Apart from being an abundant weak anion which decreases the strong anion difference, salicylate toxicity also causes lactic acidosis and ketoacidosis.
Meet the salicylates
Chemically speaking, they are all variations on the same theme. Aspirin is probably the most commonly overdosed upon of the group, and so will receive the greatest attention, even though methyl salicylate is actually more toxic (and seems to have been around for longer in toxicological literature; early pioneers of the field report on the case from 1832, of a man who drank 30ml of "oil of wintergreen" and died).
The salicylic acid molecule is the major common feature of all these compounds, and their therapeutic (or toxic) effects rely on their catabolism into salicylic acid and some other boring daughter molecules.
Why cant we just use raw untreated salicylic acid? Well, because its horribly toxic to the gastric mucosa. Topical preparations exist, which people smear on themselves in hope of wiping out acne bacteria, or with the expectation that the corrosive effect will lend them an unnaturally youthful appearance by burning away the dead skin cells.
Speaking of rubbing yourself with toxins. The abovementioned "oil of wintergreen" is smeared on our patients on a daily basis, in copious quantities. How much is too much? We know it penetrates the skin; furthermore every 10g of metsal cream contains about 2.83 grams of salicylate, equivalent to about 9 normal aspirin tablets.
Thankfully, transcutaneous salicylate overdose is rare (albeit not unheard of). One only absorbs about 2% of a topical dose after 30 minutes; which means that in order to soak up the toxic 150mg/kg dose, a 70kg person must coat themselves completely with 525g of methyl salicylate ointment. As a method of suicide, this is inconvenient.
The mechanism of high anion gap acidosis in salicylate poisoning
After the initial respiratory alkalosis (driven by a stimulation of respiratory centres) the salicylate overdose victim develops a profound metabolic acidosis, with a high anion gap.
Some of this, we may surmise, comes from the salicylate anion itself. Aspirin is rapidly hydrolysed in the liver (its halflife is only 20 minutes) and produces salicylate anions. On top of that, because it is a weak drug, we take large quantities of it (and large quantities are available in bulk from places like Walmart).
Whereas a normal therapeutic salicylate level is only about 0.2-0.7 mmol/L, the typical toxic level of salicylate which is said to require dialysis is 5.8mmol/L, or about 0.8g/L. Yes, even though salicylate is highly protein bound, dialysis seems to be a better means of clearing it than haemoperfusion. This is probably because at high doses the serum protein binding sites are saturated, and more free salicylate becomes available.
Given that serum levels in the literature range as high as 8.9mmol/L, one can see how these extra anionic millimoles might contribute to the high anion gap.
But that is not the whole picture.
Mechanism of lactic acidosis and ketoacidosis in salicylate poisoning
Salicylate is a weak acid with a pKa of around 2.7, which means that it is incompletely ionised at physiological pH. Furthermore, what with all this acidosis going on, the pH falls and even more of the salicylate becomes undissociated and thus lipid-soluble. This lipid-soluble form easily penetrates cell walls and wreaks havoc among the mitochondrial enzymes, uncoupling oxidative phosphorylation and interfering with Krebs cycle enzymes. As oxidative glucose metabolism is interfered with, Krebs cycle no longer provides enough oxaloacetate for the acetyl-CoA to bind with; the resulting excess of acetyl-CoA is converted into ketones.
Mechanism of respiratory alkalosis in salicylate toxicity
In salicylate intoxication, the earliest signs are usually those of a respiratory alkalosis, due to a centrally-driven tachypnoea. The mechanisms of this seem to be poorly understood. Most papers on salicylate toxicology make mention of it as the "classical early sign", and mutter something about medullary sensitivity to CO2 or "direct stimulation of respiratory centres" before trailing off in embarrassment.
As with many such physiology questions, everybody knows the answer but they forgot where it came from. Fortunately, the internet maintains good records. For instance, in 1955 Tenney and Miller published a paper on the respiratory and circulatory actions of salicylate. They in turn have references to describe the animal experiments which yielded this explanation.
According to these authors and their referees, in normal man prolonged salicylate administration increases the sensitivity of the respiratory center to carbon dioxide, thereby downregulating the homeostatic setpoint for PaCO2. Later investigators (Fleming et al, 1991) demonstrated that this effect on the medulla is reversed by atropine (they actually washed a mongrel dog medulla with salicylate and atropine), suggesting that the mechanism is cholinergic. It is so potent that the administration of non-overdose quanitites of salicylate (8-10g) has been demonstrated to reverse the nocturnal respiratory disturbances of sleep apnoea (Pillar et al, 1992). The direct respiratory stimulant action of acutely ingested salicylate can be an effect on the medulla alone; however it is enhanced by the effects of salicylate on aortic and carotid chemoreceptor areas. McQueen et al (1989) found that anaesthetised rats had a more vigorous respiratory response to salicylate if their carotid receptors were intact.
Mechanism of metabolic alkalosis in salicylate toxicity
Question 3.2 from the first paper of 2013 presents the candidates with a blood gas which is was taken in some mysterious circumstances, the history being left left open to interpretation. The examiners wanted one example of a clinical scenario which might explain a high anion gap with a mixed metabolic / respiratory alkalosis. The college model answer produced salicylate toxicity as their favoured explanation, which brings up the question: what the hell is the mechanism of metabolic alkalosis?
I thank one attentive reader for pointing out that this is weird, and requiring explanation. Surely, only metabolic acidosis is to be expected from salicylate metabolism. Salicylic acid, gentisic acid, acetate - these all contribute to the raised anion gap. Studies of adult salicylate toxicity patients never mention any metabolic alkalosis (Gabow et al, 1978).
Explanation #1. It was the vomiting.
Protracted emesis often follows salicylate overdose. Quite apart from the local gastric irritation which occurs, there is also some sort of central effect. For instance, Bhargava et al (1963) describes some horrific animal experiments by Hatcher and Weiss, who (in 1926) were able to induce emesis by direct application of sodium salicylate to the floor of the fourth ventricle. Some might counter that the forceful application of just about anything to the floor of your fourth ventricle might induce emesis; but the authors tested 27 substances, of which only 13 induced vomiting. Bhargava and colleagues went one further, and mechanically destroyed the chemoreceptor trigger zones in their experimental dogs. The dose of salicylate administered to these animals was so high that many of them died, but none of them vomited. In contrast, relatively modest doses induced vomiting in dogs whose chemoreceptor centers remained intact.
Explanation #2. The intensivist did it.
Another possible mechanism is purely iatrogenic. Let us behold the reality of critical care medicine. The vast majority of salicylate toxicity patients you will meet in the course of your ICU practice are going to be receiving some sort of treatment. Practically never will you encounter a completely deranged "wild-type" salicylism, lousy with florid acid-base features of aspirin intoxication - that sort of thing would require the patient to be admitted and then ignored willfully for some number of hours. Of course every salicylism patient you will see will have had some sort of management, and the mainstay of management (short of dialysis) is actually alkaline diuresis. The reliance of salicylate excretion on urinary pH is well known, and maintaining an alkaline urinary pH is a one of the cheap convenient ways to enhance its clearance. The linked study (Prescott et al, 1982) is an excellent demonstration of this effect: the control group (mean urinary pH ~ 6.1) excreted 0.16 grams of salicylate over 4 hours, whereas the patients with a urinary pH of 8.0 excreted 2.44 grams, over fifteen times more.