Evaluation of the European Society of Cardiology/European Respiratory Society derived three- and four-strata risk stratification models in pulmonary arterial hypertension: introducing an internet-based risk stratification calculator

Abstract

Aims

Estimation of prognosis in pulmonary arterial hypertension (PAH) has been influenced by that various risk stratification models use different numbers of prognostic parameters, as well as the lack of a comprehensive and time-saving risk assessment calculator. We therefore evaluated the various European Society of Cardiology (ESC)-/European Respiratory Society (ERS)-based three- and four-strata risk stratification models and established a comprehensive internet-based calculator to facilitate risk assessment.

Methods and results

Between 1 January 2000 and 26 July 2021, 773 clinical assessments on 169 incident PAH patients were evaluated at diagnosis and follow-ups. Risk scores were calculated using the original Swedish Pulmonary Arterial Hypertension Registry (SPAHR)/Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension (COMPERA) three-strata model, the updated SPAHR three-strata model with divided intermediate risk, and the simplified three-parameter COMPERA 2.0 four-strata model. The original SPAHR/COMPERA and the updated SPAHR models were tested for both 3–6 and 7–11 available parameters, respectively. Prognostic accuracy [area under the receiver operating characteristic (ROC) curve (AUC)] and Uno’s cumulative/time-dependent C-statistics (uAUC) were calculated for 1-, 3-, and 5-year mortality. At baseline, both the original SPAHR/COMPERA and the updated SPAHR models, using up to six parameters, provided the highest accuracy (uAUC = 0.73 for both models) in predicting 1-, 3-, and 5-year mortality. At follow-ups, the updated SPAHR model with divided intermediate risk (7–11 parameters) provided the highest accuracy for 1-, 3-, and 5-year mortality (uAUC = 0.90), followed by the original SPAHR/COMPERA model (7–11 parameters) (uAUC = 0.88) and the COMPERA 2.0 model (uAUC = 0.85).

Conclusions

The present study facilitates risk assessment in PAH by introducing a comprehensive internet-based risk score calculator (https://www.svefph.se/risk-stratification). At baseline, utilizing the original or the updated SPAHR models using up to six parameters was favourable, the latter model additionally offering sub-characterization of the intermediate risk group. Our findings support the 2022 ESC/ERS pulmonary hypertension guidelines' strategy for risk stratification suggesting the utilization of a three-strata model at baseline and a simplified four-strata model at follow-ups. Our findings furthermore support the utility of the updated SPAHR model with divided intermediate risk, when a more comprehensive assessment is needed at follow-ups, complementing the three-parameter COMPERA 2.0 model. Larger multi-centre studies are encouraged to validate the utility of the updated SPAHR model.

Take home message

By introducing an internet-based risk score calculator (https://www.svefph.se/risk-stratification), risk assessment is facilitated. Our results support the 2022 ESC/ERS pulmonary hypertension guidelines' risk stratification strategy, additionally suggesting the updated SPAHR three-strata model with divided intermediate risk, as a promising complement to the new simplified three-parameter COMPERA 2.0 four-strata strategy, when a more comprehensive overview is needed.

Graphical Abstract

Pulmonary arterial hypertension, Prognosis, Risk stratification, Follow-up

Introduction

Pulmonary arterial hypertension (PAH) is an angio-proliferative vasculopathy characterized by progressive increase in pulmonary vascular resistance (PVR), leading to right heart failure (HF) and premature death.¹ Individualizing treatment for patients with PAH is challenging as disease progression is not always preceded by overt changes in symptoms or deterioration of a single clinical parameter.² In the last three decades, marked progress in PAH treatment has been made, related to evolution of new drugs, and strategies for drug initiation and combination based on risk stratification.^3-5 Multi-parametric risk stratification is based on assessment of prognostic variables across demographics, biochemical markers, exercise testing, imaging, and haemodynamics, where combining many variables yields better prognostic overview.^6-9 The 2015 and 2022 European Society of Cardiology (ESC)/European Respiratory Society (ERS) pulmonary hypertension (PH) guidelines recommend regular risk assessment to guide treatment decisions.^3,4,10-13 Additionally, improvements in the risk stratification scores are associated with improved survival, deeming them as promising outcome measures in clinical PAH trials.¹⁴

Numerous risk stratification strategies have been established by utilizing different PAH registries based on variables and cut-offs defined by the 2015 ESC/ERS guidelines.^15-17 The three-tiered 2015 ESC/ERS strategy, initially based on expert opinion, was subsequently validated with different calculation models based on incident populations from the Swedish Pulmonary Arterial Hypertension Registry (SPAHR), the French Pulmonary Hypertension Registry (FPHR), and the Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension (COMPERA).^15-17 Other non-ESC-/ERS-based risk stratification models derived from the United States Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) have also been validated, including REVEAL, REVEAL 2.0, and REVEAL Lite 2, encompassing different numbers and combinations of predictive variables.^18-20

While the different risk stratification models based on the 2015 ESC/ERS guidelines varied in terms of the number of included parameters, a limitation was the large size of the intermediate risk group, encompassing 67–76% of patients at baseline and 60–64% during follow-up.^{12,13,16,17,21,22} Several approaches for dividing the intermediate risk group into ‘intermediate-low’ and ‘intermediate-high’ have thereafter been proposed by (i) new limits of the calculated average risk score using the updated SPAHR three-strata model with divided intermediate risk,²³ (ii) the addition of the ‘tricuspid annular plane systolic excursion/systolic pulmonary arterial pressure’ in conjunction with 6-min walking distance (6MWD),²⁴ and more recently (iii) the new simplified three-parameter COMPERA 2.0 four-strata model, based on World Health Organization Functional Class (WHO-FC), 6MWD, and N-terminal prohormone of brain natriuretic peptide (NT-proBNP).^22,25 The development of the COMPERA 2.0 four-strata model was based on 1655 patients and externally validated by the FPHR using 2882 patients at baseline and 2082 at follow-up.^22,25 Given the recent refined strategy for risk stratification in the 2022 ESC/ERS PH guidelines,^12,13 endorsing the use of the three-strata model at baseline and the simplified three-parameter COMPERA 2.0 four-strata model at follow-up, the guidelines also emphasize the importance to include additional variables when clinically needed for a more comprehensive multi-parametric overview, especially right heart imaging and haemodynamics.^12,13 However, the approach for incorporation of additional parameters into risk assessment is neither specified in the current 2022 ESC/ERS guidelines nor offered by the current simplified three-parameter COMPERA 2.0 four-strata model.^12,13

During assessment of patients with PAH, expert perception or clinical gestalt alone may overestimate or underestimate the risk status.^26,27 In a study based on real-world experience addressing barriers for incorporation of risk assessment into clinical practice, it was found that risk assessment in PAH is often underutilized although many clinicians attest to the use of validated risk stratification tools. Also, lack of time among clinicians was found to prevent regular use of risk assessment tools. Thus, establishing ‘technology-based solutions as time-saving tools’ was recommended to overcome this barrier.²⁷ To facilitate risk assessment in PAH, we aimed to (i) establish a comprehensive internet-based risk score calculator; (ii) evaluate the prognostic accuracy of the new simplified three-parameter COMPERA 2.0 four-strata model in relation to the original SPAHR/COMPERA three-strata model and the updated SPAHR three-strata model with divided intermediate risk, as a tool for incorporation of additional parameters when clinically needed; and (iii) investigate whether the prognostic accuracy of the original SPAHR/COMPERA and the updated SPAHR three-strata model with divided intermediate risk differ depending on the number of included prognostic parameters.

Methods

Study population

The present study is based on a well-characterized patient cohort evaluated at Skåne University Hospital in Lund, Sweden, between 1 January 2000 and 26 July 2021. In total, 2117 evaluations were performed on 402 patients. The Haemodynamic Lab at Skåne University Hospital in Lund is one of the seven PH centres in Sweden where repetitive right heart characterizations (RHCs) assessments have been performed the most extensively, i.e. in 100% at baseline and ∼80% of all follow-ups.²⁸ After applying the inclusion and exclusion criteria (Figure 1), 812 assessments on 169 incident patients with idiopathic PAH (IPAH), familial PAH (FPAH), systemic sclerosis-associated PAH (SSc-APAH), and other connective tissue disease-associated PAH (other CTD-APAH) were identified. Considering the presence of several follow-ups for a large proportion of the patients and to ensure one event per patient, the first assessments predicting 1-, 3-, and 5- year mortality were included (Figure 1). The study conforms with the Declaration of Helsinki and Istanbul and was approved by the regional ethical board in Lund, Sweden (Dnr: 2010/114, 2011/777).

Figure 1

Overview of the selection of the study population. CHD-APAH, pulmonary arterial hypertension (PAH) associated with congential heart disease; CTD-APAH, PAH associated with connective tissue disease without interstitial lung disease; FPAH, familial PAH; IPAH, idiopathic PAH; PH, pulmonary hypertension; PVOD, pulmonary veno-occlusive disease; RHC, right heart catheterization; SSc-APAH, PAH associated with systemic sclerosis without interstitial lung disease.

Diagnostic procedures

The patients with PAH were diagnosed by experienced cardiologists and defined at time of analysis in accordance with clinical practice and prevailing ESC/ERS PH guidelines, as a resting mPAP ≥ 25 mmHg, at a pulmonary arterial wedge pressure ≤ 15 mmHg and a PVR ≥ 3 Wood units, as assessed by RHC.^3,4 RHCs were performed in the supine position at rest, predominantly via the right internal jugular vein, using a Swan–Ganz catheter (Baxter Health Care Corp., Santa Ana, CA, USA). Cardiac output was measured by thermodilution. Echocardiography and/or magnetic resonance tomography, spirometry with diffusion capacity and high-resolution computed tomography, and pulmonary scintigraphy were performed to exclude Groups II, III, and IV PH.

Risk assessment

The original SPAHR/COMPERA three-strata model was first described by Kylhammar et al.¹⁶ The SPAHR/COMPERA three-strata model is based on the SPAHR equation that categorizes patients into low, intermediate, or high risk for 1-year mortality, by assigning a score (low risk = 1, intermediate risk = 2, and high risk = 3) to parameters based on thresholds in the 2015 ESC/ERS PH guidelines. The variables are classified into modifiable parameters including WHO-FC, 6MWD, NT-proBNP, right atrial area, pericardial effusion, mean right atrial pressure (mRAP), cardiac index (CI), and mixed venous oxygen saturation (SvO₂), as well as clinical observations comprising signs of right HF, progression of symptoms, and syncope. The overall risk category (low risk, intermediate risk, or high risk) is determined by the average from the scores of the available parameters for each patient, rounded off to the nearest integer (1–1.49 = low risk, 1.5–2.49 = intermediate risk, and ≥2.5 high risk) (Figures 2 and 3).^3,4,16

Figure 2

Calculation of risk scores using the European Society of Cardiology-/European Respiratory Society-based three-strata and four-strata risk stratification models. SPAHR, Swedish Pulmonary Arterial Hypertension Registry; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension.

Figure 3

Comprehensive risk assessment in pulmonary arterial hypertension (three-strata) using the original SPAHR/COMPERA model and the updated SPAHR model with divided intermediate risk. COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; SPAHR, Swedish Pulmonary Arterial Hypertension Registry.

In the new updated SPAHR three-strata model with divided intermediate risk, the calculated mean is rounded off according to modified thresholds for the intermediate risk (1.5–1.99 = intermediate-low risk, and 2.0–2.49 = intermediate-high risk), separating the risk categories into four groups.²³ The updated SPAHR three-strata model with divided intermediate risk is defined as a three-strata model, as it is based on parameter thresholds for low, intermediate, and high risk; however, it allows separation of the risk score into four separate risk groups (low, intermediate-low, intermediate-high, and high risk) (Figures 2 and 3).

The FPHR invasive model, developed after the 2015 ESC/ERS guidelines, assesses WHO-FC, 6MWD, CI, and mRAP for each patient, where the overall score is defined by the number of low-risk parameters according to the thresholds prescribed by the 2015 ESC/ERS guidelines risk stratification table. The FPHR non-invasive model is based on the number of low-risk parameters among WHO-FC, 6MWD, and NT-proBNP or BNP.^3,4,15 The FPHR models were not included in the current analyses as they were not developed to divide the risk status into three different risk categories, i.e. low, intermediate, and high risk, as described by the 2015 ESC/ERS PH guidelines.^3,4,15

The new simplified three-parameter COMPERA 2.0 four-strata model is a non-invasive risk score, based on WHO-FC, 6MWD, and BNP/NT-proBNP, categorizing patients into low, intermediate-low, intermediate-high, and high risk, by assessing a score according to thresholds in the 2022 ESC/ERS PH guidelines, in conjunction with refined BNP/NT-proBNP cut-offs.^{3,4,12,13,20,22} The overall risk category is determined by assessing a risk score ranging from 1 to 4 for each of WHO-FC, 6MWD, and NT-proBNP. The calculation is similar to the SPAHR/COMPERA models using the SPAHR equation where the sum is divided by the number of available parameters and rounded off to the nearest integer.^16,22,25 A mean score of 1, 2, 3, or 4 corresponds to low risk, intermediate-low risk, intermediate-high risk, and high risk, respectively,²² (Figure 2).

Study analyses

In the present study, we evaluated at baseline and follow-up assessments the prognostic accuracy of the original three-strata SPAHR/COMPERA model, the updated SPAHR three-strata model with divided intermediate risk, and the new COMPERA 2.0 four-strata model.

In the original SPAHR/COMPERA model based on the SPAHR equation, up to eight and six parameters were validated by the SPAHR and the COMPERA registries, respectively.^16,29 In the present study, we divided the study population into having 3–6 and 7–11 available prognistic parameters and calculated risk scores using the original SPAHR/COMPERA and the updated SPAHR models, in an attempt to evaluate how the numbers of parameters influence the prognostic accuracy.

Several exploratory analyses were performed, including (i) applying the new modified NT-proBNP cut-offs proposed in the COMPERA 2.0 four-strata model to both the original SPAHR/COMPERA model and the updated SPAHR three-strata model with divided intermediate risk; (ii) exploring the prognostic accuracy of a simplified, non-invasive score by only applying WHO-FC, 6MWD, and NT-proBNP using the original SPAHR/COMPERA- and the updated SPAHR model; and (iii) calculation of risk scores using only two variables among WHO-FC, 6MWD, and NT-proBNP, by applying the original SPAHR equation and the updated SPAHR equation with divided intermediate risk, as well as the COMPERA 2.0 four-strata equation. The bi-parametric analysis was performed to evaluate the reliability of a very simplified model in case of limited access to a more thorough evaluation, reflecting the situation during the COVID-19 pandemic where patients were not advised to perform inward evaluations.

Statistical analysis and study set-up

Statistical analyses were performed using R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism version 9.00 (GraphPad Software, La Jolla CA, USA). Data distribution was determined using histograms. Non-parametric statistics were applied due to the predominance of non-Gaussian–distributed data. Data were presented as medians (interquartile range; 25th–75th percentiles), unless otherwise stated. Receiver operating characteristic (ROC) curves were applied to test the prognostic accuracy of the risk stratification scores. The area under the ROC curve (AUC) was used to determine the prognostic accuracy of the different risk stratification models. The prognostic accuracy was defined as the ability to predict the 1-, 3-, and 5-year mortality. Thus, data were divided into three sets corresponding to the prognostic analyses of 1-, 3-, and 5-year mortality rates. All-cause mortality was considered as an event.

To ensure one event per patient, only the first assessment to predict the 1-, 3-, and 5-year mortality was included. For an assessment to qualify for analysis, the presence of ≥3 risk assessment variables defined by ESC/ERS guidelines was required. For the bi-parametric analysis, WHO-FC, 6MWD, and NT-proBNP were extracted from the included assessments. In the original SPAHR/COMPERA and the updated SPAHR models, analyses were divided into 3–6 and 7–11 parameters to ensure an adequate number of events in each group. Time-dependent C-statistics were plotted using Uno’s C-statistics to assess the cumulative/dynamic AUC, displaying the development of the mortality-predictive ability of a given risk stratification score.³⁰

Risk score calculator

An html-based webpage was developed for comprehensive calculation of risk scores, encompassing the original SPAHR/COMEPRA three-strata model, the updated SPAHR three-strata model with divided intermediate risk, the COMPERA 2.0 four-strata model, and the FPHR invasive and non-invasive models. Coding was performed using JavaScript. The webpage was based on a macro-enabled Excel spreadsheet developed by A.A., S.A., and G.R., which has been evaluated by G.R. in the clinics on patients with PAH at baseline and follow-up assessments to guide PAH-specific treatment at the Haemodynamic Lab at Skåne University Hospital in Lund since 2019. The PAH risk stratification calculator can be reached using the following URL: (https://www.svefph.se/risk-stratification).

Results

Population characteristics

The characteristics of the study population are presented in Tables 1 and 2. Patients’ data were censored at analysis on 26 July 2021. At the end of the censoring time, 68.0% (115/169) had died. After baseline within 1 year, 45.0% (76/169) had died; within 3 years, 65.1% (110/169) had died; and within 5 years, 66.9% (113/169) had died. At baseline, all patients underwent RHC. During follow-ups, RHC data were present in 77.5% (468/604), 80.4% (362/450), and 81.0% (285/352) in the 1-, 3-, and 5-year mortality cohorts, respectively. At first follow-up, data on PAH-specific therapy were available for 85.8% (145/169) of the patients, where 37.9% (55/145) had received monotherapy, 53.1% (77/145) combination therapy, and 3.4% (5/145) triple therapy. Also, 3.4% (5/145) of the patients were acute vasoreactivity responders and received calcium channel blockers only. Specific PAH therapy was discontinued for 2.1% (3/145) due to adverse events. Of the 24 patients without available data on PAH therapy at first follow-up, 11 had died and 13 had missing data. Using the original SPAHR/COMPERA 3- to 11-parameter model at baseline, 16.6% (28/169) of the patients were at low risk, 71.0% (120/169) intermediate risk, and 12.4% (21/169) high risk, with a 1-year mortality of 0%, 18.3%, and 42.9%, respectively.

Table 1

Demographics of the study population

Parameter	n (169)	Baseline	1-year		3-year		5-year
			n (604)	Follow-up	n (450)	Follow-up	n (352)	Follow-up
Sex (F)	169	116 (68.6%)	604	467 (77.3%)	450	353 (78.4%)	352	279 (79.3%)
Age at diagnosis/follow-up (years)	169	70 (59–76)	604	69 (55–76)	450	65 (49–75)	352	60 (43–72)
BMI (kg/m2)	162	25 (23–28)	546	25 (22–28)	398	26 (22–29)	312	26 (22–29)
Event (death)	169	31 (18.3%)	604	45 (7.5%)	450	41 (9.1%)	352	16 (4.5%)
Time to event (days)	169	1068 (452–1893)	604	1167 (613–2129)	450	1583 (1001–2659)	352	1915 (995–3063)
Comorbidities
Diabetes mellitus	167	27 (16.2%)
Atrial fibrillation	167	26 (15.6%)
Stroke	167	9 (5.4%)
Ischemic heart disease	167	34 (20.4%)
Previous thyroid disease	166	32 (19.3%)
Diagnosis
FPAH	169	9 (5.3%)	604	64 (10.6%)	450	62 (13.8%)	352	62 (17.6%)
IPAH	169	104 (61.5%)	604	366 (60.6%)	450	280 (62.2%)	352	216 (61.4%)
SSc-APAH	169	46 (27.2%)	604	148 (24.5%)	450	87 (19.3%)	352	60 (17%)
Other CTD-APAH	169	10 (5.9%)	604	26 (4.3%)	450	21 (4.7%)	352	14 (4%)
Medications
Anticoagulation	167	66 (39.5%)	603	255 (42.3%)	449	189 (42.1%)	352	136 (38.6%)
Diuretics	167	99 (59.3%)	602	384 (63.8%)	449	255 (56.8%)	351	180 (51.3%)
CCB	168	38 (22.6%)	601	134 (22.3%)	449	101 (22.5%)	351	101 (28.8%)
ERA	169	125 (74%)	604	504 (83.4%)	450	383 (85.1%)	352	301 (85.5%)
Prostacyclin	169	6 (3.6%)	604	62 (10.3%)	450	51 (11.3%)	352	47 (13.4%)
PDEi/sGC	169	75 (44.3)	604	475 (78.6%)	450	346 (76.9%)	352	273 (77.6%)
Haemodynamics
MAP (mmHg)	143	96 (88–106)	399	91 (83–98)	310	92 (86–98)	242	91 (86–98)
Heart rate (beats/min)	138	78 (69–88)	506	72 (64–81)	368	72 (64–81)	284	71 (63–81)
mRAP (mmHg)	168	7 (3–11)	465	5 (3–8)	359	5 (2–7)	282	4 (2–7)
mPAP (mmHg)	169	46 (39–55)	467	40 (34–48)	362	40 (33–47)	285	39 (32–47)
PAWP (mmHg)	169	8 (5–11)	465	8 (6–11)	359	8 (6–11)	282	8 (6–11)
PVR (WU)	168	9.4 (6.8–13)	466	6.5 (4.4–8.8)	361	6.3 (4.1–8.3)	284	6.1 (3.8–8.2)
CI (L/min/m2)	167	2.2 (1.8–2.7)	459	2.8 (2.3–3.2)	355	2.9 (2.4–3.3)	280	2.9 (2.4–3.4)
SvO2 (%)	110	58 (51–65)	395	66 (58–72)	316	68 (60–73)	252	70 (64–74)
Risk stratification parameters
Number of risk parameters	169	7 (5.5–7)	604	6 (4–7)	450	6 (5–8)	352	6 (5–9)
NT-proBNP (ng/L)	125	2088 (684–4316)	492	519 (178–1559)	364	332 (141–869)	288	247 (111–586)
WHO-FC	169
I		0 (%)	555	88 (15.9%)	402	84 (20.9%)	315	84 (26.7%)
II		35 (20.7%)	555	252 (45.4%)	402	225 (56%)	315	169 (53.7%)
III		113 (66.9%)	555	190 (34.2%)	402	67 (16.7%)	315	60 (19%)
IV		21 (12.4%)	555	25 (4.5%)	402	6 (1.5%)	315	2 (0.6%)
6MWD (m)	136	245 (171–350)	554	349 (240–481)	427	405 (285–508)	339	450 (347–520)
Pericardial effusion	129	20 (15.5%)	285	31 (10.9%)	228	27 (11.8%)	193	23 (11.9)
RA area (cm2)	34	24 (18–28)	163	19 (15–24)	144	18 (14–23)	131	18 (14–22)
Signs of RV failure	23	1 (4.3%)	94	6 (6.4%)	90	4 (4.4%)	88	3 (3.4%)
Progression of symptoms	23	19 (82.6%)	101	8 (7.9%)	97	6 (6.2%)	95	6 (6.3%)
Syncope	24	7 (29.2%)	109	3 (2.8%)	105	3 (2.9%)	103	3 (2.9%)

Continuous data are described as median (interquartile range), and categorical data as n (%) of the number of available parameters.

6MWD, 6-min walking distance; BMI, body mass index; CCB, calcium channel blocker; CI, cardiac index; CTD-APAH, pulmonary arterial hypertension (PAH) associated with connective tissue disease without interstitial lung disease; ERA, endothelin receptor antagonist; FPAH, familial PAH; IPAH, idiopathic PAH; MAP; mean arterial pressure; mPAP, mean pulmonary arterial pressure; mRAP, mean right atrial pressure; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; PAWP, pulmonary arterial wedge pressure; PDEi, phosphodiesterase Type 5 inhibitor; PVR, pulmonary vascular resistance; RA, right atrial; RV, right ventricular; sGC, stimulator of soluble guanylate cyclase; SSc-APAH, PAH associated with systemic sclerosis without interstitial lung disease; SvO₂, mixed-venous oxygen saturation; WHO-FC, World Health Organization Functional Class; WU, Wood units.

Table 2

Prognostic pulmonary arterial hypertension risk stratification scores of the study population

Model	Score format (x/x/x …)	1-year				3-year		5-year
SPAHR/COMPERA—(SPAHR equation)		n (169)	Baseline	n (604)	Follow-up	n (450)	Follow-up	n (352)	Follow-up
3–11 parameters, original SPAHR/COMPERA	(1/2/3)	169	28/120/21	604	261/300/43	450	245/192/13	352	227/120/5
3–11 parameters, updated SPAHR, divided intermediate risk	(1/1.75/2.25/3)	169	28/55/65/21	604	261/182/118/43	450	245/136/56/13	352	227/87/33/5
¤ 3–11 parameter, original SPAHR/COMPERA, new cut-offs	(1/2/3)	169	28/119/22	604	259/302/43	450	244/193/13	352	226/121/5
¤ 3–11 parameters, updated SPAHR, divided intermediate risk, new cut-offs	(1/1.75/2.25/3)	169	28/54/65/22	604	259/180/122/43	450	244/135/58/13	352	226/87/34/5
Bi-parametric SPAHR/COMPERA—(SPAHR equation)
NT-proBNP and WHO-FC, original SPAHR/COMPERA	(1/2/3)	125	10/48/67	474	149/225/100	347	145/168/34	276	141/117/18
¤ NT-proBNP and WHO-FC, original SPAHR/COMPERA, new cut-offs	(1/2/3)	125	10/41/74	474	149/212/113	347	145/160/42	276	141/112/23
NT-proBNP and WHO-FC, updated SPAHR, divided intermediate risk	(1/1.75/2.25/3)	125	10/15/33/67	474	149/136/89/100	347	145/117/51/34	276	141/87/30/18
¤ NT-proBNP and WHO-FC, updated SPAHR, divided intermediate risk, new cut-offs	(1/1.75/2.25/3)	125	10/14/27/74	474	149/124/88/113	347	145/107/53/42	276	141/80/32/23
NT-proBNP and 6MWD, original SPAHR/COMPERA	(1/2/3)	113	6/37/70	462	112/220/130	350	112/183/55	282	111/140/31
¤ NT-proBNP and 6MWD, original SPAHR/COMPERA, new cut-offs	(1/2/3)	113	6/31/76	462	112/203/147	350	112/170/68	282	111/132/39
NT-proBNP and 6MWD, updated SPAHR, divided intermediate risk	(1/1.75/2.25/3)	113	6/11/26/70	462	112/104/116/130	350	112/94/89/55	282	111/84/56/31
¤ NT-proBNP and 6MWD, updated SPAHR, divided intermediate risk, new cut-offs	(1/1.75/2.25/3)	113	6/10/21/76	462	112/100/103/147	350	112/90/80/68	282	111/80/52/39
6MWD and WHO-FC, original SPAHR/COMPERA	(1/2/3)	136	10/92/34	508	147/296/65	383	145/218/20	302	142/154/6
6MWD and WHO-FC, updated SPAHR, divided intermediate risk	(1/1.75/2.25/3)	136	10/21/31/74	508	147/191/105/65	383	145/157/61/20	302	142/115/39/6
COMPERA 2.0
COMPERA 2.0	(1/2/3/4)	113	11/15/68/19	447	154/134/114/45	337	151/111/62/12	270	150/82/35/3
Bi-parameter (NT-proBNP and 6MWD)	(1/2/3/4)	113	6/14/32/61	462	112/135/95/120	350	112/120/64/54	282	111/103/39/29
Bi-parameter (NT-proBNP and WHO-FC)	(1/2/3/4)	125	10/14/29/72	474	149/124/91/110	347	145/107/53/42	276	141/80/32/23
Bi-parameter (6MWD and WHO-FC)	(1/2/3/4)	136	10/21/31/74	508	147/191/105/65	383	145/157/61/20	302	142/115/39/6

¤ indicates new NT-proBNP cut-offs.²²

6MWD, 6-min walking distance; AUC, area under the ROC curve; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; SPAHR, Swedish Pulmonary Arterial Hypertension Registry; WHO-FC, World Health Organization Functional Class.

Risk stratification accuracy in predicting 1-year mortality

At baseline, the original SPAHR/COMPERA and the updated SPAHR 3- to 6-parameter models exhibited the highest mean AUCs (0.73 and 0.73, respectively), followed by COMPERA 2.0 (0.69) (Table 3).

Table 3

AUCs of original and updated risk stratification models based on the numbers of included parameters

Risk stratification models	1-year mortality		3-year mortality		5-year mortality
	Baseline (AUC: 95% CI)	Follow-ups (AUC: 95% CI)	Baseline (AUC: 95% CI)	Follow-ups (AUC: 95% CI)	Baseline (AUC: 95% CI)	Follow-ups (AUC: 95% CI)
SPAHR/COMPERA—(SPAHR equation)
3–11 parameters, original SPAHR/COMPERA	0.67: 0.60–0.75	0.76: 0.70–0.81	0.65: 0.58–0.71	0.76: 0.71–0.82	0.67: 0.60–0.73	0.75: 0.64–0.85
3–11 parameters, updated SPAHR, divided intermediate risk	0.67: 0.58–0.77	0.82: 0.76–0.88	0.63: 0.55–0.71	0.79: 0.73–0.86	0.65: 0.57–0.73	0.80: 0.68–0.92
¤ 3–11 parameters, original SPAHR/COMPERA, new cut-offs	0.67: 0.60–0.74	0.76: 0.70–0.81	0.64: 0.58–0.70	0.76: 0.70–0.82	0.66: 0.59–0.73	0.74: 0.64–0.85
¤ 3–11 parameters, updated SPAHR, divided intermediate risk, new cut-offs	0.68: 0.59–0.78	0.82: 0.76–0.88	0.63: 0.55–0.71	0.80: 0.73–0.86	0.65: 0.57–0.73	0.80: 0.68–0.92
3–6 parameters, original SPAHR/COMPERA	0.73: 0.62–0.84	0.74: 0.67–0.81	0.69: 0.60–0.79	0.71: 0.63–0.79	0.73: 0.63–0.84	0.63: 0.45–0.81
3–6 parameters, updated SPAHR, divided intermediate risk	0.73: 0.60–0.86	0.79: 0.72–0.86	0.67: 0.56–0.78	0.75: 0.66–0.84	0.68: 0.56–0.81	0.72: 0.50–0.93
¤ 3–6 parameters, original SPAHR/COMPERA, new cut-offs	0.72: 0.61–0.84	0.74: 0.67–0.81	0.68: 0.59–0.78	0.71: 0.63–0.79	0.62: 0.61–0.83	0.63: 0.45–0.81
¤ 3–6 parameters, updated SPAHR, divided intermediate risk, new cut-offs	0.73: 0.60–0.86	0.80: 0.72–0.87	0.67: 0.56–0.78	0.75: 0.66–0.84	0.67: 0.54–0.80	0.72: 0.50–0.93
7–11 parameters, original SPAHR/COMPERA	0.62: 0.53–0.70	0.78: 0.68–0.88	0.58: 0.51–0.66	0.86: 0.83–0.90	0.60: 0.51–0.68	0.90: 0.87–0.93
7–11 parameters, updated SPAHR, divided intermediate risk	0.61: 0.48–0.74	0.87: 0.75–0.99	0.58: 0.46–0.69	0.88: 0.83–0.93	0.64: 0.53–0.75	0.92: 0.87–0.97
¤ 7–11 parameters, original SPAHR/COMPERA, new cut-offs	0.62: 0.53–0.70	0.78: 0.68–0.88	0.58: 0.51–0.66	0.86: 0.83–0.90	0.60: 0.51–0.68	0.90: 0.87–0.93
¤ 7–11 parameters, updated SPAHR, divided intermediate risk, new cut-offs	0.63: 0.51–0.76	0.87: 0.75–0.98	0.59: 0.48–0.70	0.89: 0.84–0.94	0.65: 0.54–0.76	0.92: 0.87–0.97
3 parametersa, original SPAHR/COMPERA	0.62: 0.54–0.69	0.75: 0.69–0.82	0.58: 0.51–0.65	0.79: 0.73–0.84	0.62: 0.54–0.70	0.74: 0.61–0.86
3 parametersa, updated SPAHR, divided intermediate risk	0.60: 0.48–0.72	0.81: 0.74–0.88	0.55: 0.45–0.65	0.82: 0.76–0.88	0.63: 0.54–0.73	0.78: 0.64–0.92
¤ 3 parametersa, original SPAHR/COMPERA, new cut-offs	0.62: 0.54–0.69	0.75: 0.69–0.82	0.58: 0.51–0.65	0.79: 0.73–0.84	0.62: 0.54–0.70	0.73: 0.61–0.86
¤ 3 parametersa, updated SPAHR, divided intermediate risk, new cut-offs	0.62: 0.51–0.74	0.82: 0.75–0.89	0.56: 0.46–0.66	0.82: 0.76–0.89	0.64: 0.54–0.74	0.78: 0.64–0.92
COMPERA 2.0	0.69: 0.58–0.80	0.86: 0.81–0.91	0.61: 0.52–0.70	0.87: 0.82–0.91	0.66: 0.58–0.75	0.84: 0.74–0.95

¤ indicates AUC based on the new cut-offs for NT-proBNP.²²

AUCs of importance are highlighted in bold.

6MWD, 6-min walking distance; AUC, area under the ROC curve; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; SPAHR, Swedish Pulmonary Arterial Hypertension Registry; WHO-FC, World Health Organization Functional Class.

a

3 parameter model including WHO-FC, 6MWD, and NT-proBNP.

At follow-ups, the highest mean AUCs were displayed by the updated SPAHR 7- to 11-parameter model with divided intermediate risk (0.87) and COMPERA 2.0 (0.86) (Table 3). The updated SPAHR 7- to 11-parameter model displayed a higher mean AUC than the original SPAHR/COMPERA 7- to 11-parameter model (0.87 vs. 0.78, respectively).

Risk stratification accuracy in predicting 3-year mortality

At baseline, the original SPAHR/COMPERA and updated SPAHR 3- to 6-parameter models remained having the highest mean AUCs (0.69 and 0.67, respectively), whereas COMPERA 2.0 displayed a slightly lower mean AUC (0.61) (Table 3).

At follow-ups, the highest and similar AUCs were displayed by the updated SPAHR 7- to 11-parameter model (0.88), COMPERA 2.0 (0.87), and the original SPAHR/COMPERA 7- to 11-parameter model (0.86) (Table 3).

Risk stratification accuracy in predicting 5-year mortality

At baseline, the original SPAHR/COMPERA and updated SPAHR 3- to 6-parameter model remained having the highest mean AUCs (0.73 and 0.68, respectively), followed by COMPERA 2.0 (0.66) (Table 3).

At follow-ups, the updated SPAHR 7- to 11-parameter model displayed the highest mean AUC (0.92), followed by the original SPAHR/COMPERA 7- to 11-parameter model (0.90) and COMPERA 2.0 (0.84) (Table 3).

The prognostic accuracy in predicting the cumulative 1- to 5-year mortality

The mean cumulative C-statistics (uAUCs) were not calculated for the SPAHR-derived models with the newly proposed NT-proBNP cut-offs as they hardly exhibited any impact on the AUCs (Table 3). At baseline, the original SPAHR/COMPERA and the updated SPAHR 3- to 6-parameter models had the highest uAUCs (0.730 and 0.733, respectively), followed by COMPERA 2.0 (0.68) (Figure 4A, Table 4).

Figure 4

Time-dependent area under the receiver operating characteristic curves, displaying the cumulative predictive ability of risk stratification models at (A) baseline and (B) follow-ups. uAUC, Uno’s cumulative C-statistics for 1- to 5-year mortality; SPAHR, Swedish Pulmonary Arterial Hypertension Registry; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension.

Table 4

Cumulative 1- to 5-year area under the receiver operating characteristic curve of risk stratification models at baseline and follow-ups

Risk stratification model	1- to 5-year mortality (uAUC; Uno’s cumulative C-statistics)
	Baseline	Follow-ups
SPAHR/COMPERA—(SPAHR equation)
3–11 parameters, original SPAHR/COMPERA	0.68	0.75
3–11 parameters, updated SPAHR, divided intermediate risk	0.69	0.80
3–6 parameters, original SPAHR/COMPERA	0.73	0.67
3–6 parameters, updated SPAHR, divided intermediate risk	0.73	0.76
7–11 parameters, original SPAHR/COMPERA	0.60	0.88
7–11 parameters, updated SPAHR, divided intermediate risk	0.62	0.90
3 parameters,a original SPAHR/COMPERA	0.62	0.73
3 parameters,a updated SPAHR, divided intermediate risk	0.62	0.78
COMPERA 2.0	0.68	0.85

6MWD, 6-min walking distance; AUC, area under the ROC curve; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; SPAHR, Swedish Pulmonary Arterial Hypertension Registry; WHO-FC, World Health Organization Functional Class.

a

3 parameter models include WHO-FC, 6MWD and NTpro-BNP.

At follow-ups, the updated SPAHR 7- to 11-parameter model displayed the highest mean uAUC (0.90), followed by the original SPAHR/COMPERA 7- to 11-parameter model (0.88) and COMEPRA 2.0 (0.85) (Figure 4B, Table 4).

Prognostic accuracy of bi-parametric scores

At baseline, in the 1-, 3-, and 5-year mortality cohorts, the combination of WHO-FC and 6MWD displayed higher mean AUCs compared to combining NT-proBNP with WHO-FC or 6MWD (Table 5). They were generally poorer compared to using more parameters (Table 3 and 5).

Table 5

C-statistics (area under the receiver operating characteristic curves) of bi-parametric risk stratification models

Risk stratification strategy	1-year mortality (AUC: 95% CI)		3-year mortality (AUC: 95% CI)		5-year mortality (AUC: 95% CI)
SPAHR-/COMPERA-derived combinations	Baseline	Follow-ups	Baseline	Follow-ups	Baseline	Follow-ups
NT-proBNP and 6MWD	0.54: 0.42–0.66	0.77: 0.71–0.83	0.54: 0.45–0.64	0.78: 0.72–0.84	0.61: 0.52–0.70	0.83: 0.75–0.91
NT-proBNP and 6MWD, updated SPAHR, divided intermediate risk	0.55: 0.43–0.66	0.80: 0.75–0.85	0.54: 0.44–0.63	0.84: 0.80–0.88	0.61: 0.52–0.70	0.84: 0.75–0.93
¤ NT-proBNP and 6MWD	0.57: 0.46–0.68	0.76: 0.70–0.83	0.56: 0.47–0.64	0.81: 0.75–0.87	0.60: 0.52–0.69	0.86: 0.79–0.94
¤ NT-proBNP and 6MWD, updated SPAHR, divided intermediate risk	0.57: 0.47–0.68	0.79: 0.73–0.84	0.56: 0.47–0.64	0.84: 0.79–0.89	0.60: 0.51–0.69	0.85: 0.76–0.94
NT-proBNP and WHO-FC	0.61: 0.51–0.71	0.84: 0.81–0.88	0.60: 0.52–0.69	0.76: 0.70–0.82	0.64: 0.55–0.73	0.79: 0.69–0.89
NT-proBNP and WHO-FC, updated SPAHR, divided intermediate risk	0.61: 0.51–0.71	0.83: 0.76–0.89	0.59: 0.50–0.68	0.82: 0.76–0.89	0.64: 0.55–0.79	0.82: 0.72–0.93
¤ NT-proBNP and WHO-FC	0.62: 0.52–0.71	0.80: 0.74–0.86	0.60: 0.52–0.69	0.79: 0.73–0.85	0.62: 0.53–0.71	0.80: 0.70–0.91
¤ NT-proBNP and WHO-FC, updated SPAHR, divided intermediate risk	0.62: 0.52–0.71	0.82: 0.77–0.88	0.59: 0.51–0.68	0.84: 0.78–0.90	0.62: 0.53–0.71	0.83: 0.73–0.94
WHO-FC and 6MWD	0.62: 0.51–0.74	0.79: 0.73–0.85	0.64: 0.56–0.72	0.73: 0.68–0.73	0.70: 0.63–0.77	0.70: 0.62–0.77
WHO-FC and 6MWD, updated SPAHR, divided intermediate risk	0.66: 0.55–0.77	0.83: 0.78–0.89	0.66: 0.58–0.74	0.79: 0.72–0.85	0.71: 0.63–0.79	0.77: 0.66–0.88
COMPERA 2.0-derived combinations
NT-proBNP and 6MWD	0.62: 0.51–0.74	0.82: 0.77–0.87	0.61: 0.52–0.71	0.86: 0.81–0.91	0.68: 0.59–0.77	0.86: 0.77–0.95
NT-proBNP and WHO-FC	0.63: 0.53–0.72	0.82: 0.77–0.87	0.59: 0.50–0.68	0.84: 0.78–0.90	0.61: 0.52–0.70	0.83: 0.73–0.94
6MWD and WHO-FC	0.66: 0.55–0.77	0.83: 0.78–0.89	0.66: 0.58–0.74	0.79: 0.72–0.85	0.71: 0.63–0.79	0.77: 0.66–0.88

¤ indicates AUC based on the new cut-offs for NT-proBNP.

6MWD, 6-min walking distance; AUC, area under the ROC curve; COMPERA, Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; SPAHR, Swedish Pulmonary Arterial Hypertension Registry; WHO-FC, World Health Organization Functional Class.

At follow-ups, in the 1-year mortality cohort, the mean AUCs did not markedly differ depending on the combination of parameters. In the 3- and 5-year mortality cohorts, combining NT-proBNP with WHO-FC or 6MWD yielded slightly higher mean AUCs compared to WHO-FC and 6MWD, independent of the calculation method. The AUCs were, however, generally poorer than the models incorporating more parameters (Tables 3 and 5).

Discussion

The 2015 and 2022 ESC/ERS PH guidelines endorse regular prognostic risk assessment of patients with PAH at baseline and follow-ups, to guide clinicians in treatment initiation and escalation, as well as to facilitate timely referral for transplantation.^3,4,21 According to the 2022 ESC/ERS PH guidelines, the new recommendations during follow-up assessments are the use of the simplified four-strata model encompassing only WHO-FC, 6MWD, and BNP/NT-proBNP, but also incorporation of additional parameters when clinically needed. There are, however, no specific instructions on which model that should be used at follow-up assessments to achieve a more comprehensive overview, utilizing more than the three parameters stated above. In the present work, we found that the COMPERA 2.0 may be complemented by the updated SPAHR model, which allows discrimination within the intermediate risk group and encompasses additional parameters not included in the simplified four-strata model—thus enabling to fulfil the 2022 ESC/ERS PH guidelines’ recommendations.

Despite sufficient evidence supporting multi-parametric risk assessment in PAH,^3,4,15-17 its application and adherence by clinicians remain suboptimal. In a survey study across Europe and the Unites States, risk assessment was only applied on 59% of PAH patients, and there was a 55% discordance between clinical judgement and calculated risk category.²⁶ In another study based on real-world experience addressing barriers for incorporation of risk assessment into clinical practice, it was found that lack of time among clinicians prevented regular use of risk assessment tools, and technology-based solutions as time-saving tools were recommended to overcome the barrier of suboptimal use of risk assessment.²⁷ The utilization of risk assessment in the clinics has been influenced by the presence of several contemporary risk stratification models advocating the use of different numbers and combinations of prognostic parameters,^21,26 as well as the lack of a time-saving and comprehensive risk calculator. We therefore aimed to establish and introduce an independent html-based webpage to facilitate risk score calculation in PAH and thereby evaluate and provide insight into various ESC/ERS risk assessment models with respect the number of included parameters, during baseline and follow-ups.

In the present study, we analysed unique clinical data with high percentages of haemodynamic assessments, i.e. 100% at baseline and ∼80% at repetitive follow-ups.²⁸ At baseline, we found that the original SPAHR/COMPERA and the updated SPAHR models utilizing three to six parameters provided the highest accuracy in predicting the 1-, 3-, and 5-year mortality in PAH. Providing more than six parameters at baseline did not markedly improve prognostication. This may be due to the absence of weighting in the ESC/ERS strategy, where some parameters may have a lower prognostic ability, diminishing the impact of highly prognostic parameters.^3-5,12,13 This does not imply that less parameters are favourable during baseline assessments, as the prognostic and the additonal impact of clinical observations (clinical signs of right HF, progression of symptoms, and syncope) need further validation and are yet to be established. Altogether, our results indicate that at baseline, the use of the original SPAHR/COMPERA or the updated SPAHR model utilizing three to six parameters is favourable in terms of prognostic accuracy. This supports the recommendation in the 2022 ESC/ERS PH guidelines of using a three-strata strategy at baseline. The updated SPAHR model incorporates an additional stratum, providing further information on patients with intermediate-high and intermediate-low risk, which may be of great importance as it may influence treatment decisions and consequently outcome.^12,13,22,23 However, further multi-centre studies are encouraged to validate the utility of the updated SPAHR model at baseline.

We furthermore found that follow-up assessments predicted mortality better than baseline assessments, consistent with previous studies.^15,16,31 Moreover, at follow-ups, we demonstrate that the original SPAHR/COMPERA 7- to 11-parameter model, the updated SPAHR 7- to 11-parameter model, and the COMPERA 2.0 model provided the highest prognostic accuracies. The updated SPAHR 7- to 11-parameter model with divided intermediate risk and COMPERA 2.0 displayed higher mean AUCs at 1-year follow-ups compared to the original SPAHR/COMPERA 7- to 11-parameter model (Table 4). This may be ascribed to the absence of an additional stratum in the original SPAHR/COMPERA model. In line with the 2022 ESC/ERS PH guidelines, our results support the use of the simplified three-parameter COMPERA 2.0 four-strata model at follow-ups. Importantly, as suggested in the 2022 ESC/ERS PH guidelines, additional variables should be considered, especially right heart imaging and haemodynamics, when a more comprehensive multi-parameter evaluation is needed. With respect to the aforementioned, the present study supports using the updated SPAHR 7- to 11-parameter model with divided intermediate risk at follow-ups, as a plausible strategy, when a more comprehensive multi-parameter overview is required.^12,13 However, future studies are encouraged to validate the updated SPAHR model and the 2022 ESC/ERS guidelines’ newly introduced parameters of cardiac imaging, which have been suggested to enhance risk stratification.^12,13,32-34

Utilizing more parameters at follow-ups may offer additional important clinical information for prognostic assessments. According to the 2022 ESC/ERS PH guidelines, this constitutes an integral part in evaluating patients with PAH, regarding disease severity, deterioration, stability, or improvement, and of importance for accurate treatment decisions.^12,13 We additionally provide data that the use of more parameters (7–11 vs. 3–6) incorporated in the original SPAHR/COMPERA or updated SPAHR models is associated with a higher prognostic accuracy during follow-up assessments to predict short- and particularly long-term mortality, potentially related to a more comprehensive clinical overview.

In the 2022 ESC/ERS PH guidelines, the 1-year mortality estimates for the three-strata strategy were revised for the intermediate and high risk to 5–20% and >20%, respectively.^3,4,12,13 Our results confirm these revised estimates for 1-year mortality in PAH, adding further evidence to previous studies displaying similar findings.^16,17 Notably, in our study, patients were included from early 2000s, which may have influenced survival, as treatment strategies have evolved. The current PH guidelines advocate the use of the three-parameter COMPERA 2.0 four-strata model at follow-ups, for discrimination within the intermediate risk group and to guide therapeutic decision-making, in a simplified manner.^12,13 As achieving or maintaining a low-risk stratum at the very first follow-up is of great importance for outcome,^15-17 the use of the updated SPAHR 7- to 11-parameter model with divided intermediate risk may at the first follow-up evaluation be of potential additional value as it incorporates more parameters across different modalities. We support that this should be evaluated in larger international cohorts and multi-centre studies. Theoretically, patients showing stability after the early follow-ups may thereafter be assessed using the simplified three-parameter COMPERA 2.0 four-strata model, which can be complemented in case of clinical deterioration with more parameters using the updated SPAHR 7- to 11-parameter model with divided intermediate risk. Although reaching treatment goals is reassuring, pathological changes within the pulmonary vasculature and right ventricle can still occur even if no signs of clinical deterioration are apparent.^35,36 In addition to performing multi-parametric risk assessment, taking into account all available prognostic information, including clinical gestalt, PAH aetiology, age, sex, and comorbidities, is also essential.³⁶

We acknowledge that the retrospective design of our study and the relatively low number of patients may provide a limitation. However, our site uniquely provides an extremely high percentage of RHCs at both baseline and follow-ups.²⁸ In the present study, each patient in the 1-, 3-, and 5-year mortality groups, in the follow-up cohorts, could have several assessments, which could introduce a survival bias. However, the same cohorts were used to calculate the prognostic accuracy for the different risk stratification models, and ranking of the C-statistics was only made within each cohort. Although no statistical comparisons were performed due to the relatively low number of independent patients and the overlap between the different mortality cohorts, containing both paired and unpaired data, we evaluate and provide information on the prognostic accuracy of the different three- and four-strata ESC/ERS PH risk stratification strategies.

Conclusion

In the present study, we established and introduced an internet-based calculator to facilitate PAH risk stratification. At baseline, the use of the original SPAHR/COMPERA or the updated SPAHR model with divided intermediate risk (both utilizing three to six parameters) provided the highest prognostic accuracy for 1-, 3-, and 5-year mortality. At follow-ups, the use of the updated SPAHR 7- to 11-parameter model with divided intermediate risk and COMPERA 2.0 provided the highest prognostic accuracies for short- and long-term mortality up to 5 years. Thus, besides supporting the 2022 ESC/ERS PH guidelines strategy for risk stratification by utilizing a three-strata model at baseline and a simplified three-parameter four-strata model at follow-ups, our results indicate that if clinically needed, increasing the numbers of prognostic parameters, by using the updated SPAHR 7- to 11-parameter model, may be a plasuible complement during follow-up assessments when a more comprehensive multi-parametric overview is needed. Larger collaborative multi-centre studies, including employing risk stratification in a prospective setting, as well as validation of cardiac imaging parameters, are encouraged to validate and further develop our findings.

Author Contribution

A.A., S.A., and G.R. designed the study. A.A. drafted the mansucript and analysed the data with the support of S.A., D.K., and G.R. A.A., S.A., and G.R. interpreted the results. D.K. programmed the web-based calculator. S.A. and G.R. revised the manuscript critically.

Lead author biography

Dr. A. Ahmed studied in medical school at Lund University, Sweden, and graduated in 2021, earning a degree of Master of Science in Medicine. He works currently as a junior doctor at the department of Internal Medicine in Helsingborg’s Hospital, Sweden. In 2017, A. Ahmed was introduced by Dr. G. Rådegran to the fascinating and enigmatic field of pulmonary hypertension, which has grown to be one of his passions. He is currently a PhD student at the Department of Clinical Sciences, the Section for Cardiology, Lund University. Apart from clinical work and research, he enjoys practicing calligraphy during his free time.

Data availability

All analyses performed in the current article were in line with the specific ethical approval obtained for this particular study. The data underlying this article can therefore not be shared publicly due to Swedish privacy- and ethical regulations. For data inquires, please contact the corresponding author. Data can, however, be shared contractually upon request, if permission and approval are obtained from the Swedish ethical board.

Acknowledgements

We acknowledge the support and assistance provided by the staff of the Haemodynamic Lab; The Section for Heart Failure and Valvular Disease, Skåne University Hospital, Lund, Sweden; and The Section for Cardiology, the Department of Clinical Sciences, Lund University, Lund, Sweden. We sincerely thank Anna Åkesson, statistician at Region Skåne and Lund University, for statistical advice with respect to study design and analysis.

Funding

The present project was supported by unrestricted research grants from ALF (‘Avtal om Läkarutbildning och Forksning’), GoRadCare AB, and Nordic Infucare. The funding organizations played no role in the collection, analysis, or interpretation of the data and had no right to restrict the publishing of the manuscript.

Conflict of interest: A.A. and S.A. report no conflicts of interest. G.R. reports unrestricted research grants from ALF, GoRadCare AB, and Nordic Infucare, as well as and a non-interventional investigator-initiated study research grant from Janssen-Cilag AB, during the conduct of the study.

A.A. and S.A. report personal lecture fees from Janssen-Cilag AB outside the submitted work. G.R. reports personal lecture fees from Actelion Pharmaceuticals Sweden AB, Bayer Health Care, GlaxoSmithKline, Janssen-Cilag AB, Merck Sharp & Dohme AB, Nordic Infucare, and Orion Pharma outside the submitted work.

G.R. is and has been the primary, or co-, investigator in clinical PAH trials for Acceleron, Actelion Pharmaceuticals Sweden AB, Bayer, Janssen-Cilag AB, Merck Sharp & Dohme AB, Pfizer, and United Therapeutics and in clinical heart transplantation immuno-suppression trials for Novartis.