Significant advances in the field of mechanical circulatory support (MCS) have yielded 1- to 3-year outcomes for continuous-flow (CF) left ventricular assist devices (LVADs) comparable with those of heart transplantation.1,2 Despite newer technology and greater familiarity with patient management,3,4 early right ventricular failure (RVF) is still an issue that significantly impacts survival post-LVAD implantation.1–19
We aimed to develop a simple and easily memorized risk stratification tool to determine whether a patient will tolerate an isolated LVAD, as opposed to requiring biventricular support (BVAD), institutionally defined as the “ALMA” score,* in the era of CF MCS devices.
This was a retrospective study of the medical records of patients who underwent either isolated LVAD (n = 170) or unplanned BVAD (n = 88) implantation between January 2006 and December 2017 at 2 MCS coworker institutions: S. Orsola University Hospital in Bologna and S. Camillo Hospital in Rome. Patients with complete data only, according to the official MCS datasets of both institutions, were analyzed. The contribution of the 2 centers has been comparable.
We excluded the pediatric population (<18 years of age) and patients who received total artificial heart support or first-generation pulsatile LVADs. The final study patients were divided into derivation (79.8%, n = 206) and validation (20.1%, n = 52) cohorts.
The device brand profiles for these patients are presented in Table 1. The majority of LVAD patients received the HeartMate II LVAS (Abbott/Thoratec Inc., Pleasanton, CA), whereas most of the unplanned BVAD patients received double CentriMag (Abbott/Thoratec Inc.) support.
Patients in the BVAD cohort included those who had sudden RVF after initial isolated LVAD implantation and required early insertion of a temporary or long-term RV assist device (RVAD).
The decision regarding RVAD insertion was based on individual patient assessment by the multidisciplinary team including experts in cardiology, cardiothoracic surgery, and cardiac anesthetic. The indication of RVAD included failure of weaning from cardiopulmonary bypass at the time of LVAD implantation and any sign of systemic low flow (i.e., low urine output, low mix venous saturation, and rising lactate level) associated with elevated central venous pressure (CVP; >18 mm Hg) and “low-flow” LVAD estimation, despite escalating dose of inotropes/pressors and inhalating nitric oxide use. The decision of RVAD insertion was made before patients developed progressive end-organ dysfunction. RVAD placement technique has been described elsewhere by the same authors.18 The mean time from primary LVAD implantation to RVAD implantation was 1.0 (0–2) day.
Demographic, clinical, and outcome data were used from the patients’ charts pre-LVAD. Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS)1 profiles were assessed just before LVAD placement. Laboratory, echocardiographic, and hemodynamic data (acquired during right heart catheterization) were obtained within 3 ± 2, 6 ± 5, and 8 ± 7 days before surgery, respectively.
Preoperative circulatory support was defined as a preoperative need for an intraaortic balloon pump or extracorporeal membrane oxygenation.
A total of 57 variables were compared between the LVAD and unplanned BVAD cohorts. All patients included in the analysis had sufficient data. The primary outcome was severe RVF within 30 days of LVAD implantation, defined as receiving short- or long-term right-sided circulatory support despite maximal dosage of continuous inotropic support and NO ventilation. The secondary outcome was all-cause mortality.
The study was approved by the institutional review boards of both hospitals.
Transthoracic echocardiographic measurements were performed using Xcelera (Philips Healthcare) and TomTec Imaging Systems software. RV systolic function was qualitatively described as (in order of severity) normal or mildly, moderately, markedly, or severely reduced. Left- and right-sided chamber dimensions and functional parameters were measured according to established guidelines.5,6,13 The right ventricle/left ventricle (RV/LV) diameter ratio was calculated as the end-diastolic RV basal diameter/end-diastolic diameter.13 The tricuspid annular plane systolic excursion was measured on the apical 4-chamber view by manually tracking the lateral wall tricuspid annulus from maximal systolic excursion via maximal diastolic relaxation and atrial contraction.5,6,12
Hemodynamics on Right Heart Catheterization
The mean right atrial pressure, pulmonary artery (PA) systolic, diastolic, and mean pressures, and the pulmonary capillary wedge pressure (PCWP) were measured using a PA catheter during right heart catheterization. Cardiac output was assessed using thermodilution. The transpulmonary gradient was computed as mean PA pressure − mean PCWP.14,15 The RV stroke work index (RVSWi) was calculated as (mean PA pressure − mean right atrial pressure) × stroke volume index, where the stroke volume index was determined as the cardiac index divided by the heart rate.14,15 The pulmonary artery pulsatility index (PAPi)14 was calculated as [(systolic PA pressure − diastolic PA pressure)/central venous pressure].
Pre-MCS laboratory parameters and end-organ function were assessed by calculating the Model for End-Stage Liver Disease Excluding International Normalized Ratio (INR) score (MELD-XI; function of creatinine and total bilirubin levels) as follows: MELD-XI = 11.76 (log creatinine) + 5.112 (log total bilirubin) + 9.44.16
Patient characteristics are described as means (SD) or medians for continuous variables and frequency (percentage) for categorical variables. Differences between patient groups were evaluated for continuous variables by the Student’s t-tests or nonparametric Mann–Whitney U tests and for categorical variables with the χ2 test.
Univariate logistic regression analysis was applied to relate a broad range of preoperative parameters to the study outcome, including demographics, clinical values, comorbidities, echocardiographic, hemodynamic, and laboratory parameters. Variables with a value of p < 0.10 entered the multivariate stage, and a logistic regression model was constructed to predict early post-LVAD right heart failure (RHF), applying the stepwise forward method, with a value of p = 0.05, a modelentry criterion.
We used the receiver-operating characteristic (ROC) curve area under the curve (AUC) analysis to calculate the best cutoff point for its association with RHF.
The relative magnitude of the model regression coefficients from statistically significant variables in the final multivariable model was not weighted. Instead, a simple and practical risk model was generated, in which each of the five variables identified in the final multivariable model was assigned a score of 0 or 1 (overall “ALMA” minimum and maximum scores of 0 and 5, respectively).
The model discrimination abilities were evaluated by the c index of the final multivariate model. ROC curve analysis of the “ALMA” risk score was compared with published risk scores and with individual known markers of RHF. Finally, we validated the risk model in the validation cohort.
We plotted Kaplan–Meier curves for the occurrence of up to 2-year all-cause mortality according to the presence or absence of post-LVAD RHF and stratified by the “ALMA” RHF risk score categories. The log-rank test was used to examine time to mortality differences in the Kaplan–Meier analyses. A 2-tailed value of p < 0.05 was considered statistically significant.
All statistics were undertaken with SPSS statistics version 24 (IBM Corp, Armonk, NY).
The individual variables with distinct differences between the isolated LVAD and unplanned BVAD cohorts were revealed by univariable analysis (Tables 2 and 3). Both the LVAD and BVAD cohorts presented mostly an INTERMACS profile 2–3 (Table 2). Patients in the unplanned BVAD cohort were more likely to be female, to have a higher body mass index, and to have undergone cardiac surgery previously. Patients who underwent BVAD implantation were more likely to require mechanical ventilation or continuous veno-venous hemofiltration. Thereafter, the destination therapy (DT) strategy was the main treatment goal compared with patients who tolerated isolated LVAD therapy (Table 2).
Concerning the preoperative echocardiographic parameters, severe tricuspid regurgitation and high RV sphericity index values were found in the unplanned BVAD cohort preoperatively (Table 3). In terms of hemodynamics, a lower RVSWi and PAPi and higher CVP/PCWP ratio were detected by right heart catheter evaluations in the unplanned BVAD cohort (Table 3). On the other hand, pulmonary hypertension seemed to be protective in the LVAD patients. Regarding laboratory values, an elevated white blood cell count and MELD-XI score were significantly more frequent among patients in the unplanned BVAD than in the isolated LVAD cohort (Table 2).
Multivariable Logistic Regression Analysis
A stepwise multivariable logistic regression model was created by incorporating the significant variables identified by univariate analysis. Variables predictive of the need for unplanned BVAD included DT intention, PAPi <2, RVSWi <300 mm Hg/ml/m2, RV/LV ratio >0.75, and MELD-XI score >17 (Table 4).
The ALMA Risk Score
Similar to the CRITT score method,11 for simplicity and efficiency of use, a five-point risk score was developed based on the clinical variables identified in the multivariable logistic regression model (Table 4). Instead of weighting variables with coefficients based on their odds ratios, each variable is given a binary response. Therefore, a patient who satisfies the at-risk criterion (e.g., PAPi <2) is assigned a score of 1 for that variable versus 0 if not. Thus, 1 or 0 point is allotted for each of the following five variables in the institutionally defined “ALMA” score:* DT intention, PAPi <2, RVSWi <300 mm Hg/ml/m2, RV/LV ratio >0.75, and MELD-XI score >17.
The model fit and predictive power of the five-point risk score were satisfactory when applied to the ventricular assist device data obtained from patients from the University of Bologna, S. Orsola Hospital, or Rome S. Camillo Hospital.
The predicted rate of RVF was significantly (p for linear trend <0.001) increased from 9% for a score of less than 2, to 57.1% for a score of 2–3, and to 100% for a score of 4–5 (Figure 1). The c index was 0.77 in the derivation versus 0.71 in the validation cohort (p = 0.063; Figure 2). In the resulting ROC curves, a score of 3.0 points provided a sensitivity of 82.5% and a specificity of 87.1% for the entire sample. The Hosmer–Lemeshow goodness-of-fit p value was 0.62 in the validation cohort, which reflects an appropriate fit for the data in this cohort. ROC curve comparison with other individual known hemodynamic, echocardiographic markers of RV failure, and other published scores (Figure 2) demonstrated a good performance and a high AUC for the ALMA-RVF score (Table 5).
Based on this model, we recommend an isolated LVAD for patients with a score of 0 or 1 and a BVAD for those with a score of 4 or 5. Patients with a score of 2 are in the gray area and may be able to tolerate an isolated LVAD with appropriate pharmacologic and/or primary temporary RVAD support, preferably associated with tricuspid valve repair (TVR).18–20 Patients with a score of 3 have a high risk of BVAD need, but it might be speculated and investigated the same radical treatment as described above for patients with a score of 2.18–20
Preoperative RV function is an established prognosticator of RVF post-LVAD implantation.1–16,18–20 RV failure is multifactorial and depends on RV preload, RV afterload, RV contractility, and ventricular geometric mechanical interdependence. Existing risk prediction models are derived from retrospective studies based on demographics, presence of end-organ dysfunction, hemodynamics, echo imaging, and prior open thoracic procedures.1–22 Notably, most scores are derived from patient populations supported by earlier-generation pulsatile-flow pumps and hence are not fully representative of the present-day LVAD population.5,17 The usefulness of RVF risk prediction models is limited primarily because of no standardized definition of RVF, small study sample sizes, and their modest discrimination in derivation cohorts.5,17 Differences in the results of risk stratification tools among studies can also be caused by differences in the study duration, indications for LVAD therapy, and surgical strategies.5,17
As incorporated in the INTERMACS definition,1 most of the hemodynamic factors were strong risk factors for postimplant RVF.17 Surrogates of reduced RV contractility, that is, low PA systolic pressure, and RVSWi remain risk markers but have not yielded substantial predictive information by themselves.5,7,14,15 A CVP/PCWP ratio >0.63 was shown to be an independent predictor of early RVF risk in the HeartMate II Bridge-to-Transplantation Pivotal Trial, with an overall low discrimination (0.68).5,9,15
Echocardiography is emerging as a feasible tool for evaluating parameters of RV dysfunction.5,6,12,13,17 Puwanant et al.12 demonstrated a 91% specificity and 46% sensitivity for a low tricuspid annular plane systolic excursion, with a cutoff of 7.5 mm, in predicting post-LVAD RVF. The RV/LV diameter ratio, as a surrogate of disproportionate RV remodeling analogous to the CVP/PCWP ratio, showed a strong association with RVF in some studies.2,5,13 Moreover, the Stanford University team14 noticed that the PAPi is a better predictor of RVF after LVAD in patients receiving inotropes when compared with the CVP/PCWP ratio. This may be because of the effect inotropes have in unloading the LV and lowering PCWP, which may increase the CVP/PCWP in patients receiving inotropes, making CVP/PCWP less predictive of RVF in that cohort. Vivo et al.13 demonstrated that based on ROC curve analysis, an RV/LV diameter ratio >0.75 (AUC = 0.68) was as optimal as those obtained from the risk scoring systems of Matthews (AUC = 0.69)8 and Kormos (AUC = 0.63)9 in predicting RVF alone or both RVF and death.
To resume all current scores, similar to the CRITT model11 but by adopting some of the current and accurate methods and metrics to assess RV function preoperatively,17 we attempted to create a reliable risk stratification tool applicable to patients with end-stage heart failure.
Although many previously published risk models for RVF involve complex calculations,5,17 the ALMA score* is very easy to use and remember. Despite the small sample size of the current study, the ALMA score* might emerge as a competitive risk stratification tool in terms of the ROC analysis results and predictive accuracy (Table 5; Figure 2).2,5–9,11–17
In terms of practice patterns, in our opinion, patients with an ALMA score* of 2 and 3 should be treated with the most recent surgical approaches and technologies to try to improve the current outcomes (Figure 3).19 Previously, we showed that primary and prompt temporary CentriMag RVAD support in long-term LVAD recipients provides encouraging results compared with patients undergoing permanent BVAD/total artificial heart.17,18 This may allow certain patients to be on LVAD support alone, despite preoperative biventricular dysfunction.17,18–20
Despite the controversy over concomitant TVR in the literature,17,21 at our institutions, primary temporary RVAD support with TVR at the time of LVAD insertion resulted in improved early RV function with consequent higher probability of weaning from temporary RVAD.17,19 Additionally, a minimally invasive approach to LVAD placement may contribute to further stabilize RV contractility as a result of the partial integrity of the pericardium.17,22
Predicting the outcome after CF LVAD placement is challenging. More sophisticated risk assessments using several continuous (rather than dichotomized) variables that build upon already established clinical tools and employ patient data from multiple centers are needed.5,17 Given the cost, morbidity, and mortality associated with RV failure after LVAD, in the near future, we should be able to identify the patients who will benefit most from CF LVAD placement.17
This was a double-center, retrospective study, which investigated not a huge volume population sample. Only patients with complete data insertion into both official institutional MCS datasets were used for this study. Preimplant hemodynamic, echo, and laboratory parameters were not collected simultaneously before LVAD, and these parameters can evolve slightly with changes in medical therapy or a patient’s clinical state. However, there was no correlation between the ALMA score* calculation and the time between clinical assessment and implantation.
Statistically, the predictors have been reduced from 57 to 19. The stepwise regression was repeated from several starting points for checking, and the procedure always converged to the same 5 variables, as well. We tried to weigh these variables as well, but the analysis provided no better predictive value thus supporting the binary model of our suggested easy scoring system.
We resulted to have a high rate of post-LVAD RVF managed mechanically with an RVAD in overall studied population if compared with the recent literature.17 This, in our opinion, depends on our historical high volume of CentriMag implants.23 Additionally, the CentriMag population has been more unstable preoperatively and at higher risk of post-LVAD RVF if compared with the implantable long-term LVAD (particularly newer generation) population, thus providing 2 potentially different kind of MCS recipients (paracorporeal vs. implantable),17,19,23 even if not significantly in the current study. The other point is our radical and quicker decision for temporary RVAD implantation by time (even according to both institutions experience, learning curve, and clinical results as already described elsewhere19) with the aim to support the RV in the very early phase, thus getting the chance to wean certain patients and leave them on implantable LVAD support only despite a preoperative moderate RV dysfunction.17,19 This resulted to provide a high volume of “unplanned BVAD” by time. For completeness, planned BVAD population, in the same study period, encountered only 44 patients, and the outcomes have been poor as partially reported elsewhere.19
Despite the above-mentioned limitations, we think our institutional score might contribute in providing a first indicative provisional screening before any current CF LVAD placement.
The authors thank the MCS team of the Deutsches Herzzentrum Berlin, Germany, for continuous support and advices.
*“ALMA” stands both for “ALMA” Mater Studiorum Bologna University and “[A]ntonio [L]oforte & [M]ontalto [A]ndrea” MCS research network.
1. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant 2015.34: 14951504.
2. Potapov EV, Krabatsch T, Ventura HO, Hetzer R. Advances in mechanical circulatory support: year in review. J Heart Lung Transplant 2011.30: 487493.
3. Slaughter MS, Pagani FD, Rogers JG, et al.; HeartMate II Clinical Investigators: Clinical management of continuous-flow left ventricular assist devices in advanced heart failure. J Heart Lung Transplant 2010.29(4 suppl): S1S39.
4. Miller LW, Guglin M. Patient selection for ventricular assist devices: a moving target. J Am Coll Cardiol 2013.61: 12091221.
5. Neyer J, Arsanjani R, Moriguchi J, Siegel R, Kobashigawa J. Echocardiographic parameters associated with right ventricular failure
after left ventricular assist device
: A review. J Heart Lung Transplant 2016.35: 283293.
6. Raina A, Seetha Rammohan HR, Gertz ZM, Rame JE, Woo YJ, Kirkpatrick JN. Postoperative right ventricular failure
after left ventricular assist device
placement is predicted by preoperative echocardiographic structural, hemodynamic, and functional parameters. J Card Fail 2013.19: 1624.
7. Fitzpatrick JR 3rd, Frederick JR, Hsu VM, et al. Risk score
derived from pre-operative data analysis predicts the need for biventricular mechanical circulatory support. J Heart Lung Transplant 2008.27: 12861292.
8. Matthews JC, Koelling TM, Pagani FD, Aaronson KD. The right ventricular failure
a pre-operative tool for assessing the risk of right ventricular failure
in left ventricular assist device
candidates. J Am Coll Cardiol 2008.51: 21632172.
9. Kormos RL, Teuteberg JJ, Pagani FD, et al.; HeartMate II Clinical Investigators: Right ventricular failure
in patients with the HeartMate II continuous-flow left ventricular assist device
: incidence, risk factors, and effect on outcomes. J Thorac Cardiovasc Surg 2010.139: 13161324.
10. Wang Y, Simon MA, Bonde P, et al. Decision tree for adjuvant right ventricular support in patients receiving a left ventricular assist device
. J Heart Lung Transplant 2012.31: 140149.
11. Atluri P, Goldstone AB, Fairman AS, et al. Predicting right ventricular failure
in the modern, continuous flow left ventricular assist device
era. Ann Thorac Surg 2013.96: 857863; discussion 863.
12. Puwanant S, Hamilton KK, Klodell CT, et al. Tricuspid annular motion as a predictor of severe right ventricular failure
after left ventricular assist device
implantation. J Heart Lung Transplant 2008.27: 11021107.
13. Vivo RP, Cordero-Reyes AM, Qamar U, et al. Increased right-to-left ventricle diameter ratio is a strong predictor of right ventricular failure
after left ventricular assist device
. J Heart Lung Transplant 2013.32: 792799.
14. Kang G, Ha R, Banerjee D. Pulmonary artery pulsatility index predicts right ventricular failure
after left ventricular assist device
implantation. J Heart Lung Transplant 2016.35: 6773.
15. Grandin EW, Zamani P, Mazurek JA, et al. Right ventricular response to pulsatile load is associated with early right heart failure and mortality after left ventricular assist device
. J Heart Lung Transplant 2017.36: 97105.
16. Matthews JC, Pagani FD, Haft JW, Koelling TM, Naftel DC, Aaronson KD. Model for end-stage liver disease score
predicts left ventricular assist device
operative transfusion requirements, morbidity, and mortality. Circulation 2010.121: 214220.
17. Loforte A, Grigioni F, Marinelli G. The risk of right ventricular failure
with current continuous-flow left ventricular assist devices. Expert Rev Med Devices 2017.14: 969983.
18. Fitzpatrick JR 3rd, Frederick JR, Hiesinger W, et al. Early planned institution of biventricular mechanical circulatory support results in improved outcomes compared with delayed conversion of a left ventricular assist device
to a biventricular assist device. J Thorac Cardiovasc Surg 2009.137: 971977.
19. Loforte A, Stepanenko A, Potapov EV, et al. Temporary right ventricular mechanical support in high-risk left ventricular assist device
recipients versus permanent biventricular or total artificial heart support. Artif Organs 2013.37: 523530.
20. Yoshioka D, Takayama H, Garan RA, et al. Contemporary outcome of unplanned right ventricular assist device for severe right heart failure after continuous-flow left ventricular assist device
insertion. Interact Cardiovasc Thorac Surg 2017.24: 828834.
21. Robertson JO, Grau-Sepulveda MV, Okada S, et al. Concomitant tricuspid valve surgery during implantation of continuous-flow left ventricular assist devices: a Society of Thoracic Surgeons database analysis. J Heart Lung Transplant 2014.33: 609617.
22. Rojas SV, Avsar M, Hanke JS, et al. Minimally invasive ventricular assist device surgery. Artif Organs 2015.39: 473479.
23. Loforte A, Montalto A, Ranocchi F, et al. Levitronix CentriMag third-generation magnetically levitated continuous flow pump as bridge to solution. ASAIO J 2011.57: 247253.