Bleeding in Continuous Flow Left Ventricular Assist Device Recipients an Acquired Vasculopathy

October 18, 2022 Post a Comment

WHAT IS NEW?

A patient's individual risk of bleeding varies widely as a result of multiple and incompletely elucidated mechanisms.
A patient's individual risk of gastrointestinal bleeding during left ventricular assist device support can be accurately predicted a priori by means of the Utah bleeding risk score predictive model.

WHAT ARE THE CLINICAL IMPLICATIONS?

Identifying an individual's bleeding risk profile may assist clinicians and researchers in designing tailored risk-based strategies aimed at reducing the burden posed by gastrointestinal bleeding to left ventricular assist device patients.
A better understanding of the individual's bleeding risk profile may facilitate the informed consent process and lead to greater patient satisfaction.

Left ventricular assist device (LVAD) support is a standard therapeutic option for select patients with advanced heart failure, either as a bridge to heart transplantation or as destination therapy.¹ The introduction of newer generation continuous-flow LVADs (CF-LVADs) has resulted in improved survival,^2–4 which has now reached over 80% at 1 year after device implant.¹ Despite such survival benefit and nearly 50% reduction in complications provided by CF-LVADs in comparison with the first-generation pulsatile-flow LVADs, bleeding remains the most frequent adverse event after CF-LVAD implant, with an incidence of gastrointestinal bleeding (GIB) of 20% to 40% (7.8 events/100 patient-months).^1,5–8 GIB can negatively impact the efficacy of CF-LVAD support by increasing healthcare cost related to repeated hospitalizations and a significant number of diagnostic and therapeutic procedures.^7,9,10 At the individual level, GIB adversely affects patient's quality of life and may lead to an increased mortality risk.¹¹ On average, a CF-LVAD patient receives 2 to 4 units of packed red blood cells per GIB-related admission,¹² which can expose them to adverse effects,¹³ including allosensitization (particularly important in bridge-to-transplant patients).^14,15 In addition, antiplatelet agents and anticoagulation are commonly suspended in the setting of GIB, subjecting patients to higher risk of pump thrombosis and thromboembolic events.¹⁶

The risk of bleeding in patients supported with CF-LVADs varies widely, and although it seems to be influenced by several pre- and postoperative factors (eg, older age, acquired von Willebrand syndrome, reduced pulsatility), its quantification remains challenging.^17–24 Risk stratification schemes that accurately and reliably predict GIB in candidates for CF-LVADs could have important implications for candidate selection, informed consent, and the development of postimplant therapeutic strategies. The aim of this study was to construct and validate a predictive model of GIB based on preoperative characteristics of CF-LVAD patients.

Methods

The data and analytic methods that support the findings of this study are available from the corresponding author on reasonable request.

Study Population

The Utah Transplantation Affiliated Hospitals program database was used to identify patients with stage D heart failure who received an implant with a CF-LVAD for either bridge-to-transplant or destination therapy between January 2004 and August 2017. All patients met the medical policy guideline of New York Heart Association class III/IV heart failure. We excluded patients who had follow-up of <8 days, reimplantations (LVAD exchange), and patients with incomplete data. The study protocol was approved by the institutional review board.

Study Data

Preoperative clinical, laboratory, and cardiopulmonary hemodynamic data were abstracted from the electronic medical records. Baseline clinical data included patient's demographics, body mass index, heart failure cause, history of hypertension, diabetes mellitus, chronic kidney disease, liver disease, tobacco use, alcohol abuse, dyslipidemia, coronary artery disease, peripheral vascular disease, previous bleeding (of any source), New York Heart Association class, use of anticoagulation or antiplatelet agents, need for inotropic support, intra-aortic balloon pump or a temporary LVAD, and Interagency Registry of Mechanically Assisted Circulatory Support profile. Laboratory measures were obtained ≤48 hours before LVAD implant and included platelet count, serum hemoglobin, hematocrit, sodium, creatinine, glucose, total bilirubin, alanine aminotransferase, aspartate aminotransferase, albumin, total protein, BNP (brain natriuretic peptide), and the international normalized ratio. Hemodynamic data were obtained within a week before LVAD implant and included right atrial pressure, pulmonary arterial pressure, pulmonary capillary wedge pressure, cardiac index, pulmonary vascular resistance, systemic blood pressure, pulmonary arterial pulsatility index, and right ventricular (RV) stroke work index. Echocardiographic data were obtained within a week before LVAD implant. Echocardiograms were performed according to American Society of Echocardiography guidelines and were reviewed by an experienced echocardiographer.²⁵ Left ventricular ejection fraction and RV function were visually estimated. RV function was classified as either normal or mildly, moderately or severely decreased. The aforementioned variables were also obtained at 1, 3, 6, 12, 24, and 36 months after LVAD implantation.

End Points

The primary end point was confirmed GIB occurring during the first 3 years after CF-LVAD implant. GIB was defined using the standard Interagency Registry of Mechanically Assisted Circulatory Support definition, which includes episodes of suspected bleeding resulting in one of the following: death, surgical intervention, hospitalization, or transfusion of packed red blood cell. We additionally required clinical evidence of bleeding (guaiac-positive stool, melena, hematochezia, hematemesis, or the presence of blood in the gastrointestinal tract on endoscopic evaluation) and a documented decrease in hemoglobin ≥2 g/dL. Observation started at the time of LVAD implant and ended at 3 years of LVAD support or with patient censoring. Patients were censored for death, transplant, and study termination. Patients who experienced a bleeding event related to increased anticoagulation and use of thrombolytics were censored.

Statistical Analysis

Continuous variables were summarized as median (interquartile range). Continuous data were compared by means of the Mann-Whitney U test. Categorical variables were summarized as frequencies and percentages and were compared by the Pearson χ² test or by the Fisher exact test when appropriate. Univariable predictors of GIB were identified using Cox proportional hazards regression analysis that included preimplant clinical, laboratory, echocardiographic, and cardiopulmonary hemodynamic variables previously described in the Study Data section. For the development of the risk score, continuous variables were dichotomized using the highest C statistic to define the optimal cut point. A stepwise multivariable Cox proportional hazards regression model was used to determine the independent effect of multiple risk factors on the hazard of the primary end point. The proportionality assumption was tested using Schöenfeld residuals, and it was found to be satisfied by this test.²⁶ Variables significant at the P<0.10 level in unadjusted analyses were considered for inclusion; only variables significant at the P<0.05 level based on the likelihood ratio test were retained in the final model. Correlated variables were not entered simultaneously, and no more than 7 variables were entered at 1 time to avoid model overfitting. Multivariable time-to-event analysis was performed using Cox proportional hazards regression models to develop a nomogram using weighted estimators corresponding to each covariate derived from Cox regression coefficients and estimates of variance.^27,28 We also built a Cox regression model including pre- and postimplant variables using time-varying analysis to determine factors associated with GIB. Hazard ratios and 95% CIs were generated for both univariable and multivariable analyses as measures of strength of association and precision, respectively. Model discrimination was evaluated with C statistic.

Derivation and Validation of the Bleeding Risk Score

The study cohort was used to develop a model for calculation of a patient-specific risk estimate for predicting 3-year incidence of GIB among CF-LVAD candidates. A risk prediction score was developed by assigning the risk factors identified by multivariable analysis weighted points proportional to the β regression coefficient values (rounded to the nearest integer). Using previously described methods,²⁹ a risk score was then calculated for each patient, and the population was divided into 3 categories: low, intermediate, and high risk for GIB. The expected probability of GIB associated with each score was also estimated. Validation of the prediction model was assessed internally with a bootstrapping procedure for a realistic estimate of the performance of the model in similar future patients. We repeated the entire modeling process, including variable selection and optimum penalty factor search, in 200 samples drawn with replacement from the original sample. Subsequently, we developed a prediction model of GIB. Survival free from GIB was calculated using Kaplan-Meier survival analysis and compared between groups with the log-rank test. All analyses were performed using STATA software, version 15 (StataCorp LP, College Station, TX).

Results

Patient Characteristics

There were 411 durable LVADs implanted during the study period. Of these, 39 patients were excluded for incomplete data, 15 for being reimplants, and 6 for having length of support <8 days. A total of 351 patients met the inclusion criteria and comprised our study group. Baseline characteristics of the study group are summarized in Table 1. The median age was 59 years, and 82% of patients were males. Approximately half of the patients had ischemic cardiomyopathy (45%) and were implanted as a bridge to heart transplant (51%). Most patients (72%) presented with Interagency Registry of Mechanically Assisted Circulatory Support profile ≤3 (high acuity). Forty-three (12%) patients received a durable LVAD with an intra-aortic balloon pump, and 53 (15%) patients received durable LVAD with a temporary mechanical circulatory support device. Axial-flow LVADs were implanted in 240 (68%) patients and centrifugal-flow LVADs were implanted in 111 (32%) patients. Patients were followed for a median time of 196 days (interquartile range, 64–529).

**Table 1.** Baseline Characteristics of LVAD Recipients
Clinical	(n=351)
Age, y	59 [48–67]
Male sex	288 [82]
White race	289 [82]
Body mass index, kg/m²	27 [24–31]
Hypertension	157 [45]
Diabetes mellitus	121 [34]
Dyslipidemia	222 [63]
CKD	126 [36]
CAD	181 [52]
Smoking	178 [51]
Alcohol use	105 [30]
History of previous bleed	30 [9]
Heart failure cause
Nonischemic	192 [55]
Ischemic	159 [45]
NYHA class
III	107 [30]
IV	243 [69]
LVAD indication
Bridge to transplant	178 [51]
Bridge to candidacy	54 [15]
Destination therapy	118 [34]
LVAD type
Axial flow	239 [68]
Centrifugal flow	111 [32]
INTERMACS profile
1	43 [12]
2	54 [15]
3	157 [45]
4	91 [26]
5	6 [2]
Hemodynamic
Heart rate, beats per minute	89 [75–100]
MAP, mmHg	76 [70–83]
RAP, mm Hg	11 [7–16]
MPAP, mm Hg	36 [29–43]
PCWP, mm Hg	24 [18–29]
PVR, dynes s⁻¹ cm⁻⁵	253 [154–393]
SVR, dynes s⁻¹ cm⁻⁵	1513 [1198–1874]
PAPi	1.5 [0.97–2.5]
RVSWI, gm·m⁻¹ m⁻² per beat	13 [9–17]
Cardiac index, L min⁻¹·m⁻²	1.7 [1.4–2.1]
Echocardiographic
LVEDD, mm	66 [60–72]
LVEF, %	19 [15–25]
RV dysfunction
None	72 [21]
Mild	119 [34]
Moderate	102 [29]
Severe	52 [15]
Laboratory
Hematocrit, %	36 [32–41]
Platelet count, 10⁹/L	198 [158–252]
Sodium, mEq/L	135 [131–138]
Creatinine, mg/dL	1.3 [1.0–1.6]
Total bilirubin, mg/dL	1.2 [0.7–1.8]
ALT, IU/L	29 [20–51]
AST, IU/L	33 [24–53]
Albumin, g/dL	3.6 [3.2–4]
INR	1.2 [1.1–1.4]
BNP, pg/mL	1104 [548–2053]
Glucose, mg/dL	113 [96–144]

GIB After LVAD Implantation

GIB occurred as a single event in 120 (34%) patients. The median time from LVAD implant to first GIB was 58 days (interquartile range, 24–172 days). Survival free of GIB was 63%, 58%, and 53% at 1, 2, and 3 years, respectively. The most common source of GIB was arteriovenous malformations (n=95, 79%), either confirmed (n=71) or presumed (n=24) (no source of bleeding was identified after extensive workup including upper and lower endoscopy, capsule study, and push enteroscopy). Other sources of GIB included upper GIB from duodenitis/duodenal ulcer and gastritis/gastric ulcer (n=12), lower GIB from diverticulosis (n=5), colitis (n=3), colonic polyps (n=3), and others (n=2). Of note, the incidence of GIB was associated with higher all-cause mortality during LVAD support, even after adjustment for patient's age (Figure I in the Data Supplement). Overall survival in this cohort was 81%, 67%, and 58% at 1, 2, and 3 years, respectively.

Risk Factors Associated With GIB

On univariable Cox regression analysis, patients who experienced GIB were older, more likely to have history of coronary artery disease, ischemic cardiomyopathy, diabetes mellitus, chronic kidney disease, previous bleeding (any source), lower heart rate, severe RV dysfunction, lower mean pulmonary arterial pressure, higher right atrial pressure, higher pulmonary vascular resistance, higher systemic vascular resistance, lower hematocrit, higher serum creatinine, and higher fasting glucose at time of LVAD implant (Table 2). On multivariable Cox regression analysis, age >54 y, coronary artery disease, chronic kidney disease, history of previous bleed, severe RV dysfunction, mean pulmonary arterial pressure <18 mm Hg, and fasting glucose >107 mg/dL were independently associated with the occurrence of GIB during LVAD support (Table 3, Figure II in the Data Supplement). There was no correlation between the variables included in the final model (Table I in the Data Supplement). The discrimination of the model for GIB was good, with a C statistic of 0.76 (95% CI, 0.71–0.80; P<0.0001).

**Table 2.** Univariable Hazard Ratio Estimates for the Risk of GI Bleed Among LVAD Recipients
Risk Factor	Hazard Ratio [95% CI]	P Value
Clinical
Age >54 y (vs ≤54 y)	1.98 [1.30–5.93]	<0.001
History of bleed (yes vs no)	4.22 [2.71–6.61]	<0.001
CAD (yes vs no)	1.54 [1.01–2.24]	<0.001
CKD (yes vs no)	2.56 [1.79–3.68]	<0.001
Male sex (vs female)	1.24 [0.78–2.01]	0.92
White race (yes vs no)	1.01 [0.82–1.25]	0.89
Body mass index <30 kg/m² (vs ≥30 kg/m²)	1.74 [0.92–2.15]	0.12
Hypertension (yes vs no)	1.10 [0.77–1.58]	0.59
Diabetes mellitus (yes vs no)	1.20 [0.83–1.74]	0.33
Dyslipidemia (yes vs no)	1.49 [1.01–2.21]	0.04
Smoking (yes vs no)	1.33 [0.92–1.90]	0.13
Alcohol use (yes vs no)	0.90 [0.60–1.35]	0.60
Ischemic cardiomyopathy (yes vs no)	1.50 [1.05–2.18]	0.02
NYHA class IV (vs III)	1.29 [0.86–1.94]	0.20
Destination therapy indication (yes vs no)	1.00 [0.90–1.35]	0.89
Axial-flow LVAD (vs centrifugal LVAD)	1.30 [0.90–2.03]	0.16
INTERMACS profiles 1–3 (vs 4–5)	1.04 [0.74–1.53]	0.92
Hemodynamic
Heart rate <71 beats per minute (vs ≥71 beats per minute)	1.58 [1.03–2.43]	0.04
MAP <68 mm Hg (vs ≥68 mm Hg)	1.43 [0.94–2.18]	0.09
MPAP <18 mm Hg (vs ≥18 mm Hg)	3.51 [1.89–6.55]	<0.001
RAP >15 mm Hg (vs ≤15 mm Hg)	1.06 [1.02–1.09]	0.001
PCWP <18 mm Hg (vs ≥18 mm Hg)	1.38 [0.94–2.04]	0.10
PVR >517 dynes/cm² (vs ≤517 dynes/cm²)	1.54 [0.97–2.45]	0.07
SVR >2080 dynes/cm² (vs ≤2080 dynes/cm²)	1.56 [1.11–2.60]	0.02
PAPi <1.05 (vs ≥1.05)	1.10 [0.70–1.64]	0.60
RVSWI <13 (vs ≥13)	1.24 [0.25–1.74]	0.37
Cardiac index <1.5 L min⁻¹·m⁻² (vs ≥1.5 L min⁻¹·m⁻²)	1.04 [0.70–1.51]	0.93
Echocardiographic
LVEDD >67 mm (vs ≤67 mm)	0.72 [0.52–1.09]	0.13
LVEF <18% (vs ≥18%)	0.78 [0.55–1.17]	0.17
RV dysfunction
None	0.72 [0.44–1.06]	0.09
Mild	1.41 [0.82–2.44]	0.22
Moderate	1.03 [0.57–1.84]	0.93
Severe	3.38 [1.91–6.01]	<0.001
Severe RV dysfunction (vs none/moderate)	4.23 [2.71–6.64]	<0.001
Laboratory
Hematocrit <33% (vs ≥33%)	1.54 [1.07–2.22]	0.02
Creatinine >1.6 mg/dL (vs ≤1.6 mg/dL)	1.73 [1.20–2.51]	0.003
Glucose >107 mg/dL (vs ≤107 mg/dL)	2.07 [1.40–3.05]	<0.001
Platelet count <305 ×10⁹/L (vs ≥305×10⁹/L)	1.51 [0.83–2.73]	0.18
Sodium <132 mEq/L (vs ≥132 mEq/L)	0.83 [0.56–1.23]	0.35
Total bilirubin >0.8 mg/dL (vs ≤0.8 mg/dL)	1.21 [0.82–1.78]	0.34
ALT >29 IU/L (vs ≤29 IU/L)	0.78 [0.54–1.12]	0.17
AST >57 IU/L (vs ≤57 IU/L)	0.68 [0.42–1.13]	0.14
Albumin <3.6 g/dL (vs ≥3.6 g/dL)	1.22 [0.86–1.78]	0.27
INR >1.2 (vs ≤1.2)	1.33 [0.89–1.92]	0.15
BNP >1866 pg/mL (vs ≤1866 pg/mL)	0.74 [0.49–1.14]	0.17

**Table 3.** Multivariable Hazard Ratio Estimates for the Risk of GIB Among CF-LVAD Recipients
Risk Factor	Incidence Rate of GIB by Prognostic Factor Present vs Absent (Event per 100 Person-Years)	Hazard Ratio [95% CI]	P Value	β Regression Coefficient	Points
Age >54 y	50.7 vs 13.4	1.84 [1.01–3.35]	0.05	0.61	1
History of previous bleed	193.4 vs 30.5	3.24 [1.99–5.26]	<0.001	1.17	2
CAD	58.1 vs 19.6	1.78 [1.15–2.77]	0.01	0.58	1
CKD	70.0 vs 23.4	1.68 [1.14–2.48]	0.009	0.52	1
Severe RV dysfunction	131.1 vs 29.4	1.76 [1.11–2.80]	0.02	0.58	1
MPAP <18 mm Hg	116.2 vs 34.1	2.63 [1.37–5.06]	0.004	0.97	2
Glucose >107 mg/dL	52.4 vs 20.9	1.82 [1.21–2.75]	0.004	0.60	1

The Utah Bleeding Risk Score and GIB After LVAD Implant

On the basis of this model, a risk score we termed the Utah bleeding risk score (UBRS) was derived by assigning each of the 7 risk factors a number of points proportional to its regression coefficient (Table 3). A score was determined for each patient by calculating a sum of the points corresponding to his or her risk factors. The score ranged from 0 to 9, and the median was 2 (interquartile range, 1–3). GIB estimates were used to define 3 groups with significantly different risk profile: a low-risk group (0–1 points), an intermediate-risk group (2–4), and a high-risk group (5–9). The corresponding 3-year GIB rates were 4.8%, 39.8%, and 83.8% (Figure 1). The observed and expected GIB rates were similar for each score and increased with increasing scores (Figure 2). Survival free from GIB at 12 months post-implant was 93% in the low-risk, 62% in the intermediate-risk, and 14% in the high-risk group (Figure 3). The performance of UBRS was good, with an area under the receiver operating characteristics curve of 0.83 (95% CI, 0.79–0.88; Figure 4). Internal validation of our predictive model with a bootstrapping procedure yielded a realistic estimate of the performance of the model in similar future patients (C statistic of 0.74; 95% CI, 0.69–0.79). In addition to the UBRS, a nomogram was developed to provide a graphical predictive tool to predict survival free from GIB (Figure 5). Of note, the rate of pump thrombosis in the UBRS high-risk group was not higher than in other UBRS risk groups (4.4 versus 7.1 event per 100 patient-years, P=0.28)

**Figure 1.** **Three-year rate of gastrointestinal bleeding (GIB) stratified by Utah bleeding risk score risk groups.**

**Figure 2.** **Expected and observed rates of** **gastrointestinal bleeding**(**GIB) according to** **Utah bleeding risk score**(**UBRS) score.**

**Figure 3.** **Survival free from gastrointestinal bleeding (GIB) stratified by Utah bleeding risk score risk groups.**

**Figure 4.** **Performance of Utah bleeding risk score on discriminating gastrointestinal bleeding (GIB).** The area under the receiver operating characteristic curve demonstrates good discrimination for predicting GIB at 3 y after left ventricular assist device (LVAD) implantation.

**Figure 5.** **Nomogram for predicting survival free of gastrointestinal bleeding (GIB) at 3 years after left ventricular assist device (LVAD) implantation.** In this predictive tool, points are tallied for each of the patient's predictor variables. It is used by first locating the patient's position on each predictor variable. Each position has corresponding prognostic points located on the points scale. Then, the points for each variable are summed to arrive at a total points value (plotted on the total points axis at the bottom). A vertical line is then drawn from the total points axis up to the Probability of Survival Free of GIB at 3 Years to indicate the patient's probability of GIB after LVAD implant.

Postimplant Risk Factors of GIB

Clinical, echocardiographic, and hemodynamic parameters can change over time, and we therefore evaluated these parameters using time-varying Cox regression analyses with variables listed in Table 1. In addition to the variables included in the UBRS model (Table 3), we identified postimplant variables that were independently associated with GIB. Full and intermittent aortic valve opening and higher dose of angiotensin-converting enzyme inhibitor were associated with lower risk of GIB. On the other hand, higher aspirin dose (325 versus 81 mg) was associated with higher risk of GIB (Table 4). The type of pump was not associated with GIB.

**Table 4.** Multivariable Pre- and Postimplant Predictors of GIB Among CF-LVAD Recipients
Risk Factor	Hazard Ratio [95% CI]	P Value
AV opening*
Intermittent vs no opening	0.06 [0.01–0.47]	0.007
Full vs no opening	0.48 [0.30–0.76]	0.002
RV dysfunction*
Mild dysfunction vs normal function	1.15 [0.54–2.43]	0.72
Moderate dysfunction vs normal function	1.12 [0.53–2.44]	0.77
Severe dysfunction vs normal function	2.00 [0.92–4.34]	0.08
Aspirin dose 325 vs 81 mg*	1.002 [1.001–1.004]	<0.001
Higher ACE inhibitor dose, mg*	0.93 [0.86–0.99]	0.04
CAD	2.39 [1.56–3.65]	<0.001
CKD	1.80 [1.23–2.64]	0.003
History of previous bleed	2.38 [1.48–3.83]	<0.001

Discussion

Bleeding is currently the most common adverse event after CF-LVAD implant, with bleeding from the gastrointestinal tract being the most common source. At a healthcare level, LVAD-associated GIB contributes to an increased overall healthcare utilization.^30,31 At the patient level, GIB is associated with greater morbidity and increases patients risk for other adverse events, including thrombotic complications, and, as our findings suggest, all-cause mortality.^{5,6,11,16,19,32,33} Although previous studies have identified clinical variables associated with increased postimplant risk of GIB, it remains clinically difficult to assimilate the different variables and practically use them to quantify a patient's individual risk of developing GIB after LVAD implantation.^{17,18,21–24,34}

The main finding of our study is that although GIB is a frequent adverse event after LVAD implantation, the individual risk of GIB varies widely and can be determined a priori by using a predictive risk model incorporating the patient's preimplantation characteristics (Figure 6).

**Figure 6.** **This study evaluated pre implant clinical factors associated with gastrointestinal bleeding (GIB) after left ventricular assist device (LVAD) implantation.** A predictive risk model was developed based on such risk factors, leading to the creation of the Utah bleeding risk score (UBRS) denoting tiered risk of GIB at 3 y after LVAD implantation.

We identified characteristics such as older age and history of previous bleeding to be associated with an increased risk of GIB, findings that have been previously reported.^5,8,17,22,23 Further, we expanded on previous reports from our group and others showing preimplant RV dysfunction as a risk factor for GIB after LVAD implantation.^22,24 Although the exact mechanism underlying the association between RV dysfunction and GIB is largely unknown, it has been hypothesized to be related to hepatic congestion leading to coagulopathy and reduced pulsatility.^9,22,24 The association of GIB and lower mean pulmonary arterial pressure, a factor that may reflect poor RV function, further supports this concept. We also found poor renal function to be a risk factor for postimplant GIB, which could be related to platelet dysfunction and increased vascular ectasias commonly observed in patients with chronic kidney disease.^35–38 Hyperglycemia in critically ill patients has been associated with worse outcome and increased risk of GIB.^39–41

We believe the clinical implications of applying this predictive model to be important. Risk stratification of LVAD candidates beforehand can be used to guide patient selection and improve the informed consent process. Recent studies have illustrated inadequacies in disclosing the risk of GIB during informed consent process for LVAD candidates and discrepancies in the perception of prognosis, treatment options, and adverse events between physicians and patients in advanced heart failure patients.^42,43 Improvement in the quality of life after LVAD placement plays a pivotal role in patient's decision to proceed with the implant. Unfortunately, quality of life after LVAD implant can be adversely impacted by GIB.⁴⁴ Our hope is that direct illustration of a patient's individual risk of GIB can enhance informed consent and shared decision making and ultimately lead to better outcomes in advanced heart failure patients.

In addition, clinicians could implement personalized therapeutic strategies based on a patient's risk of GIB. For instance, patients at high risk of GIB could benefit from reduced antithrombotic protocols. The safety of an aspirin-free regimen has been evaluated in the European TRACE study (Study of Reduced Anti-Coagulation/Anti-Platelet Therapy in Patients With the HeartMate II), which included patients using vitamin K antagonists only, primarily because of physician preference and per standard of care of the implanting center. Compared with historical data of patients on standard dual-agent antithrombotic therapy, the incidence of bleeding was low (0.10 versus 0.65 events per patient-year) and the rates of ischemic stroke (0.02 versus 0.04 events per patient-year) or pump thrombosis (0.04 versus 0.03 events per patient-year) were comparable.⁴⁵ Another strategy that seems to be effective in reducing the rates of GIB without increasing the rates of pump thrombosis was recently tested in a pilot study of HeartMate 3 recipients who were managed with a low-intensity (international normalized ratio goal of 1.5–1.9) anticoagulation protocol.⁴⁶ Finally, personalized bridging strategies with either unfractionated heparin or low–molecular weight heparin to therapeutic warfarin could be tailored based on the patient's bleeding risk, as bleeding events not infrequently occur during the bridging process.^47,48

Postimplant factors can also have an impact on the risk of GIB. We identified reduced aortic valve opening, lower angiotensin-converting enzyme inhibitor dose, and higher aspirin dose to be associated with greater risk of GIB. Our group and others have shown that reduced pulsatility, as evaluated by device pulsatility index, aortic valve opening, or patient pulse pressure, is associated with an increase in rates of GIB, mainly related to arteriovenous malformations.^24,34,49 Inhibition of angiotensin II–induced angiogenesis by angiotensin-converting enzyme inhibitors or angiotensin receptor blockers has been shown to be associated with lower rates of GIB because of arteriovenous malformations.⁵⁰ Higher aspirin dose has been previously associated with higher risk of GIB.^45,51 The potential reduction in risk of GIB by modifiable postimplant factors highlights the importance of a reliable preimplant predictive risk model to identify high-risk patients such that postimplant therapeutic strategies can be tailored toward minimizing the risk of GIB.

We recognized that there are limitations in our study. This study was based on a retrospective analysis of our institutional registry and is subject to limitations related to the study design. Our predictive model was derived from a sizable and heterogeneous LVAD patient population implanted in our program. Although validation of our findings in an independent cohort would be desirable, internal validation of our predictive model using robust statistical methods supports its good performance. Further, our study allowed for a more detailed patient characterization and greater data completeness (data were 100% available for variables included in the predictive model) in comparison with large registry data. Finally, we dichotomized continuous variables in the development of the predictive model to simplify the scoring process for clinicians at the bedside, although this strategy has the potential of providing less refined information.

Conclusions

In summary, the risk of GIB varies widely among LVAD recipients. The UBRS is a simple predictive model based on preimplant clinical factors that can effectively risk stratify LVAD patients based on their probability of GIB. Such a predictive model could have important implications on patient selection, patient's informed consent, and personalization of therapeutic strategies.

Sources of Funding

This work was funded by the following grants: American Heart Association (AHA) Heart Failure Strategically Focused Research Network, 16SFRN29020000 (Dr Drakos); National Heart, Lung, and Blood Institute grant R1 HL135121-01 (Dr Drakos); National Heart, Lung, and Blood Institute grant R1 HL132067-01A1 (Dr Drakos); Nora Eccles Treadwell Foundation (Dr Drakos); and University of Utah Program in Personalized Health and National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number 1UL1TR02538 (Dr Wever-Pinzon). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

References

1. Kirklin JK, Pagani FD, Kormos RL, Stevenson LW, Blume ED, Myers SL, Miller MA, Baldwin JT, Young JB, Naftel DC. Eighth annual INTERMACS report: special focus on framing the impact of adverse events. J Heart Lung Transplant . 2017; 36:1080–1086. doi: 10.1016/j.healun.2017.07.005CrossrefMedlineGoogle Scholar
2. Krabatsch T, Netuka I, Schmitto JD, Zimpfer D, Garbade J, Rao V, Morshuis M, Beyersdorf F, Marasco S, Damme L, Pya Y. Heartmate 3 fully magnetically levitated left ventricular assist device for the treatment of advanced heart failure-1 year results from the CE mark trial. J Cardiothorac Surg . 2017; 12:23. doi: 10.1186/s13019-017-0587-3CrossrefMedlineGoogle Scholar
3. Rose EA, Moskowitz AJ, Packer M, Sollano JA, Williams DL, Tierney AR, Heitjan DF, Meier P, Ascheim DD, Levitan RG, Weinberg AD, Stevenson LW, Shapiro PA, Lazar RM, Watson JT, Goldstein DJ, Gelijns AC. The REMATCH trial: rationale, design, and end points. Randomized Evaluation of Mechanical Assistance for the Treatment of Congestive Heart Failure. Ann Thorac Surg . 1999; 67:723–730.CrossrefMedlineGoogle Scholar
4. Slaughter MS, Pagani FD, McGee EC, Birks EJ, Cotts WG, Gregoric I, Howard Frazier O, Icenogle T, Najjar SS, Boyce SW, Acker MA, John R, Hathaway DR, Najarian KB, Aaronson KD; HeartWare Bridge to Transplant ADVANCE Trial Investigators. HeartWare ventricular assist system for bridge to transplant: combined results of the bridge to transplant and continued access protocol trial. J Heart Lung Transplant . 2013; 32:675–683. doi: 10.1016/j.healun.2013.04.004CrossrefMedlineGoogle Scholar
5. Aggarwal A, Pant R, Kumar S, Sharma P, Gallagher C, Tatooles AJ, Pappas PS, Bhat G. Incidence and management of gastrointestinal bleeding with continuous flow assist devices. Ann Thorac Surg . 2012; 93:1534–1540. doi: 10.1016/j.athoracsur.2012.02.035CrossrefMedlineGoogle Scholar
6. Morgan JA, Paone G, Nemeh HW, Henry SE, Patel R, Vavra J, Williams CT, Lanfear DE, Tita C, Brewer RJ. Gastrointestinal bleeding with the HeartMate II left ventricular assist device. J Heart Lung Transplant . 2012; 31:715–718. doi: 10.1016/j.healun.2012.02.015CrossrefMedlineGoogle Scholar
7. Smedira NG, Hoercher KJ, Lima B, Mountis MM, Starling RC, Thuita L, Schmuhl DM, Blackstone EH. Unplanned hospital readmissions after HeartMate II implantation: frequency, risk factors, and impact on resource use and survival. JACC Heart Fail . 2013; 1:31–39. doi: 10.1016/j.jchf.2012.11.001CrossrefMedlineGoogle Scholar
8. Stern DR, Kazam J, Edwards P, Maybaum S, Bello RA, D'Alessandro DA, Goldstein DJ. Increased incidence of gastrointestinal bleeding following implantation of the HeartMate II LVAD. J Card Surg . 2010; 25:352–356. doi: 10.1111/j.1540-8191.2010.01025.xCrossrefMedlineGoogle Scholar
9. Demirozu ZT, Radovancevic R, Hochman LF, Gregoric ID, Letsou GV, Kar B, Bogaev RC, Frazier OH. Arteriovenous malformation and gastrointestinal bleeding in patients with the HeartMate II left ventricular assist device. J Heart Lung Transplant . 2011; 30:849–853. doi: 10.1016/j.healun.2011.03.008MedlineGoogle Scholar
10. Pinney SP, Anyanwu AC, Lala A, Teuteberg JJ, Uriel N, Mehra MR. Left ventricular assist devices for lifelong support. J Am Coll Cardiol . 2017; 69:2845–2861. doi: 10.1016/j.jacc.2017.04.031CrossrefMedlineGoogle Scholar
11. Jabbar HR, Abbas A, Ahmed M, Klodell CT, Chang M, Dai Y, Draganov PV. The incidence, predictors and outcomes of gastrointestinal bleeding in patients with left ventricular assist device (LVAD). Dig Dis Sci . 2015; 60:3697–3706. doi: 10.1007/s10620-015-3743-4CrossrefMedlineGoogle Scholar
12. Quader M, LaPar DJ, Wolfe L, Ailawadi G, Rich J, Speir A, Fonner C, Kasirajan V; Investigators for the Virginia Cardiac Surgery Quality Initiative. Blood poduct utilization with left ventricular assist device implantation: a decade of statewide data. ASAIO J . 2016; 62:268–273. doi: 10.1097/MAT.0000000000000345CrossrefMedlineGoogle Scholar
13. Carson JL, Triulzi DJ, Ness PM. Indications for and adverse effects of red-cell transfusion. N Engl J Med . 2017; 377:1261–1272. doi: 10.1056/NEJMra1612789CrossrefMedlineGoogle Scholar
14. Drakos SG, Stringham JC, Long JW, Gilbert EM, Fuller TC, Campbell BK, Horne BD, Hagan ME, Nelson KE, Lindblom JM, Meldrum PA, Carlson JF, Moore SA, Kfoury AG, Renlund DG. Prevalence and risks of allosensitization in HeartMate left ventricular assist device recipients: the impact of leukofiltered cellular blood product transfusions. J Thorac Cardiovasc Surg . 2007; 133:1612–1619. doi: 10.1016/j.jtcvs.2006.11.062CrossrefMedlineGoogle Scholar
15. Velez M, Johnson MR. Management of allosensitized cardiac transplant candidates. Transplant Rev (Orlando) . 2009; 23:235–247. doi: 10.1016/j.trre.2009.07.001CrossrefMedlineGoogle Scholar
16. Stulak JM, Lee D, Haft JW, Romano MA, Cowger JA, Park SJ, Aaronson KD, Pagani FD. Gastrointestinal bleeding and subsequent risk of thromboembolic events during support with a left ventricular assist device. J Heart Lung Transplant . 2014; 33:60–64. doi: 10.1016/j.healun.2013.07.020CrossrefMedlineGoogle Scholar
17. Draper KV, Huang RJ, Gerson LB. GI bleeding in patients with continuous-flow left ventricular assist devices: a systematic review and meta-analysis. Gastrointest Endosc . 2014; 80:435.e1–446.e1. doi: 10.1016/j.gie.2014.03.040CrossrefGoogle Scholar
18. Elmunzer BJ, Padhya KT, Lewis JJ, Rangnekar AS, Saini SD, Eswaran SL, Scheiman JM, Pagani FD, Haft JW, Waljee AK. Endoscopic findings and clinical outcomes in ventricular assist device recipients with gastrointestinal bleeding. Dig Dis Sci . 2011; 56:3241–3246. doi: 10.1007/s10620-011-1828-2CrossrefMedlineGoogle Scholar
19. Guha A, Eshelbrenner CL, Richards DM, Monsour HP. Gastrointestinal bleeding after continuous-flow left ventricular device implantation: review of pathophysiology and management. Methodist Debakey Cardiovasc J . 2015; 11:24–27. doi: 10.14797/mdcj-11-1-24CrossrefMedlineGoogle Scholar
20. Joy PS, Kumar G, Guddati AK, Bhama JK, Cadaret LM. Risk factors and outcomes of gastrointestinal bleeding in left ventricular assist device recipients. Am J Cardiol . 2016; 117:240–244. doi: 10.1016/j.amjcard.2015.10.041CrossrefMedlineGoogle Scholar
21. Slaughter MS. Hematologic effects of continuous flow left ventricular assist devices. J Cardiovasc Transl Res . 2010; 3:618–624. doi: 10.1007/s12265-010-9222-6CrossrefMedlineGoogle Scholar
22. Sparrow CT, Nassif ME, Raymer DS, Novak E, LaRue SJ, Schilling JD. Pre-operative right ventricular dysfunction is associated with gastrointestinal bleeding in patients supported with continuous-flow left ventricular assist devices. JACC Heart Fail . 2015; 3:956–964. doi: 10.1016/j.jchf.2015.09.009CrossrefMedlineGoogle Scholar
23. Uriel N, Pak SW, Jorde UP, Jude B, Susen S, Vincentelli A, Ennezat PV, Cappleman S, Naka Y, Mancini D. Acquired von Willebrand syndrome after continuous-flow mechanical device support contributes to a high prevalence of bleeding during long-term support and at the time of transplantation. J Am Coll Cardiol . 2010; 56:1207–1213. doi: 10.1016/j.jacc.2010.05.016CrossrefMedlineGoogle Scholar
24. Wever-Pinzon O, Selzman CH, Drakos SG, Saidi A, Stoddard GJ, Gilbert EM, Labedi M, Reid BB, Davis ES, Kfoury AG, Li DY, Stehlik J, Bader F. Pulsatility and the risk of nonsurgical bleeding in patients supported with the continuous-flow left ventricular assist device HeartMate II. Circ Heart Fail . 2013; 6:517–526. doi: 10.1161/CIRCHEARTFAILURE.112.000206LinkGoogle Scholar
25. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, Flachskampf FA, Foster E, Goldstein SA, Kuznetsova T, Lancellotti P, Muraru D, Picard MH, Rietzschel ER, Rudski L, Spencer KT, Tsang W, Voigt JU. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr . 2015; 28:1.e14–39.e14. doi: 10.1016/j.echo.2014.10.003CrossrefGoogle Scholar
26. Kleinbuam DG, Klein M. Survival Analysis: A Self-Learning Text . 1st ed.New York, NY: Springer-Verlag; 1996.CrossrefGoogle Scholar
27. Lubsen J, Pool J, van der Does E. A practical device for the application of a diagnostic or prognostic function. Methods Inf Med . 1978; 17:127–129.CrossrefMedlineGoogle Scholar
28. Zlotnik A, Abraria V. A general-purpose nomogram generator for predictive logistic regression models. Stata J . 2015; 15:537–546.CrossrefGoogle Scholar
29. Sullivan LM, Massaro JM, D'Agostino RB. Presentation of multivariate data for clinical use: the Framingham Study risk score functions. Stat Med . 2004; 23:1631–1660. doi: 10.1002/sim.1742CrossrefMedlineGoogle Scholar
30. Agrawal S, Garg L, Shah M, Agarwal M, Patel B, Singh A, Garg A, Jorde UP, Kapur NK. Thirty-day readmissions after left ventricular assist device implantation in the United States: insights from the nationwide readmissions database. Circ Heart Fail . 2018; 11:e004628. doi: 10.1161/CIRCHEARTFAILURE.117.004628LinkGoogle Scholar
31. Li F, Hinton A, Chen A, Mehta NK, Eldika S, Zhang C, Hussan H, Conwell DL, Krishna SG. Left ventricular assist devices impact hospital resource utilization without affecting patient mortality in gastrointestinal bleeding. Dig Dis Sci . 2017; 62:150–160. doi: 10.1007/s10620-016-4379-8CrossrefMedlineGoogle Scholar
32. Greenstein RJ, McElhinney AJ, Reuben D, Greenstein AJ. Colonic vascular ectasias and aortic stenosis: coincidence or causal relationship? Am J Surg . 1986; 151:347–351.CrossrefMedlineGoogle Scholar
33. Patel SR, Vukelic S, Jorde UP. Bleeding in continuous flow left ventricular assist device recipients: an acquired vasculopathy? J Thorac Dis . 2016; 8:E1321–E1327. doi: 10.21037/jtd.2016.10.81CrossrefMedlineGoogle Scholar
34. Joly JM, El-Dabh A, Kirklin JK, Marshell R, Smith MG, Acharya D, Rajapreyar IN, Tallaj JA, Tresler M, Pamboukian SV. High right atrial pressure and low pulse pressure predict gastrointestinal bleeding in patients with left ventricular assist device. J Card Fail . 2018; 24:487–493. doi: 10.1016/j.cardfail.2018.03.003CrossrefMedlineGoogle Scholar
35. Becq A, Rahmi G, Perrod G, Cellier C. Hemorrhagic angiodysplasia of the digestive tract: pathogenesis, diagnosis, and management. Gastrointest Endosc . 2017; 86:792–806. doi: 10.1016/j.gie.2017.05.018CrossrefMedlineGoogle Scholar
36. Chalasani N, Cotsonis G, Wilcox CM. Upper gastrointestinal bleeding in patients with chronic renal failure: role of vascular ectasia. Am J Gastroenterol . 1996; 91:2329–2332.MedlineGoogle Scholar
37. Ishigami J, Grams ME, Naik RP, Coresh J, Matsushita K. Chronic kidney disease and risk for gastrointestinal bleeding in the community: the Atherosclerosis Risk in Communities (ARIC) Study. Clin J Am Soc Nephrol . 2016; 11:1735–1743. doi: 10.2215/CJN.02170216CrossrefMedlineGoogle Scholar
38. Lutz J, Menke J, Sollinger D, Schinzel H, Thürmel K. Haemostasis in chronic kidney disease. Nephrol Dial Transplant . 2014; 29:29–40. doi: 10.1093/ndt/gft209CrossrefMedlineGoogle Scholar
39. Falciglia M, Freyberg RW, Almenoff PL, D'Alessio DA, Render ML. Hyperglycemia-related mortality in critically ill patients varies with admission diagnosis. Crit Care Med . 2009; 37:3001–3009. doi: 10.1097/CCM.0b013e3181b083f7CrossrefMedlineGoogle Scholar
40. Farrokhi F, Smiley D, Umpierrez GE. Glycemic control in non-diabetic critically ill patients. Best Pract Res Clin Endocrinol Metab . 2011; 25:813–824. doi: 10.1016/j.beem.2011.05.004CrossrefMedlineGoogle Scholar
41. Krinsley JS. Glycemic variability: a strong independent predictor of mortality in critically ill patients. Crit Care Med . 2008; 36:3008–3013. doi: 10.1097/CCM.0b013e31818b38d2CrossrefMedlineGoogle Scholar
42. Ambardekar AV, Thibodeau JT, DeVore AD, Kittleson MM, Forde-McLean RC, Palardy M, Mountis MM, Cadaret L, Teuteberg JJ, Pamboukian SV, Xie R, Stevenson LW, Stewart GC. Discordant perceptions of prognosis and treatment options between physicians and patients with advanced heart failure. JACC Heart Fail . 2017; 5:663–671. doi: 10.1016/j.jchf.2017.04.009CrossrefMedlineGoogle Scholar
43. Fedson SE, MacKenzie KK, Delgado ED, Abraham MN, Estep JD, Blumenthal-Barby JS, Bruce CR. Mapping the informed consent process for left ventricular assist devices. ASAIO J . 2017; 64:630–635.CrossrefGoogle Scholar
44. Maciver J, Ross HJ. Quality of life and left ventricular assist device support. Circulation . 2012; 126:866–874. doi: 10.1161/CIRCULATIONAHA.111.040279LinkGoogle Scholar
45. Netuka I, Litzler PY, Berchtold-Herz M, Flecher E, Zimpfer D, Damme L, Sundareswaran KS, Farrar DJ, Schmitto JD; EU TRACE Investigators. Outcomes in HeartMate II patients with no antiplatelet therapy: 2-year results from the European TRACE Study. Ann Thorac Surg . 2017; 103:1262–1268. doi: 10.1016/j.athoracsur.2016.07.072CrossrefMedlineGoogle Scholar
46. Netuka I, Ivák P, Tučanová Z, Gregor S, Szárszoi O, Sood P, Crandall D, Rimsans J, Connors JM, Mehra MR. Evaluation of low-intensity anti-coagulation with a fully magnetically levitated centrifugal-flow circulatory pump-the MAGENTUM 1 study. J Heart Lung Transplant . 2018; 37:579–586. doi: 10.1016/j.healun.2018.03.002CrossrefMedlineGoogle Scholar
47. Bouillon K, Bertrand M, Boudali L, Ducimetiere P, Dray-Spira R, Zureik M. Short-term risk of bleeding during heparin bridging at initiation of vitamin K antagonist therapy in more than 90 000 patients with nonvalvular atrial fibrillation managed in outpatient care. J Am Heart Assoc . 2016; 5:e004065. doi: 10.1161/JAHA.116.004065LinkGoogle Scholar
48. Rechenmacher SJ, Fang JC. Bridging anticoagulation: primum non nocere. J Am Coll Cardiol . 2015; 66:1392–1403. doi: 10.1016/j.jacc.2015.08.002CrossrefMedlineGoogle Scholar
49. Edwards AL, Fitzmorris P, Pamboukian SV, George JF, Wilcox CM, Peter S. Association of pulsatility with gastrointestinal bleeding in a cohort of HeartMate II recipients. ASAIO J . 2018; 64:472–479. doi: 10.1097/MAT.0000000000000766CrossrefMedlineGoogle Scholar
50. Houston BA, Schneider AL, Vaishnav J, Cromwell DM, Miller PE, Faridi KF, Shah A, Sciortino C, Whitman G, Tedford RJ, Stevens GR, Judge DP, Russell SD, Rouf R. Angiotensin II antagonism is associated with reduced risk for gastrointestinal bleeding caused by arteriovenous malformations in patients with left ventricular assist devices. J Heart Lung Transplant . 2017; 36:380–385. doi: 10.1016/j.healun.2016.12.016CrossrefMedlineGoogle Scholar
51. Katz JN, Adamson RM, John R, Tatooles A, Sundareswaran K, Kallel F, Farrar DJ, Jorde UP; TRACE Study. Safety of reduced anti-thrombotic strategies in HeartMate II patients: a one-year analysis of the US-TRACE Study. J Heart Lung Transplant . 2015; 34:1542–1548. doi: 10.1016/j.healun.2015.06.018CrossrefMedlineGoogle Scholar

moralesbove1969.blogspot.com

Source: https://www.ahajournals.org/doi/abs/10.1161/CIRCHEARTFAILURE.118.005267

Morales Bove1969