Pharmacokinetics in critical illness

This chapter answers parts from Section B of the 2017 CICM Primary Syllabus, which expects the exam candidate to understand "the fate of drugs in the body, including ...  how it is affected by extremes of age, obesity, pregnancy (including foetal) and disease (particularly critical illness)". So this is the "(particularly critical illness)" section. This topic appears to straddle across the CICM training program, as both Question 13 from the second Primary paper of 2013, Question 1 from the first Primary paper of 2022, and  Question 5 from the first Fellowship paper of 2011 asked about it. In the Required Reading section for the CICM Part II, there is a brief point-form entry about the influence of critical illness on pharmacokinetics, but overall this topic seems to be very Part I material, and the bulk of the discussion is carried out here. 

There is little of use in the official textbook, but fortunately other resources do exist, and are excellent. Apart from the free article by Boucher et al (2006), there is also a much better piece by Roberts and Hall (2013). If one is unable to get institutional access to it, one needs to develop friends with institutional access. It is by far the most detailed review of the topic, and this summary was significantly improved by its structure and content.

How to answer this sort of SAQ

"Candidates were asked to only outline the details of this topic", the college mutters in their comments for Question 13, and the unsuspecting exam candidate might take that as an indication that only broad brushstrokes were expected. However, in the same paragraph the examiners complain about those useless trainees who "often used vague statements", omitting "details of the direction or mechanism of change". Apparently examples were also expected. That's hardly an outline of the details. 

According to the college, "most candidates answered the question under the subheadings absorption, distribution, metabolism and elimination". In view of this, and so as not to be too different, a similar structure is adopted below. Specific attention was paid to addressing the shortcomings identified by the college. Major effects, mechanisms and direction of change with examples are offered wherever possible in the mechanism = effect (example) template.

• Absorption
  • Delayed gastric emptying = decreased drug absorption (paracetamol)
  • Gastric pH is higher = decreased drug absorption (clopidogrel)
  • Increased GIT permeability = increased drug absorption (electrolyte replacement)
  • Decreased mesenteric perfusion = decreased drug absorption (paracetamol)
  • Increased preabsorptive interactions = decreased drug absorption (NG feeds and phenytoin)
  • Decreased active efflux = increased drug absorption (tacrolimus)
  • Decreased skin/muscle perfusion tissue = decreased absorption from subcutaneous or intramuscular injections (heparin) 
• Distribution
  • Decreased protein = increased free drug levels of highly protein-bound drugs (phenytoin)
  • Altered protein binding due to changes in pH = altered free drug levels (calcium)
  • Increased volume of distribution = decreased elimination of drugs, decreased plasma concentration (aminoglycosides)
  • Microvascular haemodynamic dysfunction = decreased tissue penetration (piperacillin)
  • Impaired barrier functions = increased tissue penetration (benzylpenicillin into the CNS)
• Metabolism
  • Decreased hepatic blood flow = decreased clearance of high extraction ratio drugs (propofol)
  • Hypothermia = decreased hepatic clearance (midazolam)
  • Hyperthermia = increased clearance due to increased enzyme activity (suxamethonium)
  • Downregulation of hepatic enzymes = decreased clearance (theophylline)
  • Decreased hepatic syntheic function = decreased levels of soluble enzymes, decreased clearance of drugs in the plasma (suxamethonium)
• Elimination
  • Decreased renal blood flow and function = decreased clearance of unchanged drugs or metabolytes (everything! pick any example)
  • Increased glomerular filtration in hyperdynamic circulatory states = increased clearance of drugs (vancomycin)

Effect of critical illness on enteral drug absorption

Gastric emptying is delayed. Even though that means drugs spend longer in the stomach and therefore gastric drug absorption may be enhanced, few drugs are absorbed in that way and the net effect is usually a decreased overall enteric absorption rate. Given that gastric emptying is usually the major rate-limiting step in drug absorption (once you're in the small bowel you have a massive surface area to work with) this is actually quite an important aspect of the ways critical illness affects drug absorption. Heyland et al (1996) used paracetamol as a marker, and discovered that in critically ill patients the maximal achieved blood concentration was halved, and the time to peak concentration was delayed by a factor of three (from 30 min in health controls to 105 min in the critically ill).

Gastric pH is higher, impairing absorption. The critical care specialists dutifully funnel PPIs into their patients, and the resulting alkaline gastric pH tends to impair the absorption of weak bases. Lahner et al (2009) were able to demonstrate this effect among a non-ICU population. Drugs most affected were itraconazole, dipyridamole, clopidogrel and some anti-HIV NRT drugs.

At a risk of making vague directionless statements, it is important to point out that changes in gastrointestinal pH can also increase bioavailability.  Digoxin and nifedipine both undergo pre-absorptive hydrolysis at low pH; with an alkaline stomach both of these drugs increase in bioavailability (for digoxin, by 10% with PPI use).

Gastrointestinal permeability is increased. This has been demonstrated in studies like Johnston et al (1996), where septic patients were fed a mixture of sugars, each with different molecular masses. Some of these were then recovered in the urine. The ratio of smaller sugar molecules to larger ones had increased, suggesting that gastrointestinal barrier function was disrupted and uninvited small molecules were border-jumping into the bloodstream. The larger molecules which were supposed to be actively transported had decreased in their proportion, suggesting that functional absorptive capacity had decreased.

This has implications for the critically ill patient who is in need of electrolyte replacement. Sure, the gut is sluggish and silent, and the villi are denuded; but the absorption of electrolyte solution should still be rapid. The large concentration of tiny electrolyte molecules favours diffusion into the bloodstream. Overall, the poor functional absorptive capacity should be offset by the increased permeability.  DeCarolis et al (2016) were able to demonstrate this in the context of replacing potassium chloride. The administration of enteral KCl to critically ill patients was functionally identical to giving it IV, in terms of achieving the desired change in plasma concentration.

Mesenteric perfusion decreases.  The normal vertebrate response to shock is to maximise the perfusion of essential organs and sacrifice the performance of non-essential ones, among them the gut. In addition to basic sympathetic reflexes, the intensivist then adds splanchnic vasoconstrictors like noradrenaline and vasopressin, and decreases venous return from the gut by putting the patient on positive pressure ventilation. One might predict that these things all impair enteral drug absorption. Smith et al (2012) documents the patchy evidence in support of this hypothesis. Apparently paracetamol behaves as you'd expect (its absorption is decreased and delayed) but oseltamivir does not (bioavailability in wrecked H1N1 patients was no different to ambulatory controls). 

Erratic post-operative absorption.  Major abdominal surgery gives rise to unpredictable oral pharmacokinetics. For example, were able to demonstrate that oral dosing of fluconazole resulted in a 30%-100% range of bioavailability among patient with open and closed abdomens post laparotomy (the closed abdomen patients were slightly better absorbers).

Increased preabsorptive interactions. Unexpected gut content (enteral feeds, multivitamins,  oral contrast, three litres of blood) will have an effect on gastrointestinal absorption. Without a doubt, that effect will be to impair absorption. There will be drug-drug, drug-feeds and drug-mineral interactions. A classical example of significantly altered absorption is phenytoin, where nasogastric feeds need to be turned off for 1-2 hours before and after administration (Yeung et al, 2000). Another well-known example is the interactions of tetracyclines or fluoroquinolones with calcium.

Decreased active efflux. Apical efflux transporters like the P-glycoprotein family act to decrease bioavailability of susceptible drugs. With critical illness, the organism's attention is elsewhere and synthesis of apical efflux pumps is neglected, Drugs which would otherwise be getting pumped back into the gut lumen are absorbed unmolested, and their bioavailability actually increases in critical illness. One example of such a drug is tacrolimus, which is a substrate for the P-glycoprotein transporter. Lemahieu et al (2005) was able to demonstrate this in renal transplant patients.

Effect of critical illness on parenteral drug absorption

Decreased absorption from subcutaneous administration is to be expected, particularly in shock states and with vasopressor use. For instance, Jochberger et al (2005) were able to associate vasopressor use with poor absorption of subcutaneously administered anticoagulants.

Decreased absorption from intramuscular administration is probably also going to be a problem because muscle tissue perfusion will likely suffer in shock and with vasopressor use. Absorption from muscle is directly proportional to muscle blood flow. Unfortunately, there does not seem to be any data regarding this. De Paepe et al (2002) were left to "speculate about the potential changes". They sensibly predicted that sluggish and erratic blood flow is going to give rise to sluggish and erratic absorption.

Effects of critical illness on drug distribution

Decreased protein binding and increased free drug levels will result from the tendency of albumin to disappear during critical illness. This makes drug levels less accurate, because they typically measure the concentration of total drug. The classical example of this is phenytoin. Whereas one can typically expect 90% of phenytoin to be protein bound in any given measured level, in the context of critical illess that might be only 80% - effectively doubling the concentration of active drug. In fact that is what Boucher et al (1988) found when they measured the free fraction of phenytoin in a group of trauma patients: over seven days, the free fraction increased by 28 to 90%.

Altered protein binding due to changes in pH could be in either direction (increased or decreased) depending on the pKa of the drug. Protein conformational changes may also conceal some binding sites, or expose others. At least that's what is supposed to happen theoretically. Gonzalez et al (2011) discussed this in some detail but were unable to bring up any actual examples or human-level evidence. It's not really a drug, but ionised calcium could be a good example. 

Increased volume of distribution for hydrophilic drugs results from the increase in total body water (athat's usually our fault, from all the fluid resuscitation). "Third space" fluid volume increases, and some of the drug molecules will vanish into this shadowy void. This has implications for clearance; less drug ends up in the circulating compartment which is available to organs of clearance and extracorporeal clearance mechanisms. Though the Vd for lipophilic drugs will remain unchanged, but their clearance will still be slowed. Consider the fact that their few non-lipid-bound molecules are the only ones which would usually undergo clearance; and now these few molecules will be shared between the circulating volume and the expanded third space fluid. Roberts et al (2009) produces some excellent examples from the world of antibiotics. For instance, aminoglycosides will have increased Vd and decreased clearance with fluid shifts. Clindamycin, on the other hand, will have a largely unchanged Vd (it is very liophilic) but hepatic clearance will be slowed by the wide distribution of the free drug. Though all of this sounds terrible, in fact it could be of benefit because half-lives and thus time above MIC will be increased, enhancing the killing power of time-dependent antibiotics.

Decreased tissue penetration due to microvascular dysfunction  could be a serious problem in the context of antibiotics and septic shock, a situation where penetration into infected tissue is vital and microvascular haemodynamics are famously messed up. This is not the case for all antibiotics. Joukhadar et al (2001) found piperacillin was very poor at penetrating the muscle and fat of septic patients;  the same team in 2002 found that cefpirome were having much less trouble. 

Increased tissue penetration due to impaired barrier functions  is classically demonstrated with the example of benzylpenicillin, which only penetrates into the CNS when the blood-brain barrier is impaired. Weirdly, this is not because of some sort of "leakyness" but because the inflammation disables various efflux pumps. This is inferred from the fact that during CNS inflammation few other molecules escape into the CSF, whereas if the barrier was somehow genuinely broken then the CSF would approach the composition of plasma (Roberts et al, 2009).

Effect of critical illness on drug metabolism

Decreased hepatic blood flow decreases the clearance of drugs with a high hepatic extraction ratio. Whereas drugs which have poor intrinsic clearance in the liver are substrates of lazy enzymes and changes in hepatic blood flow will have much less effect on their clearance. Hepatic blood flow will diminish due to multiple reasons, only some of which will be related to the disease state. For example  Bonnet et al (1982) found a linear decrease in hepatic perfusion proportionally to increasing PEEP (it seems it was all related to decreased cardiac output). 

Increased hepatic blood flow may increase the metabolism of drugs with a high hepatic extraction ratio. Early stages of sepsis feature a hyperdynamic circulation and this can increase the clearance of drugs (McKindley et al, 1998). Alternatively, the intensivist might interfere with the natural process of dying and infuse the patient with inotropes, generating an artificially over-vigorous circulatory state. 

Decreased metabolism due to hypothermia is largely due to hepatic enzyme dysfunction at lower temperatures. The hypothermia chapter deals with this in greater detail. For some examples, propofol metabolism is reduced to half of its normal rate, and the metabolism of midazolam to 1% of the normothermic rate. This is not unique to drugs which underego hepatic metabolism - for instance, cisatracurium degrades spontaneously and this is a temperature-dependent process.

Increased metabolism due to hyperthermia is a logical extension of the above. Ballard (1974) published an early review of this which remains a definitive resource. In it, there are great pearls. For instance, the damage to organs which occurs with hyperthermia tends to result in the release of enzymes form broken cells into the circulation, which could increase their availability and increase drug metabolism. Existing enzymes also work harder with hyperthermia. Apparently, cat plasma can crack through suxamethonium at a much faster rate if it is heated to somewhere between 40° and 42.5° C. 

Decreased metabolism due to loss of soluble enzymes  may occur because the liver is dead, or just lazy and post-hypoxic, or perhaps fighting to conserve synthetic function for more vital proteins. Plasma esterases are diminished in their number as a result. Vaja et al (2009) discuss the implications of this for anaethetic practice, and offer suxamethonium as an example of a drug which can be expected to have a much lounger half-life in patients with end-stage liver disease. Again, the intensivist may make matters worse by sucking out all the plasma with plasmapheresis, and replacing it with something useless like albumin.

Decreased hepatic metabolism due to downregulation of metabolic enzymes  is seen in critical illness and particularly in septic shock. Shedlofsky et al (1994) injected several human volunteers with purified LPS endotoxin and found them significantly less able to metabolise barbiturates and theophylline. TNF-α seems to be the main culprit.

Increased hepatic metabolism due to upregulation of hepatic enzymes could also occur. The enzymes may be activated by some other ICU drugs, or it may be a normal response to inflammation (for example, Roh et al found that interleukin-1 stimulates hepatic metabolism).

Effect of critical illness on drug and metabolite clearance

Decreased renal clearance due to decreased renal blood flow and generally diminished renal function is an expected norm. Aminoglycosides are a reasonable example to use. Not only does glomerular filtration rate suffer, but active secretion and passive reabsorption all take a hit. The hypoxic ischaemic kidney will not be inclined to help you clear your drugs.

Increased renal clearance due to increased renal blood flow can also happen. Udy et al described it as "augmented renal clearance",  and documented glomerular filtration rates in excess of 160 ml/min/1.73 m². Vancomycin is a good example of a drug which may require dose adjustment in this situation; mainly because we do drug levels for vancomycin. Clearance of carbapenems and β-lactams is probably also increased, but we don't measure the plasma concentration and are therefore oblivious to this