Your patient is a middle-aged, moderately obese man scheduled for a renal transplant. As you are preparing your drugs for induction of anesthesia, you wonder which neuromuscular blocking agent would be most suited for this case. Cisatracurium has a predictable duration of action in patients with end-organ failure, so you think at first that it might be a good choice here because renal function is absent. Rocuronium is partly excreted by the kidney, so it is expected to provide prolonged neuromuscular blockade in your patient. However, you remember that cisatracurium has a much slower onset of action than rocuronium, and you worry that oxygen desaturation might occur before intubation if you use cisatracurium. This obese patient has reduced oxygen reserves in his lungs even after optimal preoxygenation, and you might choose rocuronium after all, because shortening onset time is probably a more important consideration than a predictable duration of action.
The differences in onset times between the 2 drugs are huge. For example, in a study involving 848 patients, mean onset time for rocuronium 0.6 mg/kg was 1.7 minutes and for cisatracurium 0.1 mg/kg, it was as long as 4.1 minutes.1 Why such a large difference in onset times between 2 agents with comparable duration of action? Onset time is a reflection of how long a drug takes from the time you push it into the bloodstream until it gains access to the neuromuscular junction. The various steps involved, which include mixing with venous blood, transiting through the right heart into the pulmonary circulation, gaining access to the left heart, the arteries, and then the capillaries, should be the same for all agents administered IV. Early pharmacokinetics is a question of mixing, not elimination.2 Then, why are there differences in onset time between 2 drugs given in equipotent doses? Since the journey from the syringe until the neuromuscular junction is the same for all drugs, the difference should be in how access is gained to the receptors once the neighborhood of the junction is reached. The article by Dilger3 in this issue of Anesthesia & Analgesia provides some insights into the mechanisms involved.
Qualitatively, the differences in onset between drugs have been explained as follows. The neuromuscular junction is packed with receptors and lies within a very restricted space. A typical junction is oval shaped, approximately 40 µm long and 30 µm across, for an area of 1000 µm2 or so, and the synaptic cleft is only 0.05 µm wide. The total number of receptors in this small volume is 10 million, or 10,000 per µm2.4 Complete neuromuscular block does not occur until at least 90% of receptors, or 9 million, are occupied, and this refers to the concept of margin of safety.4 So, irrespective of the neuromuscular blocking agent given, complete paralysis does not occur until at least 9 million molecules have gained access to the neuromuscular junction. Rocuronium 60 mg is approximately equipotent with cisatracurium 10 mg, a factor of 6.1,5 When molecular weights are considered (that of cisatracurium is approximately 1.5 times that of rocuronium bromide), 9 times as many molecules are injected with rocuronium 60 mg than cisatracurium 10 mg. These drugs are diluted in the same blood volume, so the molar concentration of cisatracurium is only one-ninth that of rocuronium. It follows that delivery of the critical 9 million cisatracurium molecules to the neuromuscular junction takes longer than the same number of rocuronium molecules. Unbound molecules must also be added in the synaptic cleft, in proportion to the molar potency ratio, that is, 9 times fewer molecules for cisatracurium than rocuronium. However, because the synaptic cleft volume is so tiny, the number of free molecules is small: only 50,000 rocuronium and 6000 cisatracurium molecules are free. This is negligible compared with the 9 million bound.3 During onset, most of the drug reaching the junction is busy occupying receptors; the process is completed faster if more molecules, that is, a higher concentration or less-potent drug, are present. The relationship between onset and potency has been verified in clinical studies, with high-potency drugs (i.e., low effective dose 95% or ED95) having long onset times.5 Cisatracurium produces neuromuscular blockade more slowly than rocuronium.1,5 Doxacurium, the most potent neuromuscular blocking agent developed so far, has an onset time of 10 minutes or more,6 and rapacuronium, which was withdrawn from the market at the turn of the century, was a low-potency agent with a fast onset.7
This correlation between potency and onset time does not mean, however, that one explains the other. Although the logic seems impeccable, the explanation does not provide a time frame for the onset–potency effect. In other words, we do not know whether these 9 million molecules are actually delivered over several minutes or just a fraction of a second. In a classic paper, Armstrong and Lester8 studied the rate at which d-tubocurarine interfered with acetylcholine, when both agents were given with micropipettes in the vicinity of the neuromuscular junction. They found that the effect of d-tubocurarine was not immediate, and its onset of action could be made faster by removing the nerve terminal from the end plate, suggesting that it is difficult for drugs to enter the neuromuscular junction. This phenomenon was called “buffered diffusion,” and the time lag associated with d-tubocurarine application was attributed to its limited diffusion into the cleft. The delay for d-tubocurarine was only a few seconds, but then d-tubocurarine is not very potent.
The next step was to test whether onset of potent drugs was slower than that of less-potent agents in vitro. Law-Min et al.9 tested this hypothesis, applying neuromuscular blocking agents of different potencies with a pipette near a mammalian neuromuscular junction. Onset was beautifully correlated with potency, but the action took place over a short time period (1–8 seconds, depending on the drug). This could be because the neuromuscular blocking agent was delivered via a pipette, at high concentrations, instead of via the bloodstream, at lower concentrations. However, it is difficult to design an experiment that mimics the type of blood delivery seen in practice in which only the delay due to buffered diffusion is measured.
If experiments cannot be performed, simulation can help, and this is what Dilger3 accomplished in his article. The Salk Institute for Biological Studies (http://www.mcell.cnl.salk.edu) has produced a computer simulation of cell signaling, and this sophisticated tool can be applied to acetylcholine release, binding, and hydrolysis at the neuromuscular junction. Acetylcholine release within the particular geometry of a mammalian neuromuscular junction as shown in Figure 1A of Dilger’s article was modeled, by inserting the proper variables, such as density of receptors, number of vesicles released and vesicle content, binding constants for acetylcholine and neuromuscular blocking agents, thickness of the synaptic cleft, number and depth of folds, etc. Then, the association and dissociation constants for acetylcholine and neuromuscular blocking agents of various potencies were fed into the model (Table 1 of Dilger’s article), and simulations were performed. Dilger first made simulations with an instantaneous change in neuromuscular blocking agent concentration in the vicinity of the neuromuscular junction, and the principle of buffered diffusion was verified (Figure 4D of Dilger’s article), that is, onset time of twitch depression is proportional to drug potency. However, for a drug like cisatracurium (50% effective dose [EC50] close to 178 nM in the graph), time until maximum block was not 5 minutes as reported in studies but rather in the 20- to 30-second range. This demonstration seemed no better than those of Armstrong and Lester8 and Law-Min et al.,9 who also got the principle right but the time scale wrong.
Obviously, the devil is in the details. Plasma concentrations do not change instantaneously in the vicinity of the neuromuscular junction, but as shown in Figure 5 of Dilger’s article, smoothing out the changes in plasma concentration to mimic the clinical situation had the effect of almost obliterating all buffered diffusion effect. However, other factors could increase the influence of buffered diffusion on onset time, such as the rate at which neuromuscular blocking agents are transferred from plasma to the interstitial space around the junction (a term called keo) and the size and thickness of the synaptic cleft. Dilger’s simulations were based on slabs of junction 3 µm in length and 2 µm in width, with the idea that an adult neuromuscular junction is not a uniform oval 40 × 30 µm in size, but rather a 40 × 30 µm inhomogeneous pretzel, with lots of discontinuities and branches.4,10 Since adductor pollicis pretzel geometry is not known with certainty, simulations with branch widths of 4 and 6 µm were also made. Compared with 2 µm, simulations with 6-µm widths led to the prediction of longer onset times, corresponding to those measured in the clinic. For example, onset times to 50% block in the 3-minute range, and of course longer for maximum blockade, can be predicted for a drug like cisatracurium (Fig. 7 of Dilger’s article).
Supporters of the buffered diffusion theory will cheer! Opponents will dismiss the whole exercise as useless, saying that fudge factors were applied until the “correct” answer was found. The discrepancy between initial results with an instantaneous rise in concentration and those with a more “realistic” concentration profile and a widened junction indicates that simulations based on many input parameters, all of which are estimated with a rather wide degree of uncertainty, can miss the target by a long shot. Does this mean simulations are useless? Certainly not. They are an opportunity to isolate the key factors that contribute to an observation. In this case, Dilger’s simulations indicate that neuromuscular junction geometry, the rate at which drugs reach the junction, and estimates of drug potency are important elements. Moreover, they prompt us to reconsider hypotheses and incorporate other factors that may have been neglected. In this particular case, perhaps the next step would not be so much repeating the study with a better model of neuromuscular junction geometry but rather refining the simulations by incorporating not 1 but several neuromuscular junctions, each with slightly different characteristics and access to neuromuscular blocking agents. When we inject our drugs, we do not target the neuromuscular junction, but rather many of them, each with a different geometry. Twitch depression is not the result of partial transmission failure at every neuromuscular junction, but rather the summation of failures and successes at many junctions, each of which responds in an all-or-none fashion. It is likely that the differences in onset times predicted for the drugs in Dilger’s study3 would be accentuated by a more refined model, incorporating many neuromuscular junctions.
Name: François Donati, PhD, MD.
Contribution: This author wrote the manuscript.
Attestation: This author approved the final manuscript.
This manuscript was handled by: Steven L. Shafer, MD.
1. Adamus M, Gabrhelik T, Marek O. Influence of gender on the course of neuromuscular block following a single bolus dose of cisatracurium or rocuronium. Eur J Anaesthesiol. 2008;25:589–95
2. Krejcie TC, Avram MJ. Recirculatory pharmacokinetic modeling: what goes around, comes around. Anesth Analg. 2012;115:223–6
3. Dilger J. Simulation of the kinetics of neuromuscular block: implications for speed of onset. Anesth Analg. 2013;117: 792–802
4. Wood SJ, Slater CR. Safety factor at the neuromuscular junction. Prog Neurobiol. 2001;64:393–429
5. Kopman AF, Klewicka MM, Kopman DJ, Neuman GG. Molar potency is predictive of the speed of onset of neuromuscular block for agents of intermediate, short, and ultrashort duration. Anesthesiology. 1999;90:425–31
6. Koscielniak-Nielsen ZJ, Law-Min JC, Donati F, Bevan DR, Clement P, Wise R. Dose-response relations of doxacurium and its reversal with neostigmine in young adults and healthy elderly patients. Anesth Analg. 1992;74:845–50
7. Kopman AF, Klewicka MM, Ghori K, Flores F, Neuman GG. Dose-response and onset/offset characteristics of rapacuronium. Anesthesiology. 2000;93:1017–21
8. Armstrong DL, Lester HA. The kinetics of tubocurarine action and restricted diffusion within the synaptic cleft. J Physiol. 1979;294:365–86
9. Law Min JC, Bekavac I, Glavinovic MI, Donati F, Bevan DR. Iontophoretic study of speed of action of various muscle relaxants. Anesthesiology. 1992;77:351–6
10. Marques MJ, Conchello JA, Lichtman JW. From plaque to pretzel: fold formation and acetylcholine receptor loss at the developing neuromuscular junction. J Neurosci. 2000;20:3663–75