The pulmonary artery pulsatility index in patients with severe aortic stenosis undergoing valve replacement

Abstract

Background and aims

The pulmonary artery pulsatility index (PAPi), that is, the pulmonary artery pulse pressure (PAPP) divided by the mean right atrial pressure (mRAP), is an increasingly used invasive index of right ventricular function. We sought to assess the prognostic impact of the PAPi in unselected patients with aortic stenosis (AS) undergoing aortic valve replacement (AVR).

Methods

We studied consecutive patients with severe AS (n = 487, 74 ± 10 years, 58% males) undergoing right heart catheterization prior to AVR with post‐AVR follow‐up of several years.

Results

The mean PAPi was 4.7 ± 3.3, and the mean values in the four PAPi quartiles were 2.1 ± 0.5, 3.2 ± 0.3, 4.5 ± 0.5 and 8.9 ± 4.2. Patients in the lowest PAPi quartile had similar AS severity, symptoms, B‐type natriuretic peptide and surgical risk compared with patients in higher quartiles. The lowest PAPi quartile had the lowest PAPP and the highest mRAP and only a slightly reduced stroke volume index (SVI) but the highest pulmonary artery capacitance (PAC). After a median post‐AVR follow‐up of 45 months mortality did not differ across PAPi quartiles (log rank P = 0.50), which was independent of the AVR mode. However, all contributors of the PAPi equation, that is, higher PAPP, lower PAC (i.e., stroke volume divided by PAPP), lower SVI and higher mRAP were associated with increased mortality.

Conclusions

In unselected patients with severe AS, the PAPi did not predict post‐AVR mortality. This may be explained by the fact that the low PAPP in those with low PAPi was mainly a reflection of a high PAC rather than a low SVI.

Introduction

The pulmonary artery pulsatility index (PAPi) is an increasingly used invasive index of right heart function. ¹ It was originally described in patients with inferior myocardial infarction for the prediction of right ventricular (RV) failure ² and has then typically been applied in patients with advanced heart failure to prognosticate RV failure after left ventricular (LV) assist device implantation. ³ , ⁴ , ⁵ The PAPi is defined as the pulmonary artery pulse pressure [PAPP, i.e., the systolic pulmonary artery pressure (sPAP) minus the diastolic pulmonary artery pressure (dPAP)] divided by the mean right atrial pressure (mRAP). The underlying concept is based on the idea that the PAPP (numerator) represents a measure of RV contractile function against a certain afterload, and high mRAP (denominator) reflects RV failure. ¹ From a practical point of view, the PAPi is a very attractive parameter because it is extremely simple and can be measured with any catheter, that is, even without a Swan Ganz catheter, because measurement of the pulmonary artery wedge pressure (PAWP) is not needed. In addition, its calculation does not include cardiac output whose measurement is always prone to errors in clinical practice (i.e., by the use of indirect Fick or thermodilution). However, the PAPP depends not only on RV stroke volume but also on pulmonary artery capacitance (PAC; i.e., RV stroke volume divided by PAPP), and therefore, interpretation of PAPi is not straightforward. In settings other than the post‐assist device implantation situation (e.g., pulmonary arterial hypertension and heart failure), studies regarding the prognostic utility of the PAPi have revealed mixed results, ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ and its role in clinical practice is still incompletely defined.

Few studies have evaluated the PAPi in patients with aortic stenosis (AS) undergoing aortic valve replacement (AVR), ¹¹ , ¹² where RV function is also very important in terms of prognosis. ¹³ In these patients, RV dysfunction represents the most advanced stage of the AS‐associated ‘cardiac damage’, a term that refers to the progressive backwards evolution of an unfavourable adaptation of the left ventricle (including the secondary involvement of the mitral valve), the left atrium, the pulmonary vasculature and the right ventricle (including the secondary involvement of the tricuspid valve) to the chronic pressure overload. ¹³ In one study among patients with severe AS referred for transcatheter AVR (TAVR), a PAPi threshold of <2.1 predicted an increased risk of heart failure hospitalizations within 2 years. ¹² In addition, a study in patients with decompensated AS undergoing emergency TAVR revealed a higher risk of in‐hospital mortality in those with lower PAPi. ¹¹ However, these studies were relatively small and included selected AS patients. We therefore measured the PAPi in a relatively large unselected population of patients with severe AS undergoing left and right heart catheterization prior to TAVR or surgical AVR (SAVR). The aim of the study was to provide a detailed non‐invasive and invasive characterization of severe AS with low versus high PAPi and to assess the value of the PAPi to predict short and long‐term mortality after AVR. We hypothesized that a low PAPi would identify AS patients with particularly unfavourable haemodynamics and increased long‐term mortality.

Methods

Study population

This is a retrospective analysis of prospectively and systematically collected cardiac catheterization data in cohort originally consisting of 503 consecutive patients with severe AS undergoing cardiac catheterization prior to AVR in a single center between January 2011 and January 2016 and with a post‐AVR follow‐up of several years. ¹⁴ At our institution, all patients with severe AS undergo right heart catheterization and coronary angiography on a routine basis prior to AVR. Accordingly, this is an unselected all‐comers cohort of AS patients evaluated prior to AVR. Complete data on sPAP, dPAP and mRAP (required for PAPi calculation) were available for 487 patients, which formed the study population. The study was approved by the ethics committee of the Canton of St. Gallen. Owing to its retrospective design, a waiver of consent was granted for this study.

Cardiac catheterization

Procedures were generally (>95%) performed in the morning in the fasting state and after withholding loop diuretics and renin–angiotensin system inhibitors. Patients underwent right heart catheterization using 6 French Swan Ganz catheters via femoral or brachial access followed by coronary angiography using 5 or 6 French catheters via the femoral or radial artery. The midthoracic level was used as zero reference point. Right atrial pressure, RV pressure, pulmonary artery pressure [sPAP, dPAP and mean pulmonary artery pressure (mPAP)] and PAWP were measured. The wedge position was confirmed by fluoroscopy and waveform analysis. Measurements were obtained at end‐expiration, the mean PAWP (mPAWP) was calculated over the entire cardiac cycle and v waves were included to determine mPAWP. This practice leads to higher values compared with the measurement of the end‐diastolic PAWP. However, for the estimation of the impact of the left heart contribution to pulmonary pressures and calculation of PVR respectively, the mPAWP is preferred. ¹⁵ In patients with atrial fibrillation, at least five cardiac cycles were used to assess pulmonary artery pressure and pulmonary artery wedge pressure. Cardiac output was assessed by the indirect Fick method based on blood gases, with blood samples taken in duplicate via arterial access and pulmonary artery. The transpulmonary gradient was calculated as difference between mPAP and mPAWP, and PVR as transpulmonary gradient divided by cardiac output. The PAC was calculated as stroke volume divided by the PAPP (i.e., difference between sPAP and dPAP). The PAPi was calculated as the PAPP divided by the mRAP. If the aortic valve was crossed, the LV end‐diastolic pressure (LVEDP) was recorded. All pressure readings were double‐checked by the operator using manual review of the pressure tracings before recording them into the report.

Echocardiography

In all patients, an echocardiogram was carried out prior to performing cardiac catheterization, and it was used as a basis for the referral. Echocardiograms were performed by experienced cardiologists in line with contemporary guidelines, but without adhering to a specified protocol. The data were retrospectively retrieved from patients' medical reports. Accordingly, most parameters were not available for all patients (indicated in Table 2), and we did not perform subgroup analysis or outcome analyses based on echocardiography data.

Table 2.

Data from echocardiography and cardiac catheterization of haemodynamic groups according to pulmonary artery pulsatility index (PAPi) quartiles (Q1–4) (n = 487 except for echocardiography, where detailed information is provided).

	PAPi Q1 (n = 124) ≤2.67	PAPi Q2 (n = 122) ≥2.68–3.81	PAPi Q3 (n = 120) ≥3.82–5.33	PAPi Q4 (n = 121) ≥5.34	P value
Echocardiography
Left ventricular end‐diastolic diameter (mm) (n = 367)	47 ± 7	46 ± 8	47 ± 8	46 ± 8	0.63
Indexed left ventricular end‐diastolic diameter (mm/m2) (n = 367)	24 ± 4	24 ± 4	26 ± 5	25 ± 4	0.05
Septal wall thickness (mm) (n = 366)	13 ± 3	13 ± 3	13 ± 3	13 ± 3	0.98
Posterior wall thickness (mm) (n = 366)	11 ± 3	11 ± 3	11 ± 2	11 ± 3	0.77
Left ventricular mass index (g/m2) (n = 366)	109 ± 30	109 ± 37	115 ± 39	113 ± 38	0.64
Left ventricular end‐diastolic volume index (mL/m2) (n = 363)	45 ± 16	44 ± 16	46 ± 19	44 ± 16	0.75
Left ventricular ejection fraction (%) (n = 487)	56 ± 11	57 ± 12	57 ± 12	52 ± 10	0.37
E/e′ (n = 229)	16 ± 7	17 ± 9	17 ± 9	18 ± 8	0.62
Left atrial diameter (mm) (n = 369)	41 ± 7	41 ± 8	40 ± 7	41 ± 7	0.62
Indexed left atrial diameter (mm/m2) (n = 369)	21 ± 4	22 ± 4	22 ± 4	23 ± 5	0.51
Left atrial area (cm2) (n = 376)	23 ± 7	24 ± 8	23 ± 7	24 ± 6	0.83
Indexed left atrial area (cm2/m2) (n = 376)	12 ± 3	12 ± 4	13 ± 4	13 ± 4	0.49
Left atrial volume index (mL/m2) (n = 340)	46 ± 14	43 ± 18	43 ± 19	44 ± 16	0.96
Indexed RV basal diameter (mm/m2) (n = 349)	15 ± 4	16 ± 3	17 ± 4	17 ± 4	0.01
Indexed right atrial area (cm2/m2) (n = 357)	9 ± 3	9 ± 2	9 ± 4	9 ± 3	0.91
Right atrial volume index (mL/m2) (n = 357)	24 ± 12	24 ± 11	24 ± 17	24 ± 10	0.96
TAPSE (mm) (n = 181)	20 ± 4	22 ± 6	20 ± 5	22 ± 4	0.13
Estimated sPAP (mmHg) (n = 219)	32 ± 9	33 ± 9	34 ± 10	33 ± 9	0.74
Mean aortic valve gradient (mmHg) (n = 487)	46 ± 16	45 ± 17	50 ± 19	48 ± 17	0.08
Aortic valve area (cm2) (n = 449)	0.82 ± 0.25	0.81 ± 0.20	0.75 ± 0.22	0.80 ± 0.27	0.17
Indexed aortic valve area (cm2/m2) (n = 449)	0.42 ± 0.13	0.43 ± 0.10	0.41 ± 0.12	0.44 ± 0.15	0.32
Mitral regurgitation (n = 487)	124	122	120	121	0.93
No	57 (46%)	56 (46%)	54 (%)	66 (55%)
Mild	56 (45%)	54 (44%)	51 (%)	44 (36%)
Moderate	9 (7%)	10 (8%)	12 (%)	9 (7%)
Severe	2 (2%)	2 (2%)	3 (%)	2 (2%)
Coronary artery disease					0.51
No coronary artery disease	60 (48%)	67 (55%)	67 (56%)	66 (54%)
1‐vessel disease	23 (19%)	23 (19%)	15 (13%)	17 (14%)
2‐vessel disease	14 (11%)	16 (13%)	21 (17%)	18 (15%)
3‐vessel disease	27 (22%)	16 (13%)	17 (14%)	20 (17%)
Invasive haemodynamics
Mean right atrial pressure (mmHg)	10 ± 4	8 ± 3	6 ± 2	4 ± 2	<0.001
Right ventricular end‐diastolic pressure (mmHg)	11 ± 4	9 ± 4	8 ± 3	6 ± 3	<0.001
sPAP (mmHg)	39 ± 14	41 ± 15	40 ± 15	40 ± 16	0.79
dPAP (mmHg)	18 ± 7	16 ± 7	15 ± 7	12 ± 7	<0.001
PAPP (mmHg)	20 ± 8	25 ± 10	26 ± 9	28 ± 11	<0.001
mPAP (mmHg)	27 ± 10	26 ± 10	25 ± 10	24 ± 10	0.07
mPAWP (mmHg)	18 ± 8	18 ± 2	15 ± 7	13 ± 7	<0.001
Transpulmonary gradient (mmHg)	8 ± 5	9 ± 4	10 ± 5	10 ± 6	0.002
Pulmonary vascular resistance (Wood units)	1.9 ± 1.4	2.0 ± 1.2	2.3 ± 1.4	2.3 ± 1.9	0.07
Pulmonary artery compliance (mL/mmHg)	4.0 ± 2.3	3.4 ± 1.5	3.2 ± 1.8	2.9 ± 1.1	<0.001
PAPi	2.1 ± 0.5	3.2 ± 0.3	4.5 ± 0.5	8.9 ± 4.2	<0.001
Left ventricular end‐diastolic pressure (mmHg) (n = 335)	23 ± 7	22 ± 8	20 ± 7	20 ± 7	<0.001
Systolic aortic pressure (mmHg)	141 ± 23	147 ± 26	144 ± 25	149 ± 26	0.10
Diastolic aortic pressure (mmHg)	70 ± 11	69 ± 11	67 ± 12	69 ± 12	0.005
Mean aortic pressure (mmHg)	98 ± 13	99 ± 14	97 ± 14	98 ± 15	0.67
Systemic vascular resistance (Wood units)	19.6 ± 4.3	19.9 ± 5.0	20.4 ± 5.3	20.4 ± 4.9	0.54
Arterial oxygen saturation (%)	95 (94–97)	95 (94–97)	95 (93–96)	95 (94–97)	0.72
Mixed venous oxygen saturation (%)	67 (62–72)	69 (64–72)	68 (64–72)	70 (66–73)	0.003
Cardiac output (L/min)	4.6 ± 0.9	4.8 ± 1.0	4.7 ± 1.1	4.8 ± 0.9	0.43
Cardiac index (L/min/m2)	2.4 ± 0.4	2.5 ± 0.5	2.5 ± 0.5	2.6 ± 0.5	<0.001
Stroke volume (mL)	68 ± 19	73 ± 19	70 ± 20	71 ± 18	0.34
Stroke volume index (mL/m2)	35 ± 9	38 ± 10	38 ± 9	38 ± 10	0.008

Note: Data are given as numbers and percentages, mean ± standard deviation and/or median (interquartile range).

Abbreviations: E/e′, ratio of peak early mitral inflow velocity to peak early mitral annular velocity, mPAP, mean pulmonary artery pressure; mPAWP, mean pulmonary artery wedge pressure; PAPP, pulmonary artery pulse pressure; sPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion.

Follow‐up

All patients underwent surgical AVR (SAVR; 72%) or TAVR (28%) following a median interval of 21 (12–35) days post‐catheterization. Among patients undergoing SAVR, 178/349 (51%) patients had at least one additional procedure including coronary artery bypass grafting (n = 121), mitral valve repair/replacement (n = 25), tricuspid valve repair (n = 2), surgery of the ascending aorta (n = 41) and/or a MAZE procedure (n = 16). We assessed 30 day all‐cause mortality and long‐term mortality after several years. Information on long‐term follow‐up was obtained from patients, general practitioners and hospital or practice cardiologists. The endpoint was all‐cause mortality.

Statistical analysis

Categorical data are presented as numbers and percentages and continuous data as mean ± standard deviation or median (interquartile range), as appropriate. Patients in different PAPi quartiles were compared using χ ² tests, analysis of variance or Kruskal–Wallis tests, as appropriate. Correlations of interest were described by Pearson correlation coefficients. Survival of patients in different PAPi quartiles was investigated using Kaplan–Meier estimates and compared using log‐rank tests. In addition, the prognostic impact of the key haemodynamic parameters contributing to PAPi, that is, sPAP, dPAP, mRAP, PAC, PVR; stroke volume index (SVI) and mPAWP was similarly assessed using Kaplan–Meier plots of the respective quartiles. The rationale for the evaluation of mPAWP in this context is based on the observation that the pulmonary artery time constant (i.e., the product of PAC and PVR) is affected by changes in mPAWP. ¹ , ¹⁶ Cox proportional hazards regression was applied to describe the association between variables of interest and mortality. The PAPi and the other haemodynamic parameters were tested both as categorical (quartiles) and continuous variables. A P value <0.05 was considered statistically significant. Analyses were performed by the senior author using the SPSS statistical package Version 20.0 (SPSS Inc., Chicago, IL, USA).

Results

Study population

The mean age of the 487 patients was 74 ± 10 years, and 283 (58%) were male. The mean indexed aortic valve area was 0.42 ± 0.12 cm²/m², and the mean left ventricular ejection fraction was 57 ± 12%. The mean sPAP, dPAP and PAPP were 40 ± 15 mmHg, 15 ± 7 mmHg and 25 ± 10 mmHg. The mean mRAP and PAPi were 7 ± 4 mmHg and 4.7 ± 3.3, respectively.

Clinical characteristics of patients with low PAPi

Patients in the lowest PAPi quartile were more likely to be male and to have atrial fibrillation and had higher body mass index and larger body surface area and lower serum potassium and forced expiratory volume within the first second (Table 1). Otherwise, there were no significant differences across PAPi quartiles. In particular, medical history, medication, symptoms, B‐type natriuretic peptide, logistic EuroScore, STS score and the proportion of patients undergoing TAVR or SVAR were similar in the four PAPi quartiles (Table 1).

Table 1.

Clinical characteristics of the haemodynamic groups according to pulmonary artery pulsatility index (PAPi) quartiles (Q1–4).

	PAPi Q1 (n = 124) ≤2.67	PAPi Q2 (n = 122) ≥2.68–3.81	PAPi Q3 (n = 120) ≥3.82–5.33	PAPi Q4 (n = 121) ≥5.34	P value
Age (years)	73 ± 11	73 ± 10	74 ± 10	76 ± 9	0.07
Sex (male)	86 (69%)	76 (62%)	62 (52%)	59 (49%)	0.003
Body mass index (kg/m2)	29.3 ± 5.2	28.2 ± 5.1	26.8 ± 5.9	27.3 ± 4.9	<0.001
Body surface area (m2)	1.96 ± 0.23	1.90 ± 0.22	1.84 ± 0.21	1.84 ± 0.21	<0.001
eGFR (mL/min/1.73 m2)	66 ± 19	66 ± 20	67 ± 18	65 ± 18	0.74
eGFR <30 mL/min/1.73 m2	2 (2%)	6 (5%)	1 (1%)	3 (2%)	0.19
Haemoglobin (g/L)	136 ± 17	135 ± 19	135 ± 16	133 ± 19	0.64
Albumin (g/L)	38 ± 5	39 ± 6	38 ± 4	39 ± 4	0.40
Sodium (mmol/L)	138 ± 3	137 ± 3	138 ± 3	137 ± 3	0.27
Potassium (mmol/L)	3.9 ± 0.4	4.0 ± 0.4	4.1 ± 0.5	4.0 ± 0.4	0.009
Diabetes	24 (19%)	21 (17%)	23 (19%)	31 (26%)	0.39
Insulin‐dependent	5 (4%)	6 (5%)	5 (4%)	5 (4%)	0.99
Stroke	7 (6%)	9 (7%)	8 (7%)	5 (4%)	0.73
Chronic obstructive pulmonary disease	12 (10%)	20 (16%)	12 (10%)	13 (11%)	0.32
Previous PCI	17 (14%)	11 (9%)	7 (6%)	10 (8%)	0.19
Previous CABG	8 (6%)	8 (7%)	4 (3%)	5 (4%)	0.57
FEV1 (% predicted)	82 ± 21	83 ± 19	88 ± 19	91 ± 20	0.002
Heart rhythm a					0.045
Sinus rhythm	93 (75%)	105 (86%)	104 (87%)	113 (93%)
Atrial fibrillation	21 (17%)	11 (9%)	10 (8%)	5 (4%)
Pacemaker	9 (7%)	4 (3%)	5 (4%)	2 (2%)
Other rhythm	1 (1%)	2 (2%)	1 (1%)	1 (1%)
Heart rate (bpm)	71 ± 15	68 ± 11	69 ± 11	70 ± 14	0.28
Medication
Oral anticoagulation	33 (27%)	21 (17%)	22 (18%)	16 (13%)	0.06
Aspirin	69 (56%)	72 (59%)	71 (59%)	85 (70%)	0.10
Loop diuretics	69 (56%)	60 (49%)	55 (46%)	60 (50%)	0.48
Beta‐blocker	66 (53%)	52 (42%)	59 (49%)	54 (45%)	0.34
ACEI/ARB	68 (55%)	68 (56%)	63 (53%)	72 (58%)	0.74
Digoxin	9 (7%)	6 (5%)	10 (8%)	7 (6%)	0.71
Spironolactone	7 (6%)	10 (8%)	4 (3%)	4 (3%)	0.26
B‐type natriuretic peptide (ng/L)	205 (106–480)	185 (57–408)	229 (92–670)	243 (92–446)	0.46
Symptoms
Dyspnoea NYHA class					0.15
I	19 (15%)	21 (17%)	31 (26%)	26 (21%)
II	73 (59%)	61 (50%)	51 (42%)	54 (45%)
III	26 (21%)	38 (31%)	32 (27%)	34 (28%)
IV	6 (5%)	2 (2%)	6 (5%)	7 (6%)
STS score	2.4 (1.3–4.2)	2.5 (1.3–3.4)	2.3 (1.3–3.7)	2.4 (1.6–3.9)	0.72
Logistic EuroScore	2.5 (1.4–4.3)	2.6 (1.4–3.9)	2.5 (1.4–4.4)	2.7 (1.6–4.7)	0.60
Mode of AVR					0.35
Surgical AVR	93 (75%)	80 (66%)	89 (74%)	87 (72%)
Transcatheter AVR	31 (25%)	42 (34%)	31 (26%)	34 (28%)

Note: Data are given as numbers and percentages, mean ± standard deviation or median (interquartile range).

Abbreviations: ACEI/ARB, angiotensin‐converting enzyme inhibitor/angiotensin receptor blocker; AVR, aortic valve replacement; eGFR, estimated glomerular filtration rate; FEV1, forced expiratory volume within the first second (percent predicted); NYHA, New York Heart Association; STS, Society of Thoracic Surgeons.

^a Rhythm at the time of cardiac catheterization.

Echocardiography and invasive haemodynamics in patients with low PAPi

The severity of AS, cardiac dimensions and RV and LV function parameters did not differ across PAPi quartiles with the exception of a smaller indexed RV basal diameter in patients with lower PAPi. The severity of mitral regurgitation and the extent of coronary artery disease were also similar in the four PAPi quartiles (Table 2). There was no significant correlation between PAPi and tricuspid annular plane systolic excursion (TAPSE; r = 0.05; P = 0.50).

As expected from the definition of PAPi, those in the lowest PAPi quartile had the lowest PAPP and the highest mRAP (Table 2). The lower PAPP in patients with lower PAPi was due to higher dPAP while sPAP did not differ across PAPi quartiles. In addition, patients with lower PAPi had higher RV end‐diastolic pressure, higher mPAWP and higher LVEDP. In contrast, the transpulmonary gradient was lower with a trend towards lower PVR and the PAC was significantly higher (i.e., better) in patients with lower PAPi. Cardiac output and stroke volume in absolute terms did not differ across PAPi quartiles. However, given the significant difference in body surface area, cardiac index and SVI were somewhat lower in patients with lower PAPi, and this was consistent with the lower mixed venous oxygen saturation in those with lower PAPi (Table 2). Quartiles for PAPi and corresponding data for key haemodynamic parameters in this context are shown in Figure 1. It becomes evident that patients with the lowest PAPi also had the lowest PAPP and the highest mRAP but only a slightly reduced SVI and the highest PAC and lowest PVR.

Figure 1.

Error bars showing mean and standard deviations for quartiles of the pulmonary artery pulsatility index (PAPi; Panel A) and corresponding data for key haemodynamic parameters in this context including pulmonary artery pulse pressure (PAPP; Panel B), mean right atrial pressure (mRAP; Panel C), stroke volume index (SVI; Panel D), pulmonary artery capacitance (PAC; Panel E) and pulmonary vascular resistance (PVR; Panel F). WU, Wood units.

Prognostic impact of the PAPi and contributing key haemodynamics

Thirty‐day mortality did not show a clear association with PAPi quartiles [Q1: n = 2 (1.6%), Q2: n = 10 (8.2%), Q3: n = 3 (2.5%), Q4: n = 5 (4.1%); P = 0.053]. For 386 patients, there is information on an echocardiographic follow‐up 15 (12–17) months after AVR. There was an improvement in left ventricular ejection fraction (LVEF) from 58 ± 11 to 61 ± 9% and a reduction in the mean aortic valve gradient from 48 ± 17 to 11 ± 4 mmHg. There was no significant correlation between PAPi and the change in LVEF (r = −0.02; P = 0.74). After a median post‐AVR follow‐up of 45 (31–69) months, 44 (9%) deaths had occurred. There was no long‐term mortality difference in patients in different PAPi quartiles (Figure 2). When used as a continuous variable, PAPi was not associated with mortality either [hazard ratio (HR) 1.01 (95% confidence interval [95% CI] 0.92–1.10) per dimensionless unit increase; P = 0.90]. There was no difference between patients undergoing SAVR or TAVR regarding the association between the PAPi and mortality (Figure S1).

Figure 2.

Kaplan–Meier plots (cumulative hazard) comparing survival of patients in different quartiles (Q1–4) for the pulmonary artery pulsatility index (PAPi).

In contrast, patients in the highest sPAP, dPAP and PAPP quartiles had significantly higher mortality than those in the first quartiles (Figure 3A–C). Patients in the highest mRAP quartiles also had a numerically higher mortality than those in the first quartile but this was not significant (Figure 3D). When used as continuous variables, higher sPAP [HR 1.03 (95% CI 1.02–1.05) per mmHg increase; P < 0.001], higher dPAP [HR 1.05 (95%CI 1.02–1.09) per mmHg increase; P = 0.001], higher PPAP [HR 1.05 (95% CI 1.02–1.07) per mmHg increase; P < 0.001] and also higher mRAP [HR 1.09 (95% 1.02–1.16) per mmHg increase; P = 0.008] were associated with higher mortality. Thus, a low numerator of the PAPi equation (i.e., the PAPP, leading to low PAPi) was associated with low mortality, and a high denominator of the PAPi equation (i.e., the mRAP, leading to low PAPi) was associated with high mortality.

Figure 3.

Kaplan–Meier plots (cumulative hazard) comparing survival of patients in different quartiles (Q1–4) for the systolic pulmonary artery pressure (sPAP; Panel A), the diastolic pulmonary artery pressure (dPAP; Panel B), the pulmonary artery pulse pressure (PAPP; Panel C) and the mean right atrial pressure (mRAP; Panel D). sPAP: Q4 versus Q1: hazard ratio (HR) 3.01 [95% confidence interval (95% CI) 1.29–7.09]; P = 0.01. Q3 versus Q1: HR 1.18 (95%CI 0.43–3.26); P = 0.75. Q2 versus Q1: HR 0.97 (95% CI 0.34–2.76); P = 0.95. dPAP: Q4 versus Q1: HR 3.46 (95% CI 1.39–8.56]; P = 0.007. Q3 versus Q1: HR 1.69 (95% CI 0.61–4.65); P = 0.31. Q2 versus Q1: HR 1.57 (95% CI 0.53–4.68); P = 0.42. PAPP: Q4 versus Q1: HR 3.79 (95% CI 1.53–9.39]; P = 0.004. Q3 versus Q1: HR 2.21 (95% CI 0.82–5.99); P = 0.12. Q2 versus Q1: HR 1.12 (95% CI 0.36–3.49); P = 0.84. mRAP: Q4 versus Q1: HR 1.85 (95% CI 0.85–4.01]; P = 0.12. Q3 versus Q1: HR 0.94 (95% CI 0.36–2.48); P = 0.91. Q2 versus Q1: HR 0.92 (95% CI 0.37–2.26); P = 0.85.

Patients in the first SVI quartile (Figure 4B) and those in the first and second PAC quartiles (Figure 4C) had higher mortality than those in the fourth quartiles, and patients in the fourth PVR quartile had higher mortality than patients in the first quartile (Figure 4D). In contrast, survival among patient in different mPAWP quartiles did not significantly differ (Figure 4A).

Figure 4.

Kaplan–Meier plots (cumulative hazard) comparing survival of patients in different quartiles (Q1–4) for the mean pulmonary artery wedge pressure (mPAWP; Panel A), the stroke volume index (SVI; Panel B), the pulmonary artery capacitance (PAC; Panel C) and the pulmonary vascular resistance (PVR; Panel D). mPAWP: Q4 versus Q1: hazard ratio (HR) 2.15 [95% confidence interval (95% CI) 0.93–4.98]; P = 0.08. Q3 versus Q1: HR 1.68 (95% CI 0.68–4.18); P = 0.21. Q2 versus Q1: HR 1.20 (95% CI 0.45–3.19); P = 0.72. SVI: Q1 versus Q4: HR 2.52 (95% CI 1.05–6.05]; P = 0.04. Q2 versus Q4: HR 1.79 (95% CI 0.71–4.50); P = 0.22. Q3 versus Q4: HR 0.81 (95% CI 0.27–2.40); P = 0.70. PAC: Q1 versus Q4: HR 4.90 (95% CI 1.67–14.4]; P = 0.004. Q2 versus Q4: HR 3.58 (95% CI 1.18–10.89); P = 0.22. Q3 versus Q4: HR 1.90 (95% CI 0.56–6.47); P = 0.31. PVR: Q4 versus Q1: HR 3.08 (95% CI 1.31–7.26]; P = 0.01. Q3 versus Q1: HR 1.41 (95% CI 0.54–3.72); P = 0.48. Q2 versus Q1: HR 0.89 (95% CI 0.30–2.64); P = 0.83.

When based on the above findings, patients were categorized into four groups using supramedian (high) versus inframedian (low) PAPi and supramedian (high) versus inframedian (low) PAC (Figure S2), patients with low PAPi/low PAC (highest mortality) and patients with high PAPi/low PAC had significantly higher mortality than patients with low PAPi/high PAC (lowest mortality; referent). Mortality did not differ between patients with high PAPi/high PAC and those with low PAPi/high PAC.

Discussion

In this study, evaluating the haemodynamic contributors and the prognostic value of a low PAPi in a relatively large and unselected patient cohort with severe AS undergoing a detailed non‐invasive and invasive pre‐AVR assessment with post‐AVR follow‐up of several years, we obtained the following key findings: first, a low PAPi was not a marker of more severe AS or LV or RV dysfunction, more severe symptoms or higher surgical risk. Second, based on its definition, a low PAPi was the result of a low PAPP and high mRAP, but this was associated with only a mildly reduced SVI but a high PAC. Third, a low PAPi was not associated with increased short‐term or long‐term post‐AVR mortality (graphical abstract). This is in line with the observation that a high PAPP (numerator of the PAPi equation; rather than a low PAPP), a high mRAP (denominator of the PAPi equation) and low SVI and a low PAC (determinants of PAPP; PAPP = SVI/PAC) were markers of increased mortality, that is, the haemodynamic contributors of PAPi equation had ‘opposing’ prognostic effects.

Because the PAPi is determined by PAPP (and thereby SVI and PAC) and mRAP, and the PAWP can affect the PAC, ¹⁶ it is a useful surrogate for (RV) SVI only in settings were PAC, mRAP and mPAWP are relatively uniform, for example, in patients with RV infarction (PAC normal, mPAWP normal and mRAP high) or patients with advanced heart failure and post‐capillary pulmonary hypertension and RV failure (PAC low, mPAWP high and mRAP high). ¹ The PAPi substantially varies between patients with different cardiac conditions. In patients with isolated RV infarction, the PAPi is lower (low SVI, normal, i.e., high PAC, high mRAP) than in patients with advanced heart failure and RV dysfunction (low SVI and abnormal, i.e., low PAC, high normal to high mRAP). Although we studied an apparently homogenous population (all patients with severe AS), we still found a wide range of PAPi values, and even within the lowest PAPi quartile, there was a broad spectrum of haemodynamic profiles. Interestingly, neither the low nor the high PAPi patient had a clearly ‘poor’ or a ‘favourable’ haemodynamic constellation but both the low (high mRAP but high PAC) and the high (low mRAP but low PAC) PAPi patients had both ‘poor’ and ‘favourable’ haemodynamic features.

Based on the above findings and consideration, it is not surprising that we found no association between PAPi and mortality. On the other hand, all key haemodynamic contributors of the PAPi equation provided strong prognostic information, that is, the PAPP (numerator) and SVI and PAC, respectively (determinants of the numerator), on the one hand and the mRAP on the other hand (denominator). These parameters obviously ‘neutralized’ each other in terms of mortality prediction. When we categorized patients into four groups based on low versus high PAPi and low versus high PAC, it became obvious that patients with both low or high PAPi can have a favourable prognosis as long as the PAC is high (Figure S2). Interestingly, this observation is exactly in line with a very recently published study in patients with heart failure and preserved left ventricular ejection fraction. ¹⁷ Thus, in the present population of unselected AS patients evaluated prior to AVR, the PAPi was not a sensitive prognostic tool because this parameter can incorporate various combinations of SVI and PAC (and mRAP). The question arises why other studies among AS patients found an association between low PAPi and poor prognosis. ¹¹ , ¹² Oshima et al. ¹² studied an elderly TAVR population (n = 227, approximately 13 years older than our patients) with severe AS, normal LVEF and favourable haemodynamics (lower mRAP, sPAP and mPAWP than in the present study, higher cardiac index), Patients with a PAPi <2.1 had a higher risk (HR approximately 7) of heart failure hospitalizations within 2 years than those with a PAPi above this threshold. Thus, the study used a different endpoint. Otherwise, there are no obvious explanations for the discrepancy. However, data on SVI and PAC were not shown, which makes a more detailed interpretation difficult. Huang et al. ¹¹ studied 31 patients undergoing emergency TAVR (mainly because of cardiogenic shock), and PAPi <1.8 was a predictor of in‐hospital mortality. These patients with severe AS were of similar age as our patients but had clearly reduced LVEF (approximately 30%) and higher mPAP and mPAWP but similar cardiac index as our patients. However, these patients needed vasopressors and often even mechanical circulatory support to achieve this cardiac index and therefore were in different haemodynamic situation. Thus, we cannot exclude that there is role for the PAPi in selected very sick AS patients.

The literature on PAPi outside of the AS setting revealed conflicting results. The PAPi has been found to predict RV failure among LV assist device recipients, ³ , ⁴ , ⁵ acute kidney injury after cardiac transplantation, ¹⁸ mortality and hospitalization at 6 months among selected patients with advanced heart failure ⁶ and 1 year mortality in patients with pulmonary arterial hypertension. ⁹ On the other hand, the PAPi was not associated with 30 days mortality in patients with cardiogenic shock. ⁷ In two recent studies, the PAPi did not predict long‐term mortality after cardiac transplantation, ⁸ and it was a predictor of 1 year mortality as a dichotomized but not as a continuous variable in a large population of contemporary pulmonary arterial hypertension patients. ¹⁰ The by far largest study on the PAPi is based on a cohort of 8285 unselected patients undergoing right heart catheterization in one single center over a period of 12 years. ¹⁹ In this study by Zern et al., ¹⁹ patients in the lowest PAPi quartile (median value 1.7) had higher risk of death, heart failure hospitalizations and major adverse cardiac events than those in the fourth quartile (median value 8.7) after a mean follow‐up of 6.7 years. Interestingly, the haemodynamic constellations were overall relatively similar as in our study: those in the first PAPi quartile had the lowest PAPP and highest mRAP and also the highest (‘best’) PAC (not explicitly reported but approximately 3.8 mL/mmHg) and lowest (‘best’) PVR (1.6 WU) and those in the fourth quartile had the highest PAPP, the lowest mRAP and the lowest (‘worst’) PAC (approximately 3.0 mL/mmHg) and the highest (‘worst’) PVR (1.9 WU). It is somewhat counterintuitive that the patients with the highest PAC and the lowest PVR had the worst outcome when considering the consistent data on the poor prognostic impact of a low PAC and high PVR, respectively, in many settings, ¹⁴ , ²⁰ , ²¹ , ²² , ²³ but the study included a large number of patients and therefore had a high power. To explain the discrepant findings of our study one could argue that our much smaller study compared with the work by Zern et al. ¹⁹ was underpowered but there was not even a signal of a worse long‐term mortality in patients in the lowest PAPi quartile (Figure 2). We think that this underscores that the prognostic value of the PAPi strongly depends on the clinical setting. The patients in the study by Zern et al. ¹⁹ had a broader spectrum of mPAWP values compared with the present data and the mPAWP impacts on the position of the hyperbolic relationship between PAC and PVR. ¹ , ¹⁶ Whether this may play a role remains speculative however. While the study by Zern et al. ¹⁹ included an impressive number of patients and produced significant results, it included a population with a broad variety of conditions, and therefore, the clinical applicability of the data remains somewhat limited, particularly when considering the conflicting data on PAPi even in setting with relatively uniform haemodynamic patterns.

We therefore think that the PAPi is not ready for clinical use in patients with severe AS although the parameter would be attractive because it is easy to assess. At least in AS patients, the PAPi is not nuanced enough to identify the full haemodynamic profile of an individual patient because a certain PAPP value can reflect various combination of SVI and PAC. Several studies have shown the prognostic importance of PAC and PVR among patients with AS, ¹⁴ , ²¹ , ²⁴ , ²⁵ and this cannot be replaced by PAPi.

Limitations

First, the number of patients and the number of events were relatively small, and we studied a mixed SAVR and TAVR population. Thus, power was limited, particularly in the TAVR subgroup. Therefore, a type II error cannot be excluded. Still, this is one of the largest cohorts of unselected AS patients with detailed invasive haemodynamics reported in the literature. Inspection of the Kaplan–Meier plots in Figures 2 and S1 shows that limited power is unlikely to be the main reason for the neutral result and that the mode of AVR had not substantial impact on the association between PAPi and mortality. Second, at the first glance, this is a negative study. However, we provide very detailed characteristics of patients with low versus high PAPi and thereby an intuitive explanation of the mechanisms underlying the neutral prognostic effect of the PAPi in this setting and potentially also other settings. Third, to assess cardiac output, we have employed the indirect Fick method, which may be subject to error, as oxygen consumption is often inaccurately estimated. ²⁶ This likely affects all cardiac output‐based measurements, including PVR, SVI and PAC. Thus, our analysis and discussion of the PAPi contributors should be interpreted cautiously with this limitation in mind. It must, however, be noted that this technique is routinely used in clinical practice. Fourth, as outlined above, echocardiography data were of limited quality and incomplete. Therefore, they were used for descriptive purposes only. The strength of the study is the availability of detailed invasive haemodynamic data from a relatively large unselected AS population.

Conclusions

In unselected patients with severe AS, the PAPi did not predict post‐AVR mortality. This may be explained by the fact that the low PAPP in those with low PAPi was mainly a reflection of a high PAC rather than a low SVI.

Funding

None.

Conflict of interest

No conflict of interest.

Supporting information

Figure S1. Kaplan–Meier plots (cumulative hazard) comparing survival of patients in different quartiles (Q1–4) for the pulmonary artery pulsatility index (PAPi) in patients undergoing surgical (panel A) or transcatheter (panel B) aortic valve replacement.

Figure S2. Kaplan–Meier plots (cumulative hazard) comparing survival of patients in different strata for the pulmonary pulsatility index (PAPi) and the pulmonary artery capacitance (PAC). For definitions please see text. PAPi low/PAC low versus PAPi low/PAC high: hazard ratio (HR) 4.32 [95% confidence interval (95%CI) 1.62–12.07]; P = 0.004. PAPi high/PAC low versus PAPi low/PAC high: HR 3.26 (95%CI 1.19–8.90); P = 0.02. PAPi high/PAC high versus PAPi low/PAC high: HR 1.97 (95%CI 0.63–6.21); P = 0.25.

Data S1. Supporting Information.

Rechsteiner, L. A. , Weber, L. , Haager, P. K. , Rigger, J. , Chronis, J. , Ammann, P. , Brenner, R. , Schmiady, M. O. , Rickli, H. , and Maeder, M. T. (2025) The pulmonary artery pulsatility index in patients with severe aortic stenosis undergoing valve replacement. ESC Heart Failure, 12: 3483–3493. 10.1002/ehf2.15378.