FormalPara Take-home message

The use of NMBAs should be considered during the early phase of severe ARDS for patients who despite deep sedation cannot be ventilated according to lung protective ventilation criteria or patients in prone position only after adjusting mechanical ventilation settings and sedation. The use of NMBAs should be integrated into a global strategy involving low tidal volume, a judicious use of PEEP, prone positioning and the use of a ventilatory mode allowing spontaneous ventilation as soon as possible.

Introduction

Mechanical ventilation (MV) is the basis of the treatment of patients presenting with acute respiratory distress syndrome (ARDS). The respective roles of mechanical ventilation with preserved spontaneous breathing (SB) and completely controlled mechanical ventilation using neuromuscular blocking agents (NMBAs) need to be clarified at the very early phase of ARDS. However, these two seemingly opposing strategies should be complementary, and defining the appropriate timing using these two strategies successively is warranted.

The current SARS-CoV-2 pandemic is associated with NMBA shortage in different countries, suggesting their widespread use [1]. Even before the publication of specific studies regarding the use of NMBAs in ARDS patients, their use was not trivial. In a sub-study of the ALVEOLI trial comparing a high PEEP strategy to a low PEEP strategy, continuous NMBAs were used at baseline in 30% and 25.4% of the lower and higher PEEP groups, respectively, and in 45% and 33% of patients with lower and higher PEEP between day 0 and day 28 [2]. Factors that were found to be associated with NMBA use are mainly related to disease severity, as assessed by a high APACHE III score, a large alveolar–arterial oxygen gradient, and a high plateau pressure [2]. Moreover, the use of prone positioning [3], permissive hypercapnia to ensure protective ventilation, extra-corporeal membrane oxygenation (ECMO) [4] or high-frequency oscillatory ventilation may require the use of NMBAs [5]. The purpose of this narrative review is to present an updated discussion on the role of muscle paralysis during mechanical ventilation in ARDS patients, based both on pathophysiological concepts and data obtained from clinical studies.

Pharmacology of NMBAs

Physiologically, acetylcholine (ACh) is released from the presynaptic motor nerve terminus, diffuses across the synaptic cleft, and binds to ligand-gated nicotinic acetylcholine receptors (AChRs) on the postsynaptic motor endplate. The binding of ACh increases the membrane permeability, which decreases the transmembrane potential. When the threshold potential is reached, the action potential is propagated, resulting in skeletal muscle cell contraction. The action of ACh is rapidly terminated by the enzyme acetylcholinesterase. NMBAs cause skeletal muscle relaxation by blocking the acetylcholine receptor neuromuscular junction [6]. These agents are classified by their mechanism of action and chemical structure. Depolarising NMBAs bind and activate AChRs, whereas non-depolarising NMBAs bind and competitively antagonize AChRs. Succinylcholine is the only depolarising agent but is not used as continuous infusion. The group of non-depolarising NMBAs is further subdivided according to their structure into benzylisoquinolinium—e.g., atracurium, cisatracurium and mivacurium, and amino steroidal compounds—e.g., rocuronium, vecuronium and pancuronium. Steroid compounds appear to further favor the occurrence of myopathies because of their structural analogy [7]. Moreover, they are metabolized into active metabolites; the elimination of these metabolites can be disturbed in renal or liver failure, and there is a risk of accumulation, especially if administered for several days. On the contrary, benzylisoquinolines are metabolized to inactive compounds by plasma esterases, depending upon the plasma temperature and pH. There is no risk of a prolonged muscular block after ending the infusion of these agents even in critically ill patients with renal or liver failure [8]. The choice of the adequate non-depolarising NMBA depends on both the indication and patient’s comorbidities (e.g., renal and liver failure). Pancuronium was the first amino steroidal compound introduced clinically. It stimulates muscarinic receptors, especially cardiac receptors with an atropine-like effect (vagal blockade with tachycardia). Atracurium and cisatracurium are preferred agents for continuous infusions due to the fact that their metabolism is unrelated to renal or hepatic function. They are both intermediate-acting NMBAs. Atracurium is metabolized through nonspecific plasma esterase-mediated hydrolysis as well as by the Hofmann elimination reaction, which is independent of hepatic and renal function, making this agent an attractive option in the intensive care unit in patients with renal and/or hepatic dysfunction. Cisatracurium is an isomer of atracurium, with a fourfold increased potency and without the associated histamine release. It is metabolized through organ-independent mechanisms via the Hofmann elimination reaction, making this benzylisoquinolinium drug one of the most commonly utilized NMBAs in critically ill patients who require neuromuscular blockade [9, 10] Main differences between benzylisoquinoliniums and other NMBAs are reported in Table 1.

Table 1 Pharmacology of commonly used neuromuscular blocking agents
Full size table

Selecting a specific NMBA in the critically ill patient depends on the indication, patient’s comorbidities (liver or renal failure), interactions with other drugs, physiological changes and risk factors that may affect the pharmacokinetics of NMBAs, such as age-related changes, hypothermia, sepsis, and metabolic or electrolyte disturbances. Tachyphylaxis has also been documented with NMBAs use, and clinical guidelines recommend that patients who develop tachyphylaxis to one NMBA should try another drug (rather from another class) if neuromuscular blockade is still required [11].

Plausible beneficial effects of NMBAs in ARDS patients

Several pathophysiological hypotheses have been proposed to explain why NMBAs used during the acute phase of moderate-to-severe ARDS might improve the outcomes. Figure 1 summarizes the key mechanisms. The main effects probably involve the following:

Fig. 1
figure 1

Plausible beneficial effects of neuromuscular blocking agents (NMBAs) in ARDS patients

Full size image

Reduction of patient-to-ventilator asynchronies and better adaptation to protective ventilation

Better control of the tidal volume by limiting inspiratory efforts and inhibition of active expiration help to decrease baro, volu and atelectrauma [12]. Thus, NMBAs limit the occurrence of high and large swings of transpulmonary pressure related to strong inspiratory efforts, as well as expiratory collapse by inhibiting active expiration. Strong expiratory efforts can generate negative transpulmonary pressure (when pleural pressure is higher than PEEP) and lead to alveolar collapse [13]. By the way, NMBAs could limit derecruitment and allow the maintenance of PEEP [13, 14]. These events are associated with a decrease in lung blood flow and alveolar–capillary permeability. NMBAs also prevent breath stacking, a patient–ventilator interaction in which consecutive machine inspiratory cycles occur in close succession with incomplete exhalation between them, typically due to inspiratory muscle effort early during the machine expiratory phase. Breath stacking can result in regular exposure to potentially injurious and occult high Vt despite ventilator settings consistent with a lung-protective strategy. By eliminating breath-stacking dyssynchrony, NMBAs ensure provision of the intended low-Vt strategy [15]. Inspiratory efforts can be clinically undetectable or associated with undiagnosed reverse triggering (a breath delivered by the ventilator triggering a contraction of the diaphragm responsible for a spontaneous breath [16]), which can be frequent even under deep sedation in non-paralysed patients [17]. Finally, an elevated rate of asynchronies has been shown to be associated with higher ICU and hospital mortality [18].

Lastly, paralysing ventilatory muscles to allow controlled ventilation could facilitate the tolerance of permissive hypercapnia.

Decrease in oxygen consumption

NMBAs have been shown to decrease oxygen consumption, mainly by eliminating muscular activity and improving systemic oxygenation, particularly in muscles implicated in respiratory function [19]. Through this mechanism, NMBAs likely reduce respiratory demand and cardiac output, followed by an increase in the mixed venous partial pressure of oxygen and the partial pressure of arterial oxygen. Reducing the work of breathing during mechanical ventilation by neuromuscular paralysis may lower the whole-body oxygen consumption in a significant manner (a 25% reduction has been reported [20]) and redistribute the blood flow to the splanchnic and other non-vital vascular beds [20]. Sparing oxygen consumption, NMBAs could reduce respiratory demand, contributing to reduce VILI from high minute ventilation and excessive patient effort, whereas decreasing cardiac output may decrease VILI from pulmonary vascular strain [21].

Increased thoraco-pulmonary compliance and functional residual capacity

This might be associated with a decrease in the intra-pulmonary shunt due to PEEP maintenance and lower atelectasis in dependant regions of the lungs [22]. NMBAs improve the mechanical viscoelastic properties of the chest wall. An improved ventilation–perfusion ratio may also be related to a more uniform distribution of lung perfusion due to the application of lower pulmonary pressure, favouring the perfusion of ventilated areas and decreasing the intra-pulmonary shunt [23].

Better regional distribution of the tidal volume

NMBAs could avoid or limit the overdistension of high compliance territories and promoting the recruitment of areas with smaller compliance.

Anti-inflammatory effects

The lower production of proinflammatory cytokines in the lung and the blood reported in patients receiving cisatracurium [24] has suggested an “anti-inflammatory” role for NMBAs. Two mechanisms could be involved: first, a reduced inflammation through the reduction of ventilator-induced lung injury (VILI). The second hypothesis is a direct “anti-inflammatory” effect of cisatracurium (see “Biological effects” section).

Biological effects of NMBAs

NMBAs have multiple, potentially positive, biological effects in humans [25]. In patients with moderate-to-severe ARDS, NMBAs administration has been associated with lower local and systemic release of inflammation, epithelial dysfunction, and endothelial injury biomarkers, such as IL-8, surfactant protein-D, and von Willebrand factor [24, 26].

The biological plausibility of a direct anti-inflammatory effect of cisatracurium is based on the broad expression of its receptor, alpha-1-nicotine receptor (nAChRα1), on epithelial cells, endothelial cells, leukocytes and fibroblasts [27,28,29,30] (Table S1, supplementary material 1). In the lungs, nAChRα1 signals as an alternative receptor for urokinase on neutrophils, leading to the release of inflammatory cytokines such as IL-1α, TNF-α, and macrophage inflammatory protein-2 [27]. An in vitro and in vivo animal study has been conducted to test the hypothesis that NMBAs are protective against biotrauma by their anti-inflammatory effects mediated by blocking the activity of nAChRα1 on epithelial cells, endothelial cells, and leukocytes [31]. Cisatracurium had intrinsic anti-inflammatory properties and its lung protection was primarily independent of the effects of synchrony. Surrogates of lung injury, such as the wet to dry ratio and protein concentration in bronchoalveolar lavage fluid (BALF), were lower in rats treated with NMBA than in controls, in which perfect synchrony was achieved with anaesthesia alone. The anti-inflammatory effects of cisatracurium, as defined by the lower release of inflammatory cytokines by several cell types (epithelial, endothelial, and CD14 + cells) after challenge with LPS, BALF or plasma from ARDS patients, was mediated by nAChRα1 blockade. cisatracurium lacked its protective effects when nAChRα1 was stably knocked down in cell clones.

Nevertheless, the putative direct anti-inflammatory effect of cisatracurium needs to be counterbalanced by muscular atrophy and weakness associated with prolonged use of heavy sedation and NMBAs and addressed for its clinical implications.

Risks of spontaneous breathing in ARDS

The maintenance of physiological respiratory muscle activity in ventilated patients has recognized benefits compared to controlled ventilation including improved alveolar recruitment, increased cardiac output, increased blood flow to vital organs, prevention of peripheral muscles withering and reduced risk of diaphragm disuse atrophy [32]. However, accumulating evidence has alerted physicians, either directly or indirectly, to the risks of spontaneous breathing in various clinical settings—e.g., non-intubated patients with acute respiratory failure [33], patients with ARDS [26, 34,35,36,37,38], patients with severe sepsis [39], patients with ARDS under ECMO [40], and paediatric patients with ARDS [41, 42]. Spontaneous breathing efforts may worsen lung injury, especially when the spontaneous effort is vigorous and/or lung injury is severe, termed patient self-inflicted lung injury (P-SILI) [43, 44]. Several potential mechanisms to explain the harm of spontaneous breathing efforts are summarized as follows: (1) global and local overdistension, (2) increased lung perfusion, (3) patient–ventilator asynchrony, and (4) derecruitment with expiratory muscle activity (Fig. 2).

Fig. 2
figure 2

Schema to explain the harm of vigorous spontaneous effort in ARDS. When vigorous spontaneous effort is preserved during mechanical ventilation, transpulmonary pressure (Paw—Ppl = PL) reaches injuriously high, thereby increasing tidal volume and causing global overdistension. In addition, negative Ppl distends pulmonary capillary vessels and increases perfusion; the transmural pressure across pulmonary capillary vessels is increased (Pcap—Ppl =  + 30), promoting interstitial edema formation (right magnified panel). In the presence of injury, permeability is increased and, therefore, alveolar edema formation is accelerated (Right magnified panel). In ARDS, atelectasis is often present in dorsal (dependent) lung regions (dotted black area in lung). Since the presence of atelectatic ‘solid-like’ lung tissue may block the pressure transmission of negative ∆Ppl following diaphragmatic contraction, more negative ∆Ppl is localized in dorsal lung regions (negative ∆Ppl − 20cmH2O vs. − 10cmH2O in dorsal vs. ventral lung regions). In this way, greater local (dorsal) lung stress causes local overdistension by drawing gas from other lung regions—e.g., ventral lung (blue line; this is called pendelluft phenomenon [45]) or directly from a ventilator. Thus, the bulk of effort-dependent lung injury occurs in dorsal lung regions, where vigorous spontaneous effort causes greater inspiratory lung stress and stretch (three red lines in lower panel). In addition, vigorous spontaneous effort causes patient–ventilator asynchrony and derecruitment with active exhalation. Abbreviations: ARDS acute respiratory distress syndrome; Pcap capillary pressure; Ppl pleural pressure

Full size image

First, a vigorous spontaneous effort will increase global transpulmonary pressure by decreasing pleural pressure and thus increasing tidal volume (i.e., global overdistension). Notably, vigorous spontaneous efforts will also carry the potential risk to increase local lung stress and strain (i.e., local overdistension) of dependent lung regions by drawing gas from other lung regions (called the Pendelluft phenomenon [45]) or directly from the trachea, despite a limiting tidal volume. The cause may be that, in the ‘solid-like’ atelectatic lung, negative inspiratory pleural pressure following diaphragmatic contraction is not transmitted but rather localized in the dependent lung regions where negative inspiratory pleural pressure is first generated [45]. Recent experimental data have confirmed that the bulk of effort-dependent lung injury occurs in the dependent lung regions, the same region where vigorous effort causes greater inspiratory stress and stretch [46]. Second, a spontaneous effort will increase lung perfusion and potentially cause lung oedema in ARDS. Vigorous spontaneous efforts will increase transmural pressure across pulmonary vessels, i.e., a net distending pressure of intrathoracic vessels, by decreasing pleural pressure. This mechanism may explain why spontaneous effort causes lung oedema during volume-controlled mode [35], or during upper airway obstruction [47]. Third, high respiratory effort is known to be associated with breath stacking [48]. Breath stacking—i.e., the occurrence of two consecutive inspirations separated by a short expiratory time–is potentially injurious because the delivered tidal volume increases [16]. Fourth, during vigorous spontaneous breathing efforts, there is a shift of the diaphragm to the cephalad direction (to expire promptly) and a decrease in expiratory transpulmonary pressure despite the use of PEEP, causing derecruitment of dependent lung regions [13, 34]. Finally, strong respiratory efforts during MV could be deleterious not only for the lungs but also for the diaphragm. The concept of diaphragm-protective ventilation has recently been proposed as a complementary strategy besides lung protective ventilation [43].

Neuromuscular blockers and safety concerns

ICU-acquired weakness

Muscles weakness, long-term sequelae of critical illness, affects roughly two-thirds of ICU survivors [49]. This neurological complication is often associated with prolonged mechanical ventilation [50], and prolonged ICU and hospital lengths of stay [51]. Critical illness-associated neuromyopathy is likely caused by multiple factors, including systemic inflammation, metabolic disorders, and interventions [49]. Notably, historically, ICU-acquired neuromuscular weakness was first described in patients receiving a high dose of corticosteroids and neuromuscular blockade for severe asthma [52]. Neuromuscular blockade is used to paralyse the patient with the expectation that, when the drug is withdrawn, the patient resumes rapid normal neuromuscular function. Curare and non-depolarising NMBAs may cause prolonged muscular weakness [53]. The risk of persistent paralysis is higher in patients with hepatic or renal dysfunction because most non-depolarising NMBAs are cleared from the plasma by the kidneys and liver. Increased plasma clearance of NMBA also parallels the duration of drug infusion and concomitant use of aminoglycosides or corticosteroids. The chemical structure of many NMBAs incorporates a steroid moiety, which may increase the risk of ICU-associated weakness. A notable exception is cisatracurium, which is cleared by Hofmann elimination and has a different chemical structure. Indeed, no increase in ICU-associated weakness from cisatracurium was demonstrated in two large RCTs [37, 54].

Prolonged neuromuscular blockade may deregulate acetylcholine metabolism and/or the function of Ach receptors with the upregulation of receptor subtypes that are less sensitive to acetylcholine. In rats, a relatively short experimental period was not associated with significant differences in the expression of known mediators of muscle atrophy, as demonstrated by similar diaphragm mRNA changes in expression of the muscle-specific ubiquitin ligases MuRF1 and Atrogin-1 [31]. However, the potential upregulation of nAChRα1 in muscle and other cell types associated with the continuous infusion of cisatracurium warrants further investigation in the future.

Additionally, prolonged blockade of the neuromuscular junction may cause muscle atrophy, particularly in the presence of corticosteroids, ischaemia, acidosis or electrolyte disturbances [55]. Myopathy is a common complication of the exposure to corticosteroids, particularly fluorinated derivatives—e.g., dexamethasone. Notably, corticosteroid toxicity is potentiated by non-depolarising neuromuscular drugs such as pancuronium because they bind to a common corticosteroid receptor. ICU-acquired weakness is more likely to occur in patients receiving NMBA and corticosteroids concomitantly. It is also more likely to occur with a high dose of fluorinated corticosteroids and prolonged administration of a non-depolarising, long-acting NMBA. Nowadays, physicians are more likely to use short courses of short-acting NMBAs and a low dose of corticosteroids, which may be less likely to result in significant muscles weakness. In a recent meta-analysis of three trials of neuromuscular blockade for ARDS (n = 431 patients), 73/223 and 62/208 (RR, 1.08; CI 0.83–1.41) patients had muscle weakness in the neuromuscular blockade and placebo groups, respectively [56]. Likewise, a systematic review and meta-analysis of 16 observational studies suggested that the administration of NMBAs might not be associated with increased risk of critical illness-associated neuromyopathy [57]. In practice, the duration of immobilization, severity of organ dysfunction, and metabolic and electrolytes disorders, as well as the concomitant use of other drugs altering neuromuscular function, may be more important to trigger ICU-acquired myopathy (or weakness) than neuromuscular blockade per se. Finally, a trial of cisatracurium for ARDS found no evidence of an increased risk of acute quadriplegic myopathy, although a high proportion of patients also received corticosteroids [37].

Nevertheless, in routine, ICU physicians should not combine fluorinated corticosteroids and non-depolarising NMBAs. We suggest using at a dose hydrocortisone not exceeding 300 mg per day, and the duration of neuromuscular blockade should be as short as possible (ideally less than 48 h). All other risk factors of critical illness-associated neuromyopathy should be avoided—e.g., maintaining normal glucose, pH and electrolytes levels and avoiding aminoglycosides. Patients should benefit from prompt passive and active mobilization, as well as a rehabilitation programme.

Ventilator-associated pneumonia (VAP)

NMBAs have not been demonstrated to increase the risk of ventilator-associated pneumonia (VAP). An ancillary study of ACURASYS showed that NMBAs were not associated with a higher occurrence of bacterial VAP [58].

Pressure and corneal ulcers

In a prospective descriptive study focusing on more than 500 ICU long stays and evaluating pressure ulcers grade 2–4 occurrence, mechanical ventilation was associated with pressure sore occurrence but not NMBAs. Sedatives associated with turning, floating heels where negatively associated with pressure ulcers [59]. A recent international study did not identify the use of NMBAs as being associated with pressure ulcers [60]. Preventing the risk of corneal ulcers in deeply sedated patients receiving NMBAs need careful and daily ocular protection.

Monitoring of NMBAs administration

Despite the lack of robust evidence, monitoring neuromuscular blockade is recommended in ICU patients [11]. Monitoring the depth of neuromuscular blockade aims to ensure that objectives for muscle relaxation are reached in an anesthetized patient, that the lowest NMBAs dose is used (which could limit the development of ICU-acquired weakness), and, less frequently, that deleterious residual neuromuscular blockade after extubation is avoided. Different studies have shown that monitoring the level of neuromuscular blockade is associated with a reduction in the amount of NMBAs along with a decreased incidence of persistent neuromuscular weakness and that the management of blockade in ARDS patients by nurses is a secure procedure [61, 62]. The depth of neuromuscular blockade should be assessed by repeated clinical and qualitative evaluation in addition to monitoring for adequate sedation and analgesia. The clinical evaluation based on the observation of skeletal muscle movement, respiratory efforts or the detection of patient–ventilator asynchronies must be coupled with a qualitative method to assess neuromuscular blockade. Train of four (TOF) supramaximal electrical impulses at a 2-Hz frequency applied every 0.5 s to the ulnar nerve (less frequently the posterior tibial nerve) of a non-paralysed limb, or to the facial nerve, produces four visualized muscle twitches. TOF remains the easiest and most reliable method available for ICU patients. Increasing the dose of NMBA is associated with a decrease in the force of twitches. The evaluation of the decline in the twitch response can be performed by comparing the strength (TOF ratio) of the fourth twitch to that of the first twitch. However, the measurement of the TOF can be impaired by hypothermia, peripheral oedema, or incorrect positioning of electrodes. Notably, quantitative neuromuscular monitoring (using mechanomyography or acceleromyography, for example) is not routinely used in ICU patients.

When monitoring treatment with NMBAs, clinicians must be aware of the low correlation of blockade measured clinically and peripherally compared with that of the diaphragm [63]. Along these lines, differences exist in the time course of response to NMBAs between central muscles (diaphragm), which recover earlier, and peripheral muscles (abductor pollicis). Depending on the clinical situation, physicians can use the ulnar or facial site, to achieve a TOF goal of 0, 1 or 2 twitches (as recommended for all ICU patients by the Neuromuscular Blockade Task Force [62]), or a TOF ratio exceeding 0.9. Furthermore, monitoring TOF recovery in response to facial nerve stimulation (as advised by the French recommendations in ICU patients [64]) before extubation could expose patients to aspiration [65], whereas facial or ulnar sites appear adequate to assess the depth of the relaxation of the diaphragm, particularly in ARDS patients.

Sedation monitoring in patients receiving NMBAs

Neither the ARDS et Curarisation Systematique (ACURASYS) study nor the Reevaluation of Systemic Early Neuromuscular Blockade (ROSE) study used sedation surveillance monitors in the NMBA arms; all the patients received deep sedation, and neither trial allowed decreasing the sedation after the initiation of NMBAs [37, 54]. Bispectral index (BIS) monitor is a noninvasive processed electroencephalogram that can identify accidental awareness with recall (AWR) in patients undergoing general anaesthesia [66]. BIS values of 40–60 minimize the risk of AWR in operating rooms [66]. Despite minimal data to corroborate its use in the ICU [67], BIS monitoring is becoming more common in mechanically ventilated patients undergoing paralysis [68]. The current NMBA guidelines do not recommend the use of sedation-monitoring devices to measure sedation [11]. There is a concern for discordance in BIS readings in patients with critical illness-associated encephalopathy [69]. Electromagnetic fields from other devices and instruments in the ICU environment might also affect BIS readings [67, 69]. The use of BIS monitors in clinical practice has been associated with increased rates of the down titration of sedatives, and the risk of under-sedating these patients needs to be always evaluated when using any monitoring devices to prevent long-term neuro-cognitive disorders, such as anxiety or post-traumatic stress disorder [66, 70]. BIS values should, moreover, be interpreted with caution as they have been shown to decline in fully awake volunteers under neuromuscular block [71, 72]. If BIS monitors are not employed for sedation monitoring, it is important to target deep sedation before NMBA initiation. Once deep sedation is achieved, a flat dose with no de-escalation should be implemented for the sedation instructions [68]. Importantly, RCTs of NMBA infusions in ARDS are limited to 48 h, and adjunctive sedation monitors might be needed for prolonged NMBA use because of the risk of tachyphylaxis [37, 54].

Outcomes of ARDS patients receiving NMBAs

Seven randomized controlled trials (RCT) [13, 24, 37, 54, 73,74,75] have studied NMBA infusions in patients with moderate-severe ARDS (Table 2). Conflicting results from the two largest randomized controlled trials (RCTs) [37, 54] evaluating the role of NMBAs in ARDS have further tempered the enthusiasm for their use as a front-line adjunctive therapy [76]. Importantly, significant differences were found in the timing of enrolment, ventilatory and non-ventilatory strategies and the initial severity among the patients enrolled in ACURASYS compared with those in ROSE; thus, the differences in the study design and methodology might explain the differences in the reported mortality [77, 78]. In the first multicentre RCT [73], a significant improvement in the PaO2/FiO2 ratio and a strong tendency towards a lower mortality rate were observed in patients receiving NMBAs for 48 h. The same group of authors [24] confirmed the beneficial effects of NMBAs on oxygenation and decreases in the plateau pressure, FiO2 and PEEP. Again, a trend towards decreased mortality in patients receiving NMBAs was observed. This was the rationale for designing the ACURASYS study [37], which showed that a strategy including cisatracurium is associated with an improvement in the adjusted 90-day survival rate compared with placebo. The 28-day mortality was 23.7% with cisatracurium and 33.3% with the placebo (p = 0.05). The beneficial effects of cisatracurium on mortality were mainly observed in patients with a PaO2/FiO2 ratio < 120 mmHg. The cisatracurium group had significantly more ventilator-free days than the placebo group during the first 28 days. NMBAs patients also presented less barotrauma. The PETAL (prevention and early treatment of acute lung injury) network aimed to re-evaluate the beneficial effects of NMBAs on mortality by designing the ROSE study [54] comparing the use of cisatracurium with management by light sedation very early in the course of moderate-to-severe ARDS. The primary endpoint was hospital mortality from any cause at day 90. The trial was stopped at the second interim analysis for futility. Neither in-hospital mortality nor ventilator-free days at day 90 were different between the groups. NMBAs neither improved oxygenation nor decreased the rate of pneumothorax but were more frequently associated with serious cardiovascular adverse events. Significant methodological differences may explain these conflicting results between the ACURASYS and ROSE studies (Table 3) [37, 54]. First, the patients were included earlier in ROSE than in ACURASYS (the median times were 8 h and 16 h, respectively). This shorter delay in the ROSE study could have compromised the adequate adjustment of both mechanical ventilation and sedation before inclusion. This finding likely explains why a significant proportion of eligible patients (N = 658) was excluded because of oxygenation improvement from inclusion to randomization [37], although patients with a PaO2/FiO2 < 150 at inclusion could be included even if PaO2/FiO2 reached but not exceeded 200 mmHg at the time of randomization (i.e., some patients had a PaO2/FiO2 > 150). Moreover, more patients in ROSE (17.1%) than in ACURASYS (4.3%) were excluded before enrollement because already receiving NMBAs [79], suggesting that some of them might have benefited from neuromuscular blockade. Third, strategies in both studies concerning PEEP adjustment were really different: a high PEEP (ALVEOLI high PEEP strategy [80]) was applied in the ROSE study, while a moderate PEEP (ARMA strategy [81]) was used in the ACURASYS study [37, 54]. A recent study has suggested that high levels of PEEP and recruitment manoeuvres could worsen the outcomes [82]. These differences in PEEP strategy could explain at least in part the discrepancies between ACURASYS and ROSE studies. Sedation strategies were also different during the first 2 days in the control group—light sedation in ROSE, heavy sedation in ACURASYS [37, 54]. Importantly, in this latter study [37], no difference was found between the two groups (NMBAs and placebo) regarding the amount of sedatives received (no “oversedation” in the placebo group). These differences in PEEP and sedation strategies may alter the level of VILI experience in both the control and intervention arms in both studies, highlighting the complex interplay among patient effort, sedation, NMBAs, and ventilator management in ARDS [76]. Another main difference was the lower use of prone positioning in ROSE (16% vs. 29% in ACURASYS) and the quick transition towards ventilatory modes, allowing spontaneous breathing and a weaning protocol in ACURASYS; no such protocol was reported in ROSE [37, 54]. Overall, these distinct approaches might explain the large difference in mortality observed between ACURASYS and ROSE regarding NMBA groups (90-day mortalities of 31.6% and 42.5%, respectively); however, no difference was found between the two studies regarding the mortality of the control group [37, 54]. The conclusion that might be drawn from these two studies is that very early use of NMBAs, before optimizing MV and sedation, using a strategy involving high PEEP levels does not modify the outcomes [37, 54]. Conversely, NMBAs (if sedation alone fails to improve respiratory status) integrated into an overall strategy including a reasoned use of PEEP, prone positioning and the rapid implementation of spontaneous breathing might improve the prognosis [37, 54]. Including the results of ROSE, three meta-analyses showed a reduction of early (21- to 28-day) mortality [83, 84] and late (90-day or ICU) mortality [83, 85] in patients receiving NMBAs. Early improvement of oxygenation was also retrieved in three of these meta-analyses [83, 84, 86]. A lower risk of barotrauma and no effect on the occurrence of ICU-acquired weakness were constantly reported.

Table 2 Randomized controlled trials of continuous infusion of neuromuscular blocking agents in patients with ARDS
Full size table
Table 3 A comparison of ACURASYS and ROSE studies on specific methodological aspects
Full size table

Partial neuromuscular blockade in patients with ARDS

In patients with ARDS, respiratory drive may be excessive, mainly due to hypercapnic acidosis, hypoxemia, and inflammation, [87, 88]and lead to P-SILI [89]. Additionally, some preliminary data have suggested that prolonged strenuous diaphragm effort may result in load-induced diaphragm injury [90, 91]. The disadvantages of full neuromuscular blockade also include the risk for the development of diaphragm disuse atrophy [92], peripheral skeletal muscle atrophy due to immobility and the need for a high dose of sedatives. Accordingly, instead of complete diaphragm paralysis, an alternative approach would be to titrate the diaphragm effort to maintain a physiological effort. From this perspective, low-dose NMBAs (“partial neuromuscular blockade”) is an interesting compromise between total diaphragm paralysis and injurious high breathing effort. Interestingly, it has been demonstrated more than 40 years ago that low-dose neuromuscular blockers (partial neuromuscular blockade) can be used to decrease respiratory muscle strength [93] while maintaining spontaneous breathing. The authors demonstrated that low-dose NMBA-induced respiratory muscle weakness but increased respiratory effort sensation. Therefore, partial neuromuscular blockade does not reduce respiratory drive, only respiratory muscle effort. In a proof-of-concept study recruiting patients with moderate ARDS and high respiratory drive, the feasibility of partial neuromuscular blockade was evaluated in patients on pressure support ventilation [94]. Careful titration with rocuronium decreased the tidal volume from 9.3 ± 0.6 ml/kg to 5.6 ± 0.2 ml/kg, fulfilling the criteria for lung-protective ventilation. For reasons incompletely understood, the reduction in minute ventilation due to a reduced tidal volume was not completely compensated by an increase in respiratory rate. Accordingly, patients developed mild respiratory acidosis with partial neuromuscular blockade. After rocuronium titration, the transdiaphragmatic pressure was decreased to ± 5.0 cmH2O, well within the physiological range for diaphragm activity during tidal breathing in healthy subjects. Although this strategy was applied for only 2 h in highly selected patients, it suggests that partial neuromuscular blockade is a feasible and safe strategy to deliver lung- and diaphragm-protective ventilation in patients with a high respiratory effort. However, future clinical studies are warranted to confirm the safety and efficacy of prolonged partial neuromuscular relaxation in ventilated patients.

Future prospects and conclusions

Pharmacologic strategies such as NMBAs should not be used routinely in all patients with moderate-severe ARDS but need to be customized for the appropriate patient at the right time to evaluate their benefit [77]. The initial management of ARDS must follow a lung-protective strategy with the optimization of PEEP and sedation for individual patients [95]. After this initial optimization of mechanical ventilation and sedation, clinicians must integrate the use of NMBAs in a step-up fashion based on objective physiologic criteria (PaO2/FiO2) and the presence of either asynchrony or unsafe ventilation [95]. Gas exchange improvement should prompt physicians to stop NMBAs and encourage spontaneous breathing activity (Fig. 3).

Fig. 3
figure 3

Place of neuromuscular blocking agents in the ventilatory strategy of ARDS patients

Full size image

More research is needed to adjust the use of NMBAs in ARDS patients. First, the benefits observed may not apply to all NMBAs, considering that cisatracurium besylate has been used in all RCTs. Second, the optimal duration of infusion needs to be evaluated according to the patients’ profiles and/or responses to the treatments. Twenty-four hours of paralysis may be sufficient in some cases with prompt improvement. By contrast, in some patients with very severe ARDS (including those requiring ECMO with persistent high respiratory drive and/or prone positioning), longer durations are often required. Because NMBAs were used in more than 85% of patients from the PROSEVA study [96], whether the use of NMBAs is required in all moderate-to-severe ARDS patients requiring prone positioning must be specified. Third, the place of NMBAs to improve respiratory system mechanics in patients without moderate-to-severe hypoxemia but with large swings in transpulmonary pressure deserves to be further explored. Finally, clinicians can be relieved regarding the potential harmful effects of NMBAs. Indeed, the use of a short course of a recent NMBA was not associated with an increased incidence of ICU-acquired neuromyopathy [37, 54] (Table 3).

Considering the current definition of ARDS, the use of NMBAs should be considered during the early phase of severe ARDS for patients who require deep sedation to facilitate lung-protective ventilation or prone positioning [97]. However, as stated in a recent guideline [97], NMBA infusion must be discussed only after optimizing mechanical ventilation and sedation (the ACURASYS strategy). The use of NMBAs should be integrated into an overall strategy including the reduction of the tidal volume, a reasoned use of PEEP according to its impact on gas exchange and the haemodynamic status, the use of prone positioning and preferential choice of a ventilatory mode allowing spontaneous ventilation as soon as possible [95] (Table 4). Partial neuromuscular blockade needs further clinical evaluation but is a promising strategy.

Table 4 Summary of the main clinical recommendations for NMBAs use in clinical practice
Full size table