In the United States, three cyanide antidotes are available: nitrites (e.g., sodium nitrite and amyl nitrite), sodium thiosulfate, and hydroxocobalamin (vitamin B12a
. Nitrites generate met(ferric)hemoglobin, which has a high affinity for cyanide, but can no longer bind oxygen; thus, nitrites can exacerbate the carbon monoxide-induced reduction in oxygen-carrying capacity in smoke inhalation victims. Moreover, nitrites can induce vasodilatation, causing hypotension 2
. Sodium thiosulfate acts as a sulfur donor for the enzyme rhodanese, which detoxifies cyanide by converting it to thiocyanate, but rhodanese is limited both in cellular amount and tissue distribution. Hydroxocobalamin binds cyanide with a relatively high affinity (KA
, but 5–10 grams are required for cyanide poisoning.
We have shown previously that cobinamide is effective as a cyanide scavenger in cultured cells 9
, a fly model 27
, and nitroprusside-induced cyanide toxicity in mice28
. We now show that cobinamide is effective in two lethal mouse models of cyanide poisoning, and demonstrate it is superior, in our models, to currently available treatments. Although cobinamide was absorbed poorly, cobinamide sulfite was rapidly absorbed from an intramuscular site, and protected mice from cyanide-induced death, even when administered after cyanide. Evans previously showed that cobinamide neutralizes cyanide in mice and rabbits, but he administered it by intravenous injection and did not strictly compare it to other cyanide antidotes 29
. With the exception of amyl nitrite, currently approved drugs for cyanide poisoning are available only as intravenous preparations, limiting their usefulness in a mass casualty setting. The time required to start intravenous lines and administer relatively large fluid volumes would be prohibitively long in treating many cyanide-poisoned persons in the field. Therefore, an intramuscular preparation that is rapidly absorbed would be highly desirable.
To develop a model of cyanide inhalation, we needed to construct a suitable exposure chamber. Cyanide gas (hydrogen cyanide) is not commercially available, and, therefore, a flow-through exposure system with accurately-controlled cyanide concentrations is not feasible. Requirements of a sealed chamber are that gases must equilibrate rapidly, and the chamber must be maintained above the boiling point of HCN (26° C). We found that our chamber generated reproducible, stable concentrations of cyanide with a sustained level of anesthetic gas throughout the exposure period (at least 30 min). The lethal LC50
we observed for cyanide (451 ppm for 30 min) was higher than previously reported in mice 18, 30
. Three factors may contribute to this difference: 1) mouse strains vary in their sensitivity to cyanide and C57BL/6 mice are relatively resistant; 2) previous reports used measured concentrations of cyanide gas that tend to underestimate the concentration of cyanide because of gas loss or condensation at room temperature; and 3) the earlier studies were performed in awake mice that likely hyperventilated on initial cyanide exposure 18, 30
, whereas our studies were performed with anesthetized mice.
Cobinamide is the penultimate compound in cobalamin biosynthesis, lacking the dimethylbenzimidazole nucleotide tail coordinated to the cobalt atom in the lower axial position 21
. Whereas cobalamin has only a free upper ligand binding site, cobinamide has free upper and lower binding sites; moreover, the dimethylbenzimidazole group has a negative trans
effect on the upper binding site, thereby reducing cobalamin's affinity for ligands 22
. The net effect is that cobinamide binds two cyanide ions, and has a greater affinity for cyanide than hydroxocobalamin, with a KA
overall of ~1022
. In addition, cobinamide is at least five times more water-soluble than hydroxocobalamin. These three chemical differences translate into smaller volumes of administration of cobinamide than hydroxocobalamin, and we calculate that 5 ml of a 200 mM cobinamide solution should neutralize one human LD50
of cyanide (5 ml can be given intramuscularly in the gluteal region).
We found that cobinamide is considerably more effective than hydroxocobalamin, and that the difference is more pronounced in the inhaled than the parenteral model of cyanide poisoning. Thus, cobinamide is 11 times more effective in the inhaled model and 3 times more effective in the parenteral model. Although there are several possible explanations for these differences in efficacy, the most plausible are the kinetics of the two models. Because of the time needed to absorb cyanide gas into the circulation, the inhaled model leads to a slower onset of toxicity, whereas, in the intraperitoneal model, cyanide is absorbed rapidly and distributes into the vascular system at a rate approaching that of intravenous injection. Other small molecules administered to mice by intraperitoneal injection are rapidly absorbed 31
. A difference between two compounds is, therefore, more likely to be seen in the inhaled model where these compounds have a longer time to act. We should note that the inhaled model is more representative of real-life circumstances, in which people are likely to be exposed to cyanide gas.
We have shown previously that cobinamide has high affinity for nitric oxide (NO) 21, 22
. Binding NO in vivo
may lead to systemic hypertension as occurs with intravenous hydroxocobalamin 32
, and to localized vasoconstriction when given by intramuscular injection. Initial studies with intramuscular cobinamide supported this impression, because post-mortem examination of injected animals demonstrated significant quantities of residual cobinamide at the injection site (TDB, unpublished observations). We found that cobinamide sulfite does not bind NO in vitro
, and that it is rapidly absorbed and highly effective. Moreover, it exhibited no clinical toxicity, even at a dose of 2000 mg/kg (2.0 mmol/kg).
Although the precise LD50
for cyanide is not known in humans, lethal poisoning has occurred with as little as 50 mg (~ 0.5 to 1 mg/kg). Mice are more resistant to the effects of cyanide than humans with the LD50
for KCN reported to be between 2–8 mg/kg, depending on the strain and mode of cyanide exposure 20, 33
. We found that C57BL/6 mice are particularly resistant, with an LD50
for cyanide of 9.75 mg/kg (0.15 mmol/kg). Thus, the amounts of cobinamide required to rescue humans from a lethal cyanide dose is likely to be considerably less than that required in this study.
The current study has several limitations. First, anesthetized animals do not approximate real-life exposures. However, we felt that studies in awake animals would be inhumane, and current animal care guidelines strongly discourage and limit the use of awake animals in studies of toxins such as cyanide 34, 35
. Anesthetics might bias our data by inducing hypotension, which could increase the susceptibility of the animal to the cardio-depressant effects of cyanide. Alternatively, in the inhaled model, anesthetics could protect animals through decreased minute ventilation or by preventing hyperventilation in response to cyanide gas 19
. Second, in the parenteral model, the onset of death was very rapid, leaving only a narrow window for intervention. Third, the studies were not conducted in a randomized fashion. However, all animals were the same inbred strain from the same supplier, and the observed effects were highly reproducible. And, fourth, the studies were not blinded because of the complexity of doing this, and the nature of the antidotes (cobinamide and hydroxocobalamin are both intensely colored).