The Proven Antidote That Can’t Be Proven to Work (Part 2)
The jury is still out on whether this is helpful.
We are approaching 20 years of experience in the United States (and more than 30 years in France) of giving an enormous bolus of vitamin B12 to critically ill smoke inhalation patients.
The jury is still out on whether this is helpful.
Hydroxocobalamin, a naturally occurring form of vitamin B12, was introduced to the US market in 2006 as Cyanokit®, and its FDA-approved label stated it could treat cyanide poisoning after “exposure to…smoke from closed-space fires.”
As reviewed in Part I,1 smoke inhalation victims are often (though not always) found with toxic cyanide levels, and hydroxocobalamin has been shown to effectively treat cyanide poisoning—most dramatically, in a cringe-inducing experiment involving 54 anesthetized and cyanide-poisoned beagles, who either died with a placebo or recovered with hydroxocobalamin.2
We will thus likely continue injecting smoke inhalation victims with $1,000 boluses of B12, in doses so heroically huge that the patient will literally turn bright red and urinate red for days to come (Figure 1).3
But we’re still waiting for hard evidence that this helps. Instead, as the evidence has mounted, we’ve only seen a whisper of harm so far.
Sources of Cyanide
Pseudomonas pyacaneus can produce cyanide in blood.4 So can a sodium nitroprusside infusion, which is why this vasodilator is contraindicated in, among other cases, underlying B12 deficiency.5 Cyanide in the smoke of cannabis and cigarettes is partly soaked up by the body’s circulating hydroxocobalamin (levels of which are chronically lower in smokers).6 It’s in apple seeds, apricot pits, and almonds, and is used in some industrial processes.
And of course, it’s been used for terror, murder, and suicide.
It was trialed by the French as a chemical weapon in World War I, but it was so volatile that it just blew away in the open air. No deaths were attributed to that use, and the warring parties moved on to chlorine, phosgene, and mustard gas, which collectively killed more than 100,000 soldiers and maimed more than 1 million.7 Even more infamously, cyanide was used as Zyklon B in the gas chambers of the Nazi Holocaust.
Cyanide was also the poison employed in the 1978 mass suicides of the Jonestown cult, when 918 followers of preacher Jim Jones infamously “drank the Kool-Aid” laced with cyanide. And in the 1980s, after someone spiked random containers of Tylenol with cyanide, eight people died, 31 million bottles of Tylenol were recalled, and we entered the era of tamper-resistant over-the-counter medications.
Lurid historical anecdotes aside, today we mostly think of cyanide as a smoke inhalation risk. Hydrogen cyanide gas can be generated by burning not only nylon, polyurethane, and other plastics, but also wool and silk.
Smoke Inhalation Reconsidered as Cyanide Inhalation
The French pioneered the practice of empirically treating smoke inhalation patients for cyanide poisoning. A 2006 retrospective review recounted the Paris Fire Brigade’s eight-year experience of giving hydroxocobalamin 101 times. But the review provides no real information about whether this works. Of the 101 times it was given, 29 patients were lost to follow-up, many died, and the 30 known survivors were doggedly described as having survived “after the administration of hydroxocobalamin.”8
A meatier 2007 study, funded by EMD Pharmaceuticals (who brought Cyanokit® to the US market), gathered prospective data on 69 smoke inhalation patients treated with hydroxocobalamin and checked cyanide levels on most. It reported a 72% overall survival rate, and a 67% survival rate among the 42 patients who had elevated cyanide levels.9
But again, this was not a trial. There was no control group. A bunch of patients got hydroxocobalamin, a bunch survived, but who knows what it all meant.
A 67% survival rate among those with an “elevated” cyanide level at least sounds good. But it also assumes a consensus about how to interpret cyanide levels.
That is a rabbit hole I shudder to go back down.
But briefly: Cyanide tests are usually sendouts that take days to result. Some don’t differentiate between “toxic” cyanide and “body-neutralized” cyanide that’s been rendered harmless by endogenous systems10; cyanide’s one-hour half-life and the fact that it is an intracellular poison makes it hard to interpret a given serum level10,11; and across the literature, different experts use a confusing range of definitions for “toxic” or “lethal” levels.1,4 (They also toggle back and forth between reporting levels in μmol/L or mg/L).
What’s more, even a profoundly cyanide-poisoned patient can recover without any particular antidote and with just good supportive care. As a landmark paper by Dr Frédéric Baud and colleagues in Paris, published more than 35 years ago in the New England Journal of Medicine, stated: “There is no blood cyanide concentration above which the outcome is invariably fatal.”11
That’s also not even talking about the methemoglobinemia issue. Smoke inhalation patients often develop methemoglobinemia on top of their many other problems, and Methemoglobinemia (MetHb) is itself a potent cyanide scavenger. (Remember, old school cyanide antidotes use nitrites to turn hemoglobin into methemoglobinemia cyanide sponges. But in a fire, inhaled nitrogen oxides can do the same thing, so the smoke the patient inhales can include both cyanide and a cyanide antidote of sorts.) A fascinating 2002 study out of Japan looked at 22 postmortem fire victims and found not only that 12 had significant cyanide toxicity and five had lethal cyanide levels, but also that nearly all, including three of the five with “lethal” levels, simultaneously had more than enough MetHb circulating to neutralize all of that cyanide!12
Just Tell Us If It Works Already!
At this point, a practicing physician would just like an answer. Does hydroxocobalamin improve outcomes in smoke inhalation patients?
That’s never been shown. In fact, quite the opposite.
A 2017 study out of Vanderbilt compared 138 smoke inhalation patients treated with hydroxocobalamin over six years (2008-2014) to a historical control group of 135 similar-yet-untreated patients (2002-2008)13. Almost all of them (92%) were intubated. There was no difference in the primary endpoint of mortality (29% in the hydroxocobalamin group, 28% in the historical control, P = 0.9). In fact, the Vanderbilt team found only one difference: Hydroxocobalamin was associated with markedly less pneumonia (23% vs 49%, P = 0.01) and more ventilator-free days.
Pfizer, which via a subsidiary had by then moved into the hydroxocobalamin market, leapt on this with a Pfizer-authored 2022 study that blew straight past “association” and into “causation.”14 It asserted that hydroxocobalamin in a smoke inhalation case prevents pneumonia, reduces time on the ventilator, and saves hospitals $7,226 per patient.
The Pfizer study even bemoaned the unfairness of how ambulances have to pay $1,000 per dose to stock hydroxocobalamin (with a shelf life of three years), while the pneumonia-prevention savings it offers accrue only to hospitals.
All of that was highly suspect. So there was no difference in survival, only a totally unexpected difference in pneumonia rates? The Vanderbilt group hemmed and hawed to explain this, with some hand-waving about improvements in “metabolic derangements,” about vitamin B12 possibly enhancing the efficacy of pneumonia-fighting CD8 white cells, and so on.
They also nodded in passing to the utterly obvious, actual explanation: They had compared their hydroxocobalamin cases to a historical control group (2002-2008) plucked from the dawn of a national campaign to reduce ventilator-acquired pneumonias (VAPs).
Back in 2002, studies were suggesting that 10%-20% of all critical patients put on a ventilator would develop a pneumonia, and those VAPs would have up to 25%-50% mortality rates.15,16 If you took the high end of those numbers, it translated to death-by-pneumonia for every 10th intubation. This became a priority concern, and especially after a landmark “ventilator bundle” campaign launched in 2004, ICUs everywhere were elevating the head of the bed, cleaning the mouth with chlorhexidine, adopting protocolized sedation vacations for early extubations, and so on.17 This whole process even made a cameo in Atul Gawande’s 2009 bestseller The Checklist Manifesto, alongside campaigns to prevent central line infections and surgical errors.
Pneumonia rates dropped nationally across ICUs. Hydroxocobalamin had nothing to do with it.
(In fact, considering how deadly VAPs are, it’s concerning that mortality rates across the Vanderbilt study were so similar—29% in the hydroxocobalamin-treated modern era vs 28% in the pneumonia-plagued past? If VAPs were killing fewer patients, was hydroxocobalamin somehow making up the difference?)
The quest to identify any benefit from hydroxocobalamin for smoke inhalation has soldiered bravely on. A 2019 study looked at 739 smoke inhalation patients across eight years treated in 21 French ICUs. A little more than half of the patients got hydroxocobalamin. The authors found no association between hydroxocobalamin and survival. Instead, they found an increased odds ratio (OR) of 1.8 for severe acute kidney injury.18
A 2024 study of 36 smoke inhalation patients admitted over the past 10 years to a Los Angeles burn center also found no association between hydroxocobalamin and survival.19
Treatment Strategies Amid This Uncertainty
Some have suggested reserving hydroxocobalamin for the smoke inhalation patient who has either a lactate above 10 mmol/L or a carboxyhemoglobin above 10%.20 (See Figures 2 and 3 for proposed treatment algorithms.19,20)
There is some evidence for this approach. The landmark paper cited above by Baud obtained blood samples at the scenes of Paris residential fires from 109 victims before they received any treatment.11 In 39 patients without severe burns, the researchers also obtained lactates. Focusing on just the 23 patients with toxic cyanide levels and a lactate checked, all but three had lactates >10 mmol/L. The authors calculated lactate level of 10 was 87% sensitive and 94% specific, with a positive predictive value of 95% for significant cyanide poisoning.
Another large study published more than 30 years ago summarized 364 house fire deaths investigated by the New Jersey State Medical Examiner’s Office. All had blood cyanide levels checked. A lethal cyanide level (here cited as 3 mg/L) was found in 31 cases (8.5%), and all had carboxyhaemoglobin (COHgb) levels above 10%.4
A 2017 study out of Germany checked cyanide levels in 93 persons found dead in connection to fires or other smoke gas exposures. Among 53 persons who died in closed space or vehicle fires, 47% had toxic (0.5 mg/L) cyanide levels and 13% had lethal (3 mg/L) cyanide levels. But again, no cyanide toxicity was seen without a COHgb level well above 10%.21
If Emergency Medical Services were to use point-of-care lactate or carboxyhemoglobin tests to screen for cyanide poisoning, hydroxocobalamin use might plummet. Consider that among the 273 patients in the Vanderbilt “hydroxocobalamin prevents pneumonia” study, the mean COHgb level was 3% and the mean lactate level 2.5; while among the 739 patients in the French “hydroxocobalamin doubles kidney injury risk” study, mean COHgb level was 4%, and only 74 (about 10%) had a lactate level >8.
This all cries out for further study. Some 3,000 people die in US residential fires every year, many from smoke inhalation. Heroic research efforts have been made, but decades have passed, and still we are groping through the haze for clarity.
