Dialyze This – Extracorporeal Removal Techniques for the Poisoned Patient
Case
A 26-year-old woman presents with palpitations after intentionally ingesting 180 tablets (36 g) of caffeine in a suicide attempt. Her triage vitals are notable for a HR of 160 bpm, and BP of 92/60 mmHg. During your initial evaluation, she has a seizure and develops a wide-complex tachycardia, which degenerates into pulseless arrest. In the midst of resuscitation, you page the nephrologist.
Hemodialysis has long been in the toxicology toolbox for overdoses in which removal of the toxin is effective and life-saving. A therapy in a shared domain, nephrology and toxicology combine expertise to identify and treat patients for whom this therapy may save a life.1
But when is it appropriate? What method should be used? How long should it continue? What are all of those doodads on the dialysis machine?
Disclaimer: I am not a nephrologist.
First, let’s define 2 key concepts: Dialyzability and Clearance.
Dialyzability
This refers to the capacity of a drug or toxin to be removed by a dialysis membrane. Although there are features of the extracorporeal circuit which may alter a substance’s dialyzability (e.g. membrane materials, surface area, etc.), it is primarily determined by several characteristics of the substance itself:
- Molecular weight. This depends on the membrane pore size, but generally highly dialyzable toxins should weigh less than 15,000 Daltons.2 What does this mean? See table 1 for a list of common substances and their molecular weight. Most drugs are under this cutoff – but proteins, heparin, monoclonal antibodies, and a few others are not.
- Protein binding. Because proteins are so large, any drug that is primarily bound to plasma proteins will simply not be removed. Examples include phenytoin, warfarin, Amanita toxin, sulfonylureas, and many antibiotics. Highly dialyzable toxins should be less than 80% protein bound. A caveat about protein binding. A person only has so much available plasma protein (e.g. albumin) to bind drug. So a drug that may be highly protein bound in therapeutic dosing may be much less so in overdose, as plasma proteins become saturated. This is true most notably of valproic acid and acetylsalicylic acid.3,4
- Volume of distribution. This is the theoretical volume into which a drug is distributed. Put another way, it’s the degree to which a drug leaves the circulation and goes into tissues. If it stays put in the blood, Vd is very low <0.06 L/kg. If it lives mostly in extracellular fluid, it’s still <1 L/kg, but more like 0.6 L/kg (think alcohols). If, however, it is distributed primarily in the fat compartment, the Vd will be very large. Amitriptyline, for example, clocks in at around 30 L/kg.5 Good thing it’s a theoretical volume or you’d be carrying around buckets of tissue for your amitriptyline.

Clearance
Where dialyzability refers to a drug or toxin, clearance refers to a technique or process by which it is removed. Expressed in ml/min, this is the rate at which a toxin (or any solute) is removed from a given fluid volume (usually plasma), and is calculated by:
Rate of elimination from plasma (mg/min) / Plasma concentration (mg/mL)
The body naturally clears substances all the time, which is called endogenous clearance. An extracorporeal method’s clearance of a given toxin is affected by parameters such as the rate of blood flow through the circuit, membrane pore size, and dialyzability. It’s important to remember that in order to be effective, a technique’s clearance should add significantly to total clearance (the sum of endogenous and exogenous clearance). Although specific criteria are lacking, and there are many factors to consider regarding the drug and the circumstance, any extracorporeal modality should make a measurable difference (increase by 30% or more) in total clearance to be effective. If endogenous clearance is already brisk, it makes little sense mucking with a system working quite well on its own.
Case Continued
So what are the characteristics of caffeine in regards to dialysis? Caffeine (1,3,7-trimethylxanthine) with a molecular weight of 194 daltons, low plasma protein binding (36%), and a low Vd (0.6-0.8 L/kg), has favorable dialyzability characteristics.6 Evidence in support of hemodialysis for caffeine toxicity is far scarcer than for theophylline, but has been utilized with apparent clinical benefit in several reported cases.7
Method Madness: Modalities for Extracorporeal Toxin Removal
Most think first of conventional hemodialysis when talking about the treatment of poisoning, and rightfully so – but it is only part of the repertoire of extracorporeal elimination techniques which include hemodialysis, hemoperfusion, continuous renal replacement therapy, exchange transfusion, plasmapheresis, and peritoneal dialysis. These contrast with corporeal elimination techniques, where toxin removal happens internally, such as in the intestinal lumen (e.g. enteroenteric or enterohepatic circulation disruption with multiple-dose activated charcoal), in the kidneys (e.g. urinary alkalinization), or in the blood (e.g. chelation.)
- Hemodialysis (HD). Conventional high-flux hemodialysis is the process by which a solute is cleared from the blood by diffusion across a concentration gradient. This is accomplished by running blood through a dialysis circuit and over a porous membrane, on the other side of which is a solution (dialysate) into which the solute diffuses. The circuit can be manipulated in several ways to enhance clearance. You can increase the flow rate of the blood or dialysate, or change the dialyser membrane to one with an increased surface area, a greater pore number (a measure of “efficiency”) or larger membrane pore size ( a measure of “flux”).2 The concentration of electrolytes in the dialysate can also be manipulated as desired. Of all the extracorporeal elimination techniques, HD is the most common method employed in poisoned patients. And with good reason – it is unquestionably the most effective way to rapidly remove amenable toxins.
- Hemoperfusion is of mostly historical significance, due to improvements in conventional hemodialysis technology. When hemoperfusion is used, blood is passed through a resin- or charcoal-coated cartridge onto which a toxin is adsorbed, increasing the capacity for removal of larger molecular weight (MW) and protein-bound molecules. Unfortunately, the column also absorbs platelets, white blood cells, and calcium, and is hard to find in the first place (many facilities no longer have charcoal columns).8 This technique has fallen into disuse with the development of high-efficiency and high-flux dialyzers with higher MW cutoffs and fewer complications.
- Continuous Renal Replacement Therapies (CRRT). These include options such as continuous venovenous hemofiltration (CVVH), hemodialysis (CVVHD), and hemodiafiltration (CVVHDF). A slower flow rate is pumped continuously through a circuit that removes solute by convection or diffusion. The reported benefit is that patients whose blood pressure cannot perfuse a hemodialysis circuit may still be treatable with this slower-flow method. It can also follow hemodialysis for toxins that undergo redistribution, to blunt rebound of serum concentrations – acute on chronic lithium toxicity, for example.9 The primary downside is that it removes toxin much more slowly (50-80% slower, in fact) than conventional HD.
- Exchange Transfusion. Somewhat of a relic, this is usually used in the context of very young infants or severe dyshemoglobinemias. Here, blood (partial or complete volume) is removed and replaced with new blood. It is more low-tech as it does not involve a dialysis circuit. It is effective in removing some toxins, but experience is limited, both in the literature and at the bedside. Finding a toxicologist comfortable with the technique is not easy, and clearance rates are about 1/10th of HD.2
- Plasmapheresis. A concept similar to hemodialysis, plasma is separated from the blood (by filtration or centrifugation), filtered, and returned to the body. In a similar process, therapeutic plasma exchange (TPE), the plasma is instead discarded and replaced by another solution. The molecular weight cutoff for plasmapheresis is higher than a traditional hemodialysis membrane, so proteins and other substances (e.g. digoxin-antibody complexes) can be removed. However, clearance rates are slow (<50 ml/min) and complications include bleeding and hypocalcemia.
- Peritoneal dialysis. The dialysis fluid is infused via a catheter into the peritoneal cavity, and toxin diffuses from the blood into the fluid . . . theoretically. In reality, the effective removal of toxin is minimal, and PD has very little role in the acute management of poisoned patients.
Here’s an illustration of the temporal changes in plasma methanol concentration with different techniques:
Adapted from Ghannoum M, et al.2 HD = hemodialysis, CVVHD = continuous venovenous hemodialysis, TPE = therapeutic plasma exchange, PD = peritoneal dialysis
Let’s consider a patient with a methanol concentration of 100 mmol/L (320 mg/dL) as depicted above. If hemodialysis is performed, the concentration will be in the non-toxic range at 9 hours, and the patient can be medically cleared. If CVVH is used, it will take a full day. With plasma exchange it will be nearly 2 days, and with PD, nearly 3 full days. However, if no extracorporeal therapy is performed the level will still be above the toxic range at 211 hours, almost 9 days . . . the inconveniently slow endogenous clearance of methanol.
So what’s the point? What are the goals of dialysis in the poisoned patient?
- Remove toxin. Dialysis should lower the serum concentration, at a higher rate than the body would do so itself. A toxin should therefore ideally have high dialyzability and low endogenous clearance (<4ml/min). This is the rationale for recommending dialysis for the Usual Suspects: salicylate, lithium, ethylene glycol, methanol, theophylline. All are very dialyzable, with endogenous clearance which is not fast (or not fast enough on its own).
- Reverse metabolic derangements. Sometimes, the poison is no longer the problem. It’s the downstream effects that get out of hand. Metformin-associated lactic acidosis, for example, or a late-presenting methanol poisoning with formic acid accumulation. If a metabolic or electrolyte disturbance is too severe, the patient may die before you can fix it . . .
- Make the patient better. Just because a substance is coming out in the dialysate, doesn’t mean you’re getting anywhere. The following should always be true when considering hemodialysis:
- The poisoning should matter. The substance needs to have significant intrinsic toxicity, and the estimated dose or serum concentration should be high enough to cause harm. Dialysis should prevent an unacceptable adverse outcome such as prolonged ICU stay (for example, in the case of phenobarbital)10, permanent disability, or death.
- Other therapies should be unavailable or insufficient. A typical acetaminophen overdose is completely treated with timely N-acetylcysteine. Clonidine toxicity resolves quickly without sequelae with good supportive care. Hemodialysis shouldn’t be necessary.
- The serum concentration should be lowered faster than it otherwise would (exogenous vs. endogenous clearance), and lowering the serum concentration should have a clinical benefit. Getting back to the example of amitriptyline, HD may lower the level of drug in plasma, but that minimally affects total body burden and correlates poorly with clinical status. So doing it doesn’t make the patient better.11
- Other considerations. The patient’s normal physiologic defenses against poisoning may not be intact. For instance, you may be more inclined to dialyze a patient with an elevated lithium level who has renal failure since lithium is primarily eliminated by renal excretion. A patient with metformin lactic acidosis who happens to have hepatic insufficiency will accumulate more lactate, because the liver is responsible for lactate clearance.12 A salicylate poisoned patient with respiratory depression is going to lose their acid-base battle sooner, as that protective respiratory alkalosis disappears.
Case Continued
The patient described has ingested a potentially lethal quantity of caffeine, and has developed seizure activity and a cardiac dysrhythmia. Caffeine levels correlate well with toxicity and, due to caffeine’s low Vd, are a reasonable approximation of total body burden. Lowering the serum concentration, which can be accomplished faster with HD than relying on endogenous clearance alone, may make the difference between fatal and near-fatal poisoning here.
But wait! There’s more . . . A few other tidbits . . .
- Don’t forget, dialysis may remove some needed therapies, like N-acetylcysteine and fomepizole.13,14 Doses need to be adjusted accordingly.
- While the toxins we classically think of are the most common indications, here are a few which may be surprising:
- Valproic acid with extremely high levels (>1300 mg/L).4
- Acetaminophen in massive overdose, with extremely high levels (>900 mg/dL), acidosis and altered mental status.15
- Thallium, atypical for a heavy metal, is somewhat dialyzable.16
- Dabigatran.17
- Phenobarbital in the presence of prolonged coma, to shorten ICU stay.10
- Carbamazepine in the presence of refractory seizures or dysrhythmias.18
- Things we get asked about a lot but the answer is still and always “no”. . . Digoxin. Metals (except thallium). Colchicine (unfortunately. . .)
Lastly, the decision to call for hemodialysis, and the criteria for doing so, is complex. Much of the evidence in support of our practice in this area consists of observational case reports. The Extracorporeal Treatments in Poisoning Workgroup (EXTRIP) conducted systematic reviews and published guidelines on criteria for starting and stopping extracorporeal treatments in a number of toxins, including those listed above.
Case Conclusion
The patient has several more episodes of cardiac arrest with apparent polymorphic ventricular tachycardia on the monitor, but does not improve with antidysrhythmics or magnesium sulfate. When pulses return between arrests, she is extremely tachycardic to 180-190 bpm. An esmolol infusion is instituted, and a hemodialysis catheter is placed. She undergoes 4 hours of intermittent hemodialysis, during which her seizure activity and dysrhythmias abate. She survives to hospital discharge with no apparent neurologic sequelae.
Did hemodialysis save the day? Perhaps. In any overdose, lots of things happen simultaneously, and the impact of any single intervention is hard to prove. Ideally, caffeine levels would have been obtained before, during, and after HD to allow a calculation of clearance – had the samples not been lost in the lab. Nonetheless, this was a massive overdose of a highly dialyzable substance, and the clinical status improved coincident with extracorporeal therapy. From the bedside, HD made a big difference.
So next time you have a life-threatening overdose involving a highly dialyzable compound, call your friendly neighborhood nephrologist and say hey, let’s Dialyze This.
Table 1 – Compounds of interest and their molecular weight.
↩︎
Many thanks to Darren Roberts, MD, PhD (Nephrologist, Pharmacologist, Toxicologist, and fellow EXTRIP participant) for expert review of this content.
The molecular weight listed for Lithium is for lithium carbonate… but surely lithium in the body exists in ionised form, Li2+, which has a molecular weight of around 7?
Can’t believe I just now read this post. Awesome explanations/delivery. Thank you.