Novel Scoring System to Risk Stratify Patients Receiving Durable Left Ventricular Assist Device From J-MACS Registry Data

Abstract

Background: Recently, destination therapy (DT) was approved in Japan, and patients ineligible for heart transplantation may now receive durable left ventricular assist devices (LVADs). Several conventional risk scores are available, but a risk score that is best to select optimal candidates for DT in the Japanese population remains unestablished.

Methods and Results: A total of 1,287 patients who underwent durable LVAD implantation and were listed for the Japanese registry for Mechanically Assisted Circulatory Support (J-MACS) were eligible for inclusion. Finally, 494 patients were assigned to the derivation cohort and 487 patients were assigned to the validation cohort. According to the time-to-event analyses, J-MACS risk scores were newly constructed to predict 3-year mortality rate, consisting of age, history of cardiac surgery, serum creatinine level, and central venous pressure to pulmonary artery wedge pressure ratio >0.71. The J-MACS risk score had the highest predictability of 3-year death compared with other conventional scores in the validation cohort, including HeartMate II risk score and HeartMate 3 risk score.

Conclusions: We constructed the J-MACS risk score to estimate 3-year mortality rate after durable LVAD implantation using large-scale multicenter Japanese data. The clinical utility of this scoring to guide the indication of DT should be validated in the next study.

With technological innovation and improved periprocedural management, the mortality and morbidity following implantation of a durable left ventricular assist device (LVAD) have decreased significantly.¹ With a shortage of donor hearts, heart transplantation is severely limited.² Destination therapy (DT), in which durable LVADs are implanted without heart transplant listing, has long been awaited and was finally approved in 2021 in Japan.³

DT is often applied in aged patients with advanced heart failure and multiple comorbidities.⁴ Optimal patient selection is essential to maintain acceptable clinical outcomes during long-term LVAD support. In Japan, we use Japanese HeartMate risk score (J-HMRS),³ which is a modification of the HeartMate II risk score (HMIIRS),⁵ although it has not yet been validated in Japanese LVAD patients.⁶ Recently, the HeartMate 3 risk score (HM3RS) was proposed by using data from MOMENTUM 3 trial.⁷ However, the applicability of HM3RS in the Japanese cohort again remains unknown. An original risk score constructed from our own Japanese cohort may be more practical than these conventional scores.

In this study, we attempted to construct a novel risk score to predict clinical outcomes following durable LVAD implantation using a multicenter Japanese registry for Mechanically Assisted Circulatory Support (J-MACS).⁸ We further validated its utility by comparing it with other conventional risk scores.

Methods

Patient Selection

Patients who underwent standard durable LVAD implantation at Japanese institutions were enrolled in the prospectively collected J-MACS registry database. Among them, patients enrolled in the database between 2013 and 2020 were included in this retrospective study. All participants were followed for 3 years or until June 2022. All patients received a durable implantable LVAD and those with extracorporeal devices were not included. Implantation strategies were limited to bridge-to-transplantation because DT had not been approved during this enrollment period. Patients who underwent LVAD implantation as bridge-to-bridge strategy, in which the patient underwent bridge surgery from an extracorporeal LVAD to an implantable one, were excluded. All patients provided informed consent prior to surgery at each center. The use of the J-MACS registry for this study was approved by the J-MACS Committee on March 7, 2023. The procedures followed were in accordance with the “Declaration of Helsinki”.

Cohort Stratification and Study Design

Patients were randomized into 2 cohorts: derivation and validation (Figure 1). After randomization, patients with missing data were excluded. Baseline characteristics that were significant predictors of the primary endpoint, defined as 3-year all-cause death, were examined in the derivation cohort. According to the hazard ratio of each variable, a risk scoring system consisting of these predictors was constructed. Because this cohort all consisted of bridge-to-transplant patients aged <65 years, age was included in the risk scoring regardless of its statistical significance. All device explantations, irrespective of cause or purpose, and heart transplantations were censored.

Figure 1.

Flowchart showing all patients listed in J-MACS randomly assigned to the derivation and validation cohorts in the present study. Exclusion of patients with data deficiencies created the final cohorts.

Using the validation cohort, 4 scores were calculated in each patient: the novel score (J-MACS risk score), HMIIRS, J-HMRS, and HM3RS, as described below. The predictability of the J-MACS risk score was compared with the other 3 conventional risk scores in the validation cohort.

Preoperative Baseline Data

Preoperative standard clinical data were obtained to assess patients’ baseline characteristics, including demographics, comorbidities, and laboratory, echocardiographic, and hemodynamics data. Here, history of cardiac surgery was defined as previous coronary artery bypass grafting (CABG) or valvular surgeries.

Calculation of Conventional Risk Scores

HMIIRS and HM3RS in the validation cohort were calculated using data for several baseline characteristics obtained before LVAD implantation. J-HMRS was constructed by modifying the HMIIRS, and this score was also calculated.

(1) HMIIRS⁵ was calculated as: 0.0274 × (age in years) − 0.723 × (serum albumin in g/dL) + 0.74 × (serum creatinine mg/dL) + 1.136 × (international normalized ratio INR) of prothrombin time) + 0.807 × (1 for centers with <15 implants and 0 for centers with ≥15 implants).

(2) J-HMRS³ was calculated as: 0.0274 × (age in years) − 0.723 × (serum albumin in g/dL) + 0.74 × (serum creatinine mg/dL) + 1.136 × (INR of prothrombin time) + 0.807 × (1 for centers with <3 implants per 2 years and 0 for centers with ≥3 implants per 2 years).

(3) HM3RS⁷ was calculated as: 0.03496 × (age in years) + 0.53029 (if previous CABG or valve surgery) − 0.04112 × (serum sodium in mEq/L − 136) + 0.01093 × (blood urine nitrogen in mg/dL) + 0.62149 (if left ventricular end-diastolic diameter (LVDd) <5.5 cm) + 0.44785 (if central venous pressure to pulmonary artery wedge pressure [CVP/PAWP] ratio >0.6).

Follow-up Protocol

All patients received guideline-directed medical therapy including aspirin and warfarin with a target INR of 2.0–2.5 after LVAD implantation.⁹ Doses were adjusted according to bleeding/thromboembolic events. Detailed management protocols were at the discretion of each center. Device speeds were adjusted appropriately based on hemodynamic and echocardiographic data.

Statistical Analysis

Statistical analyses were performed using SPSS Statistics 23 (SPSS Inc, Armonk, IL, USA). Two-sided P values <0.05 were considered statistically significant. Continuous variables are stated as median and interquartile range and were compared between groups using the Mann-Whitney U test. Categorical variables are stated as numbers and percentages and were compared between groups using Fisher’s exact test.

In the derivation cohort, variables that were significantly different between the deceased and living patients were included in the univariable Cox proportional hazard ratio regression analyses for the 3-year mortality rate, which was defined as the primary endpoint. Variables that were significant in the univariable analyses were included in the multivariable analysis with a stepwise method. Finally, variables that were significant in the multivariable analysis were again included to construct a multivariable model that included age. Age was included in the final model regardless of its statistical significance because of its clinical importance, particularly when considering the indication of DT.¹⁰ Receiver operating characteristic (ROC) analyses were performed to calculate statistical cutoffs using the Youden method. In the risk scoring model, 2 cutoffs were calculated to stratify patients into low-, intermediate-, and high-risk groups. Log-rank tests were used to compare the Kaplan-Meier curves of each risk group.

In the validation cohort, all risk scores (i.e., J-MACS risk score, HMIIRS, J-HMRS, and HM3RS) were calculated. ROC and log-rank tests were used to evaluate and compare their ability to stratify risks.

Results

Baseline Characteristics of the Derivation Cohort

A total of 1,287 patients were randomly assigned to the derivation and validation cohorts (Figure 1). From the initial derivation cohort of 645 patients, 151 patients with missing data were excluded. Finally, 494 patients were included in the derivation cohort. Patient characteristics were not statistically different between the initial and final cohorts.

Baseline characteristics of this cohort before LVAD implantation are summarized in Table 1. The median age was 46 (36, 55) years and 349 (71%) were male. Median plasma B-type natriuretic peptide level was 427 (227, 782) pg/mL. Since we included patients implanted between 2013 and 2020, the majority of devices were HMII (implanted since 2013, replaced by HM3 around 2020).

Table 1. Baseline Characteristics of the Derivation Cohort

	Total (n=494)	Deceased (n=57)	Alive (n=437)	P value
Demographics
Age, years	46 (36, 55)	50 (32, 58)	45 (36, 55)	0.14
Male sex	349 (71%)	34 (60%)	315 (72%)	0.040*
Body surface area, m2	1.6 (1.5, 1.8)	1.6 (1.5, 1.7)	1.6 (1.5, 1.8)	0.088
Ischemic etiology	52 (11%)	5 (9%)	47 (11%)	0.43
Heart rate, beats/min	80 (70, 90)	77 (70, 85)	80 (70, 90)	0.63
Systolic blood pressure, mmHg	88 (81, 96)	90 (82, 102)	88 (80, 96)	0.86
Annual implant number/center, n	25.3 (6.5, 26.0)	25.3 (6.5, 26.0)	25.3 (6.5, 26.0)	0.68
Comorbidities
History of cardiac surgery	153 (31%)	28 (49%)	125 (29%)	0.002*
Diabetes mellitus	87 (18%)	7 (12%)	80 (18%)	0.012*
COPD	6 (1%)	1 (2%)	5 (1%)	0.71
Echocardiographic data
LVDd, mm	68 (59, 77)	67 (57, 78)	68 (59, 77)	0.067
Moderate or greater MR	264 (53%)	29 (51%)	235 (54%)	0.39
Moderate or greater TR	166 (34%)	20 (35%)	146 (33%)	0.45
Laboratory data
Hemoglobin, g/mL	11.7 (10.6, 12.9)	11.4 (10.5, 12.6)	11.7 (10.6, 12.9)	0.041*
Platelets, ×104/μL	19.4 (15.2, 24.1)	18.2 (13.1, 23.8)	19.5 (15.2, 24.2)	0.17
Serum albumin, g/mL	3.8 (3.4, 4.2)	3.8 (3.5, 4.1)	3.8 (3.4, 4.2)	0.35
Blood urea nitrogen, mg/dL	16 (12, 22)	19 (12, 30)	16 (12, 21)	0.009*
Serum total bilirubin, mg/dL	0.9 (0.6, 1.3)	0.9 (0.6, 1.3)	0.9 (0.5, 1.3)	0.89
Serum creatinine, mg/dL	0.9 (0.7, 1.2)	1.2 (0.8, 1.5)	0.9 (0.7, 1.1)	0.010*
Serum sodium, mEq/L	137 (134, 139)	137 (134, 139)	137 (134, 139)	0.60
Serum potassium, mEq/L	4.3 (4.1, 4.6)	4.2 (4.0, 4.5)	4.3 (4.1, 4.6)	0.86
PT-INR	1.5 (1.1, 2.0)	1.7 (1.2, 2.2)	1.4 (1.1, 2.0)	0.099
Plasma BNP, pg/mL	427 (227, 782)	450 (161, 884)	412 (235, 768)	0.85
Hemodynamic data
CVP, mmHg	7 (5, 12)	9 (4, 11)	7 (5, 12)	0.22
PAWP, mmHg	20 (13, 27)	19 (12, 24)	20 (13, 27)	0.31
Cardiac output, L/min	3.3 (2.8, 4.0)	3.5 (2.8, 4.5)	3.3 (2.8, 4.0)	0.070
CVP/PAWP ratio	0.38 (0.26, 0.58)	0.39 (0.28, 0.75)	0.37 (0.26, 0.57)	0.035*
RVSWI, g m/m2	6.5 (4.5, 8.7)	6.8 (4.6, 9.1)	6.5 (4.5, 8.6)	0.84
PAPI	2.5 (1.5, 4.3)	2.3 (1.1, 3.6)	2.6 (1.5, 4.3)	0.059

Continuous variables are presented as median and interquartile and compared between groups using the Mann-Whitney U test. Categorical variables are presented as numbers and percentage and compared between groups using Fischer’s exact test. *P<0.05. BNP, B-type natriuretic peptide; COPD, chronic obstructive pulmonary disease; CVP, central venous pressure; LVDd, left ventricular end-diastolic diameter; PAPI, pulmonary artery pulsatility index; PAWP, pulmonary artery wedge pressure; PT-INR, prothrombin time with international normalized rate; RVSWI, right ventricular stroke work index.

Clinical Outcomes in the Derivation Cohort

Patients were followed for a median of 1,095 (778, 1,095) days after LVAD implantation. During the follow-up period, 57 patients died, the most frequent cause being stroke (n=26), followed by device failure (n=7), infection (n=6), and heart failure (n=6). There were several significant differences in the baseline characteristics of the deceased and living patients (Table 1). Deceased patients were less often male had more often undergone previous cardiac surgery. More advanced anemia and renal impairment were observed in the deceased patients, who also had a higher CVP/PAWP ratio.

Constructing Novel Risk Scores

The variables that were significantly different between deceased and living patients (Table 1) were included in the Cox univariable analyses (Table 2). Several variables, including a history of cardiac surgery, were significantly associated with 3-year death. Variables that were significant in the univariable analyses (Table 2) were included in the multivariable analysis, and ultimately 3 variables remained significant: a history of cardiac surgery, serum creatinine level, and CVP/PAWP ratio (Table 3). Finally, a multivariable model was constructed by adding age (Table 4). According to the results, we constructed a novel risk score (J-MACS risk score) to predict the 3-year mortality rate: 0.105 × [age (years)] + 2.06 × (history of cardiac surgery) + 3.56 × [serum creatinine (mg/dL)] + 2.61 × (CVP/PAWP >0.71).

Table 2. Association Between Potential Baseline Characteristics and 3-Year Mortality Rate

	HR (95% CI)	P value
Age, years	1.01 (0.991–1.03)	0.28
Male sex	1.68 (0.99–2.85)	0.055
Body surface area, m2	0.32 (0.09–1.24)	0.10
Diabetes mellitus	0.62 (0.28–1.37)	0.24
History of cardiac surgery	2.36 (1.40–3.97)	0.001*
LVDd, mm	0.99 (0.96–1.03)	0.88
Hemoglobin, g/dL	0.83 (0.72–0.96)	0.010*
Blood urea nitrogen, mg/dL	1.05 (1.03–1.06)	<0.001*
Creatinine, mg/dL	2.68 (1.51–4.75)	0.001*
PT-INR	1.11 (0.91–1.34)	0.31
CVP/PAWP rate	1.67 (1.07–2.61)	0.023*
PAPI	0.89 (0.80–1.01)	0.057
Cardiac output, L/min	1.10 (0.96–1.26)	0.17

Baseline characteristics that were significantly different between the alive and deceased patients (see Table 1) were included in the univariable Cox proportional HR regression analyses. Age was forced to be included irrespective of its statistical significance in Table 1. *P<0.05. CI, confidence interval; HR, hazard ratio. Other abbreviations as in Table 1.

Table 3. Multivariable Analysis for 3-Year Mortality Rates

	HR (95% CI)	P value
History of cardiac surgery	2.07 (1.19–3.60)	0.010*
Hemoglobin, g/dL	NA	NA
Serum creatinine, mg/dL	3.67 (2.03–6.62)	<0.001*
CVP/PAWP rate >0.71	2.59 (1.46–4.59)	0.001*

Variables significant in the univariable analyses (see Table 2) were included in the multivariable Cox proportional HR regression analysis with stepwise method. Blood urea nitrogen was excluded due to its strong collinearity with serum creatinine. Hemoglobin was not selected in the multivariable model due to statistical non-significance. CVP/PAWP rate alone was stated as a categorical variable with a statistically calculated cutoff due to its clinical utility. *P<0.05. Abbreviations as in Table 1,2.

Table 4. Multivariable Analysis for the 3-Year Mortality Rate Including Age

	HR (95% CI)	P value
Age, ×10 years	1.05 (0.84–1.32)	0.65
History of cardiac surgery	2.06 (1.19–3.58)	0.010*
Serum creatinine, mg/dL	3.56 (1.93–6.56)	<0.001*
CVP/PAWP rate >0.71	2.61 (1.93–4.63)	0.001*

Variables that were statistically significant in the multivariable analysis (see Table 3) were included in the final multivariable analysis together with age. The novel risk score was constructed using these variables and their HRs. *P<0.05. Abbreviations as in Table 1,2.

J-MACS Risk Score in the Derivation Cohort

Our new risk score (J-MACS risk score) was calculated for all patients in the derivation cohort. Two cutoffs were determined to risk stratify patients into 3 groups: low-risk group <10.2; 10.2 ≤ intermediate-risk group < 13.7; and 13.7 ≤ high-risk group (Figure 2A). The area under the curve of this risk score was 0.68 (95% confidence interval 0.61–0.76). Cumulative death was significantly stratified by the J-MACS risk score (P<0.001; Figure 2B). The prognosis in the intermediate-risk group was similar to the average mortality rate in the overall J-MACS primary LVAD cohort (12% at 1,080 days). The 3-year cumulative mortality rate in the high-risk group was 45%, which was obviously higher than the overall J-MACS primary LVAD cohort.

Figure 2.

Predictability of the J-MACS risk score, which was derived from the derivation cohort, on the primary outcome (A) and cumulative mortality tertiled by J-MACS risk score in the derivation cohort (B). The cohorts were tertiled into low-, intermediate-, and high-risk groups by 2 cutoffs (red circles in A). *P<0.05 by log-rank test. AUC, area under the curve; CI, confidence interval; CVP, central venous pressure; LVAD, left ventricular assist device; PAWP, pulmonary artery wedge pressure.

Risk Scoring in the Validation Cohort

The baseline characteristics of the validation cohort are summarized in Table 5. Baseline variables in the final validation cohort were not significantly different from those in the original validation cohort (P>0.05 for all). A total of 487 patients were enrolled. The median age was 46 (35, 56) years and 368 (76%) were male.

Table 5. Baseline Characteristics of the Validation Cohort

	Total (n=487)
Demographics
Age, years	46 (35, 56)
Male sex	368 (76%)
Body surface area, m2	1.7 (1.5, 1.8)
Ischemic etiology	54 (11%)
Heart rate, beats/min	80 (70, 91)
Systolic blood pressure, mmHg	88 (81, 96)
Total implants <15	78 (16%)
<3 implants per 2 years	41 (8%)
Comorbidities
History of cardiac surgery	174 (36%)
Diabetes mellitus	93 (19%)
COPD	6 (1%)
Echocardiographic data
LVDd, mm	69 (60, 78)
Moderate or greater MR	282 (62%)
Moderate or greater TR	187 (41%)
Laboratory data
Hemoglobin, g/mL	11.7 (10.3, 13.1)
Platelets, ×104/μL	18.6 (14.7, 23.4)
Serum albumin, g/mL	3.7 (3.3, 4.1)
Blood urea nitrogen, mg/dL	17 (12, 22)
Serum total bilirubin, mg/dL	0.9 (0.6, 1.3)
Serum creatinine, mg/dL	0.9 (0.7, 1.1)
Serum sodium, mEq/L	137 (134, 139)
Serum potassium, mEq/L	4.3 (4.0, 4.6)
PT-INR	1.3 (1.2, 2.0)
Plasma BNP, pg/mL	450 (237, 804)
Hemodynamic data
CVP, mmHg	8 (5, 13)
PAWP, mmHg	21 (12, 29)
Cardiac output, L/min	3.4 (2.8, 4.0)
CVP/PAWP ratio >0.71	94 (19%)

Continuous variables are presented as median and interquartile. Categorical variables are presented as numbers and percentage. *P<0.05. Abbreviations as in Table 2.

During the 1,095 (784, 1,095) days after LVAD implantation, 48 patients died, with the most frequent cause being stroke (n=18), followed by infection (n=15) and heart failure (n=4). The J-MACS risk score significantly stratified the patients’ cumulative mortality rate into 3 groups (P<0.001; Figure 3A). Notably, the high-risk group had a 53% cumulative 3-year mortality rate.

Figure 3.

Cumulative deaths following LVAD implantation in the validation cohort stratified by J-MACS risk score (A), HMIIRS (B), J-HMRS (C), and HM3RS (D) and predictability of each risk scoring, including J-MACS risk score, for the primary outcome (E). *P<0.05 by log-rank test. AUC, area under the curve; CI, confidence interval; LVAD, left ventricular assist device.

HMIIRS was also calculated and only 3 patients were assigned to the high-risk group (Figure 3B). The Kaplan-Meier curves of the low- and intermediate-risk groups were not significantly different (P=0.36).

We also calculated J-HMRS, which is the HMIIRS modified for Japanese patients. Only 2 patients were assigned to the high-risk group (Figure 3C). Again, the Kaplan-Meier curves of the low- and intermediate-risk groups were not significantly different (P=0.14).

HM3RS significantly stratified the patient cohort into 3 groups (Figure 3D, P<0.001). Notably, 61 patients were assigned to the high-risk group and their 3-year cumulative mortality was 26%.

The J-MACS risk score had the highest area under the curve (0.73, 95% confidence interval 0.67–0.79) among the all calculated scores (Figure 3E).

Discussion

In this retrospective study, using data from the large-scale, multicenter J-MACS registry, we constructed a novel risk score (J-MACS risk score) to estimate 3-year mortality rates after durable LVAD implantation in a derivation cohort. We further validated the predictability of this risk score by comparing it with other conventional risk scores in a validation cohort. We constructed the J-MACS risk score using age, a history of cardiac surgery, serum creatinine level, and CVP/PAWP ratio, all of which can be easily obtained before LVAD implantation. The J-MACS risk score had the highest predictability of all the risk scores evaluated, including HMIIRS and HM3RS.

Conventional Risk Scores, Including HMIIRS

Various risk scores have been proposed to stratify patients receiving durable LVADs.¹¹ Most of them comprise several clinical parameters, including laboratory, hemodynamics, and medication data. However, given the improvement in clinical outcomes due to technological innovation and better peri-operative management,¹² most of these parameters may improve after LVAD implantation and may no longer be the dominant risk factors.

Also, the target of LVAD implantation has trended to a less sick cohort. For example, our team previously proposed a risk score to predict the 1-year mortality rate after LVAD implantation,¹³ which consisted of serum albumin <3.2 g/dL, serum total bilirubin >4.8 mg/dL, LVDd <55 mm, and central venous pressure >11 mmHg. The cohort included patients with conventional extracorporeal LVADs. Most patients were classified as INTERMACS profile 1 or 2. Classic risk scores may not be applicable to the current less sick cohort receiving durable devices, and an updated risk score applicable to the current era has been desired.⁵

The predictability of HMIIRS was not high in this study. Center volume is one of the main contributors to this risk score.⁵ An additional risk is assigned when the center volume is <15 implants. However, many Japanese institutions cannot meet this threshold and we currently use J-HMRS by modifying the center volume threshold in HMIIRS.³ Nevertheless, center volume itself was not a significant risk factor in this study. The prognostic impact of center volume may be highly confounded by the severity of disease (i.e., high-volume centers tend to treat patients with more advanced and complex diseases and vice verse).¹⁴

HM3RS

HM3RS has several similar risk factors to our J-MACS risk score,⁷ although J-MACS was derived from a cohort of patients implanted with devices other than HM3. As a result, both scores had higher predictability than either HMIIRS or J-HMRS. HM3RS is the most recent risk score derived from a cohort of patients who received HM3.⁷

A unique factor shared by both scores is the CVP/PAWP ratio, which is an index of right ventricular (RV) function and a major risk factor for post-LVAD RV failure,¹⁵ which is associated not only with death but also with a variety of hemocompatibility-related adverse events.¹⁶ Prolonged or late-onset RV failure is particularly problematic for DT candidates receiving long-term LVAD support.¹⁷ Consistently, stroke and heart failure were the major causes of death in our cohort. We highly recommend performing right heart catheterization before LVAD implantation to measure the CVP/PAWP ratio (and calculate the J-MACS risk score).

Another factor shared by both scores is a history of cardiac surgery. A previous sternotomy confers a complicated intra- and postoperative course caused by increased cardiopulmonary bypass and cross-clamping times, intraoperative bleeding, and higher likelihood of postoperative right heart failure.¹⁸

Implication of the J-MACS Risk Score

Our J-MACS risk score had the highest predictability compared with the other conventional scores. It is the first score to be derived from large-scale, multicenter J-MACS registry dataand consists of 4 simple risk factors: age, history of cardiac surgery, serum creatinine level, and CVP/PAWP ratio, all of which can be routinely obtained before LVAD implantation. We included age in this risk scoring regardless of statistical significance because age is a key factor in considering the indication for DT.³ In Japan, age >65 years is a contraindication for heart transplantation listing and older age is considered one of the main criteria for DT. Other factors, such as a history of cardiac surgery, renal impairment, and RV failure, remain important risk factors even in the era of HM3.⁷ Of note, renal failure is both a marker of end-organ dysfunction and a cause of progression of cardiovascular disease, including stroke, based on the concept of cardiorenal syndrome.

The 3-year mortality rate in the HM3RS high-risk group was 26%. Although we do not deny the usefulness of HM3RS, it cannot distinguish the high-risk cohort who should be excluded from DT, whereas the 3-year mortality rate in the high-risk group of J-MACS risk scoring was 53%. Such patients may not be good candidates for LVAD implantation. For example, a 70-year-old patient with serum creatine 0.7 mg/dL, no history of cardiac surgery, and CVP/PAWP ratio <0.71 has 9.8 points (<10.3 points) and is assigned to the low-risk group. A 70-year-old patient with serum creatine 1.5 mg/dL, no history of cardiac surgery, and CVP/PAWP ratio <0.71 has 12.7 points (<13.7 points), and is assigned to the intermediate-risk group. Another 70-year-old patient with serum creatine 1.2 mg/dL, history of cardiac surgery, and CVP/PAWP ratio >0.71 has 16.3 points (>13.7 points), and is assigned to the high-risk group.

Considering that the calculated prognosis in the intermediate-risk group was similar to the overall J-MACS survival (88% survival at 1,080 days), low- to intermediate-risk patients are reasonable candidates for DT. However, we have to be very careful about selecting high-risk patients for DT, because it was only approved in Japan in 2021 and the applicability of the J-MACS risk score to this cohort, especially with HM3, should be validated in the next study.

Study Limitations

The J-MACS registry includes large-scale multicenter clinical data, but due to the nature of large-scale registry data, we were not able to obtain detailed clinical data, including tissue Doppler echocardiographic data. The registry also lacks detailed data on adverse events during the observation period. Due to the registry’s policy, we could not report detailed device types, nor were we allowed to compare clinical outcomes between devices. The included patients underwent LVAD implantation between 2013 and 2020, and few HM3 patients were included (HM3 implantation was approved in Japan in mid-2019). According to the Japanese reimbursement policy, durable LVADs were not allowed to be implanted as DT until 2021.³ Furthermore, we did not implant LVADs in patients aged over 65 years. Therefore, the applicability of our risk scoring to the DT cohort, those who received HM3, or those aged over 65 years requires further validation studies.

Conclusions

Using large-scale multicenter Japanese registry data, we constructed a novel risk score (J-MACS risk score), simply consisting of 4 baseline characteristics, to estimate 3-year mortality rates following durable LVAD implantation.

Acknowledgments

The authors acknowledge all the participants of the J-MACS registry and all clinical institutions, surgeons, and medical staff who contributed to this project.

Disclosures

K.K., Y. Saiki, Y. Sakata, and M.O. are members of Circulation Journal’s Editorial Team. G.M. receives research funding from Terumo Corp, Century Medical Inc., and Medtronic Inc.

IRB Information

The present study was approved by the J-MACS Committee. Reference number: 2203.

Data Availability

Data are not allowed to be shared according to the policy of the J-MACS registry.

References

• 1.   Shah P, Yuzefpolskaya M, Hickey GW, Breathett K, Wever-Pinzon O, Ton VK, et al. Twelfth Interagency Registry for Mechanically Assisted Circulatory Support report: Readmissions after left ventricular assist device. Ann Thorac Surg 2022; 113: 722–737.
• 2.   Nunoda S, Sasaoka T, Sakata Y, Ono M, Sawa Y, Isobe M, Heart Transplantation Committee of the Japanese Circulation Society. Survival of heart transplant candidates in Japan. Circ J 2019; 83: 681–683.
• 3.   Kinugawa K, Sakata Y, Ono M, Nunoda S, Toda K, Fukushima N, et al. Consensus Report on destination therapy in Japan: From the DT Committee of the Council for Clinical Use of Ventricular Assist Device Related Academic Societies. Circ J 2021; 85: 1906–1917.
• 4.   Gustafsson F, Rogers JG. Left ventricular assist device therapy in advanced heart failure: Patient selection and outcomes. Eur J Heart Fail 2017; 19: 595–602.
• 5.   Cowger J, Sundareswaran K, Rogers JG, Park SJ, Pagani FD, Bhat G, et al. Predicting survival in patients receiving continuous flow left ventricular assist devices: The HeartMate II risk score. J Am Coll Cardiol 2013; 61: 313–321.
• 6.   Yoshimura A, Kikuchi N, Saito S, Suzuki A, Hattori H, Shoda M, et al. Stratification of destination therapy candidates by J-HeartMate risk score among elderly non-responders to cardiac resynchronization therapy. Circ Rep 2022; 4: 405–411.
• 7.   Mehra MR, Nayak A, Morris AA, Lanfear DE, Nemeh H, Desai S, et al. Prediction of survival after implantation of a fully magnetically levitated left ventricular assist device. JACC Heart Fail 2022; 10: 948–959.
• 8.   Nakatani T, Sase K, Oshiyama H, Akiyama M, Horie M, Nawata K, et al; J-MACS Investigators. Japanese registry for mechanically assisted circulatory support: First report. J Heart Lung Transplant 2017; 36: 1087–1096.
• 9.   Ono M, Yamaguchi O, Ohtani T, Kinugawa K, Saiki Y, Sawa Y, et al. JCS/JSCVS/JATS/JSVS 2021 guideline on implantable left ventricular assist device for patients with advanced heart failure. Circ J 2022; 86: 1024–1058.
• 10.   Goldstein DJ, Mehra MR, Naka Y, Salerno C, Uriel N, Dean D, et al. Impact of age, sex, therapeutic intent, race and severity of advanced heart failure on short-term principal outcomes in the MOMENTUM 3 trial. J Heart Lung Transplant 2018; 37: 7–14.
• 11.   Ravichandran AK, Cowger J. Left ventricular assist device patient selection: Do risk scores help? J Thorac Dis 2015; 7: 2080–2087.
• 12.   Mehra MR, Cleveland JC Jr, Uriel N, Cowger JA, Hall S, Horstmanshof D, et al. Primary results of long-term outcomes in the MOMENTUM 3 pivotal trial and continued access protocol study phase: A study of 2200 HeartMate 3 left ventricular assist device implants. Eur J Heart Fail 2021; 23: 1392–1400.
• 13.   Imamura T, Kinugawa K, Shiga T, Endo M, Kato N, Inaba T, et al. Novel risk scoring system with preoperative objective parameters gives a good prediction of 1-year mortality in patients with a left ventricular assist device. Circ J 2012; 76: 1895–1903.
• 14.   Starling RC, Xanthopoulos A. Are outcomes related to left ventricular assist device center volume?: Too complex to answer. JACC Heart Fail 2017; 5: 700–702.
• 15.   Bellavia D, Iacovoni A, Scardulla C, Moja L, Pilato M, Kushwaha SS, et al. Prediction of right ventricular failure after ventricular assist device implant: Systematic review and meta-analysis of observational studies. Eur J Heart Fail 2017; 19: 926–946.
• 16.   Imamura T, Nguyen A, Kim G, Raikhelkar J, Sarswat N, Kalantari S, et al. Optimal haemodynamics during left ventricular assist device support are associated with reduced haemocompatibility-related adverse events. Eur J Heart Fail 2019; 21: 655–662.
• 17.   Rame JE, Pagani FD, Kiernan MS, Oliveira GH, Birati EY, Atluri P, et al. Evolution of late right heart failure with left ventricular assist devices and association with outcomes. J Am Coll Cardiol 2021; 78: 2294–2308.
• 18.   Fitzpatrick JR 3rd, Frederick JR, Hsu VM, Kozin ED, O’Hara ML, Howell E, et al. Risk score derived from pre-operative data analysis predicts the need for biventricular mechanical circulatory support. J Heart Lung Transplant 2008; 27: 1286–1292.