Early Blood Pressure Variables Associated with Improved Outcomes in VA-ECLS – ELSO Registry Analysis

Abstract

Introduction

As utilization of VA-ECLS in treatment of CS continues to expand, clinical variables that guide clinicians in early recognition of myocardial recovery and therefore, improved survival, after VA-ECLS are critical. There remains paucity of literature on early post-initiation blood pressure measurements that predict improved outcomes.

Methods

We queried the ELSO registry for cardiogenic shock patients treated with VA or VVA-ECLS between 2009 and 2020. Our inclusion criteria included treatment with VA or Veno-veno-arterial ECLS, absence of pre-existing durable right, left or biventricular assist devices, no pre-ECLS cardiac arrest and no surgical or percutaneously placed left ventricular venting devices during their ECLS runs. Our primary outcome of interest was the survival to discharge during index hospitalization.

Results

A total of 2400 CS patients met our inclusion criteria and had complete documentation of blood pressures. Actual mortality during index hospitalization in our cohort was 49.5% and survivors were younger, more likely to be Caucasian, intubated for >30 hours pre-ECLS initiation, and have a favorable baseline SAVE score (p < 0.05 for all). Multivariate regression analyses adjusting for SAVE score, age, ECLS flow at 4 hours and race showed that every 10 mmHg increase in baseline SBP (HR 0.92, 95% CI 0.89–0.95, p < 0.001), and baseline PP (HR 0.88, 95% CI 0.84–0.91, p < 0.001) at 24 hours was associated with a statistically significant reduction in mortality.

Conclusion

Early (within 24 hours) improvements in PP and SBP from baseline are associated with improved survival to discharge among CS patients treated with VA-ECLS.

Introduction

Utilization of veno-arterial extracorporeal life support (VA-ECLS) to treat patients in refractory cardiogenic shock (CS) has continued to rise^1,2. VA-ECLS is a resource intensive and expensive therapy, and therefore, identification of factors that are associated with better prognosis are critical for its successful use. Several pre-ECLS factors have been associated with adverse outcomes among ECLS recipients^3–9. Studies have also reported use of various biomarkers and evidence of end-organ dysfunction while supported on ECLS to aid the prognostication process^6,10–16. However, there is a paucity of data to guide clinicians in early recognition of myocardial recovery and therefore, improved survival, after VA-ECLS deployment. In this study, we investigated the prognostic role of changes in blood pressure within first 24-hours after VA-ECLS deployment. We queried The Extracorporeal Life Support Organization (ELSO) registry to identify CS patients treated with ECLS because the registry is uniquely poised to provide a large cohort of VA-ECLS patients from across the globe.

Methods

Patient Selection

Adult (≥ 18 years) CS patients treated with ECLS between the years of 2009 and 2020 were identified using the ELSO registry. Our inclusion criteria included treatment with VA or Veno-veno-arterial ECLS, absence of pre-existing durable right, left or biventricular assist devices, no pre-ECLS cardiac arrest and no surgical or percutaneously placed left ventricular venting devices during their ECLS runs. Once our cohort of interest was identified, patients were also required to have complete documentation of blood pressure variables at baseline and at 24-hours after VA-ECLS initiation. Only the first run on VA-ECLS was included in the analysis for those patients that had multiple runs during index hospitalization (supplemental figure 1). Our primary outcome of interest was the survival to discharge during index hospitalization. This retrospective analysis of the ELSO registry was reviewed and approved by the Vanderbilt University Institutional Review Board (IRB).

Hemodynamic Variables

Data on blood pressure (systolic, diastolic and pulse pressure) were collected at baseline and 24-hours post-ECLS initiation. Owing to a great number (>50%) of patients missing complete hemodynamic data (right atrial pressure, pulmonary artery pressure, and pulmonary artery occlusive pressure), we did not analyze the significance of changes in these parameters in our analysis.

Statistical Analysis

Baseline demographic, comorbidities, and hospital characteristics were compared amongst survivors versus non-survivors using the Pearson χ2 test for categorical variables (reported as proportions) as well as normally distributed continuous variables. Non-normally distributed continuous variables were reported as medians with interquartile ranges and were analyzed using Wilcoxon rank-sum. The SAVE Score³, a validated prognostic score utilizing pre-ECLS clinical and laboratory markers, was calculated with appropriate weighting parameters for all patients included in our study as recommended by the ELSO. The changes in blood pressure parameters were reported are integers with respective negative or positive signs. For example, a systolic pressure of 70 mmHg at 24hrs from an initial systolic blood pressure of 80 mmHg, was reported as change of −10 mmHg. Each blood pressure variable was then segregated into subcategories separated by 10 mmHg increments. Association between baseline characteristics and the primary outcome variable were then assessed using univariable logistic regression analyses. Afterwards, a multivariable logistic regression analysis including all variables that were significant (p < 0.2) in univariable analyses (age, race, SAVE score, ECMO flow at 24 hours and blood pressure parameters) was performed. Subsequently, the probability of mortality was predicted using the model and plotted along the subcategories of each hemodynamic variable with their 95% confidence intervals. Finally, receiver operator curves (ROC) were obtained for change in blood pressures at 24 hours to ascertain which hemodynamic parameter best predicted outcomes. All data extraction and analyses were performed using Stata (Version 14.0. College station, TX). Two-sided p value <0.05 was used for statistical significance.

Results

A total of 4041 CS patients met our inclusion criteria. Of those 2400 CS patients had complete documentation of blood pressure variables and were analyzed in our cohort (median age 56.9 years (IQR: 43.6–66), 37% women, 12.3% Black, and median SAVE score 5 (IQR: 2–8)). Actual mortality during index hospitalization in our cohort was 49.5%, and a total 86 (3.5%) patients received durable LVAD or heart transplantation. Table 1 outlines differences in baseline characteristics between survivors and non-survivors. As shown, survivors were younger, more likely to be Caucasian, intubated for >30 hours pre-ECLS initiation, and have a favorable baseline SAVE score (p < 0.05 for all).

Table 1:

Comparison of baseline characteristics Between Survivors and Non-survivors

	Totals (N=2400)	Non-Survivors (N=1188)	Survivors (N=1163)	P value
Age in years (Median, SD)	56.9 (43.6–66)	60.2(48.9–67.9)	52.9(38.4–62.9)	<0.001
Age Group (years)
18–38	851(35.4%)	44.1%	26.9%	<0.001
39–52	621(25.8%)	26.2%	25.1%	0.55
53–62	483(20.1%)	16.3%	23.8%	<0.001
≥63	445(18.5%)	13.2%	23.9%	<0.001
Women	37.0%	35.6%	37.8%	0.26
Race
White	1364(56.8%)	55.7%	57.8%	0.005
Black	296(12.3%)	11.4%	12.8%	0.29
Asian	274(11.4%)	11.6%	11.4%	0.31
Hispanic	178 (7.4%)	7.2%	7.6%	0.89
Weight (kg)
≤ 65	936(39.0%)	39.1%	38.7%	0.85
65–88	487(20.2%)	18.7%	21.9%	0.058
≥89	977(40.7%)	42.0%	39.2%	0.16
SAVE Score Comorbidities
Liver Dysfunction	181(7.5%)	9.3%	5.6%	0.001
Chronic Renal Failure	174(7.2%)	8.5%	5.8%	0.01
Congenital Heart Disorders	97(4.0%)	4.7%	3.2%	0.07
Myocarditis	135(5.6%)	3.2%	8.1%	<0.001
VT/VF	161(6.7%)	6.3%	6.7%	0.63
Post Heart/Lung Transplant *	69(2.8%)	2.4%	3.4%	0.15
Need for Renal Replacement Therapy	210(8.7%)	10.8%	6.4%	<0.001
SAVE score (median, IQR)	5 (2–8)	4(1–6)	6(3–8)	<0.001
SAVE score (mean, SD)	4.7 (±4.1)	3.6(4.0)	5.7(4.0)
Pre-ECLS Intubation (hours)
≤10	392(16.3%)	20.1	12.6	<0.001
11–29	468(19.5%)	19.1	19.5	0.84
≥30	1540(64.1%)	60.6	67.8	<0.001
ECLS Characteristics
Run-time (hours)	119 (71–208)	118(68–228)	119.5(75–193)	0.94
Flow at 4 hrs	4(3.3–4.5)	4.0(3.32–4.57)	3.95(3.26–4.50)	0.10
Flow at 24hrs	4.0(3.3–4.6)	4.07(3.5–4.70)	4.0(3.23–4.56)	<0.001
Mode
Veno-Arterial (VA)	2346 (97.7%)	96.6%	98.9%	<0.001
Venovenous-Arterial (VVA)	54(2.2%)	3.3%	1.0%	<0.001

ECLS = extracorporeal life support, SAVE = Survival after Veno-arterial ECMO, VT/VF = ventricular tachycardia or ventricular fibrillation

^*Includes patients with heart only, lung only, and dual heart/lung transplants

Table 2 outlines the distribution of blood pressure variables across survivors and non-survivors. At baseline, median systolic blood pressure (SBP) (88 vs 85 mmHg) and diastolic blood pressure (DBP) (55 vs 52 mmHg) were significantly higher among survivors compared to non-survivors (p < 0.001), whereas the pulse pressure (PP) was not (31 vs 31 mmHg, p = 0.47). At 24-hours post-ECLS initiation, median SBP (98 vs 92 mmHg) and, DBP (65 vs 63 mmHg) were consistently higher among the survivors compared to the non-survivors; additionally, the PP was also significantly higher among the survivors (32 mmHg) compared to the non-survivors (27 mmHg, p < 0.001). Comparing changes from baseline to 24 hours, median SBP, and PP were higher by 4 and 5 mmHg, respectively, among survivors compared to the non-survivors (p < 0.05 for all).

Table 2:

Hemodynamic Measurements - Survivors and Non-Survivors

	Overall	Non-Survivors	Survivors	p-value
Initial SBP	87(73–101.5)	85(70–100)	88(76–103)	<0.001
Initial DBP	53(45–64)	52(42–52)	55(46–65)	<0.001
Initial PP	31 (22–42)	31 (21–43)	31 (23–42)	0.47
SBP @ 24hr	95(84–108)	92(80–105)	98(88–110)	<0.001
DBP @ 24hr	64(57–73)	63(56–72)	65(59–73)	<0.001
PP @ 24hr	30(18–42)	27(15–39.5)	32(21–44)	<0.001
SBP (24hr) – SBP (initial)	8((−8)-24)	6(−10-(22))	10(−5-(26))	<0.001
PP (24hr) – PP (initial)	-3((−16)-10)	-5(−18–8)	0(−13–12)	<0.001

SBP = systolic blood pressure, DBP = diastolic blood pressure, MAP = mean arterial pressure, PP = pulse pressure

Table 3 outlines the results of univariable and adjusted (for SAVE score) analyses assessing various BP parameters at 24h post-ECLS initiation; data on the association between delta changes in BP parameters are also reported. On univariable analyses, every 10 mmHg increment in baseline SBP (Odds Ratio (OR) 0.95, 95% CI 0.92–0.97, p < 0.001) and PP (OR 0.90, 95% CI 0.87–0.93, p < 0.001) at 24 hours was associated with a significant reduction in mortality. This statistical significance remained when adjusted for baseline SAVE score as well; every 10-mmHg increment in SBP (OR 0.92, 95% Confidence interval (CI) 0.89–0.95, p < 0.001), and PP (OR 0.88, 95% CI 0.84–0.91, p < 0.001). Additionally, multivariable regression analyses adjusting for SAVE score, age, ECLS flow at 4 hours and race showed that every 10 mmHg increase in SBP (HR 0.91, 95% CI 0.88–0.94, p < 0.001), and PP (HR 0.87, 95% CI 0.84–0.91, p < 0.001) was associated with a statistically significant reduction in mortality. (Supplemental Table 1) However, 10 mm Hg increment in DBP at 24 hours was not associated with improvement in mortality (OR 1.03, 95% CI 0.99–1.08, p = 0.12).

Table 3:

Post-ECLS Hemodynamic Predictors of Mortality

Hemodynamic parameters (Median (IQR); in mmHg)	Univariable OR (95% CI) (per 1mmHg)	P-value	Univariable OR (95% CI) (per 10 mmHg)	P-value	Adjusted OR (per 10 mmHg) (95% CI)†	P-value
SBP @ 24hr	0.98 (0.976–0.985)	< 0.001	0.82 (0.79–0.86)	<0.001	0.81 (0.77–0.85)	0.006
DBP @ 24hr	0.98 (0.982–0.994)	< 0.001	0.89 (0.83–0.95)	<0.001	0.91 (0.85–0.97)	<0.001
PP @ 24hr	0.98 (0.98–0.99)	<0.001	0.86 (0.82–0.90)	<0.001	0.85(0.81–0.89)	<0.001
Delta pressures
DBP(24hr) - DBP(initial)	1.00 (0.99–1.00)	0.12	1.03 (0.99–1.08)	0.12	-	-
SBP(24hr) - SBP(initial)	0.99 (0.99–0.99)	<0.001	0.95 (0.92–0.97)	0.001	0.92(0.89–0.95)	<0.001
PP(24hr) - PP(initial)	0.98 (0.98–0.99)	<0.001	0.90 (0.87–0.93)	<0.001	0.88(0.84–0.91)	<0.001

SBP = systolic blood pressure, DBP = diastolic blood pressure, PP = pulse pressure

^†Adjusted for SAVE Score

Discussion

Our study highlights that higher SBP or DBP at 24h as well as an increase in baseline SBP or PP of at least 10 mmHg by 24h on VA-ECLS is associated with a statistically significant reduction in all-cause mortality during index hospitalization. These findings may assist clinicians in early risk stratification of patients receiving VA-ECLS.

The PP is dependent on arterial resistance and compliance, as well as hemodynamic factors such as stroke volume and peak aortic blood flow. It correlates well with stroke volume and cardiac index¹⁷, and low PP is a strong independent predictor of mortality in patients with advanced heart failure^17–19. Furthermore, reduced PP as a predictor of poor outcomes is independent of baseline left ventricular ejection fraction and probably a reflection of inotropic reserve¹⁸. Hence, improvement in pulse pressure while on ECLS can be viewed as a marker of myocardial reserve or recovery, and can help identify patients likely to do well after ECLS is weaned off. In their single center retrospective analysis of patients treated with ECLS for CS, Aissaoui et al noted that the mean PP was indeed higher (52 ± 12 vs 39 ± 15 mmHg) among patients (n =33) who tolerated a full weaning trial¹³. Similarly, Pappalardo et al. showed that SBP and PP at the time of ECLS wean was significantly higher than pre-ECLS in patients that were weaned and survived till discharge¹⁶. The median number of ECLS days in this study was 5 days and unlike our study, all patients had some form of LV venting strategy (IABP or Impella) in place. While these studies confirm the utility of PP during weaning trials, they do not evaluate its role in early risk-stratification of patients on VA-ECLS. Park et al, from a small-sized single center retrospective analysis, observed that mean PP and SBP over the first 6 hours after VA-ECLS (~20% with IABP for LV venting) initiation were significantly higher in survivors (PP: 39.6 ± 7.1 mmHg, SBP: 95.8 ± 23.6 mmHg) versus non-survivors (PP: 28.1 ± 16 mmHg, SBP: 79.4 ± 41.3 mmHg; p < 0.05)¹⁴. In our analysis of this multicenter international registry, we observed that improvement in absolute blood pressures and delta PP and SBP at 24 hours after VA-ECLS initiation had statistically significant association with in-hospital mortality. (Table 3, Supplemental Table 1, Figures 1–2, Central Illustration)

Figure 1:

Probability of death when adjusted for pulse pressure at 24 hours

Figure 2:

Probability of death when adjusted for change in systolic blood pressure (SBP) over 24 hours

Central Illustration:

Probability of death when adjusted for change in pulse pressure (PP) over 24 hours

While and increase in baseline systolic blood pressure and pulse pressure at 24 hours were both associated with improved outcomes, delta PP may be the preferred marker of myocardial recovery. Vasoactive medications affect both systolic and diastolic pressures due to their varying effects on systemic vascular resistance. However, at any given SBP, PP predicts stroke volume and cardiac index more accurately than SBP or mean arterial blood pressure alone¹⁷. In our cohort, this is also confirmed with a statistically significant increase in AUC when delta PP is added to baseline SAVE score (AUC 0.64) compared to delta SBP (AUC 0.66 vs 0.65, p = 0.03).

Early assessment and recognition of myocardial recovery or so called “responders” using changes in blood pressure variables may have several clinical implications. First, such a strategy provides clinicians additional prognostic tools to guide further clinical care. For example, it may allow for timely escalation of care for those patients less likely to recover, especially with more durable therapies such as left ventricular assist device (LVAD) or orthotropic heart transplantation (HT). It may also guide clinicians to consider timely initiation of alternate management strategies such as mechanical intra-cardiac unloading that may aid in ventricular recovery or successful bridge to LVAD or HT. Second, it provides clinicians additional prognostic information very early in the treatment course that may assist coordinating further care with the caregivers and families. Thirdly, VA-ECLS is a resource intense therapy and often one with limited availability. Clinical parameters that allow for objective prognostication before as well early after ECLS initiation are especially valuable in appropriate allocation of this scarce resource to those patients most likely to benefit from it.

Limitations

There are a few important limitations of our study. The ELSO registry database uses ICD-9 and ICD-10 for coding diagnoses and procedures and thus our analyses are prone to error of coding, including miscoding, under or over coding. Furthermore, our ability to evaluate hemodynamic predictors of mortality is impacted by missing variables in the ELSO database. We have overcome this limitation by only including those patients in our analysis that had complete dataset to calculate SAVE scores and those with complete hemodynamic data at baseline and at 24 hours post initiation. Hence, it is important to note that our study findings are most applicable to only those patient characteristics that met our inclusion criteria. For example, these results are not applicable to patients who are concomitantly treated with a left ventricular venting device. Given the limitations of data captured in the ELSO database, we are unable to calculate vasoactive-inotropic score (VIS) in our cohort. Inotropes and vasopressors may certainly alter the absolute blood pressures but improvement in pulse pressure on medications still reflects myocardial reserve or recovery. Finally, ELSO database only captures outcomes during index hospitalization and hence, lacks longitudinal outcomes post discharge. However, the goal of our study was to describe early hemodynamic predictors of mortality during index hospitalization.

Conclusion

Early (within 24 hours) improvement in PP and SBP from baseline is associated with improved survival to discharge among CS patients treated with VA-ECLS.

Clinical Perspectives.

Competency in medical knowledge:

Although there are prognostication scores that allow clinicians to risk stratify patients pre-ECLS cannulation, there is limited data to guide early post-cannulation prognostication.

Translational outlook:

Further evaluation of hemodynamic as well as biomarkers will allow clinicians to better guide therapy for patients once they are already on VA-ECLS support.

Abbreviation List

VA-ECLS

Veno-arterial extracorporeal life support

CS

Cardiogenic Shock

ELSO

Extracorporeal Life Support Organization

SAVE

Survival after Veno-arterial ECMO

ROC

Receiver operator curve

IQR

Inter-quartile range

SBP

Systolic blood pressure

DBP

Diastolic blood pressure

PP

Pulse pressure

OR

Odds ratio

CI

Confidence interval

LV

Left ventricle

IABP

Intra-aortic balloon pump

AUC

Area under the curve

LVAD

Left ventricular assist devices

HT

Heart Transplantation

VIS

Vasoactive-inotropic score