Introduction

Severe neonatal hypoxia can lead to left ventricular myocardial dysfunction. These sick neonates may have decreased blood flow and oxygen delivery to vital organs leading to significant morbidity and mortality [1]. Following resuscitation with 100% oxygen asphyxiated neonates are exposed to oxygen-derived free radicals and commonly have further reduced myocardial contractility and pulmonary vasoconstriction with low cardiac output (CO), systemic hypotension, and pulmonary hypertension (PHT) [1–4].

Current treatments for hypotension in asphyxiated neonates include catecholamines such as dopamine, epinephrine, and dobutamine. The stimulation of adrenoceptors by catecholamines may be associated with tachycardia, arrhythmias, vasoconstriction, fluctuating blood pressures, aggravating PHT, and increased oxygen consumption, despite inotropic effects [5–7]. The inotrope, milrinone, is a specific phosphodiesterase III inhibitor that increases myocardial contractility by elevating the intracellular cyclic adenosine monophosphate concentration through inhibiting cyclic adenosine monophosphate degradation [8]. Phosphodiesterase III inhibitors may also improve diastolic dysfunction [9], cause vasodilation [5, 10, 11], but not affect myocardial oxygen consumption [11, 12]. To date the effects of milrinone have been examined in neonates who had low CO after cardiac surgery, preterm infants at risk for low systemic blood flow, and those with unresponsive PHT [11, 13–16]. Although management strategies for these patients are similar to those for hypoxic neonates including improving cardiac function and pulmonary hemodynamics [17], doses of milrinone used and hemodynamic parameters measured are variable in these studies. To our knowledge, there is no information on the systemic and regional hemodynamic effects of milrinone in hypoxic neonates.

Using a swine model of neonatal hypoxia and reoxygenation (H-R), we primarily examined the dose-response effects of milrinone on CO and systemic blood pressure. The secondary objectives included the effects of milrinone on the pulmonary and carotid circulations as well as oxygen transport. We hypothesized that milrinone would dose-dependently improve the H-R induced cardiac dysfunction and increase blood flow to the brain without aggravating PHT and with minimal effect on systemic blood pressure in H-R piglets.

Materials and methods

Animal preparation

Mixed breed piglets aged 1–3 days (mean 2.0 ± 0.2) and weighing 1.5–2.3 kg (mean 1.86 ± 0.04). Four piglets were euthanised prematurely due to complications secondary to surgery (two) or severe hypoxia (two). The protocol followed the 1993 guidelines of the Canadian Council on Animal Care and was approved by the institutional Health Sciences Animal Welfare Committee. Animal preparation and H-R protocol have been previously described [7, 18]. Briefly, the piglets were anesthetized with halothane, which was then switched to intravenous midazolam (0.1–0.2 mg kg−1 h−1), fentanyl (5–15 μg kg−1 h−1), and pancuronium (0.05–0.1 mg kg−1 h−1) to provide anesthesia, analgesia, and paralysis, respectively. Fractional inspired oxygen concentration (FIO2) of 0.21–0.24 kept oxygen saturations between 90–100% as measured by pulse oximetry (Nellcor, Hayward, CA, USA). Infusions of 10% dextrose (10 ml kg−1 h−1, venous) and of 0.9% normal saline (2 ml kg−1 h−1, arterial) were maintained during the experiment. Through femoral vessels 5-F Argyle catheters (Sherwood Medical, St Louis, MO, USA) were inserted to the level of right atrium and another into the infrarenal aorta to measure central venous pressure, deliver fluids and medications, and measure mean arterial pressure (MAP). Endotracheal intubation via a tracheostomy allowed mechanical ventilation (ventilator model IV-100, Sechrist Industries, Anaheim, CA, USA) at a rate of 18–20 breaths/min and pressure of 19/4 cmH2O.

The left common carotid artery was encircled by a 2-mm Transonic flow probe (2SS, Transonic Systems, Ithaca, NY, USA) to measure blood flow. Through a left anterior thoracotomy the main pulmonary artery was exposed and cannulated with a 20-G Arrow catheter (Arrow International, Reading, PA, USA) to measure pulmonary arterial pressure (PAP). A 6-mm Transonic flow probe (6SB) was placed to monitor the blood flow as a surrogate of CO. The ductus arteriosus was ligated. Incisions were covered and kept moist to minimize evaporative losses. Body temperature was kept at 38.5–39.5 °C. Hemodynamic data were recorded and digitized by a DT2801-A analogue to digital converter board (Data Translation, ON, Canada) in a Dell 425E computer equipped with custom Asyst programming software.

Hypoxia-reoxygenation and treatment procedure

Piglets were block-randomized to four groups (n = 7/group) which underwent H-R; a sham-operated group (n = 4) had no H-R or medication delivery. There were no differences among groups regarding age, weight, mortality, or hemodynamic variables at normoxic baseline. After instrumentation the animals underwent a stabilization period for at least 40 min. Arterial PCO2 was maintained between 35–45 mmHg during the experimental period. Alveolar hypoxia was initiated with the inhalation of an oxygen/nitrogen gas mixture to provide an FIO2 of 0.10–0.15 for a PaO2 of 20–40 mmHg for 2 h. Piglets were then reoxygenated with 100% oxygen for 1 h and a further 3 h of 21% oxygen. Previous studies have determined that in this piglet model this degree of hypoxemia produces clinical hypoxia followed by significant cardiac dysfunction and systemic hypotension at 2 h of reoxygenation [7, 18]. All piglets received a 10 ml/kg lactated Ringer's bolus 30 min prior to medication delivery. At 2 h of reoxygenation piglets received an intravenous infusion of either placebo (normal saline) or milrinone (Pharmaceutical Partners of Canada, Richmond Hill, ON, Canada) at low dose (25 μg/kg loading, 0.25 μg kg−1 min−1 infusion), medium dose (50 μg/kg, 0.50 μg kg−1 min−1), or high dose (75 μg/kg, 0.75 μg kg−1 min−1) at a constant dose rate for 2 h in a blinded-randomized fashion. The drug administration was given at 6.7 ml/kg/h for 15 min (medication loading bolus) followed by 1 ml kg−1 h−1 for 2 h.

Hemodynamic and oxygen transport measurements

Heart rate, MAP, PAP, central venous pressure, pulmonary artery and left common carotid artery blood flows were recorded at normoxic baseline (0 min), at 2 h of hypoxia (120 min), at 0.5 h of reoxygenation (150 min) and 2 h of reoxygenation (240 min, pretreatment baseline), and every 30 min following medication or placebo delivery. Variables were calculated as a mean over 2 min at these time points. Simultaneous arterial and venous blood samples were taken for blood gas analyses by ABL500 and OSM3 Hemoximeter (Radiometer, Copenhagen, Denmark). To assess milrinone loading stability the highest and lowest hemodynamic parameters were recorded throughout the 15 min loading of medication. Hemodynamic variables were indexed for piglet weight. Pulmonary vascular resistance (VR) was estimated by PAP ÷ CO because the left atrial pressure was not measured.

Plasma lactate, troponin-I, and milrinone levels determination

Arterial plasma samples were prepared and stored at −80 °C until assay. Plasma lactate was analyzed at 0, 120, 240, and 360 min by enzyme-coupled NADH colorimetry method with spectrophotometry at 340 nm [19]. In hypoxic control and high-dose milrinone-treated groups, plasma levels of porcine cardiac-specific troponin-I were determined at 0 and 360 min using enzyme-linked immunosorbent assay (#2010-4, Life Diagnostics, West Chester, PA, USA). Using a high-performance liquid chromatographic validated method [20], plasma milrinone was determined with standards (0–2000 ng/ml) for calibration. The limit of quantification was 5 ng/ml.

Left ventricle myocardial lactate determination and histopathological examination

At study end the piglets were euthanised with intravenous pentobarbital (100 mg/kg). Left ventricle samples were snap-frozen in liquid nitrogen and stored at −80 °C whereas another sample was fixed in 10% formalin for biochemical and histopathological analysis, respectively. Frozen heart tissue (50 mg) was crushed at −80 °C, homogenized in 6% perchloric acid/0.5 mM EGTA (50 μl) on ice, and centrifuged (11,000 rpm, 2 min, 4 °C). The supernatant was weighed, 5 M potassium carbonate added (ratio 1:10 supernatant), precipitated on ice for 30 min, then centrifuged (11,000 rpm, 2 min). The supernatant was then used for lactate determination as described above. The formalin preserved specimen was stained with hematoxylin and eosin. Two pathologists (J.L., C.G.) unaware of group assignment evaluated the specimens. Pathological injury was graded based on previously established criteria [21].

Statistics

Results are expressed as mean ± SEM. Using α < 0.05 and β < 0.2, a sample size (n = 7/group) was estimated to demonstrate a statistically significant difference in CO between groups based on our previous experience [7, 18]. Hemodynamic variables were analyzed by two-way repeated measures analysis of variance or analysis of variance on ranks (Kruskal-Wallis) if normality test failed. Biochemical variables were analyzed by one-way analysis of variance. Pearson's moment correlation was used to determine the relationship between hemodynamic variables and milrinone levels. Analysis was performed using Sigma Stat (version 2.0, Jandel Scientific, San Rafael, CA, USA).

Results

2 h of hypoxia and 2 h of reoxygenation

Two hours of normocapnic alveolar hypoxia (PaO2 34 ± 1 mmHg, FIO2 0.129 ± 0.003) resulted in metabolic acidosis with pH 7.04 ± 0.04 and plasma lactate levels of 15.2 ± 0.4 mmol/. Hypoxia significantly reduced the CO to 71 ± 4 ml kg−1 min−1 with hypotension (MAP of 28 ± 2 mmHg; 41 ± 1% and 41 ± 2% of normoxic baseline, respectively; Figs. 1a, 2a, Table 1). The stroke volume was significantly decreased to 41 ± 3% of normoxic baseline (Fig. 1b). During hypoxia PAP initially increased. At 2 h of hypoxia PAP subsequently decreased to 31 ± 1 mmHg (128 ± 6% of normoxic baseline; Fig. 2b), whereas pulmonary VR progressively increased (315 ± 16% of normoxic baseline). Carotid artery blood flow (CAF) was 14 ± 1 ml kg−1 min−1 (66 ± 5% of normoxic baseline; Fig. 3).

Fig. 1
figure 1

The effect of milrinone infusion on cardiac output (a), stroke volume (b), and heart rate (c) in newborn piglets during hypoxia and reoxygenation compared to hypoxic controls (open circles): filled circles 0.25 μg kg−1 min−1; filled triangles 0.50; filled squares 0.75 (n = 7 per group); # p < 0.05 individual group vs. normoxic baseline (0 min); ## p < 0.05 groups vs. normoxic baseline; § p < 0.05 vs. pretreatment baseline (240 min); * p < 0.05 individual milrinone group vs. hypoxic controls; ** p < 0.05 all milrinone groups vs. hypoxic controls; † p = 0.07 high-dose vs. medium-dose and low-dose milrinone groups

Full size image
Fig. 2
figure 2

The effect of milrinone infusion on mean arterial pressure (a) and mean pulmonary arterial pressure (b) in newborn piglets during hypoxia and reoxygenation compared to hypoxic controls (open circles): filled circles 0.25 μg kg−1 min−1; filled triangles 0.50; filled squares 0.75 (n = 7 per group); ## p < 0.05 all groups vs. normoxic baseline (0 min); Note: the administration of milrinone maintained the mean arterial pressure without causing significant hypotension

Full size image
Fig. 3
figure 3

The effect of milrinone infusion on carotid artery flow in newborn piglets during hypoxia and reoxygenation compared to hypoxic controls (open circles): filled circles 0.25 μg kg−1 min−1; filled triangles 0.50; filled squares 0.75 (n = 7 per group); # p < 0.05 individual group vs. normoxic baseline (0 min); ## p < 0.05 all groups vs. normoxic baseline; § p < 0.05 vs. pretreatment baseline (240 min); * p < 0.05 individual milrinone group vs. hypoxic controls; † p = 0.09 individual milrinone group vs. hypoxic controls; ‡ p < 0.05 high-dose vs. low-dose milrinone group

Full size image
Table 1 Arterial pH, partial pressure of carbon dioxide, partial pressure and saturation of oxygen in newborn piglets during hypoxia-reoxygenation, and milrinone infusion (low-dose, medium-dose, and high-dose at 0.25, 0.50 and 0.75 μg kg−1 min−1, respectively). Hypoxic controls received normal saline (n = 7 per group)
Full size table

Upon reoxygenation with 100% oxygen (PaO2 404 ± 16 mmHg) systemic and regional hemodynamics immediately recovered to baseline within 10 min (data not shown). At 2 h of reoxygenation (PaO2 61 ± 1 mmHg, pH 7.38 ± 0.01) MAP fell to 67 ± 3%, CO to 75 ± 3%, and CAF to 79 ± 3% of the respective normoxic baseline (all p < 0.05; Figs. 1, 2, 3). Pulmonary VR was higher and systemic VR was lower than the respective normoxic baseline (142 ± 5% and 85 ± 5%, respectively; both p < 0.05). Stroke volume but not heart rate was significantly diminished (67 ± 3% of normoxic baseline; Fig. 1b, c).

No significant differences in hemodynamic and physiological parameters were found between groups at the end of hypoxia and after 2 h of reoxygenation prior to medication delivery. Normoxic sham-operated piglets showed no significant change in hemodynamic and physiological parameters throughout the experimental period.

Systemic hemodynamics after milrinone or placebo infusion

The loading dose of milrinone did not cause significant changes in CO, heart rate, MAP, PAP, and CAF or arrhythmias. All milrinone-treated groups had higher CO and stroke volume after 2 h of infusion than placebo-treated hypoxic controls (p < 0.001 and p = 0.01, respectively; Fig. 1a, b). In hypoxic controls CO significantly decreased further over the next 2 h, but it was maintained in the low and medium-dose milrinone-treated groups and significantly increased over pretreatment baseline to a value not significantly different from normoxic baseline after high-dose milrinone infusion. The heart rate was not different between groups during infusion (Fig. 1c). Mean arterial pressure was not affected with milrinone infusion (vs. hypoxic controls or pretreatment baseline, both p > 0.05; Fig. 2a). Systemic VR remained low in the high-dose and medium-dose milrinone-treated groups compared to the continued increase found in hypoxic controls (0.26 ± 0.01 and 0.27 ± 0.01 vs. 0.37 ± 0.04 mmHg ml−1 kg−1 min−1, respectively, p < 0.05).

Pulmonary hemodynamics after milrinone or placebo infusion

No effect was found on PAP during milrinone infusion (vs. hypoxic controls or pretreatment baseline, both p > 0.05; Fig. 2b). At 4 h of reoxygenation there was a continued increase in pulmonary VR in hypoxic controls (0.37 ± 0.03 vs. 0.16 ± 0.02 mmHg ml−1 kg−1 min−1 at normoxic baseline, p < 0.05) while pulmonary VR of milrinone-treated animals remained significantly lower than hypoxic controls (0.20 ± 0.02, 0.20 ± 0.02, and 0.24 ± 0.02 mmHg ml−1 kg−1 min−1 for high-dose, medium-dose, and low-dose milrinone-treated groups, respectively, p < 0.05).

Carotid hemodynamics after milrinone or placebo infusion

CAF of hypoxic controls was significantly lower than normoxic baseline after 4 h of reoxygenation (Fig. 3). High-dose milrinone infusion increased CAF significantly (vs. hypoxic controls or pretreatment baseline, both p < 0.05) and the effect of medium-dose milrinone infusion was modest (p = 0.09, β = 0.4; Fig. 3). No significant change in the carotid VR was found (data not shown).

Oxygen transport after milrinone or placebo infusion

Compared to hypoxic controls, all milrinone-treated groups had higher systemic oxygen delivery (Fig. 4a) and consumption (6.3 ± 0.4, 6.3 ± 0.2, and 5.6 ± 0.5 ml O2 kg−1 min−1 in high-dose, medium-dose, and low-dose milrinone-treated groups vs. 3.8 ± 0.4 ml O2 kg−1 min−1 in hypoxic controls, respectively, p < 0.05). Carotid artery oxygen delivery was higher in high-dose and medium-dose milrinone groups (vs. hypoxic controls, p < 0.05; Fig. 4b). High-dose milrinone increased carotid oxygen delivery over pretreatment baseline (p < 0.05). There was no difference in plasma lactate levels at 2 h of infusion between groups (3.0 ± 0.4, 2.7 ± 0.3, and 2.8 ± 0.7 mmol/l in low-dose, medium-dose, and high-dose milrinone-treated groups vs. 4.7 ± 1.1 mmol/l in hypoxic controls, respectively).

Fig. 4
figure 4

The effect of milrinone infusion on systemic oxygen delivery (a) and carotid artery oxygen delivery (b) in newborn piglets during hypoxia and reoxygenation compared to hypoxic controls (open circles): filled circles 0.25 μg kg−1 min−1; filled triangles 0.50; filled squares 0.75 (n = 7 per group); # p < 0.05 individual group vs. normoxic baseline (0 min); ## p < 0.05 all groups vs. normoxic baseline; § p < 0.05 vs. pretreatment baseline (240 min); * p < 0.05 individual milrinone group vs. hypoxic controls; ** p < 0.05 all milrinone groups vs. hypoxic controls; ‡ p < 0.05 high-dose vs. low-dose milrinone group

Full size image

Plasma milrinone levels

Dose-response on plasma milrinone levels after 2 h of treatment was observed (Fig. 5). Plasma milrinone levels were correlated with CO (r = 0.6, n = 28, p < 0.001; r = 0.9, n = 4 mean plasma levels and mean CO, p = 0.05). Modest correlations were found between milrinone levels and stroke volume, CAF, and carotid oxygen delivery (all r = 0.5, p < 0.05). Significant correlations were found between milrinone levels and VR (systemic, pulmonary, carotid: all r = −0.6, p < 0.05).

Fig. 5
figure 5

Boxplot showing the plasma milrinone concentrations in low-dose, medium-dose, and high-dose milrinone groups (0.25, 0.50, and 0.75 μg kg−1 min−1, respectively); Median, 25th and 75th percentiles, 95% confidence limits, and outliners are shown; * p < 0.05 vs. low-dose group; ‡ p < 0.05 vs. medium-dose group

Full size image

Plasma troponin-I levels and cardiac tissue analysis

Significantly higher plasma troponin-I levels were found at 4 h of reoxygenation compared to normoxic baseline (0.71 ± 0.17 and 0.48 ± 0.18 ng/ml for hypoxic controls and high-dose milrinone-treated group vs. 0.09 ± 0.01 ng/ml at normoxic baseline, respectively). There were no significant differences in troponin-I levels between groups. There were no significant differences in cardiac left ventricle histopathology or tissue lactate levels between groups (data not shown). None of the myocardial tissue samples demonstrated prominent ischemic changes or overt necrosis.

Discussion

In this swine model of neonatal H-R we demonstrated the dose-related response of milrinone on systemic and carotid perfusion after hypoxia. All doses of milrinone did not cause significant hypotension or worsen PHT. Milrinone prevented the progressive increases in systemic and pulmonary VR found in hypoxic controls. In the absence of significant chronotropy it is plausible that milrinone improves aspects of systolic-diastolic cardiac function as well as left ventricular preload and afterload.

Systemic hemodynamic effects of milrinone

Neonatal asphyxia induces cardiovascular dysfunction with reduced myocardial contractility and passive dilation leading to cardiac stunning and systemic hypotension [1, 3]. This low CO state, similar to that in neonates after cardiac surgery, is evidenced by the continual decrease in cardiac function in the hypoxic control group. Doppler echocardiography in neonates has demonstrated that milrinone improved low CO with a direct myocardial effect [22]. We demonstrated the dose-response of milrinone on CO and further showed the significant correlation between cardiac variables and plasma milrinone levels, which is consistent with that in children after cardiac surgery [12, 23]. Interestingly, we found the cardiovascular effect with high-dose milrinone at plasma milrinone levels in H-R piglets (Fig. 5) that were substantially lower than the level of 180–300 ng/ml reported in studies in human subjects, although a hemodynamic effect at these levels is yet to be shown in preterm infants with low systemic blood flow [13]. Milrinone pharmacokinetics may be different in various pediatric populations and indications. Our lower levels may reflect subtle differences in the pathopharmacology of low CO syndrome between H-R and postcardiopulmonary bypass patients, different pharmacokinetics between piglets and human neonates, and the short administration to reach steady state.

We did not observe any significant effect on the heart rate and MAP during milrinone infusion. The chronotropic effect of milrinone may have been masked by the tachycardia during neonatal H-R [24] or ameliorated by the preceding fluid bolus. MAP during milrinone infusion has been found to be maintained [11, 25] or decreased [5, 11, 13]. Adequate volume status coupled with increased CO most likely helped to maintain MAP despite decreased systemic VR. The cardiac effect of milrinone was not associated with increased myocardial oxygen consumption [11, 12]. Although we did not measure coronary hemodynamics or perfusion, there was no significant increase in myocardial tissue lactate or any evidence of histological features of ischemic changes or H-R damage in the milrinone-treated groups compared to hypoxic controls in this short-term (2 h) infusion.

Pulmonary hemodynamics effects of milrinone

Neonatal hypoxia can lead to PHT, which has been previously demonstrated in this model of H-R [7, 26, 27]. Three case series recently reported milrinone use in neonates with unresponsive PHT and the drug's ability to increase ventricular output and decrease pulmonary VR [14–16]. The current study showed that milrinone might have a dose-related ability to prevent elevations in pulmonary VR, which is inversely correlated with the plasma milrinone level, with no significant change in PAP. In the current study left atrial pressure was not measured (due to increased mortality), and therefore pulmonary VR was an estimated parameter. Enhanced pulmonary arterial flow with milrinone infusion did not cause PHT exacerbation as seen in vasopressive doses of epinephrine or dopamine [28, 29].

Carotid hemodynamics and oxygen delivery

Cerebral blood flow is often impaired in hypoxic infants [30], even after resuscitation when blood pressure and oxygenation normalized [31, 32]. Milrinone at 250 μg/kg increased cerebral blood flow when given to hypoxic adult dogs [33]. We found that high-dose milrinone (0.75 μg kg−1 min−1) was more effective than lower doses at improving CAF, which has been shown to be correlated with cerebral blood flow (r = 0.84) in animal models of H-R [34]. Without significant effects found on the carotid VR peripheral VR but not MAP may be important for regional blood flow after hypoxia [35, 36].

Limitations

There are recognizable limitations in this study although piglets have similar pathophysiological responses to H-R as human neonates [37, 38]. To estimate CO we assumed negligible transatrial shunt during H-R with ductus arteriosus ligated. Unlike asphyxia, hypoxemia and acidemia was controlled in the presence of normocapnia. Our study model approximates the clinical scenario of neonatal hypoxia but is distinct from perinatal asphyxia [39]. Nonetheless, the piglets were hypotensive with cardiovascular dysfunction and decreased carotid flow after H-R. The piglet's pharmacodynamics of milrinone is unknown. Oxidative stress is related to the cardiovascular dysfunction after H-R; thus, measuring oxygen-derived free radicals production with milrinone would also be beneficial for future research. Given the short duration of the model, subsequent low plasma milrinone levels, and gross nature of the biochemical and histological examination, further long-term studies of milrinone infusions looking at beneficial and detrimental effects are needed.

In conclusion, in a swine model of neonatal H-R, milrinone improved CO and carotid flow without causing significant hypotension or aggravating PHT. Given these encouraging results, clinical investigation and further studies to compare milrinone with currently used inotropes in asphyxiated neonates are warranted.