Introduction

Sodium nitroprusside is the most widely used vasodilator drug in critically ill patients [1, 2, 3, 4, 5, 6, 7]. The recommended maximum dose varies from 2 to 10 µg kg−1 min−1 [1, 2, 3]. The most important toxic effects of sodium nitroprusside are cyanide poisoning, thiocyanate toxicity, and methemoglobinemia [4]. Heme oxygenase (HO) is the initial enzyme in hememetabolism. It opens the heme ring, releasing equimolar quantities of biliverdin, iron, and carbon monoxide [8, 9]. Three isoforms of HO have been identified, including two constitutive isoforms, HO-2 and HO-3, and an inducible isoform termed HO-1 [9]. The HO-2 and HO-3 isoforms are not affected by stress conditions. They are distributed throughout the body, with high levels of activity in the brain and testes [8, 9]. HO-1 is strongly and rapidly induced by its own substrate, heme, and by a variety of stress-associated agents, including heavy metals, hyperthermia,hyperoxia, hypoxia, heatshock, endotoxin, hydrogen peroxide, cytokines, ultraviolet radiation and nitric oxide, producing carbon monoxide [9, 10, 11].

CO has been identified as an endogenous biological messenger inthe brain and has an important role in hemodynamic regulation [8, 12]. Elevated exhaled CO levels and moderate increase in blood carboxyhemoglobin levels have also been reported in critically ill patients, particularly in sepsis and trauma [8, 12, 13, 14] and after surgery [15, 16]. Some authors suggest that it could be a marker of disease severity [12]. Several experimental studies have shown that NO donors, such as sodium nitroprusside can induce HO-1 and produce CO by breakdown of heme molecules [11, 17, 18]. A case of carboxyhemoglobinemia secondary to inhaled nitric oxide therapy has been published recently [19], but, to our knowledge, there has been no previous report of carboxyhemoglobin associated with sodium nitroprusside treatment.

Case reports

Patient 1

A 6-month old Moroccan boy with dilated cardiomyopathy developed arterial hypertension 24 h after heart transplant and was treated with nitroprusside. At the beginning of the treatment the patient received NO treatment at 10 ppm, and carboxyhemoglobin levels were 1.2%. Over the following 6 days the nitroprusside dose had to be increased to 8 µg kg−1 min−1, which was associated with a rise in carboxyhemoglobin level to 5.5% (Table 1). There were no signs of systemic toxicity or hemolysis. The nitroprusside was progressively withdrawn and replaced by urapidil and diltiazem, maintaining NO at the same concentration, which led to a rapid drop in carboxyhemoglobin level. At the time when nitroprusside infusion was terminated carboxyhemoglobin level was 2%, and free cyanide levels were measured in the blood by microdiffusion and visible spectrophotometry. The results were free cyanide, 0.2 µg/ml (normal <0.25 µg/ml, toxic levels >2 µg/ml) and thiocyanate, 5.8 µg/ml (normal <9 µg/ml). The carboxyhemoglobin level was 0.3% 24 hours after the withdrawal of nitroprusside. The patient showed a favorable course with no sequelae.

Table 1 Initial and maximum dose of nitroprusside, ventilator parameters, blood gas parameters, co-oximetry, and other treatments administered in the four patients (NR not recorded)
Full size table

Patient 2

A 2-year old Spanish boy developed arterial hypertension and signs of low cardiac output 24 h after heart transplant and was treated with nitroprusside at increasing doses up to 7 µg kg−1 min−1. After 5 days of treatment an elevated carboxyhemoglobin level of 7.7% was detected (previous measurements were not recorded; Table 1). There were no signs of hemolysis or systemic toxicity. The nitroprusside dose was slowly reduced and was replaced by urapidil, labetalol and diltiazem, which led to a drop in carboxyhemoglobin level. The treatment with nitric oxide which received the patient was maintained at the same concentration (6 ppm). When nitroprusside was withdrawn, free cyanide levels were 0.5 µg/ml, thiocyanate 6.8 µg/ml, pH 7.44, PaO2 97 torr, PaCO2 41.5 torr, hemoglobin saturation 98%, HbO2:97%, carboxyhemoglobin 2.4%, methemoglobin 0%, and lactate 0.6 mmol/l with an FIO2 of 0.65 and NO at 5 ppm. The carboxyhemoglobin concentration was 0% 24 h after the withdrawal of nitroprusside. The patient recovered with no sequelae.

Patient 3

A 4-year old Spanish girl in cardiogenic shock due to dilated cardiomyopathy received treatment with nitroprusside 2 µg kg−1 min−1 for 8 days with normal carboxyhemoglobin level. Heart transplantation was performed, and the patient was admitted to the pediatric intensive care unit with extracorporeal membrane oxygenation, nitroprusside at 2 µg kg−1 min−1 without NO. On arrival at the unit the patient was in deep coma, with anisocoiria, and the sedative drugs were withdrawn. The blood gases levels are shown in Table 1. Nitropruside was progressively increased to 8 µg kg−1 min−1. However, due to an error in the calculation of the nitroprusside infusion, the patient received 16 µg kg−1 min−1 for 12 h, leading to an increase in carboxyhemoglobin level (Table 1). Nitroprusside was withdrawn, but brain death was confirmed at 16 h after the operation; respiratory assistance was withdrawn thereafter and the patient died.

Patient 4

A 2-year old Spanish boy diagnosed with dilated cardiomyopathy was treated with nitroprusside at 3 µg kg−1 min−1 for 12 days. Pretransplantation carboxyhemoglobin level was 1.2%. After heart transplant nitroprusside treatment was resumed at 6.5 µg kg−1 min−1, and NO was administered at 10 ppm. A rise in carboxyhemoglobin to 3.7% was observed after 24 h. Despite this rise nitroprusside was maintained at the same dose. There was a subsequent increase in carboxyhemoglobin level 24 h later (Table 1). However, there were no signs of systemic toxicity or hemolysis. The nitroprusside was replaced by diltiazem, maintaining the NO treatment at the same concentration, observing a progressive reduction in carboxyhemoglobin level. The carboxyhemoglobin concentration was 0% 24 h after withdrawal of the nitroprusside. The patient recovered with no sequelae.

Discussion

To our knowledge this is the first report of carboxyhemoglobin elevation in pediatric patients after heart transplantation possibly secondary to nitroprusside treatment. Our patients showed a moderate increase in carboxyhemoglobin level after nitroprusside administration, and in three of these patients the withdrawal of the drug led to the normalization of carboxyhemoglobin level. We believe that current blood analysis devices that measure cooximetry in each blood gas sample permit diagnosis of moderate carboxyhemoglobin elevations that probably would not have been discovered in the past. All four patients had undergone heart transplantation; however, the only additional common risk factor for the rise in the carboxyhemoglobin seems to be treatment with nitroprusside. One of the patients received a higher dose than recommended in children [2], due to a calculation error. The other three patients received doses in the upper range of the recommended dose for several days. In our pediatric intensive care unit we have used the same cooximeter for the past 4 years. In the most recent year 43 critically ill children were treated with nitroprusside, 9 of whom underwent heart transplantation. During this time there were no changes in our protocols nor were different medications used in these patients. Only in the four patients with carboxyhemoglobin elevations was a dose of nitroprusside administered higher than 6 µg kg−1 min−1 for more than 2 days. The other five heart transplant patients received a dose less than 4 µg kg−1 min−1 of nitroprusside, and they showed normal carboxyhemoglobin level.

We believe that the most probable mechanism causing the rise in carboxyhemoglobin level was the induction of HO-1 by the nitroprusside. In experimental studies, nitric oxide (NO) and nitroprusside have been shown to be capable of inducing HO-1 and increase the carboxyhemoglobin levels [9, 10, 11, 17, 18]. Rusca [19] reported an increase in carboxyhemoglobin level coinciding with NO treatment. Three of our patients also received NO. It seems unlikely that the NO did play a primary pathogenic role in our patients as the rise in the carboxyhemoglobin resolved after nitroprusside withdrawal, despite maintaining treatment with NO at the same concentration. Slightly elevated levels of carboxyhemoglobin have also been found in the first hours after several types of surgery [15, 16]. Our four patients had undergone heart transplant, but the carboxyhemoglobin level was initially normal and rose only after nitroprusside administration. Furthermore, none of the patients presented other factors associated with HO-1 induction, such as sepsis, hypoxia, hyperthermia, hemolysis, excessive blood transfusions, and other medications that could induce carboxyhemoglobin elevation. Inmunosuppression was performed with methylprednisolone, azathioprine, and cyclosporine or tacrolimus. Patients also received dopamine, isoprenaline, dobutamine, milrinone, midazolam, fentanyl, cefazidime, and teicoplanin. Any of these drugs can theoretically induce carboxyhemoglobin elevation.

Nitroprusside can interact with oxyhemoglobin, immediately forming methemoglobin and releasing cyanide and NO. Cyanide and NO are rapidly cleared through interaction with sulfhydril groups on proteins and in erythrocytes [1, 4]. The total nitroprusside dose required to generate 10% methemoglobinemia exceeds 10 mg/kg (10 µg kg−1 min−1 for more than 16 h). In our patient 3, who received 16 µg kg−1 min−1 for 12 h, the methemoglobin level was only slightly elevated (2.7%). None of the other three patients presented increased methemoglobin levels despite receiving treatment with NO, nitroprusside, and trimethoprim sulfamethoxazole. The remaining cyanide radicals are converted to thiocyanate via transulfuration within the liver. With higher doses and more prolonged nitroprusside infusions, or when sulfur donors are reduced (malnutrition, surgery, diuretics) cyanide radicals may accumulate and produce clinical cyanide toxicity [1, 4]. Free cyanide radicals can inactivate tissue cytochrome oxidase and prevent oxidative phosphorylation and may thus precipitate tissue anoxia, anaerobic metabolism, and lactic acidosis. Cyanide gradually dissociates from methemoglobin and is converted to thiocyanate. In two of our patients the cyanide and thiocyanate levels were measured without finding toxic levels; however, the blood samples were drawn at the time of the discontinuation of the nitroprusside, thus not coinciding with the maximum dose. Some authors have reported that cyanide levels rapidly decrease after termination of the nitroprusside infusion and are undetectable by 4 h postinfusion [7]. We do not believe excessively high levels of cyanide or thiocyanate were reached in any of our patients as metabolic acidosis did not develop, and there was no rise in lactic acid. In any case the measurement of cyanide in blood is of limited clinical utility due to the time delay before the result is known [1].

We conclude that prolonged treatment with moderate or high doses of sodium nitroprusside can produce carboxyhemoglobinemia in children after heart transplant, with no other signs of toxicity. The cases presented raise the question of whether the specific medical management of children after heart transplant may contribute with a yet unknown factor to explain this phenomenon. However, in the circumstances described, frequent measurement of carboxyhemoglobin should be considered.