I just saw one of the biggest anion gap of my life and I don’t know the cause. Worse yet, the patient had this occur a few months ago, also with no explanation. So I want to figure out what is going on before admission number three.
Patient presented to the ED obtunded and was unable to give a cohesive history. The admission labs:
So lets calculate the anion gap.
Add the bicarb to the chloride
subtract that sum from the sodium
and you get OMG!
I mean
This comes in below the gap of 70 I took care of in 2011 (presumed metformin-induced lactic acidosis) but edges out the more recent 51 from propylene glycol from 2015. So the way I like to tackle these mysterious anion gaps is to create a Gamblegram named after pediatrician James Gamble who invented the concept in 1939. The idea behind the Gamblegram (and behind the concept of the anion gap itself) is that the total number of anions must equal the total number of cations, otherwise if you touched blood, it would give you a shock. Then we graph out the cations and anions that we know, while leaving “unmeasured or unstated” cations and anions as a box. This patient looks something like this:
Looking at the numbers, the gap gets so large because not only is the bicarb so phenomenally low, but they have a pathologically low chloride and a sodium which is bumping up gainst the upper limit of normal. Additionally the potassium is a bit low, shrinking the other cations box.
We have an ABG done a few minutes after the chemistries were drawn:
- pH 6.94
- paO2 179
- pCO2 6
- HCO3 1
Then we hit that with Winter’s formula: 1.5 * 1 = 1.5 plus 8±2 gives a target CO2 of 7.5 to 9.5, just a bit higher than the actual pCO2 of 6. So this patient has a metabolic acidosis with a slight respiratory alkalosis. But who can fault them for hyperventilating with a pH south of 7.
So with a massive metabolic acidosis and a ginormous anion gap, you should be itching to order a toxic alcohol screen. But first check for other causes of an anion gap metabolic acidosis:
- Aspirin: less than 2.0 mg/dl (works especially well with the concurrent respiratory alkalosis)
- Acetaminophen: less than 5 mcg/dL
- Lactic acid: 9 mmol/L
- Ketoacidosis: This hospital doesn’t do real time serum ketones. So we didn’t have data acetone, betahydroxybutyrate or acetoacetate levels. However the U/A showed ketones at 20 mg/dL
So if we start to fill in the gap:
- A normal gap is 12 mmol/L
- Lactate is 9 mmol/L
- The phosphorus is 7 mg/dL. Four of that is included in the normal gap, the extra 3mg/dl converts to 1 mmol/L
- That comes to 22, leaving an unknown gap of 31. Some of this will presumably be filled by ketones, acetoacetate and betahydroxyburyrate.
Next step is to look for an osmolar gap, because if you aren’t thinking toxic alcohol you aren’t wired to be a nephrologist, toxicologist or critical care doc. The serum osmolality is 342. Calculated osmolality:
I know that your shitty medical calculator does not include ethanol or if it does it divides it by 4.6 not the correct 3.7. For crap calculators see MedScape, GlobalRPh, (QxMD and UpToDate don’t even include alcohol?!). Even Wikipedia gets this right. Get with the program, get MedCalcX or use MD+Calc.
So the osmolar gap is 342-321= 21. High but not very impressive, especially compared to the anion gap in the stratosphere.
The thing to understand about toxic alcohol’s, anion and osmolar gap is that they move in opposite directions. Ethylene glycol (antifreeze) and methanol (fuel, incompetently distilled spirits) are both neutral alcohols. They are osmotically active so they contribute to the osmolar gap, but since they are not anions they do not contribute to the anion gap. So early after ingestion the osmolar gap is high but the anion gap remains low.
Then the toxins are metabolized (initially and most importantly by alcohol dehydrogenase) into toxic downstream metabolites. Many of these metabolites are acids that lose a proton, and thus become anions. So after metabolism the anion gap will climb as neutral alcohols are converted into anionic acids. Interestingly, the osmolar gap falls. Even though the metabolites are low molecular weight and osmotically active, they do not contribute to the osmolar gap, because the equation includes them in the calculated osmolarity. When you multiply the sodium by two, you are covering all anions in solution. Since the acids are anions they are covered by the calculated osmolality and don’t contribute to the gap.
So our patient with the big anion gap and the modest osmolar gap could just be a late presentation of a toxic alcohol. Once that has happened and the osmolar gap has retunred to normal. Even if there is a large anion gap it is probably too late to intervene with fomepizole to block alcohol dehydrogenase. Once the osmolar gap has closed the toxin has moved downstream of the alcohol dehydrogenase.
But our patient still had an osmolar gap. So we gave fomepizole and dialyzed the patient. The next morning the osmolar gap had closed and the anion gap was nearly normal. We stopped the fomepizole and dialysis. On the third hospital day we got back the toxic alcohol screen.
- Acetone 31 mg/dl
- Methanol: not detected
- Ethylene glycol: not detected
- Isopropanol 12 mg/dL
Isopropanol, is commnly called isopropyl alcohol, rubbing alcohol. It will increase the osmolar gap but is not converted to an acid and does not cause anion gap metabolic acidosis. Divide the level by its moleculatr weight, 60, and then multiply by ten to convert “per dl” to “per liter.” So the level of 12 accounts for 2 of our abnormal osmolar gap of 21.
The Acetone is interesting. It is also not a charged molecule so will not account for the anion gap, but it is in equilibrium with two charged molecules that can generate an anion gap: acetoacetate and beta-hydroxybutyrate.
I looked but could not find the expected ratio of acetone to the other two ketones in order to extrapolate from the acetone level to the concentration of the anions. I couldn’t find a reference, but I found a number of documents that said acetone was a definite minority. So if we estimate that each ketones is at 2-3 times the concentration of acetone we have a 20-30 mmol/liter combined concentration of acetoacetate and betahydroxybutyrate. (Molecular weight of acetone is 58, so 31 mg/dl is 5.3 mmol/L). This essentially fills the unknown gap.
Could this just be alcoholic ketoacidosis? The patient had a triglyceride level of 600 mg/dL which is consistent with alcoholism? The serum glucose on admission was 57, also consistent with alcohol induced ketosis.
I personally am not very satisfied, because I see drunks all the time (at work, not socially) and nion gaps this high are very unusual. I personally think there is some other ingestion stimulating the massive ketosis. Looking for ideas.
Notes: We have an oxoproline level cooking, but the negative acetaminophen level makes this less likely in my mind. No, we didn’t send a D-lactic acid level.