The shortage of donor hearts has led to an increased use of temporary mechanical circulatory support as a bridge to urgent heart transplantation (HT). This study aimed to identify predictors of in-hospital mortality in patients supported with short-term ventricular assist devices (VADs), specifically CentriMag and Impella, as a bridge to HT, used either as direct implantation or following prior extracorporeal membrane oxygenation support (ECMO).
MethodsRetrospective, single-center, observational study conducted at a tertiary care center, including all patients receiving short-term VADs, specifically CentriMag and Impella, as a bridge to HT between 2016 and 2024.
ResultsA total of 73 patients were included: 22 received VAD following prior ECMO support (27.3% underwent transplantation; 50% in-hospital mortality post-transplant), and 51 underwent direct VAD implantation (80.4% underwent transplantation; 9.8% in-hospital mortality post-transplant). Independent predictors of in-hospital mortality included: use of CentriMag as left-sided VAD, acute coronary syndrome (ACS), VAD implantation following prior ECMO support, infection, and female sex. After adjustment for confounding variables, prior ECMO support before VAD implantation remained significantly associated with increased in-hospital mortality.
ConclusionsThe complex nature of patients requiring short-term VADs and the physiological impact of different devices must be addressed early and individually, based on the etiology of the shock and urgency status. In-hospital mortality was significantly higher among patients previously supported with ECMO, which highlights the importance of timely and optimized management to mitigate complications and enhance outcomes.
A escassez de corações de dadores tem impulsionado a utilização de suporte circulatório mecânico temporário como ponte para transplante cardíaco (TC) urgente. Este estudo teve como objetivo identificar preditores de mortalidade intra-hospitalar em doentes submetidos a dispositivos de assistência ventricular (DAV) de curta duração, nomeadamente CentriMag e Impella, implantados diretamente ou após suporte com oxigenação por membrana extracorporal (ECMO).
MétodosEstudo observacional, retrospetivo e unicêntrico, realizado num centro terciário, incluindo todos os doentes submetidos a DAV de curta duração entre 2016 e 2024.
ResultadosForam incluídos 73 doentes: 22 receberam DAV após suporte prévio com ECMO (27,3% foram transplantados; 50% de mortalidade intra-hospitalar após o transplante), enquanto 51 foram submetidos à implantação direta de DAV (80,4% transplantados; 9,8% de mortalidade). Os preditores independentes de mortalidade intra-hospitalar foram: utilização de CentriMag como DAV esquerda, síndrome coronária aguda, implantação de DAV após suporte com ECMO, infeção e sexo feminino. Após ajuste para variáveis de confusão, a utilização prévia de ECMO manteve associação significativa com mortalidade aumentada.
ConclusõesA gravidade clínica dos doentes e o impacto fisiológico dos DAV requerem avaliação precoce e individualizada, com base na etiologia do choque e na urgência. A mortalidade intra-hospitalar foi mais elevada nos doentes previamente suportados com ECMO, reforçando a importância de uma abordagem oportuna e otimizada para melhorar os desfechos clínicos.
Despite significant advances in ventricular assist devices (VADs), heart transplantation (HT) remains the gold standard for end-stage heart failure (HF).1,2 Nevertheless, access is limited by donor scarcity and patient-specific contraindications that may preclude eligibility. In cardiogenic shock, characterized by rapid onset and frequent biventricular involvement, extracorporeal membrane oxygenation (ECMO) is frequently employed as an initial stabilizing strategy. In some patients, ECMO serves as a bridge to VAD implantation. However, the management of these critically ill patients remains highly complex and is associated with substantial risk.2–6
In Spain, a relatively high organ donation rate and unique healthcare system factors have led to the widespread use of ECMO and short-term VADs as bridges to urgent transplantation, depending on the patient's functional status and the presence of uni- or biventricular involvement. Nonetheless, these temporary support strategies carry significant risks that may increase morbidity and mortality.4,7–9 Despite growing clinical use, there is limited literature on ECMO followed by short-term VAD as a bridge-to-bridge strategy to HT.
ObjectivesThis study aimed to identify independent predictors of in-hospital mortality among patients who received short-term VADs, specifically CentriMag (Abbott, Chicago, IL) and Impella (Abiomed, Danvers, MA), as a bridge to HT, either via direct VAD implantation or following prior ECMO support.
MethodsStudy designThis was a single-center, retrospective, observational and analytical study conducted at a tertiary care center between January 2016 and June 2024. We included all patients listed for HT under urgent status or Code 0 designation (national priority for the first available donor heart) who received short-term VADs, specifically CentriMag and Impella. Patients were divided into those who received direct VAD implantation and those who received prior support with venoarterial ECMO (ECMO-bridge-to-VAD). Exclusion criteria included lack of ECMO weaning prior to HT (i.e., VAD used solely for ventricular unloading), including those with ECPELLA configurations (combined ECMO and Impella for left ventricular unloading). Additionally, patients were excluded who had received other types of devices, such as long-term implantable left VADs or total artificial hearts (e.g. SynCardia; Syncardia Systems, Tucson, AZ), as a bridge to HT, as well as those who received short-term VADs as a bridge to long-term VAD implantation. Patients undergoing combined organ transplantation (e.g. heart-kidney transplantation) were also excluded.
The study was approved by the institutional ethics committee. Informed consent was not required as this was a retrospective, non-interventional study and no identifiable patient data were used.
The following variables were analyzed: age, sex, weight, height, and body mass index (BMI); comorbidities including diabetes, dyslipidemia, hypertension, and chronic obstructive pulmonary disease (COPD); underlying etiology or indication for device implantation (ACS vs. chronic dilated cardiomyopathy, either idiopathic or ischemic); type of VAD (CentriMag or Impella); INTERMACS (Interagency-Registry-for-Mechanically Assisted-Circulatory-Support) profile; peri-implant cardiac arrest; duration of support (with ECMO and VAD or VAD alone in the direct VAD to HT group); laboratory parameters including lactate, creatinine, and bilirubin measured at device implantation and at 48 hours post-implantation. Complications were also recorded, including ischemic or hemorrhagic ones, major infectious complications (pneumonia, bacteremia, or sepsis), a need for hemodialysis, and neurological complications (intracranial hemorrhage, ischemic stroke, seizures, or hypoxic–ischemic brain injury).
Surgical techniqueInitially, left-sided CentriMag VADs were implanted via median sternotomy, with an apical left ventricular inflow cannula and arterial outflow in the ascending aorta. Biventricular support involved additional right atrial drainage and pulmonary artery return cannula. The approach later evolved to left thoracotomy for inflow left ventricular cannula and subclavicular access to the axillary artery for outflow. For right ventricular support, a percutaneous approach using femoral vein to right atrial inflow and jugular vein to pulmonary artery outflow was adopted. More recently, left-sided support with axillary Impella was combined with right-sided CentriMag via dual-lumen ProtekDuo cannula (LivaNova, PLC, UK) inserted through the right internal jugular vein.10
Statistical analysisQuantitative variables were expressed as mean and standard deviation or as median and interquartile range, depending on distribution. Categorical variables were expressed as frequencies and percentage. Bivariate comparisons used Student's t-test or Mann–Whitney U test for continuous variables and Chi-square test or Fisher's exact test for categorical variables. A binary logistic regression analysis was used to identify independent predictors of in-hospital mortality. Variables with p<0.05 in the bivariate analysis with clinical relevance were included using the forward stepwise selection method to obtain a parsimonious model and minimize overfitting. Forward selection offers the advantage of incorporating only statistically significant variables, halting the inclusion process when no additional variables meet the predefined entry criteria. However, it presents the limitation of not accounting for all potential interactions among variables. Statistical significance was assessed using the likelihood ratio test.
To evaluate the adjusted effect of direct VAD versus ECMO-bridge-to-VAD, a full model including the exposure and theoretically relevant cofounders was constructed. Variables were retained if exclusion altered the odds ratio (OR) of the main exposure >10%.
For handling missing data, the listwise deletion strategy was employed, as Little's test confirmed that the missing data mechanism was missing completely at random, with a non-significant Chi-squared statistic. All analyses were performed using IBM SPSS Statistics v.27 (IBM Corp., Armonk, NY). A p-value <0.05 was considered statistically significant.
ResultsBetween January 2016 and June 2024, 73 patients were analyzed: 22 underwent ECMO-bridge-to-VAD (18 CentriMag, 4 Impella), and 51 received direct VAD (37 CentriMag, 14 Impella). Isolated left-sided VAD support was used in 58 patients; 14 required biventricular support, and 1 received isolated right-sided VAD support. In the ECMO-bridge-to-VAD group, 6 out of 22 underwent HT (27.3%). The in-hospital mortality among these transplanted patients was 50% (3 survivors), and overall in-hospital mortality in this group was 86.4%. In the direct VAD group, 41 of 51 patients reached transplantation (80.4%). In-hospital mortality among these transplanted patients was 9.8% (37 survivors), with an overall in-hospital mortality of 27.5% in the group (Figure 1).
Table 1 presents the comparison between the ECMO-bridge-to-VAD and direct VAD groups. Hypertension was more prevalent in the direct VAD group. ECMO-bridge-to-VAD patients had higher rates of peri-implant cardiac arrest (31.8% vs. 5.9%; p=0.003), were all INTERMACS class 1, and had higher arterial lactate and plasma creatinine levels. This group also had a higher incidence of complications, including infection, limb ischemia, local bleeding, and need for hemodialysis. Median support duration was significantly longer (33.5 vs. 9 days; p<0.001).
Bivariate analysis of differences between patients with ECMO-bridge-to-VAD and patients with direct VAD.
| Total(n=73) | ECMO-bridge-to-VAD(n=22) | Direct VAD(n=51) | p-Value | |
|---|---|---|---|---|
| Age (years), median (IQR) | 54.5(15.5) | 51.5(15.5) | 55.5(19.8) | 0.342 |
| Male sex | 83.6% | 77.3% | 86.3% | 0.341 |
| BMI (kg/m2), mean (SD) | 24.6(4.7) | 22.8(5.1) | 25.1(4.5) | 0.111 |
| Obesity (BMI >30 kg/m2) | 12.3% | 4.5% | 15.7% | 0.386 |
| Diabetes | 27.4% | 18.2% | 31.4% | 0.246 |
| Dyslipidemia | 37% | 22.7% | 43.1% | 0.097 |
| Hypertension | 60.3% | 36.4% | 70.6% | 0.006 |
| COPD | 12.3% | 13.6% | 11.8% | 0.823 |
| Peri-implant cardiac arrest | 13.7% | 31.8% | 5.9% | 0.003 |
| INTERMACS 1 | 34.2% | 100% | 5.9% | <0.001 |
| INTERMACS 2 | 17.8% | 0% | 25.5% | |
| INTERMACS 3 | 47.9% | 0% | 68.6% | |
| Pre-implant lactate (mmol/L), median (IQR) | 1(1.2) | 2.8(2.2) | 0.8(0.5) | <0.001 |
| Lactate at 48 h (mmol/L), median (IQR) | 1(0.9) | 1.7(1.1) | 0.9(0.5) | <0.001 |
| Pre-implant creatinine (mg/dL), median (IQR) | 0.99(0.8) | 1.81(1.5) | 0.93(0.5) | <0.001 |
| Creatinine at 48 h (mg/dL), median (IQR) | 0.92(0.7) | 1.27(1.3) | 0.85(0.6) | 0.011 |
| Pre-implant bilirubin (mg/dL), median (IQR) | 1.25(1.2) | 1.48(1.16) | 0.9(1.1) | 0.084 |
| Bilirubin at 48 h (mg/dL), median (IQR) | 1.38(1.5) | 1.6(1.6) | 1.2(1.5) | 0.331 |
| Hemodialysis | 23.3% | 54.5% | 9.8% | <0.001 |
| Infection | 32.9% | 63.7% | 19.6% | <0.001 |
| Ischemia of cannulated limb | 6.8% | 18.2% | 2% | 0.012 |
| Local hemorrhage | 20.5% | 36.4% | 13.7% | 0.028 |
| Neurological complications | 16.4% | 22.7% | 13.7% | 0.341 |
| Total assistance time (days), median (IQR) | 12(15) | 33.5(34) | 9(9) | <0.001 |
BMI: body mass index; COPD: chronic obstructive pulmonary disease; ECMO: extracorporeal membrane oxygenation; INTERMACS: Interagency-Registry-for-Mechanically-Assisted-Circulatory-Support; IQR: interquartile range; SD: standard deviation; VAD: ventricular assist device.
Table 2 compared patients with in-hospital mortality (HM) to survivors (No-HM). A lower proportion of male patients was found in the HM group (72.7% vs. 92.5%; p=0.023). In-hospital mortality was associated with ACS, peri-implant cardiac arrest, ECMO-bridge-to-VAD strategy, and use of CentriMag for left-sided support. INTERMACS class 1 was also strongly associated with mortality. The HM group had significantly higher lactate at implantation and at 48 hours, and more frequent complications: infections (57.6% vs. 12.5%; p<0.001), hemodialysis (36.4% vs. 12.5%; p=0.016), and neurological complications (27.3% vs. 7.5%; p=0.023). Median support duration was longer in the HM group (20 vs. 9 days; p<0.001).
Bivariate analysis of predictors of in-hospital mortality.
| Total(n=73) | In-hospital mortality (HM)(n=33) | No in-hospital mortality (No-HM)(n=40) | p-Value | |
|---|---|---|---|---|
| Age (years), median (IQR) | 54.5(15.5) | 56.5(15.5) | 52(19.8) | 0.39 |
| Male sex | 83.6% | 72.7% | 92.5% | 0.023 |
| BMI (kg/m2), mean (SD) | 24.6(4.7) | 24.9(5.7) | 24.4(4.1) | 0.73 |
| Obesity (BMI >30 kg/m2) | 12.3% | 12.1% | 12.5% | 0.713 |
| Diabetes | 27.4% | 7% | 13% | 0.282 |
| Dyslipidemia | 37% | 37% | 27% | 0.264 |
| Hypertension | 60.3% | 57.6% | 62.5% | 0.669 |
| COPD | 12.3% | 9.1% | 15% | 0.445 |
| Acute coronary syndrome | 23.3% | 45.5% | 5% | <0.001 |
| Peri-implant cardiac arrest | 13.7% | 24.2% | 5% | 0.017 |
| ECMO-bridge-to-VAD | 30.1% | 57.6% | 7.5% | <0.001 |
| Isolated left VAD | 79.5% | 75.8% | 82.5% | 0.42 |
| Isolated right VAD | 1.4% | 0% | 2.5% | |
| Biventricular VAD | 19.2% | 24.2% | 15% | |
| CentriMag as left VAD | 74% | 87.9% | 64.1% | 0.02 |
| INTERMACS 1 | 34.2% | 63.6% | 10% | <0.001 |
| INTERMACS 2 | 17.8% | 18.2% | 17.5% | |
| INTERMACS 3 | 47.9% | 18.2% | 72.5% | |
| Pre-implant lactate (mmol/L), median (IQR) | 1(1.2) | 1.4(3.2) | 0.9(1.6) | 0.002 |
| Lactate at 48 h (mmol/L), median (IQR) | 1(0.9) | 1.2(1) | 1(0.5) | 0.02 |
| Pre-implant creatinine (mg/dL), median (IQR) | 0.99(0.8) | 1.36(1.3) | 0.94(0.5) | 0.034 |
| Creatinine at 48 h (mg/dL), median (IQR) | 0.92(0.7) | 1.06(1.1) | 0.86(0.6) | 0.096 |
| Pre-implant bilirubin (mg/dL), median (IQR) | 1.25(1.2) | 1.25(1.36) | 1.24(1.01) | 0.881 |
| Bilirubin at 48 h (mg/dL), median (IQR) | 1.38(1.5) | 1.5(1.7) | 1.15(1.25) | 0.458 |
| Hemodialysis | 23.3% | 36.4% | 12.5% | 0.016 |
| Infection | 32.9% | 57.6% | 12.5% | <0.001 |
| Ischemia of cannulated limb | 6.8% | 12.1% | 2.5% | 0.169 |
| Local hemorrhage | 20.5% | 24.2% | 17.5% | 0.478 |
| Neurological complications | 16.4% | 27.3% | 7.5% | 0.023 |
| Total assistance time (days), median (IQR) | 12(15) | 20(31) | 9(9) | <0.001 |
BMI: body mass index; COPD: chronic obstructive pulmonary disease; ECMO: extracorporeal membrane oxygenation; INTERMACS: Interagency-Registry-for-Mechanically-Assisted-Circulatory-Support; IQR: interquartile range; SD: standard deviation; VAD: ventricular assist device.
In the patients who underwent a HT, those in the ECMO-bridge-to-VAD group showed higher mortality (42.9% vs. 7.5%; p=0.035). Infection was more frequent among those who died (57.1% vs. 12.5%; p=0.018). Although the absence of invasive mechanical ventilation (IMV) at the time of transplantation did not differ between groups, the duration of IMV was longer in patients who died (6 vs. 1 day; p=0.009). Organ ischemia time (202 vs. 187 minutes; p=0.636) and donor characteristics showed no significant differences. A multivariate analysis of this subgroup of transplanted patients could not be performed due to the limited number of events and the imbalance observed in predictor variables. Hospital mortality was mainly due to major infection, multiorgan failure, or neurological complications before HT, and predominantly to multiorgan failure after transplantation.
Multivariate analysisTo address the objective of the study, the following variables were included in the multivariate model: sex (male as reference), ACS (with chronic dilated cardiomyopathy, idiopathic or ischemic, as reference), ECMO-bridge-to-VAD (with direct VAD as the reference), use of CentriMag as left-sided VAD (versus Impella), lactate and creatinine levels at implantation, hemodialysis, infection, neurological complication, and duration of support. INTERMACS class 1 (vs. classes 2 and 3) was excluded from the model due to high collinearity with the type of support. All ECMO-bridge-to-VAD patients were INTERMACS class 1. When both variables were analyzed together in the first step of the backward stepwise method, they exhibited exceptionally large standard errors, with p=1). The following were identified as independent predictors of in-hospital mortality (Table 3): use of CentriMag as left-sided VAD (OR 810.13; p<0.001), ACS etiology (OR 134.715; p=0.001), ECMO-bridge-to-VAD (OR 79.315; p<0.001), infection (OR 26.743; p<0.001) and female sex (OR 21.169; p=0.014). Statistical significance was assessed using the likelihood ratio test.
Multivariate analysis of predictors of in-hospital mortality.
| OR | 95%CI OR | p-Value | |
|---|---|---|---|
| CentriMag as left VAD | 810.13 | 3.914–167,672.564 | <0.001 |
| Acute coronary syndrome | 134.715 | 1.677–10,823.667 | 0.001 |
| ECMO-bridge-to-VAD | 79.315 | 5.778–1,088.736 | <0.001 |
| Infection | 26.743 | 3.008–237.738 | <0.001 |
| Female sex | 21.169 | 1.432–312.916 | 0.014 |
ECMO: extracorporeal membrane oxygenation; OR: odds ratio; VAD: ventricular assist device; 95%CI: 95% confidence interval.
If these significant predictors were used to build a predictive model for in-hospital mortality, which is beyond the scope of this study, the model would explain data variability with a Nagelkerke R2 of 0.802. The model demonstrated a correct classification rate of 91.2%, with a non-significant Hosmer–Lemeshow test (p=0.992). The overall model was statistically significant (p<0.001 in likelihood ratio test), explaining a substantial proportion of variability in the outcome. The area under the curve was 0.964 (95%CI 0.925–1), with sensitivity 93.8% and specificity 88.9% (Figure 2).
To assess the adjusted effect of direct VAD versus ECMO-bridge-to-VAD, a full model was constructed including the exposure variable (ECMO-bridge-to-VAD) and the following control variables: female sex, CentriMag as left-sided VAD, ACS etiology, and infection. Results of this model are presented in Table 4. This model was selected as the optimal one for adjustment, as the exposure effect estimate varied by less than 10% and yielded the narrowest confidence interval. In this final adjusted model, ECMO-bridge-to-VAD remained a statistically significant risk factor for in-hospital mortality, with an OR of 79.315 (95%CI 5.778–1088.736; p=0.001).
Adjusted analysis of the effect of direct VAD versus ECMO-bridge-to-VAD.
| OR | 95%CI OR | p-Value | |
|---|---|---|---|
| CentriMag as left VAD | 810.13 | 3.914–167,672.564 | 0.014 |
| Acute coronary syndrome | 134.715 | 1.677–10,823.667 | 0.028 |
| ECMO-bridge-to-VAD | 79.315 | 5.778–1,088.736 | 0.001 |
| Infection | 26.743 | 3.008–237.738 | 0.003 |
| Female sex | 21.169 | 1.432–312.916 | 0.001 |
ECMO: extracorporeal membrane oxygenation; OR: odds ratio; VAD: ventricular assist device; 95%CI: 95% confidence interval.
The scarcity of donor organs has extended waiting times for HT, thereby increasing the use of VADs as a bridge to urgent HT. In cardiogenic shock, ECMO is commonly used to provide biventricular and respiratory support. However, short-term VADs, such as CentriMag and Impella, have emerged as effective alternatives for bridging to HT, either directly or following initial ECMO support.4,7,11 In our cohort of 73 patients, in-hospital mortality was higher among patients supported with prior ECMO (86.4%) compared to those directly supported with VAD (27.5%). These findings suggest that patients supported by ECMO have a more severe risk profile associated with the deleterious effects of ECMO, which may increase the likelihood of clinical deterioration prior to transplantation.
Despite increasing clinical use, data on ECMO followed by short-term VAD as a bridge-to-bridge strategy to HT remain limited. Internationally, long-term VADs are more commonly employed in this context, showing comparable mortality to ECMO-to-HT strategies, but with potentially higher complication rates.12–14 In Spain, high organ donation rates shorten waiting times for HT, influencing mechanical support strategies.15 ECMO and short-term VADs are commonly used as direct bridges to HT, shaped by the logistics of the national transplant program. In our center, the use of direct ECMO support as a bridge to HT has shown excellent outcomes, as previously reported: 90.9% of patients supported with ECMO successfully underwent HT, with a 75% survival rate among those who reached HT.16 The need for a bridge-to-bridge approach does not necessarily imply ECMO failure. In 2017, Spanish prioritization criteria for HT were modified to limit ECMO support under Urgency 0 status to 7–10 days, mandating transition to other strategies and necessitating consideration of VAD use as an additional bridge. No improvement in mortality was observed. In 2023, this time limit was removed, with strict criteria applied to avoid ECMO-related complications.17
The CentriMag system is a paracorporeal centrifugal pump device capable of providing prolonged support in various configurations.18–20 Mohite et al.19 analyzed 46 patients supported with CentriMag for over 30 days: 21 recovered, 8 bridged to long-term VAD, and only 5 bridged to HT (100% one-year survival), with the remainder deceased. Conversely, Takayama et al.20 described 143 patients with shorter support durations. 18% were transplanted, with 49% one-year survival. Most were INTERMACS class 1. The Spanish multicenter registry21 with 358 patients reported 84.6% transplantation rate and 77.6% one-year survival. The most frequent complications were bleeding, infections, and stroke.20,22
Whereas the Impella device is a transaortic axial-flow pump that unloads the left ventricle, enhancing systemic and myocardial perfusion. Its less invasive surgical approach enables early withdrawal of mechanical ventilation and active rehabilitation. Various models exist: CP, 5.0, and 5.5 for left ventricular support (our series included 6, 2, and 10 patients with each type, respectively), and RP for right ventricular support (not used at our center).10,11,23 Haddad et al.11 reported 16 patients supported with Impella reaching HT with 100% one-year survival. Funamoto et al.24 found that 10 of 25 patients supported with ECMO prior to Impella were transplanted and experienced fewer complications compared to CentriMag-supported patients.11,24 Gill et al.25 described 221 patients, 105 of whom received Impella as a bridge to HT, with 81% 90-day survival and 49% were INTERMACS class 1, and 14% had prior ECMO.
Our multivariate analysis identified use of CentriMag as left-sided VAD, ACS, ECMO-bridge-to-VAD, infection, and female sex as independent predictors of in-hospital mortality. After adjusting for confounders, ECMO use remained associated with increased mortality, suggesting that factors beyond baseline severity contribute to poor outcomes. The literature emphasizes the importance of early hemodynamic optimization, as delays may disrupt micro- and macrocirculation coherence. Invasive monitoring and timely intervention are essential.26,27 In ACS, although ECMO has shown limited benefit, early use of Impella improved survival in the Danger Shock Trial.28,29 Underlying etiology, ventricular involvement, device selection, and timing are critical.30 Patients without compensatory adaptations of chronic HF benefit from early circulatory support and ventricular unloading to enhance myocardial perfusion and reduce congestion.
Of the 73 patients in our cohort, 58 received isolated left-sided VAD support, 14 required biventricular support, and 1 received isolated right-sided support. Mortality did not differ between uni- and biventricular support. However, right ventricular failure after long-term left VAD implantation has been reported in 30–40% of cases, with 10–40% mortality, which underscores the importance of careful patient selection.31,32 Impella RP was not used in our cohort: its femoral venous access limits mobilization during prolonged support. This may be overcome using CentriMag with a dual-lumen ProtekDuo cannula via the jugular vein.10 Percutaneous biventricular VAD combining Impella and ProtekDuo offers advantages over sternotomy-based strategies, including reduced transfusion requirements, earlier weaning from mechanical ventilation, and faster recovery. It outperforms other minimally invasive techniques.33,34
The main limitation of this study is the limited sample size and number of events, which may reduce statistical power to estimate parameters accurately. The single-center design provides homogeneity but limits external validity. Overfitting due to sample size and the limited number of available events may also compromise external validity, potentially leading to an overestimation of the model's area under the curve, which could not be corrected through cross-validation. This is reflected in the wide confidence intervals of the predictors and poses a challenge to the generalizability of the results. The optimal approach would have been to use a propensity score; however, in our study, the technique employed predominantly depends on the patient's condition. Thus, a multivariable logistic regression was performed for the analysis of predictor factors. Additionally, listwise deletion to handle missing data reduced statistical power. Variability in device types and transplantation indications may also have influenced results. Moreover, local factors related to the Spanish transplant program logistics may influence outcomes, and differences in transplant program organization should be considered when extrapolating these findings. We describe clinically relevant trends that require confirmation in larger studies. Independent of unmeasured confounders, assignment to the ECMO-bridge-to-VAD group was significantly associated with an increased risk of in-hospital mortality risk.
A likely predictor excluded from the model due to collinearity was INTERMACS class 1, which is associated with a higher risk of mortality and should be taken into account when interpreting the model. Patients supported with ECMO who show early signs of right ventricular failure, as well as those classified as INTERMACS 1, likely represent the most critically ill subgroup and may have the highest risk of mortality. The objective of our study was to analyze the impact of the type of strategy employed. However, because of high collinearity, the INTERMACS profile was excluded from the multivariate analysis. However, being in the ECMO-bridge-to-VAD group is essentially equivalent to being classified as INTERMACS class 1, as both convey the same information in the analysis due to their high collinearity. Moreover, this suggests that some of the effects and complications associated with this type of assistance are due not only to the specific clinical condition but also to the inherent risks of the technique itself.
ConclusionsIn conclusion, the use of CentriMag as a left ventricular assist device, ACS etiology, ECMO-bridge-to-VAD, infection, and female sex were identified as predictors of in-hospital mortality. After adjustment for confounding factors, ECMO-bridge-to-VAD compared to direct VAD implantation also showed significant differences in mortality. In-hospital mortality remains high in patients with prior ECMO support, which highlights the importance of early and appropriate management aimed at preventing complications and improving outcomes. We describe clinically relevant trends that require confirmation in larger studies.
Ethical considerationsApproved by the Ethics Committee of the University and Polytechnic Hospital La Fe of Valencia (Spain). Informed consent was not required as this was a retrospective, non-interventional study and no identifiable patient data were used.
FundingThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Conflict of interestsManuel Pérez-Guillén reports a relationship with Palex Medical SA that includes speaking and lecture fees. All other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
