Continuous-flow left ventricular assist devices (cfLVAD) have become a standard therapy for patients with advanced chronic heart failure (HF), both as a bridge to heart transplant (BTT) or as destination therapy. cfLVAD support has been demonstrated to improve functional class, organ perfusion and HF symptoms by unloading the left ventricle (LV) and improving hemodynamics.1 Despite these proven benefits, left ventricular assist device (LVAD) support is still associated with adverse events that lead to morbidity, hospital admissions,2 and considerable costs.3 Normalization of hemodynamics seems to play an important role in achieving a clinical benefit after the implantation of a cfLVAD. It has been reported that a decrease in pulmonary capillary wedge pressure (PCWP) as a consequence of left ventricular unloading (LVU) is related to better functional class, lower HF admissions, and better survival free of any hemocompatibility-related adverse events.4,5

The rotor speed setting (Rsp) for LVAD during follow-up is a critical factor that determines the amount of LVU. This speed adjustment is crucial to achieve a compromise between the degree of LVU and the operation of the LVAD within optimal physiological ranges that promote good functional class, without excessive negative pressure in the LV cavity, resulting in ventricular suction that can lead to arrhythmias and right ventricular (RV) dysfunction.

This compromise is generally achieved with noninvasive methods that avoid the potential drawbacks of repeated invasive assessments. Nevertheless, it is still a major challenge to obtain an optimal Rsp because LVU, at a given Rsp, may also vary with the stage of HF and several preload and afterload conditions. Physical examination is the most accessible tool but it has low sensitivity in assessing hemodynamics compared with right heart catheterization (RHC).6 In this scenario, clinical guidelines recommend transthoracic echocardiography (TTE) as an integral part of determining optimal Rsp.7 That the serial measurement of LV end-diastolic diameter (LVEDD) with TTE, combined with the degree of aortic valve (AV) opening, can be used as a surrogate marker for LVU seems intuitive and is supported by limited literature, derived primarily from HeartMate II studies. However robust outcome data are limited and their applicability to centrifugal cfLVAD patients has not been demonstrated at this time. Moreover, the effects of other factors that have proved their association with LVU in scenarios outside the LVAD field, ie, age or renal function, have not been studied in this clinical setting.

The aim of this study was to evaluate the likelihood of an effective hemodynamic LVU (HLVU) in ambulatory patients with a functioning centrifugal cfLVAD implanted as BTT when the current standard recommendations for Rsp optimization were applied. The association of an incomplete HLVU with different clinical, biochemical and echocardiographic conditions and its impact on mortality was also studied.

A total of 104 consecutive patients were included in the analysis (figure 1). The patients were followed up during a median time of 23±10 months from the time of LVAD implantation. No patients were lost to follow-up.

Figure 1.

Flowchart of the study population. BTT, bridge to transplant; DT, destination therapy; HW, heart ware; periop, periperative; LVAD, left ventricular assist device; RHC, right heart catheterization.

(0.23MB).

Baseline preoperative characteristics are described in table 1. Postoperative data are summarized in table 2. None of the patients were treated with sacubitril-valsartan or specific pulmonary vasodilators.

RHC was performed after a median of 10 months (Q1:5, Q3: 15) since LVAD implantation. Seventy-five patients (72%) showed a normal pulmonary capillary wedge pressure (nPCWP) (PCWP ≤ 15mmHg) and 29 patients (28%) had hPCWP (PCWP> 15mmHg).

Patients in the hPCWP group were older, had worse preoperative and postoperative renal function and were more likely to require dialysis after LVAD implantation compared with the nPCWP group. The number of patients treated with RAASi was lower in the group with hPCWP. Requirement for oral diuretics and BNP levels at the time of RHC were significantly higher in patients with hPCWP than in patients with nPCWP. Patients with hPCWP were more likely to undergo concomitant TVR at the time of LVAD implantation due to severe tricuspid regurgitation (table 2).

Postoperative RHC showed higher mean pulmonary pressure and central venous pressure and lower pulmonary artery pulsatility index in patients with hPCWP than in the nPCWP group, but pulmonary vascular resistance showed no significant differences between groups.

Neither the preoperative nor the postoperative indexed LVEDD showed significant differences between groups. Persistent dilatation of the LV under LVAD support was not statistically associated with the presence of hPCWP. The postoperative indexed LVEDD was strongly correlated with the preoperative indexed LVEDD (R, 0.75; P <.0001). However, the magnitude of the decrease of the indexed LVEDD was more pronounced in patients with nPCWP. The indexed LVEDD decreased 15.2 + −14.7% in the nPCWP group compared with 8.9 + −11.8% in the hPCWP group (P=.043), (figure 2).

Figure 2.

Differences in the preoperative and postoperative indexed left ventricular end-diastolic diameter and in the percentage of the reduction of left ventricular end-diastolic diameter after cfLVAD stratified by hemodynamic left ventricular unloading. Comparison between groups for delta-indexed LVEDD showed P <.05. LVEDD preop, preoperative left ventricular end-diastolic diameter; postop, postoperative; delta-indexed LVEDD, percentage of the reduction of left ventricular end-diastolic diameter; PCWP, pulmonary capillary wedge pressure.

(0.11MB).

We observed a marked decrease in the BNP values after LVAD implantation (median delta 1362, Q1 330, Q3 2651 pg/mL) with higher BNP values at the time of RHC in the hPCWP group.

Regarding LVAD-related complications, patients with hPCWP showed more major gastrointestinal bleeding (GIb) events than patients with nPCWP (table 2).

Age, higher central venous pressure and BNP levels were associated with higher risk of hPCWP in the univariate LRA. A higher delta-indexed LVEDD change and treatment with RAASi decreased the odds of hPCWP (table 3). LRA demonstrated that age and BNP were independent predictors of hPCWP (table 3). BNP showed good predictive discrimination in the receiver operating characteristic analysis (figure 3). BNP <300 pg/mL was associated with a reduced risk of hPCWP (odds ratio, 0.14; 95%CI, 0.05-0.4); P <.001) and could anticipate freedom from hPCWP with a predictive value of 86%, 75% specificity, and 75% sensitivity.

Figure 3.

Receiver operating characteristic curve (ROC) of BNP as a predictor of incomplete hemodynamic left ventricular unloading. Area under the ROC curve (AUC)=0.78; 95%CI, 0.65-0.82; P <.0001.

(0.06MB).

Twelve patients required hospitalization due to HF decompensation during the surveillance time (7 patients from the nPCWP group). Seven patients died during follow-up (4 patients from the hPCWP group, table 4). Twenty-one patients underwent HT, with comparable HT likelihood between groups. Overall survival (41.5 months in nPCWP group vs 39.6 months in the hPCWP group) and survival free from HF hospitalizations (37.5 months vs 36 months respectively, log-rank P=.75) showed no significant differences between groups (figure 4). Cox regression analysis did not show a significant effect from PCWP group, age or treatment with RAASi on survival.

Figure 4.

Survival rate stratified by the existence of incomplete hemodynamic left ventricular unloading during cfLVAD support. cfLVAD, continuous-flow left ventricular assist device; hPCWP, high pulmonary capillary wedge pressure> 15mmHg; nPCWP, normal pulmonary capillary wedge pressure ≤ 15mmHg.

(0.14MB).

Despite the decrease in volume and pressure LV load that follows the implantation of a cfLVAD, the effect of the centrifugal cfLVAD on the HLVU in clinically stable patients has not been fully investigated. Moreover, it is not clear if an insufficient HLVU correlates with clinical HF worsening. To the best of our knowledge, this is the largest clinical study that evaluates HLVU in ambulatory patients supported with centrifugal cfLVAD. This study allows us to conclude that an optimal HLVU can be achieved in up to 72% of the ambulatory patients supported with current LVAD models as BTT when the Rsp is based on current clinical and echocardiographic recommendations. In comparison, a previous study reported9 that a normal HLVU was observed in 43% of clinically stable patients supported with the previous HeartMate II or with Heartware LVAD. In a study of a small cohort of 16 patients with functioning HeartMate 3, Uriel et al.10 reported that 62.5% of the patients had a normal hemodynamic profile. Although the patients from these studies might be not fully comparable, it seems that an optimal HLVU is more frequently achieved in clinical practice with the centrifugal cfLVAD than with older generations of LVAD.

We observed that suboptimal HLVU was significantly associated with fluid overload as reflected in a higher need for diuretic therapy and higher BNP levels. A better understanding of the clinical conditions associated with an insufficient HLVU could help to improve diagnostic protocols during long-term follow-up, which may facilitate an earlier diagnosis and device optimization in selected patients. We observed that the presence of an elevated LV filling pressure was more frequent in older patients with reduced renal function. Moreover, the presence of hPCWP coexisted with elevated central venous pressure, pulmonary pressure, postcapillary pulmonary hypertension and lower artery pulsatility index, which supports the evidence of the strong interaction between the left and right filling pressures that occurs during long-term LVAD support and that can affect long-term RV performance.11 This strong interconnection between filling pressures, age and renal function might perpetuate a pre-existing cardio renal syndrome, despite cfLVAD support, which further aggravates the volume overload and compromises transrenal perfusion pressure and RV function; these factors, which constitute a vicious circle, might promote a maladaptive neurohormonal activation and further renal dysfunction.

Severe preoperative tricuspid regurgitation was more frequent in the hPCWP group, despite the use of TVR at the time of LVAD implantation. The benefit of TVR in protecting the vulnerable RV after LVAD implantation is controversial. Some investigators12 reported that, although preoperative moderate and severe tricuspid regurgitation were associated with poorer survival, TVR did not improve overall survival. Severe tricuspid regurgitation might simply reflect the consequences of long-lasting RV failure and dilatation in patients with HF, which might persist despite TVR. Nevertheless, it may be beneficial, as the baseline characteristics of patients undergoing TVR were suggestive of a more progressive underlying disease.13

RAASi therapy has been extensively demonstrated to produce an effective decrease in the PCWP in patients with chronic HF, and emerging data suggest that guideline-directed medical therapies for HF may also improve clinical outcomes in LVAD patients. We also observed that RAASi were associated with a decreased probability of hPCWP, which further supports its beneficial effect on hemodynamics in LVAD patients. However, while cardiorenal protective, this medication can lead to temporary worsening of renal function in patients with renal dysfunction. In this regard, it is also known that age and reduced renal function significantly increase the risk of clinical intolerance to RAASi in patients with HF, which plausibly occurred to the hPCWP group under LVAD support in our study.14,15

Although RHC during long-term LVAD support would be the reference method for evaluating LVU, clinical guidelines16 currently recommend it only in specific scenarios. In this regard, noninvasive methods for the reliable estimation of hemodynamics would allow for optimization of medical therapy and Rsp without the potential drawbacks of repeated invasive assessments. One of the most attractive candidates for this clinical role would be BNP, which is released from ventricular myocardium as a response to volume expansion or pressure overload and is a strong predictor of mortality and cardiovascular outcomes in HF patients.17,18 A recent study has demonstrated that N-terminal pro-B-type natriuretic peptide and other biomarkers substantially decrease after LVAD implantation but do not regress to normal values.19 However, the authors of that study did not correlate their results with TTE or hemodynamics. In the present study, we demonstrate that BNP levels are persistently elevated above the reference serum levels after LVAD implantation. Furthermore, we provide new evidence that BNP values are independently associated with HLVU. A BNP cutoff point <300 pg/mL was a reliable screening index to rule out hPCWP in ambulatory LVAD patients. Whether this parameter could be broadly used as an additional noninvasive tool to optimize HLVU, limiting the use of invasive RHC, should be investigated in larger and multicenter studies.

The second potential noninvasive clinical tool to estimate hemodynamics is TTE. In this respect, clinical guidelines recommend TTE as an integral component for determining optimal Rsp, with goals including adequate LVU with midline IVS and minimal mitral regurgitation and AV opening.7 However, recent reviews of the management of patients with cfLVAD describe the difficulty of evaluating the degree of LVU with TTE.20 Although some authors have described TTE algorithms to estimate left-side filling pressures,21 optimal results require an expert technique and heavily relies on the E/e′ ratio, which is poorly validated among patients supported with LVAD.22

In clinical practice, serial LVEDD TTE measurements (combined with AV opening) are frequently used as a surrogate marker for the degree of LVU.7 However this assumption is not fully supported by clinical studies. This issue has strong clinical relevance because an unnecessary increase in LVAD speed can have a deleterious effect on volume status and RV function and can trigger arrhythmias.

We observed that the magnitude of the mechanical unloading, which is the percentage of the reduction of the LV dimensions, which translates the degree of LV reversal remodeling, was the only TTE parameter that was significantly associated with the degree of HLVU. Patients with optimal HLVU showed double the magnitude of LV mechanical decompression than the group with persistent hPCWP. Reverse myocardial remodeling is central to the benefits of most HF treatments and is associated with an improved prognosis. Some authors have also demonstrated that reduced concentrations of N-terminal pro-B-type natriuretic peptide after guideline-directed medical therapy are significantly associated with reverse remodeling.23 Our study provides evidence that optimal HLVU after LVAD implantation translates into lower BNP concentrations and corresponding more favorable effects on reverse myocardial remodeling.

In our study, the postoperative indexed LVEDD was not associated with PCWP and showed a strong correlation with the preoperative indexed LVEDD. From a practical point of view, persistent dilatation of the LV may simply reflect the degree of adverse LV remodeling and dilatation prior to LVAD implantation. The presence of mitral regurgitation or the absence of AV opening do not seem to help either in this decision-making in clinically stable patients. The AV opening might reflect the degree of intrinsic improvement in LV function, for instance in patients with LV recovery. Changes in LVAD settings or in HF medication guided by surrogate markers of inefficient HLVU, such as persistent dilatation of the LV, AV opening or mitral regurgitation, seem not to be supported by our findings and might have a deleterious effect on hemodynamics. However, the magnitude of the reverse LV remodeling seems to be associated with the degree of hemodynamic LV decompression. The discriminatory value of these parameters should not be extrapolated to cases of suspected LVAD thrombosis or device malfunction because we only evaluated the diagnostic reliability of these parameters in ambulatory patients with functioning LVAD.

Patients with hPCWP showed more major GIb events. Although the investigation of risk factors for GIb was beyond the scope of our investigation, the greater number of GIb events could be explained by the higher age and central venous pressure, worse renal function and RV performance, and lower RAASi treatment in the hPCWP group.24 However whether an incomplete HLVU could enhance the risk of GIb should be evaluated in future studies.

An optimal HLVU can be achieved in up to 72% of ambulatory patients supported with cfLVAD when current standard recommendations for Rsp are applied. Our study highlights the effect of age, central venous pressure and therapy with RAASi on achieving this goal. BNP levels and the magnitude of the reverse LV remodeling seem to be useful noninvasive tools to evaluate LVU during follow-up in ambulatory patients with functioning cfLVAD support.

None.

M.J. Ruiz-Cano: Conceptualization, data curation, formal analysis, methodology, validation, visualization, writing—original draft, writing—review and editing. R. Schramm: Conceptualization, validation, visualization. L. Paluszkiewicz: Validation, visualization, data curation, writing—review and editing. L. Ramazyan: Data curation, writing—review and editing. V. Lauenroth: Resources, validation, visualization. S.V. Rojas: Validation, visualization. J. Gummert: Conceptualization, validation, visualization. M. Morshuis: Conceptualization, methodology, supervision, validation, visualization.

The authors have no conflicts of interest to declare regarding this manuscript.