic animals were accompanied by decreased hippocampal levels of synaptophysin, cAMP response element-binding protein (CREB), CREB phosphorylated at serine residue 133 (CREBpSer133), and GluA1 phosphorylated at serine residue 845 (GluA1pSer845). Expression of TNF-α increased, IRS-1 phosphorylation at serine 636 (IRS-1pSer636) increased, and phosphorylation of IRS-1 at tyrosine 465 (IRS-1pTyr465) decreased in the hippocampus of mice on day 30 after CLP surgery. Phosphorylation of Akt at serine 473 (AktpSer473) and of GSK3 at serine 9 (GSK3βpSer9) were also diminished in hippocampus of post-septic mice. Post-septic mice were treated with liraglutide for 10 days or TDZD-8 for 5 days that began on day 20 and on day 25 after CLP surgery. Both treatments prevented memory impairment evaluated by novel object recognition task [135].

  • Vitamin B6: Wistar rats subjected to CLP who received a treatment with vitamin B6 prevented BBB disruption, neuroinflammation, oxidative stress, and energy metabolism changes and decreased long-term cognitive impairments by improving learning and memory deficits. Vitamin B6 might have brought these changes by decreasing tryptophan metabolism changes via kynurenine pathway [136], Table 1.

Table 1 Characteristics of the included pre-clinical studies
Full size table

Clinical Studies

Neuropsychiatric Manifestations and Long-Term Cognitive Decline in Sepsis Survivor Patients

A study by Iwashyna et al. evaluated the total number of Medicare beneficiaries surviving at least 3 years after severe sepsis and to evaluate the burden of their cognitive impairment and disability. Severe sepsis was evaluated by a standard administrative definition. A total of 637,867 Medicare beneficiaries were alive at the end of 2008 who had survived severe sepsis 3 or more years earlier. An estimated 476,862 had functional disability, with 106,311 survivors having moderate to severe cognitive dysfunction [167]. Another study determined the changes in cognitive dysfunction and physical behavior among patients who survived severe sepsis. Individual interviews were performed with respondents or proxies using validated surveys to assess the presence of cognitive dysfunction and to determine the number of activities of daily living (ADL) and instrumental activities of daily living (IADL) with which patients needed assistance. The prevalence of moderate to severe cognitive dysfunction increased by 10.6% among patients who survived severe sepsis, and a high rate of new functional restrictions was seen following sepsis. Patients with no limits before sepsis presented a mean of 1.57 new limitations, and for those patients with mild to moderate limitations before sepsis, a mean of 1.50 new limitations was found. In contrast, non-sepsis general hospitalizations were related with no change in moderate to severe cognitive dysfunction and with the evolution of fewer new limitations [168]. In a prospective case study, a woman patient contributed in clinical interviews, comprehensive neuropsychological testing, and neurological magnetic resonance imaging (MRI) at approximately 8 months and 3.5 years after ICU discharge. Compared to pre-ICU baseline test data, her intellectual function deteriorated nearly 2 standard deviations from 139 to 106 (from the 99th to the 61st percentile) on a standardized intelligence test 8 months post-discharge. Initial diffusion tensor brain magnetic resonance imaging (DT-MRI) at the end of ICU hospitalization presented diffuse unusual hyper-intense areas connecting predominately white matter in both hemispheres and the left cerebellum. A brain MRI 3.5 years after ICU discharge confirmed the development of profound atrophy with sulcal widening and ventricular enlargement [169]. In another prospective case-control study, researchers compared the neurodevelopmental and behavioral outcomes of 50 children with sepsis-associated encephalopathy. Children with sepsis-associated encephalopathy demonstrated worse mean verbal IQ, full-scale IQ, and General Development Score, as well as the physical, adaptive, social-emotional, cognitive, and communication subscales of the latter. The proportion of sepsis cases with low intelligence was 52 versus 32% in controls. The most common behavior changes were decline in school performance (44%), disobedience (28%), and stubbornness/irritable behavior (26%). Children with Glasgow Coma Scale scores ≤ 10 and ≤ 8 presented impairments in full-scale IQ. In summary, children who had survived sepsis-associated encephalopathy presented delayed neurodevelopment, low verbal IQ, weakening in school performance, and low intelligence at short-term follow-up. Irritability, shock, and duration of sedation were associated with reduced behavioral outcomes [170]. This prospective cohort evaluated the long-term changes in neurobehavioral parameters, brain morphology, and electroencephalography of sepsis and non-septic patients. Twenty-five septic and 19 non-septic ICU survivors were enrolled to evaluate brain morphology, standard electroencephalography, cognition and psychiatric behavior, and health-related quality of life (HRQoL). Sepsis survivors presented cognitive damage in verbal learning and memory and a decrease of left hippocampal volume compared to healthy controls. The sepsis group and to some extent the non-septic ICU patients presented more low-frequency activity in the EEG indicating brain impairment. No differences were found in HRQoL, psychological functioning or depressive symptoms, and depression could be ruled out as a confusing factor [171]. In this retrospective double-blind pilot randomized controlled trial, primary outcomes of physical function and self-reported HRQoL were recorded at ICU discharge and 6 months post-hospital discharge. A significant increase in patient self-reported physical function and physical role for the SF-36 at 6 months was found in the exercise group. Physical function scores were not significantly different between the groups. IL-10 levels was higher in the exercise group; however, there were no differences between the groups for lactate, IL-6, TNF-α, muscle strength, exercise capacity, fat-free mass, or hospital anxiety [172]. This study targeted to identify diagnoses or events during a hospitalization requiring critical care that are related with a subsequent dementia diagnosis in the elderly. Over the 3-year follow-up period, dementia was again diagnosed in 4519 (17.8%) of 25,368 patients who were treated in intensive care and survived hospital discharge. Infection or a diagnosis of severe sepsis, acute neurologic dysfunction, and acute dialysis were all independently associated with a subsequent diagnosis of dementia [173]. A prospective cohort study using a battery of questions and functional status measured as the number of ADLs and IADLs for which assistance is needed evaluated a total of 1208 members. These members presented with 1548 incident severe sepsis episodes that were associated with an increase in the prevalence of moderate/severe cognitive impairment from 9.8 to 19.4%. On adjusting stable patient clinical features, in fixed-effect regression analysis, incident severe sepsis was associated with a 4.2-fold increase in the odds of developing moderate/severe cognitive impairment. Patients with normal pre-sepsis functionality developed 1.72 new I/ADL limitations post severe sepsis. Patients with mild/moderate functional limitations (requiring assistance with 1–3 I/ADLs) pre-sepsis developed 1.40 new I/ADL limitations. However, patients with severe (> 4 I/ADL) limitations before sepsis did not show much change post-sepsis [174]. In this prospective controlled observational study, evoked oscillatory responses to rhythmic visual stimuli were evaluated and analyzed to study brain synchrony via magnetoencephalography (MEG) in 26 survivors of severe sepsis or septic shock, and 23 healthy individuals and patients diagnosed with liver cirrhosis were evaluated as control group. Dynamic adaptation of cerebral neurons in terms of frequency coupling to the rhythmic flicker stimulation was reduced in sepsis survivors and in liver cirrhosis patients; however, in sepsis survivors, it augmented with time following sepsis. The cognitive injury results from pathologically desynchronized neuronal oscillations and from an altered oscillatory coupling in the brain. This study demonstrates that long-term cognitive injury is still present 1 year after severe sepsis, and it was reflected by an abnormal functional brain synchronicity [175]. Another study was a voluntary, web-based prospective survey of sepsis survivors distributed via social media and online channels. A total of 1475 completed surveys were analyzed that presented with increases in body numbness, fatigue, pain, chest pain, palpitations, visual disturbances, stomach and eating problems, memory loss, mood changes, and hair loss, together with problems with dentition and nails, compared to before sepsis. Sepsis survivors also presented with decreased ability to perform chores, walk up and down stairs, walk for at least 15 min, run errands, adequately spell when writing, and read for at least 15 min and reduced satisfactory sex drive. In summary, sepsis survivors presented mental, physiological, and functional disabilities for a significant time following their initial episode of sepsis [176].

Major Depressive Disorder in Sepsis Survivor Patients

A study evaluated factors associated with depression symptoms in a prospective cohort of 135 patients after abdominal sepsis. Depression symptoms were evaluated using the Impact of Events Scale-Revised (IES-R) and the Beck Depression Inventory II (BDI-II). Five percent of patients expressed severe depression symptoms [177]. The aim of this study was to measure self-reported HRQoL, anxiety, depression, and cognitive behavior in pediatric septic shock survivors. HRQoL was evaluated with the KIDSCREEN-52, anxiety with the State Trait Anxiety Inventory for Children, depression with the Children’s Depression Inventory, and cognitive function with the cognitive scale of the TNO-AZL Children’s Quality of Life Questionnaire Child Form. The median age of the children at pediatric intensive care unit admission was 4.2 years old, and the median age at follow-up was 10.7 years old. Depression, quality of life, and anxiety scores were equal than those of the healthy controls, whereas cognitive function was inferior than in healthy controls. Forty-four percent of the children presented cognitive scores < 25% of those of the norm population. In septic shock survivors, HRQoL, anxiety, and depression were equivalent to or superficially better than those of the age-related Dutch norm population; however, cognitive function was decreased [178]. This study evaluated whether incident severe sepsis was associated with augmented risk of depressive symptoms. A total of 439 patients were assessed with an adapted version of the Center for Epidemiologic Studies Depression Scale, and severe sepsis was recognized using an authorized algorithm in Medicare claims. Depressive symptoms were found in 28% of sepsis survivor patients at a median of 1.2 years before sepsis and persisted in 28% of sepsis survivor patients at a median of 0.9 years after sepsis. Neither incident severe sepsis nor severe sepsis-related clinical characteristics were related with succeeding depressive symptoms [179]. One study aimed to evaluate whether pre-sepsis depressive symptoms were linked with risk of new cognitive dysfunction in survivors of severe sepsis. Severe sepsis was identified using a validated algorithm in Medicare claims. A total of 447 patients were evaluated prospectively using an adapted version of the Center for Epidemiologic Studies Depression Scale, and cognitive function was measured using versions of the Telephone Interview for Cognitive Status (TICS). Depressive symptoms in patients with normal cognition before sepsis were 38%, and after severe sepsis, 18% of the survivors had incident cognitive impairment. Depressive symptoms were connected with post-sepsis incident cognitive impairment, and pre-sepsis depressive symptoms remained the strongest factor associated with post-sepsis incident cognitive dysfunction [180]. A retrospective cohort study evaluated physical and mental long-term outcomes of ICU stay for severe sepsis in patients and their spouses. They involved 55 patients who survived severe sepsis and their spouses with a median of 55 months after ICU liberation. The Hospital Anxiety and Depression Scale, the Short Form-12 Health Survey, the Post-traumatic Stress Scale-10, and the Giessen Subjective Complaints List-24 were included. Patients and spouses (26 and 42%, respectively) showed clinically pertinent scores of anxiety and depression; about two thirds of the patients and spouses informed post-traumatic stress symptoms defined as clinically significant. Patients reported anxiety, exhaustion, and poorer mental and physical HRQoL, and the spouses presented impaired mental HRQoL and increased anxiety [181].

Delirium in Sepsis Survivor Patients

In another prospective ancillary study within the SAILS trial, a randomized controlled trial evaluating mortality and ventilator-free days for rosuvastatin versus placebo for patients with sepsis-associated acute respiratory distress syndrome, delirium was evaluated with the validated Confusion Assessment Method (CAM)-ICU method, and cognitive function was evaluated with tests for executive function, language, verbal reasoning and concept formation and working, immediate, and delayed memories. The mean percentage of days with delirium was 34% in the rosuvastatin group and 31% in the placebo group. At 6 months, 19 (36%) of 53 patients in the rosuvastatin group versus 29 (38%) of 77 in the placebo group had cognitive deficiency, with no significant difference between the groups. At 12 months, 20 (30%) of 67 patients versus 23 (28%) of 81 patients had cognitive deficiency, with no significant difference between the groups. The results suggest that delirium may affect a considerable amount of patients, with approximately one third of survivors presenting cognitive deficiency over 1 year of follow-up [182].

Quality of Life in Sepsis Survivor Patients

The aim of this research strategy was to compare the HRQoL of survivors of severe sepsis and septic shock with HRQoL in others who survived serious disease without sepsis. A follow-up interview was held 6 months after ICU discharge, and a EuroQol five-dimension (EQ-5D) questionnaire was used. A total of 104 sepsis survivors and 133 survivors in the control group responded to the EQ-5D test. Survivors of severe sepsis and septic shock presented almost identical HRQoL to that of survivors of critical illness admitted without sepsis [183]. This prospective observational study aimed to assess long-term survival and quality of life of patients admitted to a surgical ICU for the reason of sepsis or trauma. The patients were separated into sepsis and trauma groups, and quality of life was measured 2 years after ICU admission using the EQ-5D questionnaire. A total of 98 trauma patients and 66 patients with sepsis were involved in the study. There was no difference between the groups in Acute Physiology and Chronic Health Evaluation II score or length of stay in the surgical ICU. Surgical ICU survival, in-hospital survival, post-hospital survival, and cumulative 2-year survival were lesser in the sepsis group than in the trauma group. There was no variation in quality of life in all five magnitudes of the EQ-5D. Sixty percent of the patients presented symptoms of depression, nearly 60% had difficulties in their normal activities, and 56% presented pain [184].

PTSD in Sepsis Survivor Patients

This study evaluated factors related with post-traumatic stress symptoms in a prospective cohort of 135 patients after peritoneal sepsis. PTSD was evaluated using the Impact of Events Scale-Revised (IES-R) and the Post-Traumatic Symptom Scale 10 (PTSS-10). The percentage of patients with moderate PTSD symptom scores was 28% and that of patients with high PTSD symptom scores was 10%. Thirty percent of patients after peritoneal sepsis reported higher levels of PTSD symptoms [177], Table 2.

Table 2 Characteristics of clinical studies included
Full size table

Very Low Birth Weight Preterm Infants and Neonatal Sepsis

The aim of another prospective cohort was to evaluate whether neonatal infections were related with an elevate risk of unfavorable neurodevelopment at 5 years of age in very preterm children. A total of 2277 live births were qualified for a follow-up assessment at 5 years of age. Cerebral palsy and cognitive injury were considered as a function of early-onset sepsis (EOS) and late-onset sepsis (LOS), after modification for potential confounding influences, in multivariate logistic regression models. At 5 years of age, the occurrence of cerebral palsy was 9% and that of cognitive impairment was 12%. The occurrence of cerebral palsy was higher in infants with isolated EOS or isolated LOS than in uninfected infants, and this risk was even higher in cases of mutual EOS and LOS. There was no association between neonatal infection and cognitive impairment [187]. This prospective cohort evaluated neurodevelopmental and growth deficiency among extremely low-birth-weight infants with neonatal infection. Neurodevelopmental and growth consequences were evaluated at a comprehensive follow-up visit at 18 to 22 months of corrected gestational age and compared by infection group. A total of 6093 infants were studied and classified by type of infection: uninfected (n = 2161), clinical infection alone (n = 1538), sepsis (n = 1922), sepsis and necrotizing enterocolitis (n = 279), or meningitis with or without sepsis (n = 193). The sepsis group presented a different risk factor of neurodevelopmental impairment according to the infecting microorganism. In the coagulase-negative staphylococci infection group, the OD was 1.3; for other Gram-positive infections, the OD was 1.7; for Gram-negative infections, the OD was 1.4; for fungal infections, the OD was 1.4; and for combined pathogens, the OD was 1.6 [188]. This prospective cohort aimed to identify determinants of neurodevelopmental outcome in preterm children. Gestational age, sex, outborn, illness severity, bronchopulmonary dysplasia, necrotizing enterocolitis, late-onset sepsis, retinopathy of prematurity, abnormal neuroimaging, and site were significantly associated with neurodevelopmental impairment (Bayley-III < 70, severe cerebral palsy, blind or hearing aided, and neurodevelopmental impairments or death). Neurodevelopmental impairment was associated with late-onset sepsis [189]. In a prospective cohort, the authors confirmed previous reports that neonatal sepsis increases the risk of a poor neurodevelopmental outcome in extremely low-birth-weight infants. The sepsis group was associated with poor outcomes and presented an OD of 1.7 [190]. A prospective cohort nested in a double-blind randomized controlled trial included 204 pre-term patients who had survived sepsis and 204 pre-term as a control. The patients were evaluated using Bayley-III and PARCA-R instruments. Both instruments showed cognitive delay (4.4 and 19.6%, respectively) and language delay (8.4 and 12.6%, respectively) in sepsis survivor patients (Martin). A case-control study included 102 low-birth-weight infants as a control group and 18 survivors of sepsis. These infants were prospectively followed for 36 months. Preterm infants who develop sepsis are not at significantly higher risk for triggering neurodevelopmental disability [191]. This population-based prospective cohort involved infants born before 32 weeks of gestation, and cognition was assessed with the K-ABC and behavior with the Strengths and Difficulties Questionnaire (SDQ). In contrast, in these results, 48/342 (14%) premature infants who had survived sepsis were evaluated, and the results showed a non-significant association with cognitive scores and neurodevelopmental impairment [192], Table 3.

Table 3 Characteristics of low-birth weight clinical studies included
Full size table

Mechanisms by Which Sepsis Could Induce Neurological Sequelae and Declines in Cognitive Function in Survivor Patients

There are several studies demonstrating different brain dysfunctions associated with cognitive impairment in sepsis survivor patients, such as delirium, lower cerebral blood flow index, neuroinflammation, BBB permeability, and white matter disruption, among others. Sepsis-associated delirium (SAD) is described in approximately 50% of septic patients, and it is a clinical feature of the participation of the central nervous system (CNS) during sepsis [196]. Additionally, neuroinflammation, abnormal cerebral perfusion, and neurotransmitter disproportions are the central mechanisms underlying the development of SAD that can trigger decline in cognitive functions [197]. This retrospective cohort study aimed to evaluate whether severe sepsis was associated with neuropathological findings of microvascular brain injury. There were 529 subjects who underwent brain autopsy, and among them, 296 experienced a total of 873 hospitalizations during study participation. A total of 89 individuals experienced severe sepsis hospitalizations. In analyses adjusting forage at death, sex, race, history of diabetes mellitus, coronary artery disease, cerebrovascular disease, or hypertension, prior severe sepsis hospitalization was associated with a relative risk of mild to moderate microvascular brain injury of 1.77. Those with severe sepsis were less likely to have evidence of acute or subacute macroinfarcts. Severe sepsis was associated with microvascular brain injury, a finding that may provide insight into the mechanisms of the association between severe sepsis and cognitive impairment [198]. In this prospective study, the cerebral circulatory parameters pulsatility index (PI) and cerebral blood flow index (CBFi) were evaluated based on the measured velocity of the middle cerebral artery, and Acute Physiology and Chronic Health Evaluation (APACHE) II score was assessed to evaluate the severity of illness. Forty septic patients were investigated, with transcranial Doppler on the first and third days of sepsis diagnosis. Twenty-one patients presented confusion, and the mainstream of the patients presented a PI higher than 1.1 (76%). PI on the first day could predict a positive Assessment Method for the Intensive Care Unit (CAM-ICU) test in septic patients. Multivariable analysis demonstrated that PI on the first day is associated to a positive CAM-ICU test independent of age and APACHE II score. On only the first day, the mean blood velocity in the middle cerebral artery and CBFi were identified to be lower in those patients with a high initial PI [199]. In another study from the same research group, patients presented a median pre-ICU IADL score of 6.3, 14 patients had cognitive decline at discharge, 2 patients were in persistent coma despite sepsis resolution, and information recall was the most affected mental function of the other 12 patients. Only on the first day, patients with cognitive decline had higher PI and lower CBFi compared to those without cognitive deficits. Multivariable analysis presented delirium but not PI as an independent prognostic factor for cognitive decline. In summary, delirium, but not cerebral perfusion changes, was an independent risk factor for cognitive injury in septic patients who were released from the ICU [200]. This prospective representative neuroimaging study was nested within an ongoing prospective cohort study to evaluate the association among delirium duration, white matter integrity, and cognitive impairment in ICU survivors. Delirium was evaluated with the CAM-ICU, and cognitive consequences were tested at 3- and 12-month follow-up. Greater duration of delirium of 3 versus 0 days was connected with lower fractional anisotropy in the genu and splenium of the corpus callosum and anterior limb of the internal capsule at hospital discharge. These associations persisted at 3 months for the genu and splenium. Lower fractional anisotropy in the anterior limb of the internal capsule at discharge and in the genu of the corpus callosum at three months were associated with worse cognitive scores at 3 and 12 months. In summary, delirium duration in the ICU was associated with white matter disruption, and white matter disruption was associated with worse cognitive scores up to 12 months later [201].

Elderly patients are frail and afflicted by worse outcomes, which are most likely associated with reduced functional status at baseline and serious deconditioning during acute illness. This prospective study aimed to describe the differences between nonagenarians and other age groups in patients admitted to internal medicine departments with sepsis and to assess predictors of survival in patients older than 90 years of age. One thousand eighty patients who were nonagenarians constituted 10.93% of this cohort. Of these, 70.48% had a cognitive injury and 82.60% had reduced functional state. Complications secondary to sepsis at admission and throughout hospitalization and mortality rates were higher in the nonagenarian population, at 61.86 vs. 51.14%, respectively, and survival rate was lower in the nonagenarian population, at 40.68 vs. 66.84%. In summary, nonagenarians presented worse outcomes associated with reduced functional status at baseline and strong deconditioning during acute disease [202].

Limitations

Our review may be limited by pre-clinical studies that did not present statistical data to identify the effect of the adjuvant treatment on cognition. Additionally, the majority of clinical studies are observational, and hence, causation cannot be established. The clinical articles presented moderate amounts of bias, which is expected given the designs of the included clinical studies, please see Tables 4 and 5.

Table 4 Bias summary: prospective observational studies
Full size table
Table 5 Bias summary: experimental trial
Full size table

Conclusions

Pre-clinical studies have shown an auto amplification of pro-inflammatory cytokines, such as TNF-α, IL-1β, and IL-6, in the first few hours after sepsis induction, increased BBB permeability, elevated levels of MMPs, increased levels of DAMPs, such as HMGB-1, S-100 protein, and AGEs. Additionally, NLRP-3, RAGE, and NF-kB signaling, astrocytes and microglia cells were also activated during sepsis. The rodents presented long-term cognitive impairment that was prevented by blocking the aforementioned pro-inflammatory mediators and immune pathways in the first hours after sepsis induction. Clinical studies showed that sepsis survivors presented increased bodily symptoms, such as fatigue, pain, visual disturbances, gastrointestinal problems, and neuropsychiatric problems (mood changes, relative weakness in visual memory with mild deficits in attention and concentration, development of moderate and severe cognitive impairment) compared to before sepsis. Sepsis leaves survivors with an aftermath of physiological, neuropsychiatric, and functional impairment.