INTERMACS Data Research Articles

Predicting Right Ventricular Failure Following Left Ventricular Assist Device Support: An INTERMACS Validation Study

<h3>Introduction</h3> : The Utah Right Ventricular Failure (RVF) Risk score is a 7-factor predictive model (African American, history of HTN, INTERMACS profile 1-2, Na <130mEq/L, BUN >35mg/dL, PA pulse pressure <36mmHg and RA/PCWP >0.5) that achieved a high discriminative performance (AUC 0.73) for predicting RVF in a three-center study of patients undergoing left ventricular assist device (LVAD) implantation and was validated in an external two-center cohort (AUC 0.71). <h3>Objective</h3> : Validate a modified Utah RVF (mRVF) Risk score on a larger U.S. geographical scale. <h3>Methods</h3> : INTERMACS data were requested for patients (age ≥19 years) undergoing isolated continuous flow LVAD implant between January 1, 2012 to through December 31, 2017. A modified (mRVF) score was computed without "history of hypertension" due to omission of variable on INTERMACS database. We excluded cases that were missing any of the 7-factor predictor variables. Primary outcome was early RVF, defined as the need for RVAD (BiVAD) support. Secondary outcomes were 90-day all-cause mortality. Multivariable logistic regression analysis was used to validate the predictive model. <h3>Results</h3> : The study included 12550 (59.9%) INTERMACS cases that had all variables of the novel predictive model available. Of those cases, incidence of RVF was 3.37% with a median [25^th, 75^th] predictive mRVF score of 5 [3.38, 6.62]. In the patient population who did not have RVF post-LVAD implantation, their median [25^th, 75^th] predictive mRVF score was 4 [2.22, 5.78]. The novel predictive model had modest accuracy for predicting the risk of RVF post-LVAD implantation (AUC 0.678). The modified Utah RVF score was poorly predictive of 90-day mortality (area under the ROC curve 0.571). <h3>Conclusion</h3> : The modified Utah RVF Risk score offered modest accuracy when applied to a large real world LVAD registry. Further studies examining the impact of unanticipated intra-operative and post-operative events on risk stratification are necessary.

Abstract 15771: Recovery of Left Ventricular Function on Lvad Support is Associated With Improved Outcomes

Introduction: For select patients, support with a left ventricular assist device (LVAD) can be used as a bridge to recovery of ventricular function. While rates of explant are low, improvement of left ventricular ejection fraction (LVEF) on LVAD support is more common. Hypothesis: We hypothesize that LVAD recipients whose LVEF recovers have better outcomes compared to those whose LVEF does not recover. Methods: Data from INTERMACS was merged with the Duke Echo Lab Database for all patients who underwent LVAD implantation between March 2006 and June 2017 at Duke Hospital. 286 patients with a continuous flow LVAD also had serial echocardiograms performed. Based on our prior work in HFrEF, we defined recovery as an LVEF &gt;= 40%. INTERMACS data and manual chart review was used to identify hospitalizations for recurrent heart failure (HF), gastrointestinal bleeding (GIB), and pump thrombosis along with mortality. Results: 9.8% of patients had an LVEF &gt;= 40% within 2 years of implant. LVEF recovery was associated with female gender [53.6% vs 21.7%. p &lt; 0.001], lower serum Cr prior to implant [1.30 mg/dL (IQR 1.00-1.60) vs 1.40 mg/dL (IQR 1.10-1.90), p = 0.05], and absence of an ICD [53.6% vs 79.8%, p = 0.002]. Within 1-year of the follow-up echo with the highest reported LVEF, patients whose LVEF was &gt;= 40% had a reduced incidence for a composite of hospitalizations for HF, GIB, pump thrombosis, and mortality [12.8% vs 39.8%, p = 0.013] as well as reduced rate of recurrent composite endpoint events [4.7 vs. 49 events per 100 person years, p = 0.020]. Of the individual endpoints, subjects with an LVEF &gt;= 40% had a lower, but non-significant, incidence of hospitalizations for HF [0% vs 10%, p = 0.122] and GIB [4.8% vs 18.2% p = 0.081]. Importantly, those who went on to recover their LVEF did not have a lower rate of recurrent composite endpoint events between the time of implant and time to follow-up echo (88 events per 100 person years in the recovery group vs 53 per 100 person years in the non-recovery group, p =0.052), suggesting that recovery of LVEF is temporally related to the improvement in outcomes. Conclusions: Recovery of LVEF on LVAD support in not uncommon. Strategies to facilitate recovery have the potential to improve outcomes of LVAD recipients beyond pump explant.

Predicting Post LVAD Acute Severe Right Heart Failure Using Bayesian Analysis

Purpose Right heart failure (RHF) continues to be a major adverse event post LVAD implantation. The risk of developing RHF post implant is multifactorial and difficult to predict. We have previously published Bayesian models to predict early and late RHF. In this analysis, we focused on pre-operative variables that predicted severe acute RHF as defined in INTERMACS as 'need for RVAD following LVAD or death during the VAD implants hospitalization with RHF as the primary cause'. Methods We used INTERMACS data from continuous flow LVADs from 2012-2016 (n=2,661). The model using pre-implant variables was made using Tree Augmented Naïve Bayes Analysis (GeNie) and predictors ranked by their diagnostic value. The impact of these variables on each other as well as the final outcome (RHF) were noted. Results Post LVAD implantation, severe acute RHF was noted in 126 patients (4.7%). Vast majority of patients in INTERMACS were males (80%) with over 50% patients implanted as destination therapy. 34 pre-implant variables, including both modifiable and non-modifiable, were found to impact the risk of post-implant acute severe RHF. (Figure 1) Key factors included INTERMACS Profile, major clinical event pre-implant (ECMO, IABP, mechanical ventilation), inotropes, platelets, albumin and AST. The performance assessment using areas under the curve ROC was 0.79 for the derived model. Conclusion By using a Bayesian approach, we explored pre-implant factors that were predictive of post-operative acute severe RHF. This would allow further risk stratification for patients undergoing LVAD implantation. Right heart failure (RHF) continues to be a major adverse event post LVAD implantation. The risk of developing RHF post implant is multifactorial and difficult to predict. We have previously published Bayesian models to predict early and late RHF. In this analysis, we focused on pre-operative variables that predicted severe acute RHF as defined in INTERMACS as 'need for RVAD following LVAD or death during the VAD implants hospitalization with RHF as the primary cause'. We used INTERMACS data from continuous flow LVADs from 2012-2016 (n=2,661). The model using pre-implant variables was made using Tree Augmented Naïve Bayes Analysis (GeNie) and predictors ranked by their diagnostic value. The impact of these variables on each other as well as the final outcome (RHF) were noted. Post LVAD implantation, severe acute RHF was noted in 126 patients (4.7%). Vast majority of patients in INTERMACS were males (80%) with over 50% patients implanted as destination therapy. 34 pre-implant variables, including both modifiable and non-modifiable, were found to impact the risk of post-implant acute severe RHF. (Figure 1) Key factors included INTERMACS Profile, major clinical event pre-implant (ECMO, IABP, mechanical ventilation), inotropes, platelets, albumin and AST. The performance assessment using areas under the curve ROC was 0.79 for the derived model. By using a Bayesian approach, we explored pre-implant factors that were predictive of post-operative acute severe RHF. This would allow further risk stratification for patients undergoing LVAD implantation.

Showering with the Driveline Exit Site Exposed to Well Water Does Not Increase Rates of Driveline Exit Site Infection

Purpose Determine the effects of showering on rates of driveline exit site infections (DLESI). Methods Prospective analysis of all durable left ventricular assist device (LVAD) patient implants at a single academic center from June 2011 to June 2018. All patients who were provided instructions for showering were included in this analysis (n=164). Exclusion criteria included patients that expired before 90 days of support and those not discharged prior to transplant (n=30). Patients were instructed to shower without a dressing covering their DLES, to use antibacterial soap and then immediately apply a sterile dressing using a standard kit of supplies. Infections were diagnosed with positive exit site cultures. Self-reported household water sources, well water vs municipal water, were used to compare the rates of DLESI. Our overall DLESI infection rates were compared to STS INTERMACS data. SPSS was used to analyze statistics. Results Overall at our center, 77% of the patients were free of DLESI. Demographics, duration of support, indication for LVAD and payer source for the well water and municipal water groups were similar except the well water group was older (p=0.003). Freedom from DLESI was 83% (n=38/46) & 74% (n=65/88) of patients who showered with well & municipal water respectively (p=0.254). Kaplan Meier curves are similar (p=0.151). The mean time to DLESI is 90 days and 87 days in the well and municipal water groups respectively (p=0.153). The late phase infection rate (after the first three months) per 100 pt months at our center is 1.16 compared to STS INTERMACS at 1.38. Conclusion At a single center, LVAD patients showering with different water sources have comparable rates of freedom from DLESIs. LVAD patients who follow a DLES management protocol that includes showering have comparable rates of DLESI when compared to STS INTERMACS data.

1540. Left Ventricular Assist Device Driveline Infections: Relapsed Infections and Minimum Inhibitory Concentration Changes

BackgroundTreatment of left ventricular assist device (LVAD) driveline infections (DLIs) pose difficulties given the permanent nature of the LVAD. Few studies have examined the minimum inhibitory concentration (MIC) changes over time or resistance patterns of implicated pathogens causing recurrent infections.MethodsThis retrospective descriptive epidemiology study identified patients with DLIs in the Vanderbilt LVAD registry or INTERMACS data from January 2013 to August 2017. Driveline infections met International Society for Heart and Lung Transplantation definitions in addition to positive driveline drainage, blood, or sternal wound culture. Relapse included a DLI with an organism associated with previous DLI in the preceding year and similar MICs or new resistance to an antibiotic that was utilized. The LVAD registry and chart review were utilized to collect data. Patients were followed until transplant, death, or August 1, 2017.ResultsA total of 330 patients underwent LVAD implantation. Thirty (9%) met criteria for DLI. Median duration of follow-up was 26 months (IQR 16, 39). There were 74 courses of infection, 40 new infections, and 34 relapsed infections. Median time to first DLI was 171 days (IQR 83, 403). Most common organisms in new DLIs were S. aureus (MRSA 11, MSSA 10), diptheroids (6), coagulase-negative staphylococci (6), and P. aeruginosa (5). S. aureus was the most common pathogen in patients with DLI associated bacteremia (n = 16) as well as relapsed infection (n = 11). There were 42 MIC changes in nine patients with relapsed infections from S. aureus, P. aeruginosa, and mycobacterium. Median time to first MIC change was 56 days (IQR 36, 88) and type of MIC change was an increase in five cases, decrease in two cases, and both increase and decrease in two cases. Time to first relapse from initial infection was longer in those who received suppression, 60 days vs. 83 days, P = 0.047.ConclusionFew patients had DLIs, but relapsed infections were more common with S. aureus and P. aeruginosa. MIC changes were quite variable and may not be the major contributor to relapsed infection.Disclosures All authors: No reported disclosures.

(42) - Long-Term Outcomes in Continous-Flow Left Ventricular Assist Device Patients with and without ICD

Methods: Prospectively collected STS and INTERMACS data on all LVADs implanted at our institution from Dec 2008 to Jan 2014 was analyzed by dividing patients into two groups, BMI ≥ 35 and BMI < 35. Results: Of the 157 LVADs implanted during the study period, 121 (77%) were in patients with BMI < 35 and 36 (23%) were in patients with BMI ≥ 35. The BMI ≥ 35 group were younger (47 ± 12.7 yrs. vs. 55 ± 13.4 yrs.) and had more female patients (44% vs. 24%), all P values < 0.05. Preoperative morbidities, INTERMACS class, LVAD type and implantation designation were comparable between the two groups (Table). However, a prior history of myocardial infarction and arrhythmia were significantly higher in BMI < 35 group and right heart catheterization pressures were higher in BMI ≥ 35 group, P = < 0.05. Postoperative and device specific complications were higher in BMI ≥ 35 group, but only bleeding requiring reoperation (42% vs. 22% p = 0.03) and frequency of driveline infections were statistically significant (Table). Operative mortality was higher in BMI ≥ 35 group but was not statistically significant (19.5% vs. 11% p = 0.25). At a mean follow-up of 27.3 ± 20.5 months, fewer patients with a BMI ≥ 35 received a HTx (19% vs. 33% p = 0.15) and they had longer wait times (619 ± 372 vs.403 ± 342 days, p= 0.1). After HTx, patients with a BMI ≥ 35 had significantly higher post operative bleeding requiring reoperation (57% vs. 15% p= 0.03) and a longer hospital stay (45±43 days vs. 27 ± 21 days). HTx survival and graft survival was comparable in both groups. Conclusion: Morbidly obese patients requiring LVAD support encounter higher postoperative complications and wait longer for heart transplantation, however HTx and graft survival is comparable.

An International Survey To Assess Referral Thresholds for Destination Therapy in Non-Inotrope Dependent Patients: Results of the CONSENSUS-DT Study

Purpose Criteria for the identification of “at risk” patients with advanced heart failure (HF) have been previously suggested. However, identification of appropriate non inotrope dependent (NID) patients for Destination Therapy (DT) remains a challenge and referral rates are low. We surveyed expert provider opinion about patient selection for early DT. Methods and Materials An internet-based survey was sent to international HF providers to investigate opinions about clinical indicators for referral of NID patients for DT. Subjects were identified from membership lists of HF professional organizations. Respondents were asked to score their level of agreement with 10 clinical markers describing at risk NID patients ( Figure 1 ). Results 231 respondents who had been in a position to recommend LVAD therapy for 7.6±5.8 years identified themselves as HF providers: 41% HF cardiologists, 27% HF nurse practitioners, 21% cardiothoracic surgeons, 9% LVAD coordinators and 2% general cardiologists. Responses, scored on a 5 point Likert scale, are shown in Fig 1 . More than two thirds of respondents strongly agreed or agreed with 7 of the 10 items. Similar consensus was not seen for 6 min walk and BNP. When asked which criteria would have to be present for LVAD referral only item 1 reached >50% affirmation amongst respondents. No criteria reached >50% response as being sufficient alone for DT. Conclusions Surprisingly, we found that expert providers agreed upon most surveyed clinical parameters to identify at risk HF patients. Yet, INTERMACS data shows very low referral rates of NID patients for DT. Ongoing multi-center trials may close this gap.

Improved Outcomes in Left Ventricular Assist Device Therapy for Patients between 65 and 70 Years

Purpose Left ventricular assist device therapy (LVAD) is effective management for advanced heart failure. INTERMACS data suggests increased early mortality in older recipients, but similar intermediate-term outcome among early survivors. We analyzed results of LVAD therapy in pt between 65 and 70 years and compared their outcomes with younger pt from our cohort. Methods and Materials Between 2004 and 2011, 295 pt Results Postop RV failure (OA: 4% vs. YA: 5%, p=1.00), dialysis (OA: 2% vs. YA: 9%; p=0.09), hospital LOS (p=0.68), and 90-day mortality (OA: 6% vs. YA: 8%; p=0.77) were similar. Median duration of support was 14 and 11 months for the OA and YA groups, respectively (p=0.27). Late pump-related intravascular hemolysis (p=0.33), driveline infection (p=0.78), and heart failure readmissions (p=0.23) were similar, while the OA group had more late neurological events (12% vs. 4%; p=0.04). Actuarial survival for the OA group was 81 +/- 12%, 72 +/- 14% and 66 +/- 17% at 12, 18 and 24 months, respectively. This was statistically similar to 85 +/- 5%, 82 +/- 6% and 76 +/- 8% survival at similar intervals for the YA cohort (p= 0.42). Conclusions Outcomes of LVAD therapy for pt between 65 and 70 years are similar to those of younger counterparts. While older pt have increased risk of late neurological events, late survival and other non-fatal morbidity are similar to younger pt. With demonstrated outcome improvement of LVAD therapy and paucity of donor organs for all pt, it is reasonable to revisit organ allocation strategies for older recipients.

Impact of Early Adverse Events (AE’s) on Later Mortality in Patients Supported by Continuous-Flow Ventricular Assist Devices (CF-VAD)

Purpose INTERMACS data show that select AE’s lead to higher mortality during VAD support. However, the extent to which these AE’s experienced early (within 60 days) after VAD implant effect long-term survival after this time period is not known for those with a CF-VAD. Methods and Materials We analyzed prospectively collected data from a single-center database on adults implanted with a CF-VAD (2004-2011) who survived beyond 60 days. 14 different classes of AE’s as defined by INTERMACS were assessed during the first 60 days post-implant. For each AE, 1-year survival was compared for those with and without the AE using Kaplan-Meier analysis. Any class of AEs at least marginally associated (p Results In 91 patients, the mean age was 56 years, 81% were male, 82% were white, and 65% were bridge to transplant. The occurrence of right ventricular failure (RVF), renal, neurologic, and respiratory events in the first 60 days of support were associated with significantly increased mortality at 12 months (p figure 1 ] Conclusions Of the 14 major AE’s experienced early after CF-VAD implantation, only RVF, renal, neurologic, and respiratory events had significant long-term effects on mortality that persisted beyond the initial event. The only early AE that was an independent predictor of poor late outcome was RVF.

Long-term biventricular support with rotary blood pumps: prospects and pitfalls

Congestive heart failure affects 23 million people worldwide, including 7.5 million in North America (670 000 new cases per year) and 7 million in Europe [1]. Since 10% of those over 65 suffer from systolic left ventricular dysfunction, the number of patients with heart failure will double over the next 25 years. At any time, 10% of the heart failure cohort are categorized Stage D with advanced structural heart disease and symptoms at rest, despite detailed medical and cardiac resynchronization therapy. Twenty percent of these are younger than 65 years of age (around 140 000 in both the USA and Europe). The prognosis of Stage D heart failure is grim. In the REMATCH study, only 8% of the medically treated patients were alive at 2 years and continued to suffer in the interim [2]. A recurrent sentiment expressed by the Stage D patients is that they would sacrifice some duration of survival for a period of symptomatic relief [3]. Cardiac transplantation provides this for a highly selected tiny minority [4]. In contrast, cardiac resynchronization therapy is disappointing. A meta-analysis of 14 trials that randomized resynchronization against medical treatment showed only 59% of the NYHA IV device patients to have borderline symptomatic improvement to NYHA III with no survival benefit [5]. Boyle et al. [6] compared functional outcomes for NYHA IV patients after resynchronization or implantation of a left ventricular assist device (LVAD). At 6 months, resynchronized patients achieved only an additional 6 m in the 6 min walk test (insufficient to affect their daily living) and remained NYHA III or IV. In contrast, LVAD patients improved by 341 m, achieving NYHA I or II status. The study could not be randomized because 90% of the LVAD candidates were bed-bound on intravenous inotropes and could not walk beforehand. Long-term mechanical circulatory support is an increasingly realistic ‘off the shelf’ alternative to cardiac transplantation and an effective solution for those rendered ineligible through common heart failure comorbidities [7–9]. The treatment aims are uncontroversial. The first is to provide symptomatic relief and the second to achieve at least 5-year survival with a good quality of life. The third aim is cost containment by reducing the number of hospital admissions to palliate intolerable levels of breathlessness and fatigue. Hospital events account for $20.9 billion of an overall $35 billion heart failure budget in the USA. The rationale for a blood pump deployment is clear. The failing heart beats more than 120 000 times per day pumping around 7000 l of blood against an increasing afterload. As the heart dilates, the ventricular wall tension, myocardial energy and oxygen consumption increase, while sub-endocardial blood flow decreases. The LVAD unloads the failing ventricle resulting in beneficial structural and functional changes in the diseased myocardium [10–12]. Systemic blood flow returns to physiological levels to preserve the coronary and vital organ perfusion [13]. When native heart contractility and segmental wall motion improve, there is less propensity for intraventricular thrombus formation and thromboembolism. In September 2011, the Society of Thoracic Surgeons and the Food and Drugs Administration (USA) assembled a ‘Think Tank’ to focus on the need for long-term mechanical circulatory support in the USA. The group acknowledged the increasing safety profile and reliability of contemporary implantable rotary blood pumps together with the potential for stem cells to broaden the horizons of heart failure management. A need for 40 000 long-term LVADs per year was predicted, which would account for 50% of the heart failure annual budget. Between 15 and 30% of patients require biventricular support and there is no satisfactory long-term mechanical solution for these patients. The pulsatile total artificial hearts have numerous restrictions which confine their use as a bridge to transplantation [14]. INTERMACS data indicate that all total artificial heart patients are transplanted or dead within 6 months [15].