Commentary on “Dynamic Right Ventricular Reserve May Refine Post-LVAD Risk Prediction”

Dear Editor,

We read with great interest the study by Mortada et al, which compared the Michigan, European Registry for Patients with Mechanical Circulatory Support (EUROMACS) Penn/Fitzpatrick, and Clinical Risk Tool for Right Heart Failure (CRITT) scores for predicting right ventricular failure (RVF) after continuous-flow left ventricular assist device (LVAD) implantation. The authors should be commended for evaluating widely used models in a contemporary cohort and for extending the analysis beyond discrimination to include implant-year adjustment, Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) severity, calibration, internal validation, and decision curve analysis. Their observation that EUROMACS was the best-performing score, but still showed only modest discrimination (area under the receiver operating characteristic curve, AUC, 0.670; adjusted AUC 0.713), is clinically relevant because post-LVAD RVF continues to drive morbidity and mortality.^1-4

One issue that could further sharpen interpretation is the distinction between static preimplantation burden and dynamic right ventricular (RV) reserve. In the study, patients who developed RVF had higher admission right-sided and pulmonary pressures, whereas most optimized hemodynamic variables no longer differed significantly after preoperative optimization. ¹ This pattern does not necessarily diminish the value of hemodynamics; rather, it suggests that the trajectory of response to optimization may convey information not captured by baseline or final values alone. A previous analysis similarly showed that failure to achieve hemodynamic optimization goals before LVAD implantation was associated with subsequent RVF and 1-year mortality. ⁵ Future risk models could therefore consider changes in right atrial pressure, pulmonary capillary wedge pressure, transpulmonary gradient, pulmonary artery pulsatility index, and escalation or de-escalation of inotropes/mechanical circulatory support during optimization as markers of RV adaptability.

A second consideration is that the clinical value of a score depends on the decision it is intended to inform. The Youden-derived thresholds reported by Mortada et al help summarize test performance, but bedside decisions—such as intensified decongestion, pulmonary vasodilator therapy, planned temporary RV support, or biventricular support—carry different tolerances for false-positive and false-negative classification. Current expert recommendations emphasize an individualized synthesis of hemodynamics, echocardiography, end-organ function, and operative planning. ⁶ Future work may therefore be strengthened by pairing risk categories with predefined management pathways and evaluating whether those pathways improve clinically meaningful outcomes, rather than relying primarily on AUC comparisons.

Overall, Mortada et al offer a timely reminder that existing RVF scores are useful but incomplete. Their findings support a next step in which conventional score components are supplemented by optimization-response variables and tested in decision-specific validation studies. Such an approach may move prediction closer to the practical question facing LVAD teams: not only who is likely to develop RVF, but who is most likely to benefit from earlier or more aggressive RV-directed support.