The effect of dexmedetomidine and clonidine on the inflammatory response in critical illness: a systematic review of animal and human studies

Background The α2 agonists, dexmedetomidine and clonidine, are used as sedative drugs during critical illness. These drugs may have anti-inflammatory effects, which might be relevant to critical illness, but a systematic review of published literature has not been published. We reviewed animal and human studies relevant to critical illness to summarise the evidence for an anti-inflammatory effect from α2 agonists. Methods We searched PubMed, the Cochrane library, and Medline. Animal and human studies published in English were included. Broad search terms were used: dexmedetomidine or clonidine, sepsis, and inflammation. Reference lists were screened for additional publications. Titles and abstracts were screened independently by two reviewers and full-text articles obtained for potentially eligible studies. Data extraction used a bespoke template given study diversity, and quality assessment was qualitative. Results Study diversity meant meta-analysis was not feasible so descriptive synthesis was undertaken. We identified 30 animal studies (caecal ligation/puncture (9), lipopolysaccharide (14), acute lung injury (5), and ischaemia-reperfusion syndrome (5)), and 9 human studies. Most animal (26 dexmedetomidine, 4 clonidine) and all human studies used dexmedetomidine. In animal studies, α2 agonists reduced serum and/or tissue TNFα (20 studies), IL-6 (17 studies), IL-1β (7 studies), NFκB (6 studies), TLR4 (6 studies), and a range of other mediators. Timing and doses varied widely, but in many cases were not directly relevant to human sedation use. In human studies, dexmedetomidine reduced CRP (4 studies), TNFα (5 studies), IL-6 (6 studies), IL-1β (3 studies), and altered several other mediators. Most studies were small and low quality. No studies related effects to clinical outcomes. Conclusion Evidence supports potential anti-inflammatory effects from α2 agonists, but the relevance to clinically important outcomes is uncertain. Further work should explore whether dose relationships with inflammation and clinical outcomes are present which might be separate from sedation-mediated effects.


Background
The α2 agonists dexmedetomidine and clonidine are widely used as sedative drugs during critical illness, acting by dosedependent decrease in activity of noradrenergic neurons in the brain stem via post-synaptic receptor-mediated inhibition [1]. This increases gamma-aminobutyric acid (GABA) neurone activity, which mediates central sedative effects. Dexmedetomidine, a highly selective α2 agonist (α2:α1 receptor selectivity ratio 1620:1), is licenced for intensive care unit (ICU) sedation and systematic reviews suggest it decreases duration of mechanical ventilation and ICU stay compared with propofol and/or benzodiazepines [2][3][4]. Effects on delirium are less certain [2,4], but recent trials suggest dexmedetomidine can decrease delirium in selected populations, even when used in low doses [5]. Clonidine has substantially lower α2:α1 selectivity (220:1), is unlicensed for ICU sedation, but is also widely used in some countries. Despite this popularity, evidence for clinical effectiveness is very limited and of poor quality [6].
α2 agonists decrease central sympathetic nerve activity, which might affect inflammation and immune function either directly via cell surface receptors or indirectly by altering sympathetic/parasympathetic balance [7][8][9]. Although α2 agonists cause bradycardia, they may paradoxically increase cardiovascular stability in shock [10][11][12]. A post hoc trial analysis [13] and prospective clinical trial [14] suggest trends towards improved survival in sepsis when dexmedetomidine is used for sedation but remains unproven and was not observed in the recent large SPICE III randomised trial [15].
If α2 agonists have effects via mechanisms other than sedation, these could be important for clinical practice. Any anti-inflammatory effects could plausibly explain putative effects on delirium, shock, and other organ dysfunction and might explain why low doses of dexmedetomidine decrease delirium in perioperative patients [16]. Current evidence for anti-inflammatory effects from α2 agonists in critical illness has not previously been summarised. We therefore undertook a systematic review of relevant animal and human studies of critical illness to summarise current evidence.

Inclusion criteria
Eligible studies were animal or human studies undertaken in the context of sepsis, inflammation, or both. We limited studies to those that included dexmedetomidine or clonidine. All included studies required a biological measure of the inflammatory response as an outcome.

Search
We searched PubMed, the Cochrane library, and Medline in accordance with the PRISMA guidelines; the PRISMA checklist is available as an Additional file 1. The final search was done on May 14, 2019. Search terms used were broad: dexmedetomidine, sepsis and inflammation; and clonidine, sepsis and inflammation. We had no date limits but included only studies published in English. Reference lists were screened for additional publications. After removing duplicates, titles and abstracts were screened independently by two reviewers (CAF and ASR) and full-text articles obtained wherever possible for all potentially eligible studies. Disagreements were resolved between the two reviewers, with reference to the senior author (TSW). Full-text articles were retrieved from the National Health Service, Athens, Knowledge Network, University of Edinburgh, and the British Library. Where a full-text article was unavailable and there was insufficient information in the abstract to determine eligibility, it was excluded.
For eligible studies, data were extracted into a table that summarised: type of study (animal, human), population, interventions used and comparators, outcomes, key summary findings, and any relevant additional comments or findings (see Additional files 2, 3, 4 and 5).
We grouped studies as follows:

Human studies
Any human study involving perioperative and/or critically ill patients fulfilling inclusion criteria.
As the type and range of studies varied widely and were mostly animal models, we did not undertake formal quality assessment using an existing tool. Instead, we documented potentially important limitations qualitatively for each study. Evidence synthesis was descriptive, because data were not suitable for meta-analysis.

Results
The search identified 188 records; manual reference searching identified an additional 21 records. After removing duplicates 140 study abstracts were screened, and 95 selected for full text analysis. After full text review, 53 articles did not meet inclusion criteria and 42 articles were included in the review. The PRISMA flowchart is shown in Fig. 1. We identified 33 animal studies (9 CLP, 14 LPS, 5 ALI, 5 IRI) and 9 human studies.

Animal studies
The key inflammatory molecules tested are described in Table 1 with a summary of all animal studies in Additional file 2 and a detailed description in Additional file 3.
Filos et al. used a haemorrhage model and studied the effects of clonidine pre-treatment (150 μg/kg subcutaneously) on circulating endotoxin and tissue superoxide radicals, finding both were reduced [39]. Uysal et al. used a femoral artery/vein clamp model to study effects of dexmedetomidine 10 and 30 μg/kg on an epigastric island skin flap 12 h and 7 days post-IRI finding both doses reduced tissue NO, MDA, and MPO at both time points, and reduced flap necrosis area [19]. Shen et al. used a superior mesenteric artery occlusion model to study effects of dexmedetomidine 2.5 and 5 μg/kg/h infusion pre-IRI on lung injury, finding dose-dependent reductions in lung TLR4 and MyD88 (but not phosphoinositide 3-kinase (Pi3k) or Protein kinase B (AKT)) and improved lung histology and wet/dry weight ratio, with effects unaltered by yohimbine [24]. Sugita et al. studied effects of dexmedetomidine 10 and 20 μg/kg/h from the time of reperfusion on circulating TNFα, IL-6, IL-6, and NOS, finding attenuation of IL-6 and iNOS, but not TNFα, Endothelial derived nitric oxide (eNOS), or Neuronal nitric oxide (nNOS) 6 h post-IRI [27]. Finally, Zhang et al. studied the effects of dexmedetomidine infusion on myocardial injury in a coronary occlusion model, finding a reduction in infarct size and all measures of myocardial inflammation studied; this effect was attenuated by yohimbine [43].

Discussion
We found both animal and human studies supported antiinflammatory effects from the α2 agonists used in treatment of critically ill humans, namely dexmedetomidine and clonidine. However, only four animal and no human evaluated the effects of clonidine. There was wide variability in animal study design, but overall results were consistent with a reduction in markers of inflammation and organ injury. Anti-inflammatory effects were also observed in most human studies [14,50,51,53,[55][56][57].
Animal studies were heterogeneous in design in relation to the stimulus (LPS, CLP, IRI, and VILI), route of administration and timing of α2 agonist, and dose regimen used. There was variation in the biological/pathological marker of inflammation used, but effects on pro-inflammatory cytokines were broadly consistent across studies. Studies using antagonists [4,18,20,25,29,30,33,35,38,44] mostly suggested attenuation or elimination of effects, supporting a receptor-mediated mechanism, but this was not universal. The relevance of the animal studies to human disease is uncertain, especially because doses were mostly significantly higher than licenced for human use. No study measured plasma drug concentrations. Although the relevance to established critical illness of pre-insult administration is uncertain, this might be relevant to perioperative administration during major surgery. The coadministration of drugs such as ketamine in some studies [21,26] might also interact with α2 agonist pathways, limiting relevance to human disease. Few studies measured concurrent cardiovascular responses [21,27,31,32,47], which are potentially important side effects in humans, and almost all studies were undertaken in small animals, which may not be relevant models to human inflammatory disease. The consistency of reported effects across studies supports potentially relevant modulation of inflammation and/or immune responses. However, we cannot exclude an influence from publication bias.
Of the nine human studies, only two were undertaken in critically ill patients with sepsis or other clear inflammatory conditions [14,57]; the remainder included elective surgical populations [20,50,[52][53][54][55][56]. Most studies performed serum cytokine assays until around 48 h post-operatively and observed reductions compared with control patients in the immediate post-operative period. The main study limitations were small sample size and the uncertain relevance of inflammatory mediator concentrations to clinical outcomes. All of the 9 human studies used dexmedetomidine. The largest study was undertaken in mechanically ventilated septic patients and had non-significantly different clinical primary outcomes (mortality and ventilation-free days). Dexmedetomidine doses were smaller than those widely used for clinical sedation (0.1-0.7 μg/kg/h), but despite this the observed reduction in CRP concentrations is consistent with an anti-inflammatory effect. This study did not measure pro-or anti-inflammatory pathways [14]. The size and quality of the human studies was generally low, populations were heterogeneous, and most populations did not have severe illness. Anti-inflammatory effects were observed for potentially relevant inflammatory mediators in all studies, but none linked modulation of inflammation to clinical outcomes. This is the first systematic review of the antiinflammatory effects of dexmedetomidine and clonidine that includes both human and animal studies. We included studies covering a wide range of inflammatory insults including animal sepsis, ARDS and ischemia models, human critical illness, and major surgical interventions. In animal studies, we also specifically described the dose of α2 agonist used and timing of administration to highlight relevance to human disease. The only other published systematic review described anti-inflammatory effects of perioperative dexmedetomidine as an adjunct to general anaesthesia and found significant reductions in perioperative IL-6, IL-8, TNFα concentrations, and post-operative IL-10. It also observed wide variation in the surgery studied and high statistical heterogeneity of effects between studies. However, the relevance to clinical outcomes was not included and only dexmedetomidine administered pre-operatively was reviewed [8]. Our review summarised the available animal data and also highlights the limited evidence for clonidine. Taken together, these data support a possible anti-inflammatory effect from α2 agonists but are inconclusive about the clinical importance of this finding.
There are several suggested mechanisms for an antiinflammatory/immune regulatory effect from α2 agonists. These include modulation of macrophage/ monocyte function through α2-receptor-mediated inhibition of apoptosis, central and peripheral α2-receptor agonist-mediated effects, direct or indirect central sympatholytic effects, stimulation of cholinergic antiinflammatory pathways, reduction in anxiety and stress, and indirect effects via anti-nociceptive and/or sedative effects. The studies included in our review do not enable the exact mechanisms to be conclusively elucidated, but High Mobility Group Box 1 (HMGB1) Decrease Serum Ueki et al. [56] those that included an antagonist [4,18,20,25,29,30,33,35,38,44] mostly showed attenuation of α2 receptor agonist-mediated anti-inflammatory/immune regulatory effects suggesting a receptor-mediated effect is an important mechanism. This, however, was not a universal finding.
Whether anti-inflammatory effects are on a causative pathway to improved clinical outcomes in human critical illness is unproven. In the recent SPICE trial, which compared early dexmedetomidine based sedation with usual care (mostly propofol and/or midazolam) no overall mortality benefit was found and there was also no mortality benefit in a large sepsis sub-group (in whom an anti-inflammatory benefit may be most biological plausible) [15]. However, there were beneficial effects on both ventilation duration and delirium, and in a subgroup analysis the authors observed differences in the effect on mortality between younger (greater mortality with dexmedetomidine) and older (lower mortality with dexmedetomidine) patients for which the significance is uncertain. The mechanism of differential age-related effects of these agents is unexplained, but non-sedation related actions potentially, including immune/inflammatory modulation, might contribute. Further human studies are needed to explore whether the anti-inflammatory properties of α2-agonists are clinically important and whether differences between dexmedetomidine and clonidine might influence their modulation of clinical outcomes.
Our review has several limitations. Firstly, we used broad search terms to try and capture a wide range of studies, but we may have missed potentially relevant papers. Second, we cannot exclude publication bias especially for animal studies. Third, quantitative metaanalysis was not undertaken, because we did not consider this plausible or relevant for the identified literature; our data synthesis was therefore descriptive.

Conclusion
In conclusion, although the quality of animal and human studies is low, we have shown that α2 agonist drugs may potentially modify inflammatory and immune pathways in acute inflammatory conditions. Further work is required, especially in critically unwell humans, to explore mechanisms of action and whether these translate into improved patient outcomes.