Pulmonary arterial hypertension (PAH) is a progressive, life-threatening condition driven by pulmonary vascular remodeling that culminates in right ventricular (RV) failure and premature death. Contemporary management, based on risk stratification, aims to achieve and maintain a low-risk profile; however, many patients classified outside the high-risk category still experience adverse outcomes, which highlights the limitations of current algorithms [1,2].

Non-invasive imaging, particularly echocardiography, offers clinical insights into RV structure and function and may improve prognostic assessment [3–6]. Building on this concept, REVEAL-ECHO (R-Echo) integrates echocardiographic parameters into the REVEAL framework to enhance risk stratification beyond clinical and hemodynamic data [3].

Within the spectrum of PAH, pulmonary veno-occlusive disease (PVOD) represents one of the most aggressive phenotypes [1]. It is characterized by severe hypoxemia with markedly reduced diffusing capacity of the lung for carbon monoxide (DLCO) secondary to venular and capillary remodeling, and by poor tolerance to vasodilator therapy due to the risk of pulmonary edema. In this clinical context, bilateral lung transplantation (LT) remains the only treatment alternative [7].

In Spain, PVOD has been reported as relatively more frequent, partly due to EIF2AK4-related familial clusters among Iberian Romani patients [8–10]. Data from the Spanish Registry of Pulmonary Hypertension (REHAP) reported a prevalence of 6.6% [11], which is higher than figures from REVEAL or French registries [12,13]. At the same time, Spain offers a unique setting to address this condition given its high organ donation and LT rates [14].

Although existing PAH risk models perform adequately in general populations, PVOD patients show higher mortality, even within low risk strata [15,16]. This excess mortality suggests that factors beyond those included in current risk assessment tools may contribute to prognosis in PVOD-rich cohorts, including treatment response, susceptibility to pulmonary edema, and progressive respiratory insufficiency. Despite the high incidence of adverse events, there is heterogeneity in disease course and survival [15–17]. This variability highlights the need for refined risk stratification tools capable of identifying patients at imminent risk of deterioration and optimizing timely listing for LT.

To date, no study has specifically assessed RV remodeling in PVOD or tested imaging-based risk models in cohorts enriched for PVOD. Because most risk scores analyzed mortality as the primary endpoint, incorporating LT as a competing outcome is relevant in this population. Therefore, the present study aims to: (1) describe RV remodeling in PVOD patients; (2) assess the prognostic performance of R-Echo for the combined endpoint of death or LT in a cohort representative of a population with frequent PVOD, and (3) develop a modified version of the model, adapted to this population, to improve risk stratification.

In this study, we provide a comprehensive evaluation of risk stratification tools in a large cohort of patients with PAH, with a particular focus on PVOD. Our main findings can be summarized as follows: (1) PVOD patients exhibited more adverse RV remodeling despite similar hemodynamic profiles compared with non-PVOD; (2) R-Lite demonstrated superior prognostic accuracy compared with the original R-Echo, with significantly better discrimination and calibration; (3) the inclusion of PVOD, a well-established marker of poor prognosis, as a weighted variable significantly improved the performance of the R-Echo score in a population with a high prevalence of this condition; and (4) R-Echo-PVOD and its 4-strata version improved reclassification of patients with events compared with the non-PVOD-adjusted models. Moreover, combining clinical and echocardiographic risk models yielded a high negative predictive value for adverse outcomes, especially in LT.

To our knowledge, this is the first study to characterize echocardiographic remodeling in PVOD and to evaluate imaging-based risk stratification models for predicting the relevant composite endpoint of death or LT in a cohort with a high prevalence of this condition.

Our findings reveal that PVOD patients had worse RV remodeling—greater dilation and reduced systolic function—despite similar hemodynamics compared with non-PVOD patients. This suggests that the poor prognosis in PVOD may not be driven solely by hypoxemia or vasodilator intolerance, but also by more advanced baseline RV dysfunction. In this context, echocardiography provides incremental value over clinical markers (WHO functional class or six-minute walk distance), which may be confounded by respiratory impairment, thereby improving identification of patients likely to require earlier listing for LT.

These observations align with previous evidence suggesting that hypoxemia can directly affect RV structure and function, independently of afterload. In healthy individuals, hypoxia has been associated with increased RV dimensions, impaired RV relaxation, and a higher Tei index despite only modest elevations in RV systolic pressure [22,23]. Similarly, at the molecular level, disruption of HIF-1α in cardiomyocytes worsens RV remodeling during chronic hypoxia without increasing RV systolic pressure [24]. Clinically, hypoxemia-triggered myocardial infarction with non-obstructive coronary arteries has been reported in heritable PVOD, illustrating extreme oxygen supply–demand mismatch [25]. These findings support the hypothesis that adverse RV remodeling in PVOD may reflect the combined impact of PH, chronic hypoxemia and intrinsic myocardial maladaptation.

Risk stratification analysis in our cohort demonstrated that clinical models such as R-Lite performed robustly, even in a population with a high prevalence of PVOD. These findings are consistent with previous reports, where R-Lite achieved a C-index of 0.61 for mortality prediction in a PVOD cohort [16]. Notably, our study extends prior knowledge by demonstrating that R-Lite remains effective for the composite endpoint of death or LT, although it does not fully separate intermediate- from high-risk groups for LT alone at long-term follow-up.

Although R-Lite performed well in our cohort, integrating the echocardiographic model with PVOD etiology provided additional prognostic value. R-Echo-PVOD improved event reclassification compared with R-Lite by shifting a greater proportion of patients with adverse outcomes into higher-risk categories. This effect was especially evident in transplant-oriented analyses, supporting the clinical value of incorporating echocardiographic and disease-specific information into risk assessment. In addition, the 4-strata R-Echo-PVOD model further improved both event reclassification and overall reclassification performance, while providing the best discrimination for the composite endpoint. Among transplanted patients, the mean time from baseline assessment to waitlist inclusion decreased across increasing risk categories in both models; however, the low-risk group defined by R-Echo-PVOD showed the longest time to listing (46.2 months), exceeding that observed with R-Lite (36.0 months). This finding suggests that the echocardiographic component may improve identification of patients with a more stable clinical course.

Most importantly, concordant low-risk classification by both models yielded the highest negative predictive value, reaching 93.2% for death, 93.2% for the composite endpoint, and 100% for LT alone. Clinically, this combined low-risk profile may help identify patients who could be referred early to a transplant center for specialized follow-up without immediate waitlist inclusion, provided that close and repeated reassessment is maintained. Overall, these findings support a complementary approach to risk stratification, in which clinical and imaging information are integrated to improve prognostic assessment and guide decision-making more accurately than either domain alone.

This was an observational single-center study based on retrospectively collected data. PH was defined using the hemodynamic thresholds in place at the time of registry entry. Approximately 60% of patients lacked definitive histological or genetic confirmation of PVOD. However, diagnosis was adjudicated using previously published international criteria. To reduce misclassification, patients with relevant respiratory comorbidities were excluded. The lack of external validation poses a risk of limited generalizability. However, bootstrap resampling showed minimal optimism. This internal validation does not substitute the need for external validation in independent cohorts.

In this tertiary-center PAH cohort enriched for PVOD, this etiology was associated with disproportionate adverse RV remodeling despite similar baseline hemodynamics, and with higher rates of death and LT, including urgent LT. Incorporating PVOD as a weighted variable enhanced the performance of the echocardiography-based R-Echo framework, and the resulting 4-strata R-Echo-PVOD model showed the best overall prognostic performance, with better discrimination and reclassification across all endpoints. Concordant low-risk classification by both clinical and echocardiographic models was also associated with a high negative predictive value for adverse events, particularly transplant. These findings support the complementary use of clinical and echocardiographic risk assessment to refine prognostic stratification and improve transplant-related decision-making in populations with a high prevalence of PVOD.

EGO contributed to data collection, data analysis and interpretation, and drafting of the manuscript. CDS contributed to data collection. CJL-G, AC-U, PE-S and JPE contributed to conception and design. JPE performed a complete statistical review and revised the manuscript. SAC, MV, BB, IMDM, CAQL and FAY critically revised the manuscript. All authors approved the final version of the manuscript.

The authors declare that no artificial intelligence tools were used in the generation of the manuscript.

Alejandro Cruz-Utrilla holds a research contract Juan Rodes from the Instituto de Salud Carlos III, Ministerio de Ciencia, Innovación y Universidades, Spanish Government (JR23/00071) and a research grant I+D+I from the Instituto de Salud Carlos III, Ministerio de Ciencia, Innovación y Universidades, Spanish Government (PI24/01880). Eva Gutierrez Ortiz is supported by a Río Hortega Grant from the Instituto de Salud Carlos III (ISCIII, CM25/00215). Pilar Escribano-Subias holds a research grant I+D+I from the Instituto de Salud Carlos III, Ministerio de Ciencia, Innovación y Universidades, Spanish Government (PI21/01690).

EGO reports honoraria from Janssen Pharmaceutica (Johnson & Johnson), MSD and Ferrer, and advisory board participation for Janssen. CJLG reports consulting fees from BeOne, honoraria from AOP Orphan, Ferrer, AstraZeneca, BeOne and Janssen (Johnson & Johnson), and advisory board participation for Pulmovant. ACU reports honoraria from AOP Orphan, Ferrer, Gossamer Bio, MSD and Janssen (Johnson & Johnson), and advisory board participation for Gossamer Bio, MSD and Janssen. PES reports honoraria from MSD, Ferrer, AOP Health, Gossamer and Janssen, and advisory board participation for MSD, Ferrer, Gossamer, AOP Health and Janssen. IMDM reports honoraria and advisory board participation from Janssen. MV reports honoraria from MSD, Janssen, Boston Scientific and SMT. BB reports honoraria from MSD and Janssen (Johnson & Johnson). JPE, SAC, CDS, CAQL and FAY declare no conflicts of interest.