Comparative effects of glyceryl trinitrate and amyl nitrite on pulse wave reflection and augmentation index


Professor D. J. Webb, Clinical Research Centre, The University of Edinburgh, Western General Hospital, Edinburgh, EH4 2XU, UK. Tel.: + 44 (0)131 537 2006 Fax: + 44 (0)870 134 0897 E-mail:



The influence of vasodilators on augmentation index (AIx) offers a simple, rapid and noninvasive method of evaluating vascular function. Glyceryl trinitrate (GTN) is widely used as an endothelium-independent vasodilator, although other nitrates that are shorter acting may have advantages in clinical studies. The aim of this study was to compare the effects of two short-acting nitrates, GTN and amyl nitrite, which have differing pharmacodynamic profiles.


Twenty-one healthy volunteers (15 male; mean age 35 years, range 21–56 years) attended on three occasions and received sublingual GTN (0.5 mg for 3 min), inhaled amyl nitrite (0.2 ml inhaled for 30 s), or no treatment in a randomized cross-over design. Haemodynamic responses of AIx, blood pressure and thoracic bioimpedance (heart rate, cardiac index) were assessed by measurement at baseline, every 60 s for the first 5 min, and then every 5 min for a further 55 min.


AIx was reduced by amyl nitrite (peak effect −9 ± 2% at 1 min, P < 0.002) and GTN (peak effect −12 ± 3% at 4 min, P < 0.05). Compared with amyl nitrite, the onset and offset of action of GTN was slower. Amyl nitrite initially increased heart rate by 27 ± 4% (P < 0.001) and cardiac index by 13 ± 3% (P < 0.001) whereas GTN had no significant effect (P > 0.05). Neither agent affected blood pressure.


GTN causes a slower and more sustained reduction in AIx than amyl nitrite. Although amyl nitrite causes a more rapid fall and recovery in AIx, it induces a reflex tachycardia that may limit interpretation of the initial (1 min) but not later (2 min) changes in AIx. The prolonged offset of GTN suggests that a sufficient washout period must be included when making repeated measures or when assessing the subsequent effects of other agents.


Endothelial dysfunction is an early feature of vascular disease and predicts outcome in patients with established coronary artery disease [1–3]. Currently available techniques for the in vivo assessment of endothelial dysfunction have limitations that restrict their use in large-scale clinical trials or clinical practice [4]. These techniques (particularly, flow-mediated dilatation) are technically demanding, require a high level of expertise, and are time-consuming to perform [5]. Furthermore, flow-mediated dilatation [6] requires the use of an occlusive cuff that some individuals may find uncomfortable, while intra-arterial [7] and intracoronary [8] infusions of acetylcholine are invasive and not without risk. The development of a noninvasive, quick and easy-to-learn method could allow the measurement of endothelial function to be incorporated into large-scale trials, and eventually into clinical practice [5].

Systolic pulse contour analysis is a technique which uses applanation tonometry to assess the arterial pressure waveform [5]. At any point, the pressure within an artery is a composite of the forward ejected wave and reflected waves travelling back to the heart. In healthy compliant arteries, these reflected waves travel slowly and reach the heart in diastole, thus enhancing coronary perfusion. If the vasculature is stiffened, such as in old age, hypertension or hypercholesterolaemia, the reflected wave travels more rapidly and reaches the heart in systole, augmenting systolic pressure and increasing left ventricular afterload [9]. Systemic arterial stiffness, increasingly recognized as a useful marker of cardiovascular risk [10], may be measured using augmentation index (AIx), defined as the difference between the first and second peaks of the central arterial waveform (Figure 1) (ΔP) expressed as a percentage of the total pulse pressure (PP).

Figure 1.

Schematic of the effect of arterial stiffness on the peripheral wave form. P1 represents the initial systolic pressure wave, travelling from the heart to the periphery. In compliant large arteries (as in healthy young subjects, Figure 1a), P2, composite of the forward wave and reflected pressure waves, arrive back at the central aorta in diastole, augmenting diastolic blood pressure and coronary filling. In stiff arteries (e.g. in age, with increasing cardiovascular risk, Figure 1b), wave reflection occurs earlier and thus the systolic peak is augmented. Augmentation index is calculated as the difference between P1 and P2 expressed as a percentage of the pulse pressure (PP)

AIx decreases in response to vasodilators by alteration of smooth muscle tone, independently of effects on pulse wave velocity [11], and we and others have previously demonstrated that AIx falls following administration of inhaled salbutamol [12, 13]. Consistent with its endothelium-dependent actions [14], the AIx response to inhaled salbutamol is attenuated by l-NMMA, a nitric oxide inhibitor, and reduced in patients with hypercholesterolaemia and type II diabetes mellitus, conditions known to be associated with endothelial dysfunction [15, 16]. These results demonstrate that this is a feasible technique for assessing endothelial function.

GTN is widely used as an endothelium-independent vasodilator control, and indeed causes a fall in AIx (Figure 1c) which is unaffected by l-NMMA [12]. Like inhaled salbutamol, GTN can be easily and safely administered to a wide range of individuals. However, while GTN is a short-acting drug, its effect on AIx may last at least 20 min and this may limit its clinical utility when assessing vascular function, particularly if repeated measurements are required. An alternative endothelium-independent nitric oxide donor is amyl nitrite. Originally used as an antianginal therapy [17], its action on altering the pulse waveform has been recognized since the 19th century [18]. Amyl nitrite is an ultra short-acting drug and thus may be a more suitable agent for use in acute vascular studies, allowing for a shorter protocol with no carry-over effects.

The present study was designed to characterize and compare the effects of GTN and amyl nitrite on AIx and on systemic haemodynamics. We hypothesized that a dose of amyl nitrite with a similar maximal effect on AIx to that of GTN would have a shorter onset and offset of action.



Twenty-one healthy volunteers participated in the study, which was approved by the local Research Ethics Committee. Written informed consent was obtained from each participant. None of the subjects was taking regular medications, and all avoided any medication for 1 week before each study. All subjects abstained from alcohol for 24 h and from caffeine and tobacco for at least 12 h before each study.


Radial arterial waveforms were recorded in the dominant arm using a high fidelity micro manometer (SPC-301; Millar Instruments, Houston, Texas, USA) as previously described [19]. Pulse wave analysis (AtCor Medical, West Ryde, Australia) was used to generate a corresponding central waveform using a validated transfer function [20]. Non-invasive blood pressure measurements were made in the nondominant arm using a validated oscillometric technique (HEM-705CP, Omron) [21]. Cardiac index and heart rate were measured noninvasively by transthoracic electrical bioimpedance (NCCOM3-R7, BoMed, Irvine, California, USA) as described previously [22].

Study protocol

The study was conducted over three visits, each separated by at least 48 h. Subjects rested recumbent in a quiet room maintained at a constant temperature of 22–24 °C for 30 min to allow haemodynamic stabilization. Following an initial dose-ranging pilot study to ascertain doses of GTN and amyl nitrite which would produce comparable reductions in AIx (data not shown), subjects then received either GTN (Cox Pharmaceuticals, Barnstaple, UK; 0.5 mg sublingual tablet removed after 3 min [19]), amyl nitrite (Norton Healthcare, London, UK; 30 s inhalation from a 0.2 ml vitrella), or no treatment, in a randomized order. A normal resting pattern of respiration was continued during inhalation of amyl nitrite. Haemodynamic responses to treatment were assessed every 60 s for 5 min, the shortest time interval that could be achieved in practice, and then every 5 min for a further 55 min.

Data acquisition and statistical analysis

All measurements were made in duplicate and the mean calculated. Results are expressed as mean percentage change from baseline ± SEM. Results were analyzed using analysis of variance (anova) and Student's t-test where appropriate. Significance was taken at the 5% level.


Baseline variables

Subjects were predominantly male with no risk factors for vascular disease other than smoking (Table 1). There were no differences in baseline variables between study visits (P > 0.6; data not shown). Measurements of augmentation index, heart rate, blood pressure and cardiac index did not vary significantly when subjects received no treatment (all P > 0.7; data not shown).

Table 1.  Baseline subject characteristics
ParameterMean (SD)
  • *

    number (percentage of subjects).

 Sex15M : 6F
 Age (years)35 ± 13 (range 21–56)
 BMI (kg m−2)25 (4)
Risk factors
 Total cholesterol (mmol l−1)4.8 (0.8)
 Glucose (mmol l−1)5.1 (0.5)
 Creatinine (µmol l−1)79 (18)
 Smokers2 (10)*
 Augmentation index (%)2 (16)
 Heart rate (beats min−1)68 (8)
 Mean arterial blood   pressure (mmHg)86 (8)
 Cardiac index (l min−1)4.2 (1.8)

Pulse wave analysis

AIx was reduced to a similar extent by amyl nitrite (maximum effect −9 ± 2% at 1 min; P < 0.002) and GTN (maximum effect −12 ± 3% at 4 min; P < 0.05) (Figure 2a). This effect was seen earlier for amyl nitrite, where AIx was reduced for the first 5 min (P < 0.05, anova), and the maximum response was seen at 1 min, with approximately 85% of subjects responding maximally within 2 min. No reduction in AIx to GTN was seen until 2 min (P = 0.004), reaching a maximum at 4 min (P = 0.0011), a response that persisted for 25 min (P = 0.021). The timing of maximum response to GTN was more variable, although maximum response was seen within 10 min for approximately 85% of subjects.

Figure 2.

Effect of amyl nitrite (closed triangle), GTN (closed square) and placebo (open rhomboid) on (a) augmentation index (b) heart rate (c) cardiac index and (d) mean arterial blood pressure. Placebo (◊), GTN (▪), Amyl Nitrite (▴)


Amyl nitrite increased heart rate by a maximum of 27% ± 4% (approximately 20 beats min−1; P < 0.001). GTN was associated with a maximum 8.4% rise in heart rate (approximately 5 beats min−1), but this failed to reach statistical significance (anova, P = 0.06) (Figure 2b).

Amyl nitrite initially increased cardiac index by 13 ± 3% (P < 0.001) whereas GTN had no significant effect (P > 0.05) (Figure 2c). Subsequently, both agents caused a reduction in cardiac index (P < 0.001).

Neither amyl nitrite nor GTN caused a significant change in mean arterial blood pressure (anova; P > 0.05).


We have shown here that, when using nitric oxide-donating vasodilators as a method of reducing augmentation index, amyl nitrite has a more rapid onset and offset of action than GTN. However, the utility of amyl nitrite is hindered by the associated induction of a reflex tachycardia and increase in cardiac output that may limit interpretation of the response.

Augmentation index and haemodynamics

The initial marked increase in cardiac index in association with limited change in blood pressure following administration of amyl nitrite reflects an early decrease in peripheral vascular resistance. The associated increase in heart rate is likely to be a reflex compensatory mechanism in response to this fall in peripheral resistance, rather than a direct cardiac effect of amyl nitrite [23]. This tachycardia reduces the duration of systole, causing the reflected wave to return later in the cardiac cycle and arrive in diastole, thereby lowering measurements of systolic augmentation. This theory has been borne out in clinical studies that have demonstrated an inverse linear relationship between augmentation index and heart rate in both older [24] and younger [25] patients undergoing cardiac pacing. AIx is reduced by around 4–5% for each 10 beats min−1 increment in heart rate [24, 25]. In the current study, the mean heart rate rose by almost 20 beats min−1 following amyl nitrite, and this alone may account for the initial 9% fall in AIx. However, by 2 min the heart rate had returned to baseline and could no longer account for the reduction in AIx thereafter. The interpretation of the initial peak response is clearly limited by the attendant increase in both heart rate and cardiac output. If amyl nitrite were to be used as an endothelium-independent control agent in the context of a technique using pulse wave analysis, it may be that measurements should be confined to this time (2 min postdose) when AIx is still markedly reduced, but heart rate has returned to baseline. In contrast, the 12% reduction in AIx in response to GTN cannot be explained solely by the smaller, here nonsignificant, 5 beats min−1 increase in heart rate.

The subsequent delayed fall in cardiac index seen after both amyl nitrite and GTN, but not on the control visit, is possibly a manifestation of venodilatation caused by the vasoactive agents, resulting in reduced preload in these healthy supine subjects.

Potential limitations of the study

While the magnitude of reduction in AIx was similar between amyl nitrite and GTN, it is possible that amyl nitrite actually had a greater effect given that the first measurements were taken at 1 min and that this was the maximum recorded effect. Therefore, an earlier effect could potentially have been missed, and may explain the reflex increases in heart rate and cardiac output seen following amyl nitrite administration. On the other hand, earlier work suggests that the peak action of amyl nitrite on forearm blood flow occurs at 75 s [26], implying that the current study did not miss the maximum effect.

Systemic haemodynamic variables were measured using the noninvasive technique of bioimpedance. Some studies have previously demonstrated a poor correlation between this technique and continuous thermodilution measurement, the gold standard method of assessing cardiac output [27]. However, a meta-analysis of 154 studies comparing thoracic bioimpedance with a reference method found that, while the correlation between these methods may be suboptimal for absolute values, thoracic bioimpedance is a useful method for trend analysis [28]. Furthermore, as the subject group were healthy volunteers it was deemed unjustified to use invasive haemodynamic monitoring for the sole purpose of this study. An additional limitation of noninvasive monitoring is that continuous monitoring is not feasible and it is possible that rapid changes, especially early effects of amyl nitrite, may have been missed.

In conclusion, in doses of agents that give comparable reductions in augmentation index, we have demonstrated that inhaled amyl nitrite has a more rapid onset and offset of action than sublingual GTN. Although a short duration of action is desirable, the reflex cardiac effects of amyl nitrite suggest that if used as an alternative to GTN as an endothelium independent vasodilator for use in clinical studies, the response after return of heart rate to baseline, at around 2 min, should be used.

Our thanks are due to Mark Miller for the illustration.

Dr Greig and Dr Leslie were the recipients of British Heart Foundation Junior Research Fellowships (FS/2000004 and FS/98040, respectively).

Competing interests: None declared.