eGFR Trajectory in LVAD Recipients with Right Heart Failure: Status Quo or a Steady Decline?

Two outcomes particularly dreaded after left ventricular assist device (LVAD) implantation are development of right heart failure (RHF) and worsening kidney function. Yet for better or worse, right heart function and kidney function are innately intertwined (Figure). Increased venous return, left ventricular (LV) unloading, changes in systemic afterload, intraoperative factors, the effects of chronic heart failure (HF) on right ventricular function, and conformational changes in how ventricles twist as a result of LVAD being sewn into LV can all contribute to the development of postoperative RHF. ¹ Chronic kidney disease itself may contribute to the development of RHF via reduced glomerular filtration rate (GFR), tubular dysfunction, renin‐angiotensin‐aldosterone system activation, and possibly other mechanisms such as uremic toxin build‐up and higher FGF‐23 levels. ² , ³ , ⁴ , ⁵ RHF may adversely affect kidney function predominantly through elevated systemic venous pressure, which leads to a cascade of downstream effects all culminating in reduced GFR and reduced diuretic efficiency, with cardiac underfilling likely playing a more minor role. ⁶ , ⁷

Figure . Pathophysiology of the interplay between right heart failure and kidney disease.

eGFR indicates estimated glomerular filtration rate; FGF23, fibroblast growth factor 23; HF, heart failure; IAP, intra‐abdominal pressure; LV, left ventricular; LVAD, left ventricular assist device; RAAS, renin‐angiotensin‐aldosterone system; RV, right ventricular; SNS, sympathetic nervous system; and SVP, systemic venous pressure.

In this issue of The Journal of the American Heart Association (JAHA), the study by Walther et al ⁸ examines how development of new RHF affects trajectories in estimated glomerular filtration rate (eGFR) after LVAD implantation in a retrospective analysis of the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) using data from 2014 to 2017. The authors show that lower baseline eGFR is associated with development of both early and later RHF after LVAD implantation. They also report that patients with transient or persistent RHF at 1 month have a significantly higher incidence of requiring dialysis compared with those without RHF. Finally, the authors describe various 1‐year trajectories in eGFR after LVAD implantation. Notably, they report that while those without RHF and those with transient RHF have an initial increase in eGFR in the first month after LVAD implantation followed by a subsequent decline in the following months (as has traditionally been reported in other studies), those with persistent RHF do not experience this initial increase after LVAD implantation, a novel finding. ⁹ , ¹⁰

This observation that those with persistent RHF had no increase in eGFR within the first month of LVAD implantation is a new and important finding that may even be understated by the INTERMACS data. The authors postulate that parenchymal kidney disease or surgery‐associated acute injury make some patients more susceptible to RHF after LVAD implantation. They also propose that residual confounding from pre‐existing right ventricular dysfunction may lead to persistently elevated central venous pressures resulting in stable rather than improved eGFR after LVAD implantation. These are all probable explanations. In addition, it is possible that this apparent stable eGFR among patients with persistent RHF may not reflect stable actual GFR. Rather, an early decrease in GFR may actually be present. eGFR is ascertained using serum creatinine in INTERMACS, because there are no longitudinal measurements of cystatin C. Cystatin C is less influenced by changes in muscle mass than creatinine, and trajectories of cystatin C‐based eGFR could yield different results. In support of this notion, a prospective study of 116 LVAD recipients with serial measures of serum creatinine and cystatin C found that eGFR using serum creatinine improved within the first 6 weeks of LVAD implantation followed by a subsequent decline, as we would expect. ¹¹ However, cystatin C‐based eGFR was stable throughout the first 12 months. Moreover, in a separate cohort, increases in eGFR using serum creatinine after LVAD implantation correlated with a decrease in muscle mass, whereby 1 month after LVAD implantation there was a 4% decrease in serum creatinine for every 1‐cm²/m² decrease in pectoralis muscle index from baseline. ¹¹ Because creatinine is a byproduct of muscle metabolism, the authors concluded that the increase in eGFR using serum creatinine after LVAD surgery may be the result of muscle wasting after major surgery. ¹² It is plausible that if cystatin C measurements were available in the present study by Walther et al, that we may indeed find that eGFR decreases within the first month of LVAD implantation among patients with persistent RHF. This makes the case for more routine measurement of cystatin C in clinical practice in this population and perhaps the addition of this measurement to INTERMACS. This also underscores the importance of early referral to nephrology in patients with persistent RHF and chronic kidney disease who may be at higher risk for developing kidney failure.

Also of clinical importance is the finding by Walther et al that LVAD recipients with transient or persistent RHF at 1 month had a higher incidence of requiring dialysis. The need for kidney replacement therapy is associated with abysmal outcomes in LVAD recipients. ¹³ Predicting preoperatively which patients will progress to kidney failure is challenging. eGFR before LVAD implantation can be influenced by numerous factors, resulting in large swings in eGFR within a single patient even on a day‐to‐day basis. Preimplantation eGFR can provide some insight into future trajectories in kidney function but may not be the most accurate predictor. Comorbid conditions such as diabetes may also contribute to risk. The Kidney Failure Risk Equation is used to predict 2‐ and 5‐year risk of developing kidney failure in the general chronic kidney disease population. ¹⁴ A risk score is calculated using eGFR, albuminuria, age, sex, serum albumin, phosphorus, bicarbonate, and calcium. The Kidney Failure Risk Equation has not been validated in patients with LVAD or even patients with advanced HF, and its validity in such populations may be further confounded by competing risks. Finally, intraoperative factors during LVAD implantation such as prolonged hypotension or bypass time may result in kidney damage. Such factors cannot be accounted for in a preoperative risk score.

Several risk calculators for RHF, even among LVAD recipients, also exist. The European counterpart to INTERMACS, the EUROMACS (European Registry for Patients With Mechanical Circulatory Support), developed a risk score for the development of postoperative RHF. ¹⁵ They found that higher baseline serum creatinine and blood urea nitrogen were associated with developing RHF in unadjusted analyses, although these values were not included in the final model. Several other risk scores for RHF have also been developed, but current methods used to predict risk of RHF remain nondiscriminatory due to various factors including intraoperative conditions that cannot be determined preoperatively and varying definitions of RHF. Moreover, prior risk calculators have not been validated in HeartMate 3 recipients. ¹⁶ Persistent RHF at >3 months after LVAD implantation was associated with worse prognosis while higher blood urea nitrogen before implantation was associated with increasing risk of RHF in an analysis of 5537 INTERMACS patients. ¹⁷ Compared with the previously published INTERMACS analysis, the current study provides a more in‐depth assessment of the association between kidney function and RHF in the same LVAD cohort, and adds to these data by describing differences in the 1‐year change in eGFR in those with and without RHF. However, it was unable to account for intra‐ and postoperative factors that are known limitations of INTERMACS data set analyses. These limitations highlight the need for validation in the current HeartMate 3 cohort and the development of better prediction models for long‐term RHF and kidney dysfunction after LVAD implantation.

The incomplete data from the INTERMACS registry must also be acknowledged. There were 8076 patients with baseline characteristics included in the present analysis. Information on severe right ventricular dysfunction was missing in 17.8% and temporary mechanical circulatory support in 15.3%. There was also quite a bit of missingness at follow‐up. At 1 month, information for 460 participants was missing, and at 3 months data on the incidence of RHF were missing in 3226 participants. This was predominantly due to not having any follow‐up information recorded during the 3‐month RHF ascertainment window. Of the 3226 with missing follow‐up information, ≈72% were missing due to lack of recorded data. Additionally, 9% had died by 3 months, and 5.5% had received a heart transplant. All of these together raise concern for informative censoring and integrity of INTERMACS data. It is somewhat reassuring that baseline characteristics and 12‐month survival were similar among the 4850 with 3‐month follow‐up and the 3226 with missing data.

The primary limitation of this analysis is the inclusion of predominantly HeartMate 2 and HeartWare patients. Currently, only the HeartMate 3 device is implanted after the Food and Drug Administration issued a regulatory warning to discontinue implantation of the HeartWare HVAD device due to internal pump malfunction. Walther et al point out that in the MOMENTUM 3 analysis, the degree of RHF and kidney dysfunction were similar in axial flow and centrifugal pump devices, thereby implying that their current findings can be applied to patients receiving HeartMate 3. ¹⁸ However, the current population of patients receiving HeartMate 3 devices may reflect a sicker population than those enrolled in MOMENTUM 3. This is because with the United Network for Organ Sharing (UNOS) heart allocation policy change in 2018, there has been an increase in the number of LVAD implantations for destination therapy rather than bridge to transplant. ¹⁹ Before the new allocation system, 49% of LVADs were implanted as destination therapy compared with 81% currently. ²⁰ This is largely due to the prolonged waiting time and the lower transplantation rates. ²¹ It is conceivable that some of the reasons current patients with LVAD receive destination therapy rather than heart transplantation may also make them more predisposed to develop worsening kidney function and RHF. This selection bias somewhat limits the applicability of the results reported by Walther et al, whose follow‐up period ends in December 2017, to current patients.

In conclusion, the interplay between kidney disease and RHF in LVAD recipients is remarkably complex and clinically important. These findings highlight that kidney function does not necessarily improve after LVAD implantation among patients with RHF. There is a need to inform potential LVAD patients about the possibility of worsening kidney function and kidney replacement therapy despite best attempts at improving cardiac function. Similar analyses should be replicated in a contemporary cohort of patients reflective of the current device used, current allocation practices, and less biased assessments of kidney function.

Sources of Funding

B. Roehm is supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number KL2TR003981.

Disclosures

None.