Journal of the American College of Cardiology
Production of oxidative products of nitric oxide in infarcted human heart
Author + information
- Received July 25, 1997
- Revision received April 2, 1998
- Accepted April 17, 1998
- Published online August 1, 1998.
Author Information
- Kazuhide Akiyama, MDa,
- Akio Kimura, MDa,
- Hiroshi Suzuki, MD∗,
- Youichi Takeyama, MD∗,
- Tracy L Gluckman, BSa,
- Artin Terhakopian, BSa,
- Takashi Katagiri, MD∗,
- Ki-Young Suh, BSa,
- James Roseto, BSa and
- Richard J Bing, MD, FACCa,*
- ↵*Address for correspondence: Dr. Richard J. Bing, Huntington Medical Research Institutes, Department of Experimental Cardiology, 99 North El Molino Avenue, Pasadena, California 91101
Abstract
Objectives. We sought to assess whether oxidation products of nitric oxide (NO), nitrite (NO2−) and nitrate (NO3−), referred to as NOx, are released by the heart of patients after acute myocardial infarction (AMI) and whether NOxcan be determined in peripheral blood of these patients.
Background. Previously we reported that in experimental myocardial infarction (rabbits) NOxis released mainly by inflammatory cells (macrophages) in the myocardium 3 days after onset of ischemia. NOxis formed in heart muscle from NO; NO originates through the activity of the inducible form of nitric oxide synthase (iNOS).
Methods. Eight patients with acute anterior MI and an equal number of controls were studied. Coronary venous blood was obtained by coronary sinus catheterization; NOxconcentrations in coronary sinus, in arterial and peripheral venous plasma were measured. Left ventricular end-diastolic pressure was determined. Measurements were carried out 24, 48 and 72 h after onset of symptoms. The type and location of coronary arterial lesions were determined by coronary angiography. Plasma NO3−was reduced to NO2−by nitrate reductase before determination of NO2−concentration by chemiluminescence.
Results. The results provided evidence that in patients with acute anterior MI, the myocardial production of nitrite and nitrate (NOx) was increased, as well as the coronary arterial–venous difference. Increased NOxproduction by the infarcted heart accounted for the increase of NOxconcentration in arterial and the peripheral venous plasma. The peak elevation of NOxoccurred on days 2 and 3 after onset of the symptoms, suggesting that NOxproduction was at least in part the result of production of NO by inflammatory cells (macrophages) in the heart.
Conclusions. The appearance of oxidative products of NO (NO2−and NO3−) in peripheral blood of patients with acute MI is the result of their increased release from infarcted heart during the inflammatory phase of myocardial ischemia. Further studies are needed to define the clinical value of these observations.
Nitric oxide (NO) is a free radical with an unpaired electron; it is an important physiologic messenger, produced by nitric oxide synthases, which catalyze the reaction l-arginine to citrulline and NO. There are two main categories of NO synthase, the constitutive and the inducible forms. The constitutive isoforms exists in neuronal and endothelial cells and is calcium dependent (1,2). Calcium binds to calmodulin and the calcium calmodulin complex activates the constitutive NO synthase that releases NO, relaxing smooth muscle cells through activation of guanylate cyclase and the production cGMP (3). Therefore, the NO produced has a negative inotropic effect on the heart and is instrumental in the autoregulation of the coronary circulation (4,5).
In contrast, the inducible form of NO synthase (iNOS) is mostly produced in macrophages activated by cytokines and endotoxin (6). It eliminates intracellular pathogens, damaging cells by inhibiting ATP production and oxidative phosphorylation and DNA synthesis (7). In infection, lipopolysaccharide released from bacterial walls, stimulates production of iNOS primarily in macrophages (8). The large amount of NO produced causes extensive vasodilation and hypotension. The iNOS is also activated in heart muscle under pathologic conditions, such as in heart failure, dilated cardiomyopathy and cardiac allograft rejection (9,10). It has been found that in infarcted rabbit heart muscle after ligation of a coronary artery, iNOS activity increases on the first postoperative day and persists for at least 14 days, declining to control levels 3 weeks after the onset of ischemia. iNOS was also localized in infiltrating macrophages of the human heart (11). Animals with myocardial ischemia showed consistent elevation of plasma nitrite (NO2−) and nitrate (NO3−) concentration 3 days after onset of myocardial infarction (12). Several findings suggest a causal relationship between NO production by the heart and elevated plasma levels of NOx, such as a relationship in time between NOxplasma concentration and NOxproduction and between coronary arteriovenous difference of NOxand myocardial iNOS activity.
It is the purpose of this study to demonstrate that after occlusion of the anterior descending coronary artery in humans, NOxplasma levels are elevated and that this is caused by increased NOxproduction in the infarcted human heart muscle.
Methods
Patients and methods
All participants in the study were informed of the procedure and gave written consent. The protocol was approved by the local ethics committees and is in line with the recommendations of the Declaration of Helsinki.
The study involved 16 patients, of these eight (6 men, 2 women), aged 48 to 80 years (mean 65.0 ± 3.3 years) with no evidence of acute myocardial infarction (AMI) served as control (Table 1). These patients were undergoing coronary arteriography using a Judkins catheter (6F) for evaluation of nonspecific chest pain and had no angiographic evidence of significant coronary artery stenosis (Table 1). All control patients were studied after an overnight fast and received 5,000 units of intravenous heparin at the initiation of the study. Antianginal and antihypertensive medications were discontinued at least 48 hours before this study. Routine cardiac catheterization was performed using a femoral approach and a biplane cineangio system (Toshiba). After a Cordis infinity micro pigtail catheter (6F) was inserted into the left ventricle, left ventricular end-diastolic pressure was recorded. Left ventriculography was performed and left ventricular ejection fraction was calculated (Table 1). A coronary sinus catheter (5F; Baxter) was placed in the coronary sinus (CS) through the left subclavian vein and the position was confirmed by injection of the contrast medium (iopamilone) and by typical position on fluoroscopy (13). Blood samples for analysis were drawn from the CS, femoral artery and femoral or brachial vein (henceforth referred to as the peripheral vein) before performing coronary angiography. A Cordis infinity Judkins type catheter (6F) was used for diagnostic coronary angiography.
Patient Characteristics and Hemodynamic Variables (Control and AMI) legend
Patients with acute myocardial infarction (AMI) (6 men, 2 women), aged 46 to 81 years (mean 66.0 = 4.5 years) (Table 1)displayed evidence of infarction as documented by the following criteria: 1) chest pain lasting ≥30 min; 2) electrocardiographic ST-segment elevation of ≥0.1 mV in at least two leads in either anterior or lateral distribution; and 3) elevated creatine kinase MB isoenzymes within 24 h after the onset of symptoms. Excluded were patients with electrocardiographic or historic evidence of previous myocardial infarction, coronary artery bypass surgery or percutaneous transluminal coronary angioplasty.
Tissue plasminogen activator or urokinase were not administered. Therapy included heparin and morphine when indicated. In most cases patients with AMI had sublingual nitrate once in the emergency room or catheterization laboratory before or after coronary angioplasty. Blood was drawn for determination of NOx>12 h after coronary angioplasty and the administration of nitroglycerine. Patients with AMI were subjected to cardiac catheterization within 6 h after onset of symptoms. Coronary angiography and angioplasty were performed using a standard femoral approach. All patients underwent direct coronary angioplasty (mean time from onset of chest pain to first inflation was 3.9 ± 0.6 h). The culprit artery was identified by coronary angiogram (Table 1). A significant coronary artery stenosis was defined as >75% luminal narrowing by visual estimate after intracoronary nitroglycerin (100 to 200 μg). The culprit arteries were revascularized using Scimed Bandit coronary angioplasty dilation catheter (2.5 to 3.5 mm by 20 mm, Boston Scientific) to Thrombolysis in Myocardial Infaction flow grade 3 until residual stenosis was <50% (14). Electrocardiographic ST-segment elevation decreased to normal levels in one patient (No. 2) after coronary angioplasty (Table 1). The remaining patients displayed continuous ST-segment elevation after coronary angioplasty. Sustained ventricular tachycardia, ventricular fibrillation or atrial fibrillation/flutter were not observed.
During hospitalization no patient had cardiogenic shock, reinfarction or infection. At time of discharge coronary angiography revealed that all culprit arteries had remained patent.
Measurement of plasma NO2−and NO3−
Blood was drawn in heparinized tubes for NO2−and NO3−determination, and immediately centrifuged at 1,200 gfor 10 min for removal of the formed elements by plasma separation. The plasma samples were ultrafiltered using a micropartitioning device at 4,000 gfor 1 h (Centrifree Micropartition System) to remove residual proteins. The deproteinized plasma samples were frozen at −20°C until analysis by chemiluminescence (15).
Analysis of NO2−+ NO3−required reduction of NO3−to NO2−with aspergillus nitrate reductase (Sigma Chemical). All samples were run in duplicate. Nitrate reductase 50 μl (0.2 U), FAD 5 μl (5 mmol/liter), NADPH 5 μl (6 mmol/liter) and phosphate buffer (1.2 mmol/liter) were added to 50 μl of deproteinized plasma to yield a final volume of 150 μl and subsequently incubated at 36°C for 1 h to allow for sufficient conversion of NO3−to NO2−. After incubation, 50 μl of the sample were injected into the reaction vessel. The NO3−content of the sample was calculated by subtracting the amount of NO2−in untreated samples from the amount of NO2−in the reduced samples (12). The sample containing nitrate reductase contains basal NO2−plus NO2−derived from reduction of NO3−.
Chemiluminescence was used for detection of NO2−after its reduction to NO. We used a reducing medium consisting of a ferrocene derivative dissolved in acetonitrile containing 1% perchloric acid (Sigma). In this solution, NO is generated from NO2−by a one-electron reduction pathway. In the reaction vessel, helium is used as the carrier gas of NO to the NO analyzer, at a rate of 35 ml/min. Reducing solutions (30 mg of 1,1-dimethylferrocene in 3 ml of acetonitrile, acidified with 50 μl of 70% perchloric acid) were freshly prepared and used within 2 h (15). Under these conditions, NO2−in the injected samples (50 μl) was reduced to NO and detected by chemiluminescence with a NO analyzer. Data were recorded automatically on a chart recorder (Soltec Corporation) as previously described (15).
NO2−and NO3−standard curves
Standard curves for NO2−(0 to 80 μmol/L) and NO3−(0 to 80 μmol/liter) were calculated by adding small aliquots (10 μl/50 μl sample) of sodium nitrite or sodium nitrate solution (in degassed ultrapure water) (12). A standard curve relating the luminescence produced by the added NO2−or NO3−was constructed, and the data were fitted to a straight line (linear regression) (12). All standard curves had r >0.96.
Calculation of NOxproduction equivalent
To estimate the production of NOxby the heart, the arterial–venous difference of NOxand coronary flow had to be known. NOxproduction by the heart was calculated as: Output of NOxfrom the heart (CSNOx× coronary flow) − Input of NOxinto the heart (ArterialNOx× coronary flow) (16). Because coronary flow was not measured directly in these patients, data were used from patients with AMI, in which coronary flow was calculated from positron emission tomography imaging, using [13N]ammonia (17). Czernin et al. (17)established a regression correlation (r = 0.74) relating the rate pressure product (RPP) to myocardial blood flow as defined by the equation: y = 0.00009x − 0.07, where y = myocardial blood flow and x = RPP.
Using this equation we calculated mean coronary flow in our patients. The production equivalent of NOxby the heart could then be calculated. The term production equivalent was used in preference to production because coronary flow was not determined directly (12).
Statistical analysis
Two analyses were performed to evaluate the change in mean NOxlevels over time. A two-way analysis of variance with repeated-measures was used to analyze the difference in mean NOxlevels over time in the venous, arterial and coronary sinus plasma in patients with AMI. The repeated measures model used a general covariance matrix, allowing for unequal correlations over time. The design of this study was an unbalanced two-way repeated measures analysis of variance. All parameters (differences, means, p values) were estimated via maximum likelihood. A second analysis was performed comparing the mean NOxlevels in the control patients to the patients with AMI at 24, 48 and 72 h based on between-group repeated measures variances for the peripheral venous, arterial and coronary sinus blood.
To compare the three sources of mean NOxlevels to each other within each time period (control 24, 48 and 72 h), post-hoc ttests (based on the covariance structure of the analysis of variance model) were performed using the Tukey-Fisher least significant difference method. The within-subject variances were used to calculate the standard errors.
A similar method was used to compare the mean NOxlevel of the control group to the mean NOxlevels at 24, 48 and 72 h. Post-hoc ttests (based on the covariance structure of the analysis of variance model) using the Tukey-Fisher least significant difference method were used to compare these means. The total variance estimated from the analysis of variance model using a general covariance matrix was used to calculate these standard errors. All variances used to calculate the standard errors for the within-subject and between-subject comparisons were estimated from the repeated measures of the analysis of variance model using a general covariance matrix, which allowed for unequal correlations over time. A p value of 0.05 or less was considered statistically significant. Data processing was performed using the SAS (version 6.11) software package.
Results
Hemodynamic data (heart rate, mean blood pressure, left ventricular end-diastolic pressure and left ventricular ejection fraction in patients with and without myocardial infarction are shown in Table 1. The infarction group had higher left ventricular end-diastolic pressure and lower left ventricular ejection fraction. The respective means for left ventricular end-diastolic pressure in the two groups were 9.38 ± 0.89 mm Hg in the control and 13.25 ± 1.70 mm Hg in patients with myocardial infarction (p < 0.05); left ventricular ejection fraction was 66.13 ± 2.49% in the control and 42.43 ± 2.61% in patients with infarction (p < 0.05). The possibility must be considered that some of the changes in patients with myocardial infarction were attributable to factors other than coronary occlusion such as cigarette smoking or medication. Three of the patients of the control group and five patients with myocardial infarction were smokers. Smoking as a factor, however, could be excluded. There were no statistical difference within each group for mean left ventricular end-diastolic pressure or left ventricular ejection fraction or for coronary sinus NOxconcentration. There was also no difference in heart rate or mean blood pressure. Diuretics and calcium antagonists also did not influence NOxconcentration in CS plasma. The effect of other factors such as angiotension-converting enzyme inhibitors or beta-adrenergic blocking agent could not be evaluated because all but one patient were given angiotension-converting enzyme inhibitors and only one patient was given beta-blockers.
Table 2shows that after onset of symptoms the values for NOxin CS plasma are close to significance at 48 h (p < 0.07) as compared to arterial and peripheral venous NOxconcentrations. At 24 h, CS NOxconcentrations still exceed those in arterial and peripheral venous plasma (Table 2). Compared to the control group, all patients with AMI showed statistically significant increase in NOxconcentrations in arterial, CS and peripheral venous plasma at 48 and 72 h after onset of symptoms (Fig. 1). These statistical differences were absent 24 h after onset of symptoms. In Figure 2the NOxlevels in arterial blood are plotted in each individual patient as a function of time from the presentation through the remainder of the study. The values of the control series are also shown. A general increase in arterial NOxconcentrations in individual patients is noted. Changes in individual CS and peripheral venous plasma followed a similar trend (not shown).
Mean plasma concentrations (μmol/L) of NOxfrom arterial (closed circles), peripheral venous (open triangles)and coronary sinus (open squares)plasma in control patients and in patients with AMI. Plasma NOxconcentrations are significantly elevated above control values, at 48 and 72 h after onset of symptoms (p < 0.01). Values for NOxin coronary sinus plasma significantly exceed those in peripheral venous plasma from 24 to 72 h after onset of symptoms (p < 0.05). The NOxvalues in coronary sinus plasma also significantly exceed those in arterial plasma from 24 to 72 h after onset of symptoms (p < 0.05).
NOxconcentrations in arterial plasma of individual AMI patients increased from 24 to 48 h after onset of symptoms. The peak NOxconcentration usually occurred from 48 to 72 h after onset of symptoms.
Plasma NOxin Control and AMI Patients legend
The coronary arterial–venous differences of NOxwere significantly greater when compared to their control and the systemic arterial–venous NOxdifferences (Table 2). Forty-eight hours after onset of symptoms the mean difference between the coronary and systemic arterial–venous NOxdifference was 6.0 μmol/L (p < 0.05). In control patients this difference was not noted (Table 2). This suggests a cardiac origin of NOx.
The output of NOxfrom the heart is represented as the product of coronary flow and the NOxconcentration in the CS plasma (12,16). Because coronary flow was not determined directly, it was calculated from the rate–pressure product as presented in Table 3, according to methods of Czernin et al. (17). From the coronary flow thus calculated the production equivalent of NOxby the heart was obtained (12). The cardiac NOxproduction equivalent is listed in Table 3. A multiple comparison test showed that at 48 h after onset of symptoms the NOxproduction equivalent was significantly greater than at control, again pointing to the heart as the source of NOxproduction (control 137 ± 37 nmol/100 g/min; AMI at 48 h = 472 ± 104 nmol/100 g/min; p < 0.05; Table 3).
Calculation of Cardiac NOxProduction Equivalent legend
Discussion
Patients
Patients in this study represent a clinically homogenous population; none of them had cardiogenic shock or congestive heart failure, and all had anterior myocardial infarction with the left anterior descending coronary artery as the culprit vessel. Furthermore all AMI patients had higher left ventricular end-diastolic pressure and lower left ventricular ejection fraction when compared to controls. Heart rate and blood pressure did not differ (Table 1). Other factors such as smoking, diuretics or calcium antagonists did not influence either hemodynamics or NOxplasma concentration in patients with AMI (Tables 1 and 2). The effect of angiotension-converting enzyme inhibitors could not be evaluated, because all but one patient was given angiotension-converting enzyme inhibitors.
The angiotension-converting enzyme inhibitors may have influenced NO production as inhibition of angiotension-converting enzyme localized in the luminal site of the vascular endothelium results in increased synthesis of NO and prostaglandin I2by accumulation of endogenous bradykinin. This demonstrates the potential ability to increase NO production through the kinin pathway (18).
Formation of NOx
It is well known that NO is produced in infarcted heart muscle primarily by the activity of iNOS (19). Activation of iNOS occurs in cardiomyocytes and during the inflammatory phase in activated macrophages (11). iNOS shows no requirement for Ca2+and calmodulin, although calmodulin exists as a highly bound subunit (20). Activation of iNOS depends on the release of endogenous substances, including cytokines (tumor necrosis factor-alpha, interleukin-1 beta, interferon) (1). NO itself has an unpaired electron, and is physiologically and pharmacologically active (21). In contrast, as shown by Stuehr and Nathan (22), NO2−and NO3−are inactive, except at acidic pH when NO2−is converted into more reactive species. NO2−is formed by a reaction of NO with oxygen. The primary metabolite of NO is NO3−with <10% of the total appearing as NO2−(23); NO2−concentrations remain constant, whereas those of NO3−vary widely (12,16). The sum of NO2−and NO3−has been designated as NOx(12). The product of the reaction of NO with oxygen, superoxide and transition metals support additional nitrosative reactions (24). Hibbs et al. (8)showed that l-arginine is required for activation of macrophages to a bactericidal and tumoricidal state. Apparently NO is an intermediate in the oxidation of l-arginine to NO2−and NO3−and the citrulline pathway (25).
NOxand myocardial infarction
The results of the present study provide evidence that in patients with anterior AMI at 48 h after onset of symptoms, NOxproduction equivalent by the heart is increased together with its coronary arterial–venous difference (Tables 2 and 3). This accounts for the increase of NOxin arterial and peripheral venous plasma (Table 2, Fig. 1). In experimental myocardial infarction activated macrophages are a major source of NO (26). An elevated NOxproduction by the heart and increased concentrations of NOxin peripheral venous blood have been noted previously in rabbits 3 days after occlusion of a coronary artery (16). This may have been the result of formation of NO by activated macrophages (19). It is possible that both cardiomyocytes and inflammatory cells, primarily activated macrophages, are involved (11).
Although it is likely from the evidence presented here that the elevated levels of NOxin peripheral plasma observed after myocardial infarction are the result of increased production of NOxby macrophages in infarcted heart muscle, the possibility must be considered that other factors are involved. Part of plasma NOxoriginates as a consequence of the activity of the endothelial form of NO synthase. Evidence for this is the finding of Akiyama et al. (12)that S-methylisothiourea, which primarily inhibits iNOS, only partially lowers plasma levels of NOx. It is unlikely that coronary angioplasty influences the results as collections of plasma for NOxdetermination were carried out ≥18 h after angioplasty. Another possible reason for the increase of NOxin plasma is accumulation of nitrate anion if the production of NOxexceeds its elimination within the half-life of NOx(3.8 h) (16). Similarly, diminished clearance of NOxby the kidney can lead to increased NOxplasma levels. This may be the case in severe congestive heart failure as reported by Winlaw et al. (27). However, none of the patients in this series had signs of congestive heart failure.
Another possible explanation for increased appearance of NOxin CS plasma is enhanced NO production originating in coronary vascular endothelium beginning 24 h after reperfusion. Kim et al. (28)have shown that brief ischemic episodes are responsible for enhanced coronary artery endothelial function. This effect is delayed in onset and prolonged in duration. This may be one of the factors responsible for the increased concentration of NOxin CS plasma of patients after onset of acute myocardial infarction. It also suggests that in certain instances involving multiple factors NOxin plasma may not be the exclusive reflection of iNOS in heart muscle.
Our observations of increased arteriocoronary sinus difference and of NOxmyocardial production equivalent strongly point to the heart as a source of increased plasma levels of NOxin myocardial infarction.
Nitric oxide and cardiac disorders
Cardiac diseases other than myocardial infarction are accompanied by changes in plasma concentration of NO3−(29,30). Winlaw et al. (27)reported that plasma NO3−increased significantly in patients with chronic congestive heart failure; however, because only peripheral plasma concentrations were measured, decrease in renal function may have been partially responsible. Haywood et al. (31)found increased expression of inducible form of nitric oxide synthase messenger RNA in the hearts of patients with chronic congestive heart failure. Langrehr et al. (29)demonstrated in allografted rats a significant increase in NO2−/NO3−levels in peripheral blood before critical signs of rejection were noted. Treatment with FK-506, an inhibitor of transplant rejection, abolished this increase. Winlaw et al. (32)found in rats an eightfold increased excretion of urinary NO2−in untreated allograft rejection. Akiyama et al. (12)also reported reduction of plasma levels of NOxby S-methylisothiourea, a selective inhibitor of iNOS in experimental myocardial infarction in rabbits.
Conclusions
From data presented in this report, no definite conclusions on future applications of NOxdeterminations in peripheral blood of patients with myocardial infarction can be drawn. More patients need to be studied and the various parameters responsible for increased NOxproduction should be further defined. However, the evidence of a cardiac origin in the increase in plasma NOxconcentration in patients with myocardial infarction appears convincing.
Acknowledgements
We acknowledge the support of the Sidney Stern Memorial Trust, Pacific Palisades, California, and an unrestricted grant from Pfizer Inc., New York, New York. We also thank the U.C.L.A. Biomathematics Department for their help in carrying out the statistical analysis.
Footnotes
☆ This research was supported by the Sidney Stern Memorial Trust, Pacific Palisades, California, and an unrestricted grant from Pfizer Inc., New York, New York.
- Abbreviations
- AMI
- acute myocardial infarction
- CPK
- creatine kinase
- CS
- coronary sinus
- iNOS
- inducible form of nitric oxide synthase
- LAD
- left anterior descending coronary artery
- NO
- nitric oxide
- NO2−
- nitrite
- NO3−
- nitrate
- NOx
- sum of nitrite and nitrate
- RPP
- rate pressure product
- Received July 25, 1997.
- Revision received April 2, 1998.
- Accepted April 17, 1998.
- American College of Cardiology
References
- ↵
- ↵
- ↵
- Brady A.J.B,
- Warren J.B,
- Poole-Wilson P.A,
- Williams T.J,
- Harding S.E
- Schulz R,
- Nava E,
- Moncada S
- ↵
- Stuehr D.J,
- Marletta M.A
- ↵
- ↵
- Hibbs J.B Jr.,
- Taintor R.R,
- Vavrin Z
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Zeballos G.A,
- Bernstein R.D,
- Thompson C.I,
- et al.
- ↵
- Czernin J,
- Porenta G,
- Brunken R,
- et al.
- ↵
- ↵
- ↵
- ↵
- ↵
- Stuehr D.J,
- Nathan C.F
- ↵
- Wennmmalm A,
- Benthin G,
- Edlund A,
- et al.
- ↵
- ↵
- ↵
- Wildhirt S.M,
- Dudek R.R,
- Suzuki H,
- Pinto V,
- Narayan K.S,
- Bing R.J
- ↵
- ↵
- Kim S,
- Ghaleh B,
- Kudej R,
- Hintze T,
- Vatner S
- ↵
- Ishiyama S,
- Hiroe M,
- Nishikawa T,
- et al.
- ↵
- Haywood G,
- Tsao P,
- von der Leyen H,
- et al.
- ↵