INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mechanisms that maintain high pulmonary vascular resistance (PVR) in utero and contribute to the marked increase in pulmonary blood flow at birth are incompletely understood. Past studies have demonstrated that basal production of nitric oxide (NO) modulates pulmonary vascular tone in the normal fetal lung and contributes to the fall in PVR after birth (1). Blockade of NO synthase activity increases fetal PVR, suggesting that endogenous NO is tonically synthesized and released under basal conditions in the normal fetal lung (1). In addition, enhanced NO production by various endothelium-dependent stimuli, such as acetylcholine and oxygen, causes fetal pulmonary vasodilation (1, 3). However, the fetal pulmonary vasodilator response to several stimuli, including acetylcholine and oxygen, is only transient (5). That is, although pulmonary blood flow initially increases in response to these agents, blood flow returns toward baseline values despite continued exposure to these dilators. These findings suggest that mechanisms exist in the normal fetus to oppose vasodilation and that these mechanisms may be important determinants of fetal pulmonary vasoreactivity. Mechanisms that maintain high fetal PVR and limit fetal pulmonary vasodilation are uncertain, but there are several possibilities, such as enhanced production of a vasoconstrictor (e.g., endothelin [10] or sulfidopeptide leukotrienes [11]); high myogenic tone in fetal vascular smooth muscle (12); or an inability to sustain the release or activity of endogenous vasodilators, including NO (7).

Once produced by the endothelial cell, NO diffuses to subjacent smooth muscle cells, where it activates soluble guanylate cyclase. That enzyme increases smooth muscle cell concentrations of cyclic guanosine 3',5'-monophosphate (cGMP), resulting in vasodilation (13). One factor that potentially limits the vasodilator response to endogenous or exogenous NO (16) is the rapid hydrolysis and inactivation of cGMP by cGMP-specific phosphodiesterase enzymes (PDE5). The PDE5 family is found in large amounts in crude lung homogenates and appears quantitatively to be a major source of cGMP phosphodiesterase activity in the lung (19). Past studies have shown that PDE5 inhibition augments vasodilation in the normal newborn lamb lung (20). More recently, intrapulmonary infusions of dipyridamole and zaprinast, two PDE5 inhibitors, were found to cause potent and sustained pulmonary vasodilation in the late-gestation ovine fetus, suggesting an important role for PDE5 activity in regulating basal PVR (21). In addition, preliminary biochemical studies report that lung PDE5 protein and activity is markedly elevated during fetal life, then rapidly falls after birth (22). Therefore, PDE5 activity appears to play an important role in the regulation of pulmonary vascular tone in the perinatal period. However, whether high PDE5 activity limits the fetal pulmonary vascular response to vasodilator stimuli is uncertain.

Therefore, to examine the role of PDE5 activity in fetal pulmonary vasoreactivity, we studied the hemodynamic effects of dipyridamole, a known PDE5 antagonist, on the fetal pulmonary vasodilator responses to acetylcholine, an endothelium-dependent vasodilator, and inhaled NO, which acts by directly stimulating cGMP production in vascular smooth muscle. Because past studies have shown that acetylcholine-induced vasodilation is mediated through NO release in the fetal lung (1) and that acetylcholine causes only transient vasodilation (7, 8), we hypothesized that PDE5 inhibition would augment and sustain fetal pulmonary vasodilation during prolonged acetylcholine infusion. We further hypothesized that dipyridamole infusion would augment inhaled-NO-mediated vasodilation. To test these hypotheses, we performed a series of experiments to examine whether dipyridamole-induced vasodilation was dependent on basal NO activity in the late-gestation ovine fetus. We then compared the fetal pulmonary vasodilator responses to acetylcholine (an endothelium-dependent dilator) and inhaled NO (an endothelium-independent dilator) before and during the administration of dipyridamole in the late-gestation fetus.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgical Preparation

All procedures and protocols were previously reviewed and approved by the Animal Care and Use Committee at the University of Colorado Health Sciences Center (Denver, Colorado). As outlined below, two different surgical approaches were used.

Mixed breed (Columbia-Rambouillet) pregnant ewes between 122 and 128 d gestation (term = 147 d) were fasted 24 h before surgery. Ewes were sedated intravenously with pentobarbital sodium (administered through an external jugular line) and anesthetized with lumbar intrathecal doses of 1% tetracaine hydrochloride (3 mg). Preoperative, intramuscular doses of penicillin (500 mg) and streptomycin (1 g) were also given. Pentobarbital dosing was adjusted so that ewes were sedated but breathed spontaneously throughout surgery. Using sterile technique, the uterus was exposed through a midline abdominal approach. The fetal lamb's left forelimb was delivered through a uterine incision. After local infiltration of 1% lidocaine, a skin incision was made under the left forelimb of the fetus, and the axillary vein and artery were isolated. Polyvinyl catheters were inserted into the axillary vein and artery and directed into the superior vena cava (SVC) and ascending aorta (Ao), respectively. A left thoracotomy exposed the heart and great vessels. Polyvinyl catheters were inserted into the left pulmonary artery (LPA), main pulmonary artery (MPA), and left atrium (LA) by direct puncture through purse string sutures as previously described (3). A 6-mm ultrasonic flow transducer (Transonics, Ithaca, NY) was placed around the LPA to measure LPA blood flow (Qp). After the catheters and flow transducer were secure, the fetal skin was closed and the fetus was placed back into the uterus. A catheter was left in the amniotic space, ampicillin (500 mg) was injected into the amniotic cavity, and the hysterotomy was closed. The abdominal incision of the ewe was closed in two layers, and the catheters and flow transducer cable were tunneled subcutaneously to an external flank pouch on the ewe. The ewe and fetus were given 48 to 72 h to recover from surgery before studies were initiated. Prophylactic ampicillin (250 mg) was infused into the fetus and amniotic cavity daily for 3 d following surgery, and catheter patency was ensured by daily flushes of 2-3 ml of heparinized saline (100 U/ml).

The catheters (excluding the SVC and amniotic catheters) and ultrasonic flow transducer were placed as described above in fetuses between 136 and 142 d gestation (term = 147 d). This gestational age was selected to avoid potential problems with lung immaturity and surfactant deficiency often present in less-mature lambs. After enough time for hemodynamic variables to recover from surgery, the fetus was given a dose of pancuronium bromide (0.5 mg), removed from the uterus, and left lying on the abdomen of the ewe with the umbilical cord intact. The fetus was covered and warmed with an external warming blanket to maintain normothermia (39.5° C). A tracheostomy was performed with placement of a 4.5-mm interior diameter cuffed endotracheal tube, and the study was then initiated as described below.

Physiologic Measurements

Blood flow in the left pulmonary artery was measured continuously using an ultrasonic flow transducer connected to an internally calibrated flowmeter (Transonics) with a digital display. The Ao, MPA, LA, and amniotic catheters were connected to a Gould-Statham P23 ID pressure transducer (Gould Statham, Cleveland, OH), with mean and phasic pressure measurements continuously recorded onto a Gould strip-chart recorder. The pressure transducer was calibrated at the start of each study using a mercury-column manometer. Aortic, MPA, and LA pressures were referenced to the amniotic cavity pressure in the chronic fetal preparation (protocols 1 and 2). Heart rate was determined from the flowmeter or from phasic pulmonary blood flow tracings. Because only LPA flow was measured, PVR represents resistance across the left lung reported in units of mm Hg × min × ml¹ [PVR = (MPA pressure  LA pressure)/Qp].

Blood samples for pH, PA_CO2, PA_O2, hemoglobin, and oxygen saturation were drawn from the aortic catheter and measured at 39.5° C with a Radiometer OSM-3 blood gas analyzer and hemoximeter (Radiometer, Copenhagen, Denmark).

Drug Preparation

Dipyridamole (Sigma, St. Louis, MO) was dissolved in 50 mM hydrochloric acid (HCl) and diluted with 0.9% saline to achieve the final concentration. Although this preparation of dipyridamole results in a very acidic solution, previous dose-response tests in the chronically prepared ovine fetus demonstrated no change in hemodynamic values or blood gas measurements with 10-min infusions of this carrier solution (21). Adenosine and acetylcholine (Sigma) were dissolved in 0.9% saline. Nitro-L-arginine (L-NA) was initially dissolved in 1 N HCl and diluted with saline. Sodium hydroxide (1 N) was then added to achieve a physiologic pH. All drugs used in this study were administered at a flow rate of 0.1 ml/min.

Nitric oxide gas (AIRCO, Riverton, NJ) was introduced into the afferent limb of the ventilator circuit through a ¹/₄-inch × ¹/₄-inch Luer adaptor fitted within 25 cm of the endotracheal tube, thus mixing with fixed flow of circuit gas (10 L/min). The rate of NO flow was changed to yield NO concentrations of 5 and 20 ppm verified by in-line chemiluminescence measurement (Model 14A; Thermoenvironmental Instruments, Franklin, MA).

Experimental Design

Using a precalibrated syringe infusion pump, dipyridamole was infused at 400 µg/min for 10 min into the LPA of each fetus (n = 5 animals, mean gestational age = 131 d). Hemodynamic variables were measured at baseline and after the infusion. This dose of dipyridamole was chosen from previous dose-response tests that demonstrated a doubling of Qp without changing MPA or Ao pressures (21). On a separate day, the dipyridamole infusion was repeated following administration of 1.0 mg/min L-NA for 10 min into the LPA. Previous work has shown that this dose of L-NA increases fetal pulmonary vascular tone and attentuates the response to endothelium-dependent vasodilators (1). The vasodilator responses to dipyridamole before and after L-NA were then compared. To make sure that the effect of L-NA on dipyridamole-induced vasodilation was not a nonspecific effect of L-NA on resting vascular tone, in three animals the non-cGMP-dependent dilator adenosine was studied, using a dose of 40 µg/min for 10 min, before and after administration of the same dose of L-NA. Previous work has shown that this dose of adenosine has an equivalent effect on Qp and PVR as the dose of dipyridamole used in this protocol (21). For each drug study, we recorded heart rate, mean Qp, and mean pulmonary artery (PA), Ao, and LA pressures and calculated left-lung PVR every 10 min during a 30-min baseline period and then after each 10-min infusion. After L-NA infusions, saline was infused for 5 min before the study drug (dipyridamole or adenosine) infusion.

In each animal (n = 6 animals, mean gestational age = 132 d), three 120-min infusions into the LPA were performed on separate days in random order: acetylcholine (2.5 µg/min, a dose found to increase Qp and decrease PVR by approximately 100% and 50%, respectively, following a 10-min infusion), dipyridamole (25 µg/min, a dose found from previous studies to have no effect on baseline hemodynamics following a 10-min infusion), and the two agents infused together. A precalibrated syringe infusion pump was used, and baseline hemodynamics had been stable for 30 min before each drug infusion began. During simultaneous delivery of both agents, a three-way stopcock was used. Hemodynamic variables as described in Protocol 1 were measured every 10 min during the 30-min baseline period, throughout the infusion, and for 30 min after the infusion ended. Aortic pH, PA_CO2, and PA_O2 were measured at baseline, and at 30, 60, and 120 min during the infusion.

After securing the endotracheal tube into place in the delivered ovine fetus and allowing lung fluid to drain passively, positive pressure ventilation with an FI_O2 of 0.03-0.05 (balance N₂) was initiated (n = 8 animals, mean gestational age = 139 d). The FI_O2, ventilator rate (IMV), peak inspiratory pressure (PIP), and end expiratory pressure (PEEP) were adjusted to maintain fetal arterial blood gases and provide adequate lung inflation as assessed by direct inspection of the left lung. After hemodynamic variables reached a steady state on positive-pressure ventilation (baseline 1), inhaled NO was administered for 10-min intervals at concentrations of 5 and 20 ppm. During NO exposures, FI_O2 and ventilatory settings remained constant. Hemodynamic variables were recorded at the end of each NO dosing interval. Nitric oxide was then discontinued and hemodynamic variables were allowed to reach a second stable point (baseline 2). After a new steady state was achieved, dipyridamole was infused into the LPA at 40 µg/min. This dose of dipyridamole was chosen based on the results of previous dose-response studies showing its minimal effect on basal hemodynamics in the chronically instrumented ovine fetus (21). After 10 min of the dipyridamole infusion, hemodynamic variables were recorded and exposures to NO at 5 and 20 ppm were repeated. The dipyridamole infusion was maintained throughout the second exposure to NO. Aortic pH, PA_O2, PA_CO2, and oxygen saturation were measured every 20 to 30 min throughout the study period. Methemoglobin levels were recorded at the beginning and end of each study.

To ensure that the hemodynamic effects of dipyridamole plus inhaled NO were not due to the effect of repeated NO treatment, three animals received NO without dipyridamole after the initial NO exposure.

Data Analysis

Data are presented as means ± SEM. Statistical analysis was performed with the Statview SE software package (Abacus Concepts, Berkeley, CA). Comparisons over time were made, with each animal serving as its own control, using a univariate analysis of variance for repeated measured (ANOVA) with linear contrast analysis to determine differences between study points. For the comparison of two discrete hemodynamic measurements in the same animal, a paired Student's t test was utilized. A p value < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Protocol 1: The effect of L-NA on dipyridamole-induced pulmonary vasodilation. Eight studies were done with L-NA (five with dipyridamole; three with adenosine). Infusion of L-NA caused an increase in baseline PVR from 0.596 ± 0.44 to 0.744 ± 0.68 mm Hg/ml/min (p < 0.01), mostly caused by an increase in pressure, as Qp did not change significantly (Table 1). Dipyridamole increased Qp and decreased PVR (from 77 ± 9 to 143 ± 17 ml/min and 0.582 ± 0.62 to 0.317 ± 0.39 units, respectively; p < 0.01 for both variables). As illustrated in Figure 1, dipyridamole-induced pulmonary vasodilation was completely abolished by L-NA. Infusion of adenosine caused an increase in Qp and decrease in PVR (from 76 ± 6 to 159 ± 5 ml/ min and 0.566 ± 0.062 to 0.252 ± 0.014 units, respectively; p < 0.01 for both variables). As shown in Figure 2, the vasodilator response to adenosine remained intact after L-NA. Heart rate and PA, Ao, and LA pressures did not change with infusion of either dipyridamole or adenosine.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Pretreatment with L-NA completely blocked the increase in Qp and decrease in PVR in response to dipyridamole (n = 5 animals; *p < 0.05 versus baseline by paired t test).

View larger version (27K): [in this window] [in a new window]   Figure 2.   Adenosine (40 µg/min × 10 min) caused a significant increase in Qp and decrease in PVR compared with baseline, both before and after pretreatment with L-NA (n = 3 animals; *p < 0.05 versus baseline by paired t test).

Protocol 2: Fetal pulmonary vascular response during prolonged intrapulmonary infusion of acetylcholine with and without dipyridamole. Figure 3 summarizes the Qp and PVR responses to 120-min infusions of acetylcholine, dipyridamole, and acetylcholine with dipyridamole. Infusion of acetylcholine transiently increased Qp (10-50 min; p < 0.05 versus baseline) and decreased PVR (10-70 min; p < 0.05 versus baseline), with the maximal increase in Qp of 98% at 30 min. Dipyridamole also caused a transient rise in Qp (20-80 min; p < 0.05 versus baseline) and drop in PVR (20-80 min; p < 0.05 versus baseline) with the maximum increase in Qp of 44% occurring at 70 min. Administration of acetylcholine with dipyridamole resulted in a more sustained vasodilatory response, with an increase in Qp from 10 to 100 min (p < 0.05 versus baseline) and a decrease in PVR for the entire infusion (10- 120 min; p < 0.05 versus baseline). The maximal increase in Qp, 132% occurring 30 min into the infusion, was greater than that achieved with either acetylcholine or dipyridamole alone, but it was similar to the sum of the peak flow responses to each agent. Heart rate, arterial blood gas values, and LA, PA, and Ao pressures did not differ from baseline values during any of the infusions.

View larger version (24K): [in this window] [in a new window]   Figure 3.   Changes in Qp and PVR during 2-h intrapulmonary infusions of acetylcholine (Ach; 2.5 µg/min), dipyridamole (D; 25 µg/ min), and Ach + D. Acetylcholine and dipyridamole both caused transient pulmonary vasodilation when given alone. When administered together, the vasodilator response was sustained for the duration of the infusion (PVR, p < 0.05 versus baseline for entire 120-min study period when Ach was given with D [n = 6 animals; significance determined by ANOVA]).

Protocol 3: Hemodynamic effect of inhaled NO, with and without dipyridamole, in the transitional fetus ventilated with low FI_O2. Beginning positive pressure ventilation without changing PA_O2 resulted in an increase in Qp (94 ± 8 to 150 ± 14 ml/ min; p < 0.05) and decrease in PVR (0.610 ± 0.58 to 0.347 ± 0.033 units; p < 0.05), which then reached steady-state values. Heart rate, arterial blood gases, and oxygen saturations remained constant during the course of the experiment. Table 2 illustrates PA, Ao, and LA pressures at different times during the experiment. Pulmonary artery and Ao pressures tracked each other very closely and differed from baseline values at the study points noted in the table. Both PA and Ao pressures decreased when inhaled NO (at both doses) was given concomitantly with dipyridamole, compared with NO alone. Left atrium pressure did not change during the course of the experiment.

View larger version (30K): [in this window] [in a new window]   Figure 4.   Left pulmonary artery blood flow (LPA Q) and PVR responses to NO, without (control) and with co-administration of dipyridamole (D) (n = 8 animals; *p < 0.05 by ANOVA).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We report that dipyridamole-induced fetal pulmonary vasodilation is dependent upon endogenous NO release in the fetal lung, and that dipyridamole augments and prolongs the vasodilator response to endothelium-dependent (acetylcholine) and -independent (inhaled NO) dilators in the late-gestation fetus. Although past studies in the normal newborn lamb (20) and adult sheep (23) and cat (24) reported that PDE5 inhibition with zaprinast can augment or prolong vasodilation by several agonists, our findings suggest that PDE5 activity plays an important physiologic role in maintaining high fetal PVR and modulating pulmonary vasoreactivity in utero.

Endogenous NO is an important modulator of vascular tone in the fetal and transitional pulmonary circulations (1), with its putative mechanism involving increased cGMP formation within vascular smooth muscle cells (13). Intracellular cGMP levels reflect the balance of cGMP production by the enzyme guanylate cyclase and cGMP degradation by the phosphodiesterases, specifically PDE5. These findings show that PDE5 activity modulates high PVR in the normal fetal lung; past studies have demonstrated that PDE5 contributes to pulmonary vascular tone in pathologic models of high PVR due to chronic hypoxia (25) or closure of the ductus arteriosus in utero (26). One mechanism contributing to pulmonary hypertension, especially when associated with decreased responsiveness to inhaled NO, might be upregulation of PDE5 activity. Thus, the use of PDE5 inhibitors with inhaled NO provides a potential means of maximizing pulmonary vascular smooth muscle cell concentrations of cGMP and, perhaps, improving responsiveness to inhaled NO or other cGMP-dependent dilators (27). Because dipyridamole is a commercially available PDE5 inhibitor approved for use in humans, it holds promise as a therapeutic agent for accomplishing these goals. However, dipyridamole is purported to have other mechanisms of eliciting vasodilation, including inhibition of adenosine re-uptake resulting in increased circulating levels of adenosine, nonspecific activity on cAMP-phosphodiesterases, and antiplatelet effects (28- 31). Previous work has shown that dipyridamole-induced pulmonary vasodilation in the ovine fetus is not attenuated by adenosine receptor blockade (21). In the present study, we provide further evidence that dipyridamole's vasodilator activity is cGMP-dependent, since inhibition of endogenous NO (and thus cGMP) formation with the NO synthase inhibitor L-NA completely blocked dipyridamole-induced pulmonary vasodilation. We further found that dipyridamole potentiated the vasodilator responses to the endothelium-dependent agonist, acetylcholine, and the endothelium-independent agonist, inhaled NO. This potentiation was more additive than synergistic, and our results do not allow us to definitively conclude that cGMP is the central signal transducer of NO-induced pulmonary vasodilation or that dipyridamole is not working through some other mechanism.

We have previously shown that dipyridamole and zaprinast, another PDE5 inhibitor, cause equivalent vasodilation (at equimolar doses) in the fetal lung (21). This finding supports in vitro enzyme inhibition data that show very similar PDE5 inhibition profiles for these two drugs (17, 19). It also suggests an important role for the PDE5 family in the regulation of fetal pulmonary vascular tone. Doses of dipyridamole used in the present study with acetylcholine and inhaled NO were determined by previous dose-response tests to have insignificant hemodynamic effects when given as 10-min intrapulmonary infusions in the chronically instrumented ovine fetus (21). In the present study, this dose of dipyridamole, when administered as a 120-min infusion in Protocols 1 and 2 and as a 10-min infusion in Protocol 3, had significant effects on basal pulmonary hemodynamics. In the fetus, this may represent a cumulative effect of prolonged dipyridamole infusions. Limited studies using lower doses of dipyridamole (data not included) showed no effect on the pulmonary vasodilator response to intrapulmonary acetylcholine. In the transitional fetus circulation, it is possible that increased release of endogenous NO in response to birth-related stimuli accounts for the greater response to dipyridamole. The presence of increased pulmonary vascular smooth muscle cell concentrations of cGMP in response to increased endogenous NO might result in a greater vasodilator response to a given amount of phosphodiesterase inhibition.

In testing dipyridamole in combination with prolonged intrapulmonary acetylcholine infusions, we sought to determine the effect of dipyridamole on the autoregulatory response of the fetal pulmonary vascular bed. In the fetus, multiple vasodilating stimuli cause time-dependent pulmonary vasodilation, with hemodynamic variables returning to baseline despite continued exposure to the vasodilating stimulus (5). Teleologically, this active opposition to prolonged pulmonary vasodilation in the fetal lung prevents a "steal" of blood flow from the placenta, the fetal organ of gas exchange. Mechanisms contributing to this time-dependent autoregulatory response are unknown but may include decreased production of endogenous NO by the endothelial cell, uncoupling of NO between the endothelial and smooth muscle cell, downregulation of guanylate cyclase, or increased release of vasoconstricting mediators such as endothelin-1 (10). Another plausible mechanism might be upregulation of PDE5 activity. However, this seems unlikely because some endothelium-independent agents that directly increase smooth muscle cell cGMP levels cause sustained pulmonary vasodilation during prolonged infusions (7). The finding that co-administration of acetylcholine and dipyridamole results in sustained vasodilation suggests that decreased cGMP production over time, which can be partially compensated for by PDE5 blockade, contributes to the autoregulatory response.

Few studies have examined the influence of PDE5 inhibitors on cGMP-mediated vasodilator responses. Zaprinast has been shown to augment cGMP-dependent vasodilation in both the adult cat and neonatal lamb pulmonary vascular beds (20, 23, 24). In addition, zaprinast augments the pulmonary vasodilator response to inhaled NO in sheep models of pulmonary hypertension (23, 26). Dipyridamole potentiates the platelet anti-aggregating and vasodilator activity of NO in rabbit aortic rings (27), but we are unaware of any studies examining the effects of dipyridamole in combination with cGMP-dependent dilators in the pulmonary vascular bed in vivo. We did not find prolongation of NO-induced pulmonary vasodilation with dipyridamole, as noted by other investigators with the other PDE5 inhibitor, zaprinast (23), and the reason for this remains unclear.

The results of the present investigation suggest that further study of dipyridamole's impact on cGMP-dependent vasodilation is indicated, particularly in conditions associated with an abnormal pulmonary vascular bed, such as experimental models of pulmonary hypertension (25, 26). In addition, PDE5 inhibition may offer a potential approach to the treatment of pulmonary hypertension in humans. Low-dose inhaled NO has proven efficacy in some patients with pulmonary hypertension, but not all patients respond to inhaled NO as manifested by acute changes in hemodynamics or oxygenation. In selected patients, dipyridamole may augment the clinical response to inhaled NO (32), but further studies are required to determine its potential role in clinical settings. In addition, concern remains regarding potential toxicity with high doses or prolonged treatment. The use of a submaximal dose of a PDE5 inhibitor might provide a pharmacologic means of improving responsiveness to NO at lower doses, minimizing toxicity. As both of our experimental models had a widely patent PDA and thus open communication between the systemic and pulmonary circulations, our results do not allow for accurate assessment of selectivity for the pulmonary circulation. Successful clinical use of dipyridamole would be partly dependent upon selection of a dose that did not adversely affect systemic hemodynamics.

In summary, dipyridamole-induced pulmonary vasodilation is dependent upon endogenous NO activity and thus endogenous cGMP production. In addition, dipyridamole potentiates the vasodilator responses to endothelium-dependent and -independent dilators. Further studies with direct tissue measurements of cGMP and studies in experimental models of pulmonary hypertension are indicated to better clarify dipyridamole's mechanism of action and potential therapeutic utility. We speculate that dipyridamole, and possibly newer PDE5 inhibitors, may offer another approach to the treatment of clinical pulmonary hypertension, especially in combination with inhaled NO.