Mortality Risk Assessment Using the REVEAL 2.0 Score in Pulmonary Hypertension Secondary to Left Heart Disease

Background Pulmonary hypertension (PH) frequently complicates the course of patients with left heart disease (PH-LHD) and is associated with worse clinical outcomes. Mortality calculators for PH-LHD are lacking, and it is unclear whether any risk prediction tools originally derived from other forms of PH can accurately predict outcomes in patients with PH-LHD. Methods We retrospectively analyzed data from 161 patients diagnosed with PH-LHD referred to our pulmonary hypertension center from 2016 to 2022. We calculated the Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL 2.0) risk score and categorized patients as low, intermediate, or high-risk. We assessed survival at 1 and 3 years using Kaplan-Meier and Cox proportional hazards, as well as classification performance using a concordance index. Results At the first outpatient visit, 15% of patients were stratified as low-risk, 27% as intermediate, and 57% as high-risk. Cumulative 1-year survival rates were 100%, 94%, and 91% for the low, intermediate, and high-risk strata, respectively. Cumulative 3-year survival rates were 96%, 89%, and 70% for the low, intermediate, and high-risk strata, respectively. We found no difference in outcomes at 1 year between risk groups. High-risk patients had an increased risk of death at 3 years using REVEAL 2.0 (HR 5.32, p < 0.001). However, while REVEAL 2.0 accurately discriminated high-risk patients, the hazard ratio was not statistically different between patients classified as intermediate-risk compared to low-risk. Conclusion REVEAL 2.0 accurately predicted 3-year survival in PH-LHD patients with high-risk features. However, the mortality risk between patients classified as intermediate-risk was not different from the low-risk stratum, suggesting inaccurate classification for this group of patients.


Introduction
Pulmonary hypertension (PH) associated with left heart disease (LHD), de ned by a mean pulmonary arterial pressure (mPAP) > 20 mmHg and a pulmonary arterial wedge pressure (PAWP) or left ventricularend diastolic pressure (LVEDP) > 15 mm Hg 1 , is the most common type of PH, with an estimated prevalence ranging between 40%-80% in all patients with heart failure (HF) 2 3 .In addition to its high prevalence, numerous studies have shown an inverse relationship between PH and survival in patients with HF, particularly when associated with right ventricular (RV) dysfunction (i.e.biventricular dysfunction) 45 6 .Interestingly, despite its high frequency and impact on disease progression, risk prediction tools are lacking in this PH subtype.
Several risk prediction scores have been derived and validated to estimate the probability of death in patients with pulmonary arterial hypertension (PAH), including the Registry to Evaluate Early and Longterm PAH Disease Management Registry (REVEAL) and its derivatives (REVEAL 2.0 and REVEAL-Lite 7 ).
However, whether any of these tools perform with precision in a different population of patients without PAH (Group 1 PH), speci cally PH-LHD, remains to be investigated.Here we aimed to assess the prognostic utility of the REVEAL 2.0 model in predicting survival in a cohort of patients with PH-LHD at 1 and 3 years.

De nition of Study Population and Clinical Variables
We retrospectively analyzed data from patients referred to the Pulmonary Hypertension Clinic at the University of Cincinnati from 2016-2022.Patients were diagnosed with PH-LHD using the 2022 ESC/ERS pulmonary hypertension guidelines 8 , with the following hemodynamic thresholds, obtained by right heart catheterization (RHC) at an optimized euvolemic state: 1) mPAP > 20 mmHg at rest and 2) PAWP > 15 mmHg 9 .According to ERS/ESC guidelines, isolated post-capillary pulmonary hypertension (ipcPH) was de ned as mPAP > 20 mmHg, PAWP > 15 mmHg and a PVR ≤ 2 Wood Units (WU).Combined pre and post-capillary pulmonary hypertension (cpcPH) was de ned as mPAP > 20 mmHg, PAWP > 15 mmHg and a PVR > 2 WU.All RHC tracings were reviewed by a pulmonary hypertension specialist (J.M.E, J.G-A) to con rm the validity of the PCWP tracings.HF with preserved ejection fraction (HFpEF) and HF with reduced ejection fraction (HFrEF) were de ned based on signs and symptoms consistent with HF failure and a left ventricular ejection fraction (LVEF) estimated by echocardiography of ≥ 50% or LVEF ≤ 40%, respectively. 10Co-morbidities such as hypertension, dyslipidemia, diabetes, COPD and OSA chronic kidney disease were de ned as per ICD-10 codes in the electronical medical record system.Coronary artery disease was de ned as a combination of ICD-10 or evidence by left heart catheterization or CTbased imaging.Obesity was de ned as having a BMI higher or equal to 30.Atrial brillation, as well as other electrocardiographic abnormalities were determined by ICD-10 code and independent EKG review (D.S.D).

Risk Strati cation using the REVEAL 2.0 Model
As previously reported, the REVEAL 2.0 risk score is comprised of thirteen variables 11 (Online Supplemental Material).Each variable contributes to a weighted risk; a minimum of 7 variables is required to generate an accurate score.The REVEAL 2.0 score assigns patients to a risk category based on the weighted score as follows: Low-risk (score of ≤ 6), intermediate-risk (score of 7 or 8), and high-risk (score of ≥ 9. To assign patients a risk score, we collected demographic data, clinical characteristics, laboratory, and hemodynamic data closest to the rst outpatient clinic visit (OV-1) after a diagnosis of PH-LHD was established.Speci cally, echocardiographic, relevant laboratory (natriuretic peptide levels [BNP and NTproBNP], and renal function tests), and clinical (demographics, vital signs, New York Heart Association-World Health Organization Functional Class (NYHA/WHO-FC), six-minute walk distance [6MWD]) data collected within 3 months of the OV-1 was used, and RHC hemodynamic data within 1 year of the OV-1 was used for calculation of each subjects' initial REVEAL 2.0 risk score.
The University of Cincinnati Institutional Review Board approved this study protocol and all related activities (IRB Protocol Number 2020 − 0466).

Statistical Methods
Categorical variables are reported as percentages, while continuous variables are reported as median and interquartile range (IQR).Between-group differences were compared by Wilcoxon rank-sum test, or chi 2 tests as appropriate.Spearman's coe cient was used to calculate correlations.The concordance index was calculated to evaluate the performance of the REVEAL score in classifying survivors versus non-survivors at 1 and 3 years from the OV-1.Kaplan-Meier curves were used to compare mortality at 3 years following OV-1.Cox proportional hazard models were used to examine the relationships between risk strati cation and 3-year survival.Continuous variables (BNP, 6MWD, and pulmonary vascular resistance [PVR] were standardized to z-scores to allow comparisons of odds ratios based upon a one standard deviation change in a given variable.Cox proportional hazards models were adjusted for relevant demographic data in a multivariable fashion.All tests were two-sided, and statistical signi cance was set at p < 0.05.Analysis was performed using IBM SPSS version 28.0 (Armonk, NY) and R Studio (R version 4.2.3), using the survival, Epi and ROC packages.

Results
Cohort Demographics, Clinical Characteristics, and Hemodynamic Data at the First Outpatient Visit We included a total of 161 individuals who met criteria for pulmonary hypertension due to left heart disease (PH-LHD), all of whom had comprehensive clinical and hemodynamic data available.Table 1 presents a summary of the baseline demographic characteristics, laboratory ndings, and hemodynamic parameters at the initial outpatient visit (OV-1).The median age of the participants was 66 years (IQR 18 years).The cohort was predominantly female sex (68%).HFpEF was the most frequent subtype of HF (85.1% [n = 137]).In the subset of patients with HFrEF, the median left ventricular ejection fraction (LVEF) by echocardiography was 33% (IRQ 17.5%).Most people were categorized as a NYHA/WHO FC III (77.6% [n = 125]) and had a 6MWD of 244 meters (IQR 169.136]) and chronic obstructive lung disease (COPD) (33% [n = 53]) were the two most frequent pulmonary co-morbidities.The median forced expiratory volume percentage (FEV1%) predicted in patients carrying a diagnosis of COPD was 61% (IQR 25%).Eighteen patients (11%) had an FEV1% predicted below 50%.
Unsurprisingly, high-risk patients exhibited a worse hemodynamic pro le compared to low-risk strata.As illustrated in Fig. 1, mPAP, PCWP, and PVR were signi cantly higher in high-risk compared to low-risk patients.However, no survival differences were found between patients in low versus intermittent-risk strata.
Figure 1E-G illustrates the distribution of values for systolic blood pressure, heart rate and GFR between REVEAL risk groups.Notably, there were no differences in systolic blood pressure, heart rate of GFR between low and intermediate-risk groups, which could explain why the mortality risk intermediate-risk stratum was not statistically different from the low-risk category.Finally, multivariable Cox proportional hazard analysis including all REVEAL 2.0 variables demonstrated that only Z-standardized BNP/NT-ProBNP (HR 1.56, 95% C.I, 1.05-2.31,[p < 0.05]) was independently associated with higher mortality (Table 5).

Discussion
In this single-center retrospective analysis, we found that in adults with PH-LHD, the REVEAL 2.0 score accurately separated patients at the highest risk of death at 3 years from the rst outpatient visit.
However, patients classi ed as intermediate-risk did not show a signi cant difference in survival compared to the low-risk strata.The c-index for the REVEAL 2.0 score was 0.68, which is below the threshold for good discrimination and lower than what has been reported for patients with PAH (0.73, 95% CI, 0.71-0.757 ).We observed a signi cant overlap in the distribution of values for systolic blood pressure, heart rate, and GFR between low and intermediate-risk groups, which may affect the overall classi cation performance.Like published data for PAH, BNP/NT-pro BNP was the strongest biomarker driving risk strati cation and outcomes in our sample of patients with PH-LHD.Collectively, our results suggest that the REVEAL 2.0 risk score may have prognostic value in predicting survival in PH-LHD patients, however also underscores certain limitations that could help improve classi cation performance.
Despite a high prevalence and signi cant effect on morbidity and mortality, PH in the context of HF tends to be underestimated 13 .Indeed, most clinically relevant mortality calculators for HF do not consider the presence of PH or right ventricular dysfunction as co-variates when assessing risk 14 15 .Prognostic equations and risk prediction models originally developed to predict mortality in PAH have increasingly become an important tool for the management of these patients.However, virtually all calculators were derived from PAH registries and their performance in other types of PH, including PH-ILD remains largely undetermined.Our data suggest the REVEAL 2.0 can accurately discriminate PH-LHD patients at highrisk of death, but it may not accurately distinguish patients with intermediate-risk features.We believe that the signi cant overlap in the distribution of values between groups for key variables (i.e.variables selected for abridged versions of the score) such as systolic blood pressure, heart rate and GFR plays a signi cant role in the score's performance.Indeed, we did not nd a difference in any of these 3 parameters between patients strati ed as low versus intermediate-risk.Similarly, we did not nd a signi cant difference in these 3 parameters between survivors and non-survivors (data not shown).In contrast to patients with PAH, where low systolic blood pressure is a biomarker of progressive right ventricular dysfunction, there is a signi cant amount of evidence to support the bene cial effect of reducing systemic vascular resistance in patients with left heart disease, in order to improve forward ow 10 .In addition, more than 50% of our cohort was receiving beta-blocker therapy which likely in uenced heart rate.In a similar way, as shown in Fig. 1, we found a broad distribution of hemodynamic values that resulted in signi cant overlap between low and intermediate-risk patients, suggesting that these parameters are not different enough to differentiate patients with more severe disease, as it would be the case for PAH.Whether or not these clinical or hemodynamic parameters ought to be exchanged or the thresholds modi ed to re ect a decompensated state in a PH-LHD patient population warrants further investigation.
In contrast, we found that BNP/NT-pro BNP was the only parameters from the REVEAL 2.0 score that independently predicted higher mortality.This is not surprising given that BNP has repeatedly been shown to predict outcomes in patients with left-sided heart failure 16 and in PAH 17 .Similar ndings were reported when the REVEAL Lite 2 score was reported 7 , where Benza and collaborators showed the most highly predictive parameter was BNP/NT-proBNP.
One interesting nding was that, although numerically higher, we did not see a statistical difference in outcomes between patients with ipcPH and cpcPH.This nding may have been in uenced by our sample size and will require further investigation.However, Nasri and collaborators 18 , reported similar results in a larger sample of patients with heart failure.We found that elevated mPAP was strongly associated with worse outcomes, irrespective of a diagnosis of ipcPH/cpcPH.Similar ndings have been reported by Titz, Ulrich and collaborators 19 where in their cohort of patients with cpcPH, they found a 58% increase in risk of death when the mPAP was higher than 46 mmHg.Surprisingly, they did not nd a relationship between elevated PVR (> 5.8 Wood units) and mortality.Titz et al. also failed to nd an association between age > 68 yrs and mortality.Compared to patients with PAH, a most common form of HF (i.e., HFpEF]) is almost exclusively a disorder of persons aged 60 yrs or older 20 , and it has been estimated that post-capillary PH is most frequent in HF patients older than 65 yrs 21 .Thus, it is likely that age > 65 years is a parameter without utility for risk strati cation in patients with PH-LHD.
In our cohort, a striking majority of patients were classi ed as high-risk on their rst outpatient visit.Most importantly, the survival rates at 3-years were remarkably similar to patients with PAH receiving pharmacotherapies 7 , which is comparable to data published by others 19 21 .We believe this result is likely multifactorial and re ects our incomplete understanding of the mechanisms driving disease progression in patients with PH-LHD, including 1) the effect of biventricular (RV and LV) failure, 2) the impact of highly prevalent co-comorbidities in HF such as coronary artery disease, conduction abnormalities and 3) the lack of speci c pharmacotherapy and the effect of beta-blockers, which are frequently avoided in patients with precapillary PH.It is essential to mention that non-survivors exhibited lower TAPSE, highlighting the contribution of RV dysfunction in the prognosis of these patients, even when the absolute difference in TAPSE was not large (average difference 0.198 cm between survivors and nonsurvivors).Similar results have been shown by Melenovsky et al. 22 in patients with HFpEF with and without concomitant right ventricular dysfunction where even a 15% difference in RV fractional area change was associated (45% vs 30%) with worse outcomes 22 .

Limitations
Our study had both strengths and limitations.On the one hand, the single center cohort, limited sample size, cross-sectional design, and retrospective nature of our study certainly limits the interpretation of the causality of associations and mechanistic relationships found in this group of patients.Selection bias may have been present, given that only certain patients with HF are referred to a pulmonary hypertension clinic and thus, our results may not be generalizable to a broader population of patients with PH-LHD.However, one important strength of our study was data completion and deep phenotypic characterization.

Conclusions
The REVEAL 2.0 risk assessment score accurately predicted 3-year survival in PH-LHD patients with high-risk features.However, the mortality risk between patients classi ed as intermediate-risk was not different from the low-risk stratum, suggesting inaccurate classi cation for this group of patients.Our ndings underscore a signi cant overlap of clinical and hemodynamic features between intermediate and low-risk patients, which may play a role in classi cation performance.Further studies are warranted to modify the existing PAH scores to better stratify and estimate mortality in patients with PH-LHD.

Figure 1 Box
Figure 1