Idiopathic pulmonary arterial hypertension phenotypes determined by cluster analysis from the COMPERA registry
The term idiopathic pulmonary arterial hypertension (IPAH) is used to categorize patients with pre-capillary pulmonary hypertension of unknown origin. There is considerable variability in the clinical presentation of these patients.Using data from the Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension, we performed a cluster analysis of 841 patients with IPAH based on age, sex, diffusion capacity of the lung for carbon monoxide (DLCO; <45% vs ≥45% predicted), smoking status, and presence of comorbidities (obesity, hypertension, coronary heart disease, and diabetes mellitus). A hierarchical agglomerative clustering algorithm was performed using Ward's minimum variance method. The clusters were analyzed in terms of baseline characteristics; survival; and response to pulmonary arterial hypertension (PAH) therapy, expressed as changes from baseline to follow-up in functional class, 6-minute walking distance, cardiac biomarkers, and risk.Three clusters were identified: Cluster 1 (n = 106; 12.6%): median age 45 years, 76% females, no comorbidities, mostly never smokers, DLCO ≥45%; Cluster 2 (n = 301; 35.8%): median age 75 years, 98% females, frequent comorbidities, no smoking history, DLCO mostly ≥45%; and Cluster 3 (n = 434; 51.6%): median age 72 years, 72% males, frequent comorbidities, history of smoking, and low DLCO. Patients in Cluster 1 had a better response to PAH treatment than patients in the 2 other clusters. Survival over 5 years was 84.6% in Cluster 1, 59.2% in Cluster 2, and 42.2% in Cluster 3 (unadjusted p < 0.001 for comparison between all groups).The population of patients diagnosed with IPAH is heterogenous. This cluster analysis identified distinct phenotypes, which differed in clinical presentation, response to therapy, and survival. The term idiopathic pulmonary arterial hypertension (IPAH) is used to categorize patients with pre-capillary pulmonary hypertension of unknown origin. There is considerable variability in the clinical presentation of these patients. Using data from the Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension, we performed a cluster analysis of 841 patients with IPAH based on age, sex, diffusion capacity of the lung for carbon monoxide (DLCO; <45% vs ≥45% predicted), smoking status, and presence of comorbidities (obesity, hypertension, coronary heart disease, and diabetes mellitus). A hierarchical agglomerative clustering algorithm was performed using Ward's minimum variance method. The clusters were analyzed in terms of baseline characteristics; survival; and response to pulmonary arterial hypertension (PAH) therapy, expressed as changes from baseline to follow-up in functional class, 6-minute walking distance, cardiac biomarkers, and risk. Three clusters were identified: Cluster 1 (n = 106; 12.6%): median age 45 years, 76% females, no comorbidities, mostly never smokers, DLCO ≥45%; Cluster 2 (n = 301; 35.8%): median age 75 years, 98% females, frequent comorbidities, no smoking history, DLCO mostly ≥45%; and Cluster 3 (n = 434; 51.6%): median age 72 years, 72% males, frequent comorbidities, history of smoking, and low DLCO. Patients in Cluster 1 had a better response to PAH treatment than patients in the 2 other clusters. Survival over 5 years was 84.6% in Cluster 1, 59.2% in Cluster 2, and 42.2% in Cluster 3 (unadjusted p < 0.001 for comparison between all groups). The population of patients diagnosed with IPAH is heterogenous. This cluster analysis identified distinct phenotypes, which differed in clinical presentation, response to therapy, and survival. Pulmonary arterial hypertension (PAH) consists of heterogenous conditions that share a similar pulmonary arterial vasculopathy and a common therapeutic approach. PAH can be associated with well-defined diseases such as connective tissue disease, HIV infection, liver disease, or congenital heart disease, but the most common form of PAH is idiopathic (IPAH). A diagnosis of IPAH is made once the presence of pre-capillary pulmonary hypertension (PH) has been determined by right heart catheterization and other forms of PH and PAH have been excluded.1Frost A Badesch D Gibbs JSR et al.Diagnosis of pulmonary hypertension.Eur Respir J. 2019; 531801904Crossref PubMed Scopus (178) Google Scholar Originally, IPAH (formerly known as primary PH) was described as a disease affecting primarily younger females without typical risk factors for cardiopulmonary diseases.2Rich S Dantzker DR Ayres SM et al.Primary pulmonary hypertension. A national prospective study.Ann Intern Med. 1987; 107: 216-223Crossref PubMed Scopus (1709) Google Scholar Contemporary registries from developing countries have shown a similar pattern.3Zhang R Dai LZ Xie WP et al.Survival of Chinese patients with pulmonary arterial hypertension in the modern treatment era.Chest. 2011; 140: 301-309Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar,4Alves Jr, JL Gavilanes F Jardim C et al.Pulmonary arterial hypertension in the southern hemisphere: results from a registry of incident Brazilian cases.Chest. 2015; 147: 495-501Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar However, more recent registries from Europe and the US have demonstrated a change in the demographics of patients diagnosed with IPAH characterized by increasing age and an accompanying surge in cardiopulmonary comorbidities.5Ling Y Johnson MK Kiely DG et al.Changing demographics, epidemiology, and survival of incident pulmonary arterial hypertension: results from the pulmonary hypertension registry of the United Kingdom and Ireland.Am J Respir Crit Care Med. 2012; 186: 790-796Crossref PubMed Scopus (384) Google Scholar, 6Frost AE Badesch DB Barst RJ et al.The changing picture of patients with pulmonary arterial hypertension in the United States: how REVEAL differs from historic and non-US contemporary registries.Chest. 2011; 139: 128-137Abstract Full Text Full Text PDF PubMed Scopus (231) Google Scholar, 7Hoeper MM Huscher D Ghofrani HA et al.Elderly patients diagnosed with idiopathic pulmonary arterial hypertension: results from the COMPERA registry.Int J Cardiol. 2013; 168: 871-880Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar, 8Hoeper MM Huscher D Pittrow D Incidence and prevalence of pulmonary arterial hypertension in Germany.Int J Cardiol. 2016; 203: 612-613Abstract Full Text Full Text PDF PubMed Scopus (39) Google Scholar, 9Rådegran G Kjellström B Ekmehag B et al.Characteristics and survival of adult Swedish PAH and CTEPH patients 2000-2014.Scand Cardiovasc J. 2016; 50: 243-250Crossref PubMed Scopus (67) Google Scholar Comprehensive phenotypic analysis of large cohorts of patients diagnosed with IPAH have not yet been performed. Here, we present data from a cluster analysis of patients diagnosed with IPAH who have been enrolled in the Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension (COMPERA). Cluster analysis is a data-driven approach that groups subjects into so-called clusters such that the subjects in each cluster are more similar to each other than to those in other clusters. It belongs to the unsupervised statistical learning methods, which means that it can be applied when groups are not known in advance. Few cluster analyses have been performed in patients with PH, but these analyses did not incorporate risk factors for other cardiopulmonary diseases.10Badagliacca R Rischard F Papa S et al.Clinical implications of idiopathic pulmonary arterial hypertension phenotypes defined by cluster analysis.J Heart Lung Transplant. 2020; 39: 310-320Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar, 11Parikh KS Rao Y Ahmad T Shen K Felker GM Rajagopal S Novel approach to classifying patients with pulmonary arterial hypertension using cluster analysis.Pulm Circ. 2017; 7: 486-493Crossref PubMed Scopus (9) Google Scholar, 12Launay D Montani D Hassoun PM et al.Clinical phenotypes and survival of pre-capillary pulmonary hypertension in systemic sclerosis.PLoS One. 2018; 13e0197112Crossref PubMed Scopus (28) Google Scholar The primary objective of this investigation was to identify clinical phenotypes of adult patients with IPAH from COMPERA. As secondary objectives, responses to medical therapy and survival were analyzed and compared in the obtained groups. Details of COMPERA (www.COMPERA.org; registered at Clinicaltrials.gov under the identifier NCT01347216) have been described previously.7Hoeper MM Huscher D Ghofrani HA et al.Elderly patients diagnosed with idiopathic pulmonary arterial hypertension: results from the COMPERA registry.Int J Cardiol. 2013; 168: 871-880Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar,13Hoeper MM Behr J Held M et al.Pulmonary hypertension in patients with chronic fibrosing idiopathic interstitial pneumonias.PLoS One. 2015; 10e0141911Crossref PubMed Scopus (53) Google Scholar, 14Opitz CF Hoeper MM Gibbs JS et al.Pre-capillary, combined, and post-capillary pulmonary hypertension: a pathophysiological continuum.J Am Coll Cardiol. 2016; 68: 368-378Crossref PubMed Scopus (166) Google Scholar, 15Hoeper MM Kramer T Pan Z et al.Mortality in pulmonary arterial hypertension: prediction by the 2015 European pulmonary hypertension guidelines risk stratification model.Eur Respir J. 2017; 501700740Crossref PubMed Scopus (329) Google Scholar In brief, COMPERA is an ongoing web-based PH registry launched in 2007 that collects baseline, follow-up, and outcome data from patients who receive targeted therapies for PH. Specialized centers from several European countries participate (Austria, Belgium, Germany, Greece, Hungary, Italy, Latvia, Lithuania, Netherlands, Slovakia, Switzerland, United Kingdom), with about 80% of the enrolled patients coming from German PH centers. Patients were selected from the COMPERA database using the following criteria: (1) treatment-naive patients newly diagnosed with IPAH between June 1, 2007 and November 11, 2019; (2) mean pulmonary arterial pressure (mPAP) ≥25 mm Hg, pulmonary arterial wedge pressure (PAWP) ≤15 mm Hg, and pulmonary vascular resistance (PVR) >240 dyn·s·cm−5 at the time of diagnosis; (3) complete data concerning the parameters selected for clustering (described later); and (4) at least 1 follow-up visit available. Patients with other forms of PH or PAH were excluded, as were patients with hereditary or drug-associated PAH and patients younger than 18 years at diagnosis. Hierarchical agglomerative clustering was performed using Ward's minimum variance method as linkage criterion. For calculation of the pairwise dissimilarities between observations, Gower's formula was used, as it is able to deal with variables of mixed type. The number of clusters was determined to be 3 based on the clinical observations outlined in the introduction. Cluster analysis requires complete data of all patients. The baseline parameters considered in the cluster analysis were age (continuous), sex (dichotomous), smoking history (dichotomous; never vs current/former), diffusion capacity of the lung for carbon monoxide (DLCO) (dichotomous; <45% vs ≥45% of the predicted value),16Trip P Nossent EJ de Man FS et al.Severely reduced diffusion capacity in idiopathic pulmonary arterial hypertension: patient characteristics and treatment responses.Eur Respir J. 2013; 42: 1575-1585Crossref PubMed Scopus (95) Google Scholar,17Hoeper MM Meyer K Rademacher J Fuge J Welte T Olsson KM Diffusion capacity and mortality in patients with pulmonary hypertension due to heart failure with preserved Heart 2016; PubMed Scopus Google Scholar and comorbidities (dichotomous; vs at least 1 of the arterial hypertension, coronary heart disease the database coronary heart disease was present or without diabetes and defined by a cluster analyses were performed with the parameters but for the of comorbidities, and in Patients with Pulmonary Hypertension based on vs AE et of in pulmonary arterial J Med. 2015; PubMed Scopus Google Scholar The variables were selected based on with disease data are as mean or as median and and data as number and Survival was using and to for the obtained and were applied for and the was < and were performed to the clusters of the of this no for was to PAH therapy was determined by changes from baseline to follow-up treatment For 6-minute walking the in between follow-up and baseline were For and of the change at follow-up in of baseline was in functional and risk were in the and to the at follow-up compared with was by the approach using MM Kramer T Pan Z et al.Mortality in pulmonary arterial hypertension: prediction by the 2015 European pulmonary hypertension guidelines risk stratification model.Eur Respir J. 2017; 501700740Crossref PubMed Scopus (329) Google D Kjellström B C et risk stratification at follow-up in pulmonary arterial hypertension.Eur Heart J. 2018; 39: PubMed Scopus Google Scholar using the by the European of PH M JL et guidelines for the diagnosis and treatment of pulmonary hypertension: the for the diagnosis and treatment of pulmonary hypertension of the European of and the European for European and for Heart and Lung Respir J. 2015; PubMed Scopus Google M JL et for the diagnosis and treatment of pulmonary hypertension: the for the diagnosis and treatment of pulmonary hypertension of the European of and the European for European and for Heart and Lung Heart J. 2016; PubMed Scopus Google statistical analyses were performed using R A patient from the COMPERA database is in patients with 841 patients the and were for this The baseline characteristics of selected and patients were similar at The characteristics of the patients in the 3 clusters are in Cluster 1 of patients with a median age of 45 years and a to of patients had of the risk factors for heart of the patients were or with a median of Lung was and all patients had a DLCO ≥45% of the predicted at the of IPAH = = = of the 3 groups by for variables and for on between each can be obtained from the = years sex, smokers, (44) years mm 5 mm mm 3 dyn·s·cm−5 and predicted predicted predicted predicted DLCO <45% mm mm hypertension, heart disease, least 1 of 1, (28) 2, of 6-minute walking cardiac diffusion capacity of the lung for carbon in 1 idiopathic pulmonary arterial mean pulmonary arterial of arterial carbon arterial pulmonary arterial wedge pulmonary vascular right lung data are shown as and of the data are as mean or median of the 3 groups by for variables and for on between each can be obtained from the in a 6-minute walking cardiac diffusion capacity of the lung for carbon in 1 idiopathic pulmonary arterial mean pulmonary arterial of arterial carbon arterial pulmonary arterial wedge pulmonary vascular right lung data are shown as and of the data are as mean or median Cluster 2 was of patients. patients were females with a median age of 75 of these patients had a smoking history, but all had at least 1 of the risk factors for heart The cluster = 434; was by patients with a median age of 72 As in Cluster 2, most of the patients in Cluster 3 had risk factors for heart the patients in Cluster 3 from those in Cluster 2 was the the of patients with a smoking history and the large of patients with a DLCO <45% of the predicted the cluster analysis was based on the presence of vs risk factors for heart disease as in the AE et of in pulmonary arterial J Med. 2015; PubMed Scopus Google JL RJ et with pulmonary arterial hypertension with and without risk results from the Heart Lung Transplant. 2019; Full Text Full Text PDF PubMed Scopus Google Scholar it was not to distinct when we used vs risk factors for heart disease as cluster analysis did not between phenotypes and and the in age, sex, comorbidities, and smoking described were distinct baseline the 3 clusters 1 and 3 clusters were characterized by pre-capillary PH. Pulmonary expressed as lung capacity capacity and in 1 was in the in all 3 clusters but with the in Cluster patients in this cluster had arterial than patients in the 2 other clusters. The of patients with a history of was and in 1, 2, and baseline characteristics patients with a history of were and had a prevalence of comorbidities than patients without 2 PAH to clusters. Patients in Cluster 1 were with of PAH therapies were used in of the patients the 3 of diagnosis 2007 and the of therapy in this cluster was between 2015 and the was 1 of these patients were Patients in 2 and 3 were with at baseline and at of PAH were used in and of the patients in Cluster 2 and Cluster at baseline and in and of the patients in Cluster 2 and Cluster at Therapies to = = = = 3 diagnosis (3) or (2) at 1 diagnosis (3) or pulmonary arterial in a pulmonary arterial responses by changes in and from baseline to follow-up are shown in The by in Cluster 1, by in Cluster 2, and by in Cluster from baseline to follow-up by a median of to in Cluster 1, by to in Cluster 2, and by to in Cluster in of patients in Cluster 1, in and in In Cluster 2, in in and in of the patients. In Cluster in in and in of the patients. as by the in Cluster 1 patients but to a in Cluster 2 and Cluster 3 patients Survival at 1, and 5 years in Cluster 1 were and and in Cluster and and in Cluster 3 The between the groups were < for age, the between 1 and 3 and 2 and 3 = and p < the survival between 1 and 2 was no A analysis patients without a history of similar results and This that the population of patients diagnosed with IPAH is cluster analysis identified 3 distinct (1) patients with a IPAH characterized by age, of risk factors for cardiopulmonary disease, and preserved (2) patients with a heart failure with preserved that mostly with risk factors for heart disease but without a history of smoking, most of with preserved and (3) patients with a cardiopulmonary of mostly with a history of smoking who with risk factors for heart disease but with low DLCO. Patients in the 3 clusters not differed in demographics and comorbidities but a response to PAH therapy and survival It these were to the pulmonary vascular disease or to other such as treatment or Patients with the IPAH of This low number from of patients without risk factors for heart of these patients had a history of smoking, which is in with the of in data The median number of years was it that smoking had in a to the pulmonary vasculopathy of these patients. Patients in Cluster 1 were but to the US of of patients with primary PH described in S Dantzker DR Ayres SM et al.Primary pulmonary hypertension. A national prospective study.Ann Intern Med. 1987; 107: 216-223Crossref PubMed Scopus (1709) Google Scholar with the other patients in Cluster 1 were younger and with of and at baseline but the response to medical therapy and had the survival However, when for age, the survival between 1 and 2 was no Cluster 2 of with a median age of 75 years who had no smoking history but other risk factors for heart patients with an M MM A approach to diagnosis of heart failure with preserved 2018; PubMed Scopus Google JL RJ S et al.Pulmonary hypertension due to heart Respir J. 2019; PubMed Scopus Google Scholar yet the of IPAH the presence of pre-capillary PH with a of 3 mm it is that of these patients had 2 PH, that PH of heart disease than It is known that at can be in these right with in heart failure with preserved Heart J. 2016; PubMed Scopus Google P MM diagnosis of heart failure with preserved Heart PubMed Scopus Google Scholar on or that JL RJ S et al.Pulmonary hypertension due to heart Respir J. 2019; PubMed Scopus Google were not available. to between patients with IPAH with a and those with an the an of vs AE et of in pulmonary arterial J Med. 2015; PubMed Scopus Google JL RJ et with pulmonary arterial hypertension with and without risk results from the Heart Lung Transplant. 2019; Full Text Full Text PDF PubMed Scopus Google Scholar this to this cluster we the phenotypic between the clusters. on these we that the is to patients with a IPAH from those with a heart disease cluster analysis a population of patients with IPAH when we between patients with or without risk factors for heart observations be considered when clinical The and cluster of patients was the most in terms of age, the patients were to Cluster 2 72 vs 75 The patients in Cluster 3 had risk factors for heart However, in to the 2 other patients in Cluster 3 were and most of had a history of lung by and was DLCO and arterial were than in the 2 other clusters. to the patients in Cluster 2, patients in Cluster 3 the for However, these patients to have at least 2 to the of PH, that an of risk factors for heart disease similar to Cluster 2 with an pulmonary The of a pulmonary vasculopathy is In to a of pulmonary of the of M A et and pulmonary hypertension in 2011; 147: Full Text Full Text PDF PubMed Scopus Google Scholar It has been that a similar pulmonary in patients with a history of P Nossent EJ de Man FS et al.Severely reduced diffusion capacity in idiopathic pulmonary arterial hypertension: patient characteristics and treatment responses.Eur Respir J. 2013; 42: 1575-1585Crossref PubMed Scopus (95) Google Scholar, KM Fuge J Meyer K Welte T Hoeper MM on idiopathic pulmonary arterial hypertension with a low Respir J. 2017; PubMed Scopus Google Scholar, MM a pulmonary Respir Med. 2017; Full Text Full Text PDF PubMed Scopus Google Scholar, et lung disease low diffusion capacity survival and treatment response in patients diagnosed with idiopathic pulmonary arterial hypertension.Eur Respir J. 2020; PubMed Scopus Google Scholar this pulmonary vasculopathy can in patients with or without clinical and of lung KM Fuge J Meyer K Welte T Hoeper MM on idiopathic pulmonary arterial hypertension with a low Respir J. 2017; PubMed Scopus Google Scholar Patients in Cluster 3 had a which is with on patients with IPAH and low P Nossent EJ de Man FS et al.Severely reduced diffusion capacity in idiopathic pulmonary arterial hypertension: patient characteristics and treatment responses.Eur Respir J. 2013; 42: 1575-1585Crossref PubMed Scopus (95) Google KM Fuge J Meyer K Welte T Hoeper MM on idiopathic pulmonary arterial hypertension with a low Respir J. 2017; PubMed Scopus Google et lung disease low diffusion capacity survival and treatment response in patients diagnosed with idiopathic pulmonary arterial hypertension.Eur Respir J. 2020; PubMed Scopus Google Scholar are to the the survival between the clusters. PAH between the clusters. of PAH were used more in Cluster 1 than in the 2 other clusters. Patients were in this response in terms of and risk was in all but treatment were more in Cluster 1 than in the 2 other clusters. It is this was due to the more treatment approach or to a better of these patients. has it and that are phenotypes patients diagnosed with data least in the European patients diagnosed with IPAH are from the patients S Dantzker DR Ayres SM et al.Primary pulmonary hypertension. A national prospective study.Ann Intern Med. 1987; 107: 216-223Crossref PubMed Scopus (1709) Google Scholar that the survival the of age and comorbidities, the that distinct and pathophysiological to the pulmonary vascular disease in the patients under has and The is the large of are the the prospective and data and the follow-up It is all patients were but all the for COMPERA is a European registry with from about 80% of the patients were enrolled from Germany, with the age and demographics of the patients. that the data were over a PAH not the contemporary which the of the of registries such as and a data it to data on variables such as comorbidities, or It is unknown the patients with DLCO had of lung pulmonary by and was preserved in these and a recent by et MM a pulmonary Respir Med. 2017; Full Text Full Text PDF PubMed Scopus Google Scholar has shown that patients with this present with or et lung disease low diffusion capacity survival and treatment response in patients diagnosed with idiopathic pulmonary arterial hypertension.Eur Respir J. 2020; PubMed Scopus Google Scholar to the cluster approach The of cluster analysis is not to have This was by of the patients with IPAH enrolled into which have in In cluster analysis not an a the variables of have to be selected in which a of in as the number of variables considered in this cluster analysis was the have been In cluster analysis 3 distinct IPAH The were patients with that females of younger age without risk The were without a history of smoking but with risk factors for heart The were males, most of had a history of the 3 patients with IPAH had the the survival of the of patients was It is that the to PAH these 3 and data that the be for the response to PAH Hoeper has for from and has for from has for from United and Olsson has for from United and has from and United has for from and a with has from and in the of of and pulmonary arterial hypertension. In has a for for the treatment of systemic has from and J. Gibbs has for from and has from and Ghofrani has from and United Huscher has for and from Pittrow has for from and has for from and United for from and and for on from the from and the of the and from and the Behr from and for or from and from and the from and United the from and the from and the from and and and from Johnson and the The have no of to The are to the COMPERA and This was by the German of Lung COMPERA is by from and were not in data analysis or the of this with