You Can't Always Get What You Want

Corresponding Author

In this issue of JACC: Asia, Jiang et al¹ report the development of a simplified clinical score, which they termed “Prediction Of raISEd wedge pressures” (POISE), to predict postcapillary pulmonary hypertension (PH)^2,3 in an Asian cohort undergoing right heart catheterization. Using 4 routinely available variables, that is, body mass index, atrial fibrillation, mitral E/e′, and left atrial enlargement, the model achieved modest discrimination (C-statistic ≥0.72) and high specificity at a prespecified threshold, with performance exceeding that of the OPTICS risk score⁴ and H2FPEF score⁵ within this cohort. The study addresses a clinically relevant problem, namely the differentiation of postcapillary from precapillary PH in contexts where invasive hemodynamic assessment is not routinely performed or is limited.

Several aspects of the work have notable strengths. The focus on a multiethnic Asian population addresses an important gap, given the limited validation of existing risk scores outside Western cohorts. The cohort is clinically relevant, derived from patients undergoing invasive hemodynamic assessment, thereby anchoring the analysis to a definitive reference standard. The study also evaluates a diagnostically challenging population without overt left-sided disease, reflecting a common and clinically important scenario. The variables selected are clinically intuitive and grounded in established pathophysiology of elevated left-sided filling pressures, further reinforcing biological plausibility. Perhaps most importantly is that the model is intentionally parsimonious, relying on variables that are readily available in routine clinical practice, which enhances usability and facilitates bedside application without reliance on specialized testing or computational tools. The analytic framework is transparent, with appropriate reporting of discrimination, calibration, and internal validation using bootstrapping. Finally, comparison with established scores provides a pragmatic benchmark and underscores potential population-specific differences in model performance.

However, important limitations temper the interpretation. The model represents a refinement of existing approaches rather than a substantive conceptual advance, as the included variables are well-established correlates of elevated left-sided filling pressures and overlap substantially with prior scores. Not surprisingly, the POISE score outperformed the OPTICS⁴ and H2FPEF⁵ scores by just a minimal extent, hardly of clinical significance. As such, an important and somewhat hidden message in the study is that the 3 scores all performed less well in the Asian population than in a Western population. A further limitation of all 3 scores is that they were developed in a population undergoing right heart catheterization, with a relatively high prevalence of precapillary PH. The pretest likelihood of precapillary PH in patients who would benefit from the application of such tools would probably be several fold lower and would significantly alter performance characteristics. The modest sample size, combined with stepwise variable selection, also raises concerns regarding model stability and overfitting. Internal validation does not address generalizability, and the absence of external validation is a significant limitation. Missing data were not missing completely at random, yet the primary analysis relied on complete case methods, introducing potential selection bias. The selected threshold prioritizes specificity at the expense of sensitivity, resulting in a tool that identifies only a small proportion of affected patients. In addition, the outcome definition is based on a binary threshold for resting pulmonary capillary wedge pressure, which does not fully capture the dynamic and continuous nature of left-sided filling pressures or the complexity of mixed precapillary and postcapillary PH phenotypes.

Within the broader clinical context, the role of the POISE score is likely to be focused rather than generalizable. Contemporary diagnostic approaches to suspected postcapillary PH are increasingly anchored in structured echocardiographic assessment of left ventricular diastolic function and filling pressures,⁶ alongside comprehensive right heart evaluation that incorporates right ventricular size and function and estimates of pulmonary pressures.⁷ These frameworks emphasize multiparametric integration rather than reliance on individual variables. In this setting, a simplified score derived from a limited number of inputs may offer supportive information, particularly when findings are concordant, but is unlikely to replace a comprehensive echocardiographic assessment. A highly specific, low-sensitivity approach may be most useful in selected settings where the goal is to identify patients with a high likelihood of postcapillary physiology, particularly when access to invasive hemodynamic assessment is limited. In routine practice, however, where discordant findings and mixed phenotypes are common, the incremental contribution of such a score may be more modest. Importantly, it should be interpreted as complementary to guideline-directed imaging assessment and not as a substitute for invasive hemodynamic evaluation when clinical suspicion for precapillary or combined disease persists.

Future work should focus on external validation across diverse populations and clinical settings, with particular attention to performance in diagnostically ambiguous cohorts. Methodologically, approaches that avoid dichotomization of continuous variables and incorporate physiologically grounded metrics may improve discrimination. Integration of multimodality imaging and hemodynamic phenotyping, rather than reliance on simplified binary classification, is likely to be necessary to advance the field. A shift toward multidimensional frameworks that better capture right ventricular–pulmonary vascular–left heart interactions may ultimately provide greater clinical utility than incremental refinements of existing risk scores.

Funding Support and Author Disclosures

Dr Mukherjee has received funding from the National Institutes of Health, National Heart, Lung, and Blood Institute (R01HL162851) and the U.S. Department of Defense (PR231648). Unrelated to this work, Dr Mukherjee serves on the Data Safety Monitoring Board for Advarra, Inc and AllRock Bio, Inc. Dr Bogaard has received funding the Royal Netherlands Academy, Dutch Federation of University Medical Centres, Netherlands CardioVascular Research Initiative: the Dutch Heart Foundation, Netherlands Organisation for Health Research and Development, and the Royal Netherlands Academy of Sciences (CVON-2017-10 Dolphin-Genesis; CVON-2012-08 PHAEDRA and CVON-2018-29 PHAEDRA-IMPACT). Unrelated to this work, Dr Bogaard is a part-time employee of Linxis Biopharmaceuticals and received research grants from Ferrer, MSD, and Janssen Pharmaceuticals.