FormalPara Key Summary Points

Why carry out this study

Assessment of 1-year mortality risk is important for treatment decision-making in pulmonary arterial hypertension (PAH).

The EXPOSURE study is an ongoing, prospective, international, observational study enrolling patients with PAH at initiation of a new PAH-specific therapy in clinical practice, with > 2500 patients enrolled to date.

This analysis describes the characteristics, treatment patterns and outcomes of patients with PAH, categorised by risk status when initiated with the oral selective prostacyclin receptor agonist selexipag in clinical settings.

What was learned from the study?

In clinical settings, selexipag is initiated across all risk groups, predominantly as triple combination therapy with an endothelin receptor antagonist and phosphodiesterase 5 inhibitor; excellent 1-year survival was observed in the intermediate-high, intermediate-low and low risk groups.

Our study suggests an opportunity for more frequent patient monitoring and earlier treatment escalation, given that the 4-strata risk assessment was prognostic for hospitalisation and survival in this large, contemporary PAH cohort.

Introduction

Assessment of 1-year mortality risk provides a basis for the management of patients with pulmonary arterial hypertension (PAH) and aids clinical decision-making [1, 2]. The overall therapeutic goal is for patients to survive and display PAH disease characteristics associated with a 95% survival rate at 1-year, which is considered low mortality risk in PAH [1, 2]. While comprehensive patient evaluation including right heart catheterisation is recommended at treatment initiation, at follow-up a pragmatic approach based on evaluation of three non-invasive parameters—6-min walk distance (6MWD), World Health Organization functional class (WHO FC) and brain natriuretic peptide (BNP)/N-terminal (NT)-proBNP—is proposed [1, 2]. Thresholds reached for these parameters are the basis of a calculated score used to estimate the risk of death within 1-year. The score allows patients to be categorised as having low risk (0–3%), intermediate-low risk (2–7%), intermediate-high risk (9–19%) and high risk (> 20%) of death at 1-year, and supports decisions about treatment escalation [1, 2].

The pragmatic 4-strata risk approach is particularly relevant to the oral selective prostacyclin receptor agonist selexipag. Selexipag is indicated to delay the progression of PAH in patients with WHO FC II and FC III symptoms [3, 4] and is typically added to an endothelin receptor antagonist (ERA) and phosphodiesterase 5 inhibitor (PDE5i) at follow-up [5], in line with both treatment recommendations and the approved indication [1, 2, 6]. In addition to the established long-term benefits [7, 8], treatment escalation with selexipag has been shown to improve risk status at 6 months follow-up, regardless of background therapy [9]. The ongoing observational study EXPOSURE (EXPloratory Observational Study of Uptravi in Real-lifE) provides insights into the management of patients with PAH in clinical practice with particular focus on selexipag. This analysis categorises patients by their 1-year mortality risk using the recommended 4-strata risk approach and describes their characteristics, treatment patterns and outcomes at the time of treatment escalation with selexipag and over the long term.

Methods

Data Sharing Statement

These data are not currently publicly available for sharing, requests for sharing can be sent to the corresponding author and will be evaluated on an individual basis.

Study Design

EXPOSURE (EUPAS19085) is an ongoing, multicentre, prospective, observational study of patients with PAH initiating a PAH-specific therapy in Europe and Canada. Recruitment started in 2017, and study details have been described previously [10]. Adult patients with Group 1 pulmonary hypertension (PAH) initiating a new PAH-specific therapy per decision of the treating physician, within 1 month of enrolment or at enrolment, were eligible. Calcium channel blockers were not considered PAH-specific therapy for this analysis. Patients initiating a PAH-specific therapy must not have been previously treated with that same drug. Patient visits were not mandated and occurred as per clinical practice.

Patient Population

This analysis included and followed consecutively enrolled patients initiating selexipag with a known selexipag initiation date, from 17 September 2017 to 30 November 2022. Patients with available baseline data to calculate a risk score were grouped as low, intermediate-low, intermediate-high or high risk of 1-year mortality according to the 4-strata method recommended in the European Society of Cardiology/European Respiratory Society (ESC/ERS) guidelines [1, 2]. For completeness, data are also presented for all patients who initiated selexipag up to the data cutoff, including patients for whom the risk assessment was not able to be calculated.

Analyses

Patients were observed during the selexipag exposure period, defined from the time of selexipag initiation (i.e. baseline) up to date of last available information, selexipag discontinuation (> 7 days without therapy) or death, whichever occurred first. All analyses were descriptive; no formal statistical comparisons were made. Risk of 1-year mortality was calculated from baseline data using the 4-strata risk score recommended in the 2022 ESC/ERS guidelines [1, 2], with at least two of the following parameters: BNP/NT-proBNP and WHO FC and/or 6MWD.

The titration method for selexipag has been described previously [11]. Selexipag titration was conducted independently of the study design and per physician's judgement and experience. For the purposes of this analysis, the duration of selexipag titration and individualised dose were identified and defined in a stepwise manner for each patient (Supplementary Methods). To summarise the dose data, three dose groups were defined based on patients’ individualised dose at the end of titration: low dose (200 or 400 µg twice daily [b.i.d.]), medium dose (600, 800 or 1000 µg b.i.d.) and high dose (1200, 1400 or 1600 µg b.i.d.). Following titration, the number of dose adjustments, as well as selexipag discontinuations and reasons for discontinuation were described. Treatment regimen was described as monotherapy for those treated with selexipag alone. Treatment regimen was described as combination therapy if concomitant PAH-specific therapy overlapped with selexipag for more than 1 day at baseline or within 30 days after baseline. If selexipag was taken in addition to one, two, or more concomitant PAH-specific therapies, the combination was described as double, triple, or other combination therapy, respectively.

Incidence rates for all-cause and PAH-related hospitalisation and all-cause mortality rates were calculated as the number of patients experiencing an event after baseline divided by the overall sum of patients’ exposure periods while at risk for an event. Rates were expressed as 100 person-years, with associated 95% confidence intervals (CI) and calculations based on a Poisson distribution model with log (exposure time) as an offset. Kaplan–Meier (KM) estimates with 95% CIs were used to present time to first event for hospitalisations and all-cause death during the selexipag exposure period.

The following known adverse reactions frequently associated with the mode of action of drugs targeting the prostacyclin pathway (headache, diarrhoea, jaw pain, nausea, myalgia, vomiting, pain in extremity, flushing, arthralgia) were not reported as an adverse event (AE) unless they fulfilled at least one of the following: any seriousness criteria, lead to selexipag discontinuation or dose reduction, introduction of symptomatic treatment or were reflective of an unusual pattern of severity based on the physician’s medical judgement. The imputation methods used have been described previously [10]. Imputations were made for partially missing dates with day imputed (if year and month present) for start/end of hospitalisation, date of death, and date of PAH diagnosis, and day and/or month imputed (if year present) for start/discontinuation of PAH therapy. For completely missing dates, no imputations were made.

Monitoring and Ethics Statement

The study was conducted in compliance with the Declaration of Helsinki and the protocol was approved at each study site (and at the national level in Sweden; Table S1) by an institutional review board or independent ethics committee and the study was conducted in accordance with the International Society for Pharmacoepidemiology (ISPE) Guidelines for Good Pharmacoepidemiology Practices (GPP) [12] and in accordance with the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance (ENCePP) Guide on Methodological Standards in Pharmacoepidemiology [13]. Written informed consent was obtained for each patient for accessing, processing and using anonymised patient data in scientific research.

Results

Demographics and Baseline Characteristics

By 30 November 2022, 535 out of 698 (77%) patients had data allowing for risk assessment calculation at baseline. Of these, 14% (N = 76) were in the low risk group, 31% (N = 168) in the intermediate-low risk group, 34% (N = 182) in the intermediate-high risk group and 20% (N = 109) in the high risk group. The demographics and baseline characteristics of these patients are outlined in Table 1.

Table 1 Patient characteristics at baseline
Full size table

Overall, the population was predominantly female (71%), with idiopathic/heritable PAH (I/HPAH; 56%) or PAH associated with connective tissue disease (CTD-PAH; 27%), with a median age of 60 years and prevalent (2 years) PAH disease. Across the groups from low to high risk, there was an increase in the proportion of CTD-PAH and in age—ranging from 12%–40% and 46–68 years, respectively, and a decrease in the proportion of I/HPAH and in the time from diagnosis—ranging from 71%–52% and 3.2–1.6 years, respectively. There was a history of cardiovascular risk factors across all risk groups. The most prominent were history of smoking, systemic hypertension and obesity, which were reported in 13–18% of patients in the low risk group and 29–47% of patients in the higher risk groups (Fig. S1). Overall, the haemodynamic parameters worsened with increasing risk group. Diffusing capacity of lung for carbon monoxide, although only captured in a third of patients, decreased with increasing risk group. Values for the parameters used to calculate the risk score (WHO FC, 6MWD and NT-proBNP) worsened with increasing risk group, as expected. Demographics and baseline characteristics for the overall selexipag group are provided in Table 1 and Fig. S1.

Selexipag Titration, Dosing and Treatment Patterns

The median (interquartile range [Q1, Q3]) duration of selexipag titration was comparable across all four risk groups, with a slightly shorter duration for patients in the high risk group compared to the other risk groups (low risk 1.9 [0.9, 3.0] months; high risk 1.6 [0.6, 2.4] months) (Table 2). The majority of patients in all risk groups completed their titration (85–93%), 1–10% were ongoing and 5–10% discontinued selexipag (Table 2). The median individualised dose was 800 µg b.i.d. for patients in the low, intermediate-low and intermediate-high risk groups, with a slightly lower dose for patients at high risk (median [Q1, Q3]: 600 [400, 1000] µg b.i.d.) (Table 2). The distribution of dose groups is outlined in Table S2. The proportion of patients with further dose adjustments after titration was comparable between the risk groups (28–35%) (Table 2), over a median of 4.2–6.1 months.

Table 2 Selexipag titration and dosing
Full size table

At 6 months prior to baseline, the majority of patients in all risk groups were on double combination PAH-specific therapy (57–63%). At baseline, the vast majority of patients in all risk groups initiated selexipag as part of triple combination therapy (range 74–81%), mainly with both an ERA and a PDE5i (Fig. 1). A detailed overview of therapies at baseline is provided in Table S3. This escalation to triple combination therapy with selexipag tended to occur within a shorter time from diagnosis for patients at high risk versus low risk (median [Q1, Q3]: 1.6 [0.5, 4.1] vs 3.2 [1.4, 9.3] years, respectively). Reasons for the initiation of selexipag are outlined in Table S4; expectation of further improvement was the main reason across all risk groups. Selexipag titration, dosing and treatment patterns for the overall selexipag group are displayed in Table 2 and Fig. 1.

Fig. 1
figure 1

PAH-specific treatment patterns at baseline. *Includes 163 patients who did not have sufficient data to allow for risk evaluation. †Includes patients with therapies that have missing start and end dates, and patients for whom it cannot be determined if the treatments are prior or concomitant. ERA endothelin receptor antagonist, PAH pulmonary arterial hypertension, PDE5i phosphodiesterase 5 inhibitor, PGI2 prostacyclin and its analogues, sGC soluble guanylate cyclase

Full size image

Hospitalisations

The median (Q1, Q3) selexipag exposure duration was slightly longer for patients in the lower risk groups compared to the higher risk groups (low risk 11.5 [4.0, 25.3] months; intermediate-low risk 11.4 [4.0, 26.4]; intermediate-high risk 10.1 [3.7, 24.1]; high risk 9.0 [2.8, 17.2]) (Table 3). Over the selexipag exposure period, the proportion of patients hospitalised increased with increasing risk status (16%, 23%, 34%, 42% for the low, intermediate-low, intermediate-high and high risk groups, respectively) (Table 3). A greater proportion of hospitalisations were deemed PAH-related for patients in the high risk group (71%) compared to patients in the other risk groups (47–54%) (Table 3). The incidence rate for PAH-related hospitalisations was 6.0 (95% CI: 2.2, 13.1) per 100 person-years for patients at low risk and increased incrementally to 33.6 (95% CI: 22.2, 48.4) per 100 person-years for patients at high risk (Table 3). This indicates that for 100 patients over 1 year, 6.0 and 33.6 patients will have at least one PAH-related hospitalisation in the low risk and high risk groups, respectively. At 1-year, 88%, 84%, 69% and 58% patients were free from hospitalisation in the low, intermediate-low, intermediate-high and high risk groups, respectively (Fig. 2). Hospitalisations for the overall selexipag group are provided in Table 3 and Fig. S2.

Table 3 Hospitalisations and mortality during selexipag exposure
Full size table
Fig. 2
figure 2

Time to first all-cause hospitalisation during selexipag exposure illustrated using KM curves. Each curve is cut at the first timepoint where < 10% of patients in the group are left at risk. Patients with a hospitalisation ongoing at baseline were excluded from the KM analysis. KM estimates (95% CI) shown at 12 and 24 months. For the high risk group, patient numbers were insufficient at 24 months to calculate a KM estimate. CI confidence interval, KM Kaplan–Meier

Full size image

Survival

There were 1, 7, 22 and 23 deaths reported in the low, intermediate-low, intermediate-high and high risk groups, respectively. Of these, 0%, 71%, 65% and 80% were deemed PAH-related, as per judgement of the treating physician (Table 3). The majority of deaths were described as disease progression/PAH worsening for patients in the intermediate-high and high risk groups and right heart failure for patients in the intermediate-low risk group (Table 3). The mortality rate per 100 person-years (95% CI) increased incrementally with increasing risk group: 1.0 (0.0, 5.5), 3.1 (1.2, 6.3), 9.5 (5.9, 14.3) and 20.3 (12.9, 30.5) for patients at low, intermediate-low, intermediate-high and high risk, respectively (Table 3). KM estimates of survival in the respective risk groups were 98%, 98%, 93% and 80% at 1-year and 98%, 92%, 81% and 67% at 2-years. KM estimates of survival at 3-years were 98%, 92% and 76% for patients at low, intermediate-low and intermediate-high risk, respectively (Fig. 3). Survival for the overall selexipag group is provided in Table 3 and Fig. S3.

Fig. 3
figure 3

Time to all-cause death during selexipag exposure illustrated using KM curves. Each curve is cut at the first timepoint where < 10% of patients in the group are left at risk. KM estimates (95% CI) shown at 12, 24 and 36 months. For the high risk group, patient numbers were insufficient at 36 months to calculate a KM estimate. CI confidence interval, KM Kaplan–Meier

Full size image

Tolerability and Safety

The proportion of patients that discontinued selexipag because of tolerability/AE during the exposure period was 12%, 14%, 15% and 23% for patients at low, intermediate-low, intermediate-high and high risk, respectively (Table 4). Other main reasons for discontinuation were PAH disease progression (5–9%) for all risk groups, and death for patients in the intermediate-high and high risk groups (12% and 21%, respectively) (Table 4). The most frequently reported AEs in all risk groups were consistent with targeting the prostacyclin pathway: diarrhoea and headache (Table 4). For patients in the high risk group, cardiac failure was also one of the most frequently reported AEs (6%; n = 6) (Table 4). Safety for the overall selexipag group is provided in Table 4.

Table 4 Selexipag discontinuations and adverse events during the selexipag exposure period
Full size table

Discussion

A number of observational studies have established the prognostic value of characterising patients with PAH based on WHO FC, 6MWD and NT-proBNP [14, 15]. Such characterisation has been associated with 1-year mortality [14, 15] and is the recommended approach to aid treatment decision-making for patients already receiving PAH-specific therapy [1, 2]. These studies provide characterisation of patients at diagnosis and within 6 months of diagnosis, as well as long-term survival ranging from 7 to 10 years [14, 15]. As they describe the initial treatment regimen ranging from monotherapy with oral agents to combination therapy including treatment with parenteral prostacyclin or its analogues [14, 15], a detailed understanding of how the 4-strata risk assessment approach applies to the use of selexipag in clinical practice and in a more prevalent patient group is lacking. The EXPOSURE study enrols patients at new PAH-specific therapy initiation and offers a unique opportunity to understand the impact of treatment escalation with selexipag on survival in patients treated in clinical practice. Importantly, this study allows us to better understand the majority of patients seen in clinic who have already received double oral therapy for an extended period of time when characterised as at low risk, intermediate-low, intermediate-high or high risk of death at 1-year.

This population in EXPOSURE is predominantly prevalent with a median time from diagnosis of 2.0 years; overall, there were approximately 175 patients with a time since diagnosis of 8.4 months or less. Most patients (78%) initiated selexipag as part of a triple oral combination therapy regimen. The ESC/ERS 4-strata risk score (and not 3-strata) was chosen as the most applicable method for this population as it is recommended for patients at follow-up in clinical practice [1, 2]. When patients were grouped according to this 1-year mortality risk score derived from WHO FC, 6MWD and NT-proBNP thresholds [1, 2], we see across the risk groups that the values for each of these parameters, as well as the values for haemodynamic parameters with prognostic relevance, provide interesting insights. Among patients in the low risk group the values not only meet the thresholds of low risk (WHO FC I/II, 6MWD > 440 m, NT-proBNP < 300 ng/L [1, 2]) but they are far above/below them, as are the values for cardiac index, mean right atrial pressure and mixed venous oxygen saturation. For patients categorised as intermediate-low the split by WHO FC is 52% FC II and 45% FC III, while the 6MWD and haemodynamic values are within or trending to the low risk thresholds. In our population, for the intermediate-high and high risk groups the split by WHO FC and delineation across the more objective parameters are also generally aligned with the recommended thresholds, or are better than expected [1, 2]. Of note, although the high risk group are severe, they appear to be less severe that many of the patients at high risk described in disease registries [14, 15]. Taken together, these observations suggest that this population from EXPOSURE comprises patients who trend towards the less severe in each risk group than those seen in disease registries [14, 15].

The majority of patients reached their individualised selexipag dose, regardless of risk group, and overall 6% patients discontinued selexipag during titration. Approximately one third of patients had further dose adjustments following titration, regardless of risk group. As expected and similar to other observations from EXPOSURE [11], the titration patterns and dosing here do not follow those observed in the selexipag pivotal trial, GRIPHON [7]. Overall, there was a slightly higher proportion of patients in the low and medium dose groups (25% and 41%, respectively, vs 20% and 36% observed in GRIPHON [16]). The distribution of dose groups was similar across the risk groups, with patients at high risk tending to have a lower median individualised dose compared to other risk groups. While discontinuations due to tolerability/AEs were generally similar across all risk groups, discontinuations due to death were higher in the higher risk groups. As such, overall discontinuations of selexipag during the exposure period increased with increasing risk. The known AEs associated with the mode of action of selexipag include headache, diarrhoea, pain in jaw, and nausea and these were the most frequent AEs reported for selexipag-treated patients in EXPOSURE in all risk groups. No new safety signals were detected.

Current treatment guidelines recommend addition of selexipag to double combination therapy at follow-up for patients at intermediate-low risk status and for patients at intermediate-high/high risk for which parenteral prostacyclin agents are unfeasible. This analysis shows that in clinical settings selexipag was mostly initiated as part of triple combination therapy with an ERA and PDE5i, across patients in all risk groups, with approximately half of patients being at low or intermediate-low risk and approximately half with I/HPAH. While selexipag does seem to be used in patients categorised as high risk by the method described in clinical practice, as described earlier, the patients at high risk in this analysis (based on NT-proBNP, 6MWD and/or WHO FC thresholds) appear to be less severe than those described in disease registries elsewhere [14, 15].

With respect to reasons for selexipag initiation, disease progression was a major reason in the intermediate-high and high risk groups, given for 30% of these patients compared with approximately 15% of patients in the lower risk groups. This reasoning, along with the length of time from PAH diagnosis for patients in the low and intermediate-low risk groups (2.7–3.2 years) and the substantial proportion of patients in the intermediate-high and high risk groups, suggests a delay in the initiation of selexipag and reflects an opportunity to start treatment earlier in the patients’ disease course. Evidence from randomised controlled trials (RCTs) supports the earlier use of selexipag, with greater benefits on hospitalisation and mortality shown when initiated within 6 months of diagnosis [8].

We followed patients over a period of up to 3 years in EXPOSURE and the survival observed at 1-year across the risk groups is at the upper end of the range described in the literature [1, 2], indicating promising survival rates for patients treated with triple oral therapy across each of the risk groups. Our study supports the prognostic value of the 4-strata risk assessment method for all-cause hospitalisation and death in prevalent patients with PAH with comorbidities on combination therapy. While a survival bias may have been introduced by the prevalent nature of our population, the separation of the KM curves may be as informative as the observed 1-year estimates. During the first 6 months we observe a difference in survival, i.e. separation in the curve, only for the patients categorised as high risk. After this, the high risk group continues to separate with 80% survival at 1-year and 67% survival at 2-years. For the other three groups, survival remains above 90% at 1-year, with no separation between intermediate-low and low risk groups (both 98% survival) and limited separation only for those in the intermediate-high group (93% survival). Indeed, it is only after 2-years that we observe clear separation between all four groups. Though not articulated in other papers, these observations are consistent with those across the literature. Extended observation periods might provide a false sense that the difference in survival among risk categorised patients with PAH is obvious; however, these differences are less clear early on [14, 15]. These findings have substantial clinical significance; they suggest that treatment escalation should be based on careful evaluation of the patient, including 1-year mortality risk assessment, and that the benefits of early escalation or the consequences of therapeutic inertia may not be apparent within a short timeframe but rather over a period of years. This poses a challenge in clinic as it may lead to a scenario of “too little, too late”. In such a scenario, our results suggest that the occurrence of hospitalisations among patients may prove useful, as our KM curves for hospitalisation show earlier and more distinct separation than those for survival, indicating that hospitalisations can be considered an earlier sign of worsening. While hospitalisation data from disease registries are not described in the literature, our observations are supported by analyses from RCTs [17], other risk assessment tools [18] and expert consensus [19].

This analysis is subject to limitations. Our dataset comprises patients who initiate selexipag in clinical practice and as selexipag is typically used as an add on therapy, this introduces a number of biases compared with disease registries of consecutive patients. Firstly, our population is enrolled at the time of initiation with selexipag and may reflect patients experiencing deterioration. Secondly, our population is prevalent and therefore subject to survivor bias. Finally, our population likely excludes the most severe patients who are not candidates for selexipag. While these factors may limit the generalizability of our results, the findings remain particularly relevant to selexipag use in clinical practice. In addition, while available for the majority of patients, baseline data for risk score calculation were not available for all.

Conclusions

In the clinical setting, selexipag is initiated in patients across all risk groups and predominantly as part of triple oral combination therapy with an ERA and PDE5i. Our study supports the prognostic value of the 4-strata method and the use of risk status to trigger treatment escalation as per the 2022 ESC/ERS guidelines. We confirmed that 1-year survival is not particularly distinct across the three lower risk groups and distinction in survival is clear only at later timepoints. Interestingly, all-cause hospitalisation showed an earlier and more distinct separation across risk groups than survival. Together with data from RCTs [7, 17], our data suggest that all-cause hospitalisation has prognostic relevance. We conclude that being vigilant and escalating treatment early and in accordance with risk evaluation is critical to prevent deterioration among patients with PAH. The safety/tolerability profile of selexipag was consistent across all risk groups and was in line with the known side effects described in the approved product information documents.