Detection of exhaled hydrogen sulphide gas in healthy human volunteers during intravenous administration of sodium sulphide

Abstract

INTRODUCTION

Hydrogen sulphide (H₂S) is an endogenous gaseous signaling molecule and potential therapeutic agent. Emerging studies indicate its therapeutic potential in a variety of cardiovascular diseases and in critical illness. Augmentation of endogenous sulphide concentrations by intravenous administration of sodium sulphide can be used for the delivery of H₂S to the tissues. In the current study, we have measured H₂S concentrations in the exhaled breath of healthy human volunteers subjected to increasing doses sodium sulphide in a human phase I safety and tolerability study.

METHODS

We have measured reactive sulphide in the blood via ex vivo derivatization of sulphide with monobromobimane to form sulphide-dibimane and blood concentrations of thiosulfate (major oxidative metabolite of sulphide) via ion chromatography. We have measured exhaled H₂S concentrations using a custom-made device based on a sulphide gas detector (Interscan).

RESULTS

Administration of IK-1001, a parenteral formulation of Na₂S (0.005–0.20 mg kg⁻¹, i.v., infused over 1 min) induced an elevation of blood sulphide and thiosulfate concentrations over baseline, which was observed within the first 1–5 min following administration of IK-1001 at 0.10 mg kg⁻¹ dose and higher. In all subjects, basal exhaled H₂S was observed to be higher than the ambient concentration of H₂S gas in room air, indicative of on-going endogenous H₂S production in human subjects. Upon intravenous administration of Na₂S, a rapid elevation of exhaled H₂S concentrations was observed. The amount of exhaled H₂S rapidly decreased after discontinuation of the infusion of Na₂S.

CONCLUSION

Exhaled H₂S represents a detectable route of elimination after parenteral administration of Na₂S.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Hydrogen sulphide (H₂S) is a biological mediator and a potential therapeutic agent. In animal studies, the metabolism and pharmacokinetics of H₂S have been characterized.

WHAT THIS STUDY ADDS

• This study is first to demonstrate the pharmacokinetics of an intravenously administered H₂S formulation in humans, and to characterize the exhaled H₂S response in humans.

Introduction

Hydrogen sulphide (H₂S) is a colourless, water-soluble gas with the characteristic smell of rotten eggs. For many decades, H₂S was primarily viewed as a toxic gas and as an environmental hazard [1, 2]. However, recent work has identified H₂S as a gaseous biological mediator that is produced in mammalian species, including humans. H₂S is synthesized endogenously by two pyridoxal-5′-phosphate-dependent enzymes responsible for metabolism of l-cysteine, cystathionine beta-synthase (CBS) and cystathionine gamma-lyase (CSE) [3–7].

H₂S appears to be a mediator of key biological functions including life span and survivability under severely hypoxic conditions [8, 9]. In addition, H₂S, as a vascular relaxant agent, may be a participant in the regulation of cardiovascular function [10–14]. Recent studies have also shown that in many pathophysiological conditions, intravenous administration of H₂S formulations (generally, as sodium sulphide [Na₂S]) may be of therapeutic benefit (overviewed in [5, 15]). For instance, parenteral sulphide administration has been shown to be of therapeutic benefit in various experimental models including myocardial infarction [16–20], acute respiratory distress syndrome [21, 22], liver ischaemia and reperfusion [23], and various forms of inflammation [24–26].

We have recently demonstrated in a rat model that exhalation of H₂S gas can occur when Na₂S, or other H₂S donors are administered intravenously [27]. In the present report we have used IK-1001 (sodium sulphide for injection), a parenteral injectable formulation of H₂S [5, 27], and demonstrated exhaled H₂S gas as a route of elimination in human subjects.

Methods

Human subjects

The research described here was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association, and was approved by the local Ethics Committee of the institution in which the work was performed (Alfred Hospital of Monash University and Nucleus Networks Phase I Trial Centre, Melbourne, Australia). Informed consent was obtained from each subject after full explanation of the purpose, nature and risk of all procedures. During a screening period, a complete medical history was obtained, clinical chemistry, haematologic and serologic tests were completed and a full physical examination was conducted. Subjects who had satisfactorily passed the screening tests and satisfied relevant inclusion/exclusion criteria were eligible for randomization to receive intravenous sodium sulphide (0.005, 0.01, 0.03, 0.06, 0.1, 0.15 or 0.2 mg kg⁻¹) or a volume-matched placebo (0.9% normal saline). Subjects were randomized 3:1 (active : placebo) beginning with enrolment to the 0.01 mg kg⁻¹ cohort.

Following screening and randomization, subjects checked into the facility the evening before the administration of study drug and eligibility to participate in the study was reaffirmed.

Clinical procedures and study drug administration

On the morning of study drug administration, peripheral venous catheters were placed in both arms for study drug administration and blood sampling. The radial artery was cannulated and connected to a pressure transducer to enable continuous and direct monitoring of arterial blood pressure during study drug administration. A board certified anaesthesiologist was present to monitor arterial pressures and to terminate study drug administration if systolic arterial pressure were to decrease by ≥30 mmHg (the protocol-defined criterion).

Subjects were confined to their study beds while study specific physiologic measures, vital signs, blood samples and other data were obtained to establish a pre-dose baseline for each subject. After obtaining baseline parameters and samples, study subjects remained in their beds during study drug administration and for the first 4 h following study drug administration. The indwelling arterial line was removed at approximately 30 min after administration of study drug.

Study drug was administered intravenously by a healthcare professional over exactly 1 min using a syringe pump and an intravenous infusion set. Dosage of study drug was administered on a body weight basis and according to cohort assignment. Where needed, the study drug was administered after dilution with saline, in order to keep the volume of the test article administered between 1 and 5 ml throughout the study. Study drug dosage, volumes, dilution factors and infusion rates were all determined according to prespecified instructions supplied to the research pharmacist assigned to the study.

Arterial pressure during study drug administration

Beginning 15 min prior to the administration of study drug and continuing until 30 min after the dose administration, arterial blood pressure was continually monitored and recorded by the attending anaesthesiologist.

Measurement of blood concentration of hydrogen sulphide by monobromobimane derivatization

Concentrations of reactive sulphide in freshly drawn blood samples (pre-dose, 1, 5, 15, 30 and 45 min) were measured using a bedside sulphide derivatization method involving the reaction of sulphide with monobromobimane to the fluorescent product sulphide dibimane, which was extracted from blood, separated by HPLC and quantified by its fluorescence against an external sulphide dibimane calibration standard [19, 28]. The method to quantify sulphide dibimane in clinical human blood samples was validated and performed by the analytical laboratory ITR (Montreal, Canada) under GLP. The CV of the precision for inter- and intra-assay variability was within ±15% for quality control samples above the LLOQ and within ±20% for the LLOQ quality control sample. Accuracy was between 85 and 115% for quality control samples above and between 80 and 120% for the LLOQ quality control sample. LLOQ of the assay was 0.166 µg ml⁻¹ sulphide dibimane, which corresponds to a sodium sulphide concentration of 31 ng ml⁻¹.

Measurement of blood thiosulfate by ion chromatography

Concentrations of thiosulfate, a major oxidative metabolite of sulphide, were determined in freshly drawn blood samples (pre-dose, 1, 5, 15, 30 and 45 min) by analysis of centrifuge-derived blood plasma. Plasma samples were subsequently filtered (10 000 MWCO) and stored frozen until analyzed by ion chromatography.

Monitoring exhaled hydrogen sulphide (H₂S)

During the 5 min prior to delivery of study drug, the 1 min during study drug administration, and the 5 min following administration of study drug (11 min total), subjects wore a face mask connected to numerous components (gas mixing chamber, airflow sensing pneumotachometer transducer to measure airway flow) of a human respiratory physiology system (AD Instruments, Colorado Springs, CO, USA). Exhaled air was directed to the sampling ports of two Interscan RM-17 sulphide detectors (Interscan, Chatsworth, CA). The two sulphide detectors contained two different sensing elements with overlapping ranges to establish a broader analytical range (10 ppb–5 ppm) between the two instruments than with either instrument alone. PowerLab analog-to-digital converters (AD Instruments) were used to record and archive real time data to computer. Chart5 Pro software (AD Instruments) was used to process data further to generate summary parameters and values to quantify exhaled sulphide measurements.

Human subject assessment of odour

The human olfactory detection threshold for sulphide is estimated to be in the 10–50 ppb range [29, 30]. Therefore, at the design of the study, the possibility was considered that if exhaled sulphide were present in sufficient quantity, then subjects would detect odour emanating from within as exhaled air crossed their nasal passages. Therefore, a human subject odour assessment protocol was incorporated into the study design. Within the first hour following administration of study drug, subjects were asked whether they detected any unusual odour during the time that the study drug was being administered. For those subjects answering in the affirmative, they were then presented with an odour assessment instrument to aid in the identification of the odour by broad categories (floral, medicinal, fruity, chemical, etc.) and more specifically to highly distinct odours within these categories (vanilla, menthol, orange, turpentine). Under the ‘offensive’ odours category, the testing instrument included the specific term ‘rotten eggs’ as one of the 15 specific odours named in this category. Subjects were also asked to rate their perceived intensity of the odours on a visual analogue scale.

Statistical analysis

All values are reported as group mean ± standard deviation (SD) unless noted otherwise. Results obtained from placebo recipients were combined to create a single placebo group for comparison with active cohorts, which consisted of n= 13 subjects across the seven cohorts who received normal saline. Probability values of <0.05 were considered statistically significant.

Results

Subject enrolment

A total of 52 subjects were enrolled in the study. Cohort sizes were planned for eight subjects in each cohort (six active, two placebo). However, the initial starting cohort (0.005 mg kg⁻¹) started at half of the planned normal cohort size (three active, one placebo). At the completion of the study a total of 13 subjects had received placebo across the escalating active cohorts and they have been combined into a single placebo group (n= 13) for analysis purposes. Three subjects received active therapy (IK-1001) at 0.005 mg kg⁻¹ in the starting cohort. Thereafter, six subjects received active therapy (IK-1001) in each of the 0.01, 0.03, 0.06, 0.1, 0.15 and 0.20 mg kg⁻¹ cohorts.

Vital signs, ECG, clinical chemistry, haematology, coagulation and arterial blood pressure

Administration of active study drug or placebo did not provoke any acute changes in vital signs (heart rate, blood pressure, respiration, temperature), the electrocardiogram, clinical chemistry, haematology and coagulation function. During the course of the studies, there were no documented cases of an abrupt change in arterial blood pressure, or any other conditions that required an unscheduled cessation of IK-1001 dose administration.

Arterial blood pressure and heart rate data for the 30 min preceding and 30 min following study drug administration are presented in Figures 1 and 2. Systolic and diastolic pressures were recorded at 1 min intervals in the timeframe immediately before and after administration of drug and mean arterial pressure was calculated as 1/3 systolic + 2/3 diastolic. There were no clinically significant acute changes in arterial pressure (systolic, diastolic or mean) or heart rate with any active dose of IK-1001 administered or with the administration of placebo.

Figure 1.

Direct measure of radial artery pressure during a 1-min delivery of placebo, or the three highest doses of IK-1001 used (0.10, 0.15 and 0.20 mg kg⁻¹). Blood pressure values were taken at 30 and 15 min prior to delivery of study drug (baselines) and recorded each minute for the 10 min preceding and following administration of study drug. At 15 and 30 min following administration of study drug, follow-up blood pressure readings were recorded before removal of the arterial cannula. Systolic, diastolic and mean arterial pressures are shown for the placebo group (n= 13) (panel A) and for the highest three doses (0.10, 0.15 and 0.20 mg kg⁻¹; n= 6 each) of IK-1001 (panels B, C and D, respectively). Data are group means ± SD. (A) SYS placebo (); DIAS placebo (); MEAN placebo (); (B) SYS 0.1 (); DIAS 0.1 (); MEAN 0.1 (); (C) SYS 0.15 (); DIAS 0.15 (); MEAN 0.15 (); (D) SYS 0.2 (); DIAS 0.2 (); MEAN 0.2 ()

Figure 2.

Heart rate during 1 min delivery of the three highest doses (0.10, 0.15 and 0.20 mg kg⁻¹) of IK-1001 or placebo. Values taken at 30 and 15 min prior to delivery of study drug (baseline) and recorded each minute for the 10 min preceding and following administration of study drug. At 15 and 30 min following administration of study drug, a follow-up value for heart rate was recorded. Heart rate is shown for the placebo group (closed symbols, n= 13) and the top three doses (0.10, 0.15 and 0.20 mg kg⁻¹) of IK-1001 (open symbols, n= 6 each). Data are group means ± SD. placebo (); 0.01 (); 0.15 (); 0.20 ()

Effect of IK-1001 infusion on plasma concentrations of sulphide

Given the relatively short half-life of sulphide in the bloodstream and the dosages administered, IK-1001 was not expected to produces changes in biologically available sulphide in any sample other than the 1 min blood sample which coincided with the end of administration of study drug. As shown in Table 1, the administration of placebo did not alter the concentration of blood sulphide as determined by the monobomobimane derivitization method. IK-1001 administered at 0.005 and 0.01 mg kg⁻¹ did not alter blood sulphide values. With IK-1001 administered at 0.03 and 0.06 mg kg⁻¹, blood sulphide concentrations were nominally elevated, but with only marginal significance or a trend toward significance. In contrast, IK-1001 at 0.10 and 0.20 mg kg⁻¹ clearly elevated the blood sulphide by 86% and 150%, respectively, at the 1 min time point in comparison with the pre-dose sample. By the 5 min time point in both of these cohorts, blood sulphide concentrations had declined greatly from the peak observed at 1 min.

Table 1.

Blood concentrationsof sodium sulfide (ng ml⁻¹) by cohort in subjects receiving IK-1001 or placebo

Cohort	n	Pre-dose	1 min	5 min	15 min	Pre-dose vs. 1 min comparison
Placebo	11	41.3 ± 15.8	42.4 ± 17.4	40.3 ± 11.8	52.5 ± 18.5	NS
IK-1001 0.005 mg kg−1	3	23.6 ± 1.3	24.8 ± 2.2	96.6 ± 102.4	24.9 ± 3.1	NS
IK-1001 0.01 mg kg−1	6	57.0 ± 15.3	59.5 ± 18.7	53.0 ± 16.3	57.4 ± 15.3	NS
IK-1001 0.03 mg kg−1	6	39.1 ± 17.1	59.3 ± 28.1	46.6 ± 13.0	55.2 ± 20.3	P= 0.051
IK-1001 0.06 mg kg−1	6	52.4 ± 20.5	60.2 ± 21.0	44.1 ± 10.6	47.4 ± 6.5	P= 0.203
IK-1001 0.10 mg kg−1	6	31.9 ± 9.0	59.3 ± 21.3	43.4 ± 13.4	40.2 ± 6.8	P= 0.0013
IK-1001 0.20 mg kg−1	6	42.5 ± 8.9	106.2 ± 54.1	60.5 ± 14.5	48.5 ± 8.6	P= 0.0004

Data are group means ± SD with number of patient samples available for analysis from each cohort. Due to loss of samples during processing, there are no data available for the 0.15 mg kg⁻¹ cohort. Within each cohort one-way anova was used and where overall significance was found (P < 0.05) pairwise comparisons of time points were made using Student's t-test.

Effect of IK-1001 infusion on plasma concentrations of thiosulfate

The i.v. administration of IK-1001 also produced a discernable effect on blood thiosulfate. At the 1 min time point, blood thiosulfate in the 0.10, 0.15 and 0.20 mg kg⁻¹ cohorts was nominally elevated over the pre-dose time point, but did not reach statistical significance (P values of 0.13, 0.36 and 0.07, respectively). As shown in Figure 3, blood thiosulfate was elevated at the 5 min time point when compared with the pre-dose time points in the 0.10, 0.15 and 0.20 mg kg⁻¹ cohorts, an effect which was statistically significant (P < 0.05) at the 5 min time point in the 0.10 and 0.20 mg kg⁻¹ groups and marginally significant (P= 0.08) in the 0.15 mg kg⁻¹ group. At the 15 min time point and beyond, blood thiosulphide had returned to concentrations comparable with those at the pre-dose time point.

Figure 3.

Blood concentrations of thiosulfate following intravenous administration of IK-1001 or placebo. Thiosulfate data available for n= 11 in placebo group; IK-1001 at 0.10, 0.15 and 0.20 mg kg⁻¹ cohorts are n= 6 each. * is P < 0.05 for 5 min time point vs. pre-dose (time = 0) for the 0.10 and 0.20 mg kg⁻¹ groups. For the same comparison in 0.15 mg kg⁻¹ group, P= 0.08. placebo (); 0.01 (); 0.15 (); 0.20 ()

Effect of IK-1001 i.v. administration on exhalation of H₂S gas

In the two highest dose cohorts, the study protocol incorporated an assessment of exhaled sulphide. Figure 4 depicts 4 min segments of raw data obtained from two subjects (one active and one placebo) in the 0.15 mg kg⁻¹ cohort. The data in panel A show sulphide concentrations detected by the monitor in air sampled at the subject's face mask. The data in panel B is the simultaneous readout of air sampled from a 3 l expired air mixing chamber. The 3 l expired air mixing chamber has the effect of mixing and averaging expired air over several respiratory cycles. In both subjects, the detection of basal quantities of exhaled H₂S gas in humans becomes evident when the face mask was first affixed to the subject. The sulphide detectors reflected an abrupt change when the electrochemical detector responded to a sample stream that abruptly transitions from ambient room air to the first samples being exhaled by the study subject. Upon the intravenous administration of IK-1001 (0.15 mg kg⁻¹ over 1 min), a sizable increase in exhaled sulphide was clearly demonstrated from samples taken at either the facemask (showing the oscillatory pattern occurring in each respiratory cycle) or the expired gas mixing chamber.

Figure 4.

A) Raw data waveforms recorded from two individuals where exhaled sulphide was measured from the subject's face mask assembly. Sampling intervals include the time interval where the sulphide detector was not connected to the subject and drawing in ambient air (room air), when the face mask was first attached to the subject (affix mask) and then during the 1 min administration of study drug. B) Raw data waveforms recorded from two individuals where exhaled sulphide was measured from the exhaled gas mixing chamber. Sampling intervals include a time interval where the sulphide detector was not connected to the subject and drawing in ambient air (room air), when the face mask was first attached to the subject (affix mask) and then during the 1 min administration of study drug. IK-1001 (); placebo ()

In Figure 5, exhaled sulphide values are reported for 14 subjects in the 0.15 and 0.20 mg kg⁻¹ cohorts during the time when the face mask was first affixed to the subject. Two subjects (one placebo, one active) in the 0.20 mg kg⁻¹ cohort were excluded from analysis due to improper connection of a one-way airflow value which allowed exhaled air from the subject to be exhausted to room air instead of being directed to the sulphide monitor. In all subjects with valid recordings there was a clearly distinguishable increase in the measured sulphide concentration when the sample stream being directed to the sulphide monitor abruptly transitions from ambient room air (indicated H₂S concentration of 6.4 ± 2.4 ppb) to the first samples being exhaled by the study subject (26.9 ± 4.6 ppb, P < 0.01 for paired comparison). This is clear evidence that normal human subjects exhale a measurable quantity of hydrogen sulphide.

Figure 5.

Exhaled sulphide from 14 subjects in the 0.15 and 0.20 mg kg⁻¹ cohort combined. Values are hydrogen sulphide detected prior to the subject wearing the face mask (room air) and then exhaled sulphide values recorded when the subject wore the face mask assembly (affix mask). Open bars are subjects who would go on to receive placebo; filled bars are subjects who would go on to receive active IK-1001. Data are individual values from each subject

Figure 6 depicts the change in exhaled sulphide from the subjects during their own baseline period (when their endogenous quantity of exhaled sulphide was being monitored) and through the 5 min interval immediately following infusion of IK-1001 or placebo. Exhaled air sampled from the gas mixing chamber was used to compute integrated areas of exhaled sulphide by time yielding an area-under-the-curve for each time interval indicated. The 1 min infusion of placebo did not produce any significant change in the measured exhaled sulphide in the subsequent 5 min period where exhaled sulphide was being monitored. In the subjects receiving IK-1001, the exhaled sulphide AUCs rose two–three times over baseline levels in the first minutes following the 1 min infusion of IK-1001 and had nearly returned to baseline within 5 min. Despite the rapid dissipation of exhaled sulphide that can be observed when sampling real-time at the patient's face mask (Figure 4A), the apparent exhaled sulphide in the mixing chamber was cleared more slowly as residual sulphide present in the mixing chamber was purged by the ongoing passage of exhaled air from the subject after the administration of study test article had ceased.

Figure 6.

Mean increase in average exhaled sulphide in placebo (n= 3) and IK-1001 recipients in the 0.15 (n= 6) and 0.20 mg kg⁻¹ (n= 5) cohorts. Exhaled sulphide recordings were processed to generate area under the curve (AUC) values in 1 min intervals and are reported in units of ppb × s. Data are group means ± SD. placebo (□); 0.15 mg kg⁻¹ (); 0.20 mg kg⁻¹ ()

Subjective responses to the odour scale

In the placebo cohort and IK-1001 cohorts up to 0.03 mg kg⁻¹, it was generally rare (≤33%) for a subject to report having detected an any odour at all during the time in which they received study medication (Figure 7A). Beginning with the 0.06 mg kg⁻¹ cohort, the majority of subjects (≥50% of each cohort) reported the detection of an odour during the time at which they received intravenous IK-1001. Among the 24 subjects enrolled into the 0.06 through 0.20 mg kg⁻¹ cohorts who received IK-1001, 18 (67%) reported detection of an odour during the time of IK-1001 administration. Of these same 18 subjects, 12 of them (67%) correctly and specifically identified this odour as ‘rotten eggs.’ These results suggest that the majority of subjects at the 0.06 mg kg⁻¹ cohort and higher can qualitatively identify an odour that they detect when receiving intravenous IK-1001. There were three instances in the 0.15 mg kg⁻¹ cohort where subjects had reported an odour but fell short of correctly identifying ‘rotten eggs’ as the specific odour. In one instance the subject correctly identified the ‘offensive’ category but named ‘raw meat’ as the specific odour. Another subject correctly identified the ‘offensive’ category of the odour testing instrument but neglected to name a specific odour within that category. A third subject identified the ‘chemical’ category of the odour test and selected ‘sulphur’ of 18 possibilities as the specific odour. It is possible that these instances are cases where, in actuality, the subject experienced an olfactory sensation that was anticipated to occur but was inadequately captured or characterized by the selected testing instrument. To the extent that these aforementioned subject perceptions were correct, the specific identification of ‘rotten eggs’ might have been in greater proportions that we have reported.

Figure 7.

A) Proportion of subjects within each cohort who reported a sensation of odour during the time of study drug administration. Cohort sizes are n= 13 for placebo, n= 3 for 0.005 mg kg⁻¹ cohort and n= 6 for all cohorts thereafter. B) Self-reported assessment (visual analogie scale) of odour intensity in the subjects who identified ‘rotten eggs’ as the specific odour. Number of respondents are n= 1 (placebo), n= 1 (0.03 mg kg⁻¹), n= 3 (0.06 mg kg⁻¹), n= 5 (0.10 mg kg⁻¹), n= 1 (0.15 mg kg⁻¹) and n= 3 (0.20 mg kg⁻¹). Data are individual reported values or group means ± SD (where number of observations permit)

In order to add a semi-quantitative measure to the subject odour assessment, those subjects correctly reporting detection of an odour were asked to rank the perceived intensity of the odour on a visual analogue scale. Figure 7B depicts the mean VAS score from subjects correctly identifying ‘rotten eggs’ as the specific odour. In the placebo cohort, a single individual (of 13) reported rotten eggs with a VAS score of 25. No subjects in the 0.005 and 0.01 mg kg⁻¹ cohorts correctly identified ‘rotten eggs’ as the specific odour and no VAS scores were reported. In the 0.03 mg kg⁻¹ cohort, a single individual reported ‘rotten eggs’ with a VAS score of 50. Beginning with the 0.06 mg kg⁻¹ cohort, subjects correctly reported ‘rotten eggs’ with greater incidence, allowing arithmetic means to be calculated for VAS scores. However, the group mean VAS did not appear to increase as the administered dose increased approximately three-fold from 0.06 to 0.20 mg kg⁻¹.

Discussion

In a recent study, we demonstrated in a rat model that detectable amounts of exhaled H₂S can be measured in response to intravenously administered sodium sulphide [27]. We also provided evidence for the pharmacological modulation of this response; for instance, the amount of exhaled H₂S was found to be significantly increased when endogenous nitric oxide synthesis was inhibited with the NO synthase inhibitor N^G-nitro-L-argentine-methyl ester (L-NAME). We noted a significant baseline production of H₂S in the rats, and cysteine, the amino acid precursor of endogenous H₂S biosynthesis was found to give a slight, but detectable increase in the exhaled H₂S concentration [27].

The current results extend these observations into humans. At the doses used in the current study (up to 0.2 mg kg⁻¹ administered over 1 min), no adverse effects were noted. Even though H₂S has been reported to exert vasodilator and anti-hypertensive effects [3, 10, 13, 14, 31, 32], at the current doses no statistically significant haemodynamic alterations were noted, despite sulphide being available instantaneously here in an inorganic form, whereas sulphide release from organic sulphide donors is presumably more gradual. In the human subjects, significant levels of H₂S production were noted under baseline conditions, and there was a significant increase in exhaled sulphide concentrations when a brief (1 min) infusion of sodium sulphide was performed. Following the decline of the exhaled H₂S signal over time after the discontinuation of the short infusion indicated a short half-life (<5 min) of H₂S in humans after the short bolus infusions of sodium sulphide. The characteristic odour of H₂S was also noticed by the human volunteers during and shortly after the 1 min of the infusion of IK-1001. They typically pointed out the ‘eggy’ smell on the odour chart used in the study. It is important to keep in mind that the human nose detects H₂S with sensitivity in the parts per billion range [29, 30].

From the comparison of the exhaled concentrations of H₂S gas and the total amount of sulphide molecules administered parenterally, it is clear that exhaled H₂S is a minor (<1%) route of elimination in humans. However, under our experimental conditions, the method for measurement of exhaled H₂S appeared to be as sensitive as the method for measurement of plasma concentrations of sulphide, or its major metabolite, thiosulfate. Thiosulfate is produced from H₂S via a series of reactions involving mitochondrial enzymes [33]. The observation that thiosulfate shows a good correlation with IK-1001 exposure appears to be in line with prior studies demonstrating that thiosulfate is a useful marker of H₂S exposure in humans [34–36]. However, it should be noted that while exhaled sulphide is able to measure reactive sulphide in real-time (i.e. during the time window of its administration), the measurement of sulphide and thiosulfate requires processing, sample transfer and analytical methods.

In recent years, multiple lines of investigations have suggested that parenteral administration of hydrogen sulphide formulations (see: Introduction), H₂S-releasing prodrugs [37, 38], natural polysulphides that produce H₂S upon reaction with intracellular glutathione [39, 40] or molecules that combine a known pharmacological entity with a sulphide-releasing moiety [15, 24, 41–43] may have significant therapeutic potential for the therapy of various cardiovascular and inflammatory diseases and various forms of critical illness. Based on the current findings, we speculate that measurement of exhaled sulphide may serve as a potential safety marker for future clinical trials involving sulphide and sulphide-releasing compounds.

Hydrogen sulphide has previously been shown to play an important role in the regulation of vascular tone and blood pressure in multiple studies. However, in the current study, no changes in haemodynamic parameters were observed. The difference may be due to the fact that the current doses (up to 0.2 mg kg⁻¹) and duration of administration (1 min) were lower and shorter than the doses and durations in previously published preclinical studies. In rats, dose of 1.5 mg kg⁻¹ min⁻¹ caused a fall in blood pressure [44] (i.e. a dose that was more than seven times higher than the top dose used currently). On the other hand, a lower dose of hydrogen sulphide (0.6 mg kg⁻¹ min⁻¹) in rats did not cause a hypotensive effect, and in fact it slightly elevated blood pressure [44]. This latter effect was attributed to the quenching of endogenous nitric oxide by exogenously applied hydrogen sulphide in the vasculature [44]. It is important to point out that blood pressure regulation by hydrogen sulphide is a complex phenomenon, which also appears to involve central regulatory mechanisms [45].

How do the currently tested doses of hydrogen sulphide compare with the previously reported, therapeutically effective doses of hydrogen sulphide? In studies in mice, doses of 0.05–0.5 mg kg⁻¹, administered in a single, intravenous bolus, were shown to have a marked beneficial effect (reduction in infarct size, reduction of neutrophil infiltration, changes in inflammatory mediator production) in models of myocardial infarction [16] and myocardial preconditioning [20]. In a model of cardiac arrest in mice, a single intravenous dose of 0.55 mg kg⁻¹ was found to be effective [46]. In a model of liver ischaemia and reperfusion, single intravenous bolus doses of 0.3–1 mg kg⁻¹ were therapeutically effective [23]. Although one must exercise caution before drawing a direct comparison between therapeutic doses in mice (where the doses of sulphide were injected in a rapid bolus fashion) with controlled, 1 min infusion administrations in humans (i.e. the current study), one may conclude that all of the above doses, in general, are in a comparable range. One must also point out that multiple animal studies, especially the ones using large animal models, have been more successful when utilizing infusion regimens of hydrogen sulphide administration, as opposed to single bolus doses (e.g. [18, 19, 22, 47]). Clearly, additional work is needed, both preclinically and clinically, to establish the optimal dosing regimens for therapeutic hydrogen sulphide administration.

The finding that significant basal production of H₂S was noted in humans is worthy of additional discussion. The sources of this H₂S may be multiple, including the oral cavity [48–51], the intestinal bacterial flora [52, 53], as well as endogenous sulphide-generating enzymes (CSE and CGS) that are expressed in many cells and tissues of the human body [4, 5]. The determination of the relative contribution of bacterial vs. mammalian sources to the production of basal H₂S production in humans remains to be determined in future studies. It is important to point out that several disease conditions, including systemic inflammation or circulatory shock and certain local inflammatory diseases such as pancreatitis have been shown to be associated with the up-regulation of H₂S-producing enzymes [54–56]. According to a recent study, this up-regulation can be blocked by pretreatment with glucocorticoids [55]. It remains to be tested in future studies whether various pathophysiological conditions are associated with an up-regulation of endogenous H₂S production, detectable by the measurement of exhaled H₂S concentration. In this context, it is noteworthy that a recent study has reported an increase in exhaled H₂S in patients suffering from chronic pancreatitis [57], but not in lung transplant patients in the acute rejection phase [58]. Other volatile sulphur compounds have been detected in exhaled breath from patients with cystic fibrosis [59]. Along these lines, it will be interesting to evaluate in the future whether measurement of exhaled H₂S may also have potential diagnostic uses.

Competing interests

The authors are employees of Ikaria, a for-profit organization involved in the clinical development of IK-1001.