Predictors of change in estimated glomerular filtration rate post-impella 5.5 implantation in heart failure patients bridged to transplant

Background

Renal dysfunction in advanced heart failure (HF) patients critically influences heart transplant eligibility and the choice between heart-only transplantation and simultaneous heart-kidney transplantation (SHKT). In many patients, kidney injury reflects potentially reversible cardiorenal syndrome driven by hemodynamic derangements, particularly elevated venous congestion, rather than intrinsic renal disease. Temporary mechanical circulatory support (tMCS) devices, such as the Impella 5.5, offers rapid hemodynamic stabilization and may unmask reversible renal dysfunction. Early identification of renal recovery during support could meaningfully impact transplant strategy.

Purpose

To evaluate whether short-term changes in invasive hemodynamics during Impella 5.5 support are associated with renal function recovery.

Methods

We retrospectively analyzed patients with HF supported with Impella 5.5 as bridge-to-transplant. Invasive hemodynamics and renal function were assessed at baseline and 72 hours post-Impella 5.5 implantation.

The primary outcome was the change in estimated Glomerular Filtration Rate (ΔeGFR). Univariate and multivariate linear regression analyses evaluated associations between ΔeGFR and changes in Cardiac Index (ΔCI), Pulmonary Artery Pulsatility index (ΔPAPi), Stroke Volume (ΔSV), Mean Arterial Pressure (ΔMAP), and Right Atrial Pressure (ΔRA), adjusting for baseline clinical characteristics.

Results

The study cohort included 61 patients supported with Impella 5.5 with complete pre- and post-Impella hemodynamic profiles. The majority were male (80.3%), white (58.3%), and 57.4% had hypertension. The median baseline eGFR was 47 mL/min/1.73 m² (27-63).

On univariate analysis, ΔGFR was positively associated with ΔCI (β 8.17, p=0.002), ΔPAPi (β 0.86, p=0.020), and ΔSV (β 0.22, p=0.032), but not with ΔMAP or ΔRA (Table 1).

In multivariate hemodynamic modeling, both ΔCI (β 13.29, p=0.011) and ΔPAPi (β 0.72, p=0.036) remained independently associated with renal recovery (model R²=0.24) (Table 2). However, in fully adjusted analyses incorporating baseline clinical characteristics, ΔCI emerged as the dominant predictor (β 15.44, p=0.005), whereas baseline hypertension attenuated renal recovery (β −9.80, p=0.023). Other pressure-based indices were not independently predictive. The final model demonstrated moderate explanatory power (R²=0.321, p=0.0066).

Additionally, to evaluate potential effect modification, we tested the interaction between ΔCI and baseline hypertension (Table 3). We found no evidence of effect modification by baseline hypertension.

Conclusion

Collectively, these data reinforce the concept that dynamic hemodynamic response at 72 hours, particularly improvement in cardiac index, may serve as a biologically meaningful marker of reversible cardiorenal physiology and may help inform transplant strategy and timing decisions.Tables