Inhaled Nitric Oxide Therapy and Changes in Pulmonary Artery Pulsatility Index During Venoarterial Extracorporeal Membrane Oxygenation Weaning in Cardiogenic Shock Requiring ECPELLA

Abstract

Background

Inhaled nitric oxide (iNO) selectively dilates pulmonary vessels and reduces pulmonary vascular resistance, potentially improving right ventricular (RV) function. However, its clinical benefit in patients with cardiogenic shock requiring ECPELLA (i.e., combined Impella® (ABIOMED) and venoarterial extracorporeal membrane oxygenation (VA-ECMO) remains unclear. The aim of this study was to evaluate the impact of iNO on RV function in this critical population.

Methods and Results

Between September 2017 and June 2023, we retrospectively analyzed 54 consecutive patients with cardiogenic shock treated with ECPELLA. Patients were divided into 2 groups based on the use of iNO. RV function was assessed using the pulmonary artery pulsatility index (PAPi) at the time of ECPELLA initiation (Pre-PAPi) and VA-ECMO weaning (Post-PAPi). The iNO group had a significantly lower Pre-PAPi, prompting propensity-score matching to adjust for baseline differences. After matching, the Post-PAPi was significantly higher in the iNO group than in the non-iNO group (1.95 [1.63–3.00] vs. 1.40 [0.50–2.05], P<0.05). A significant interaction between iNO use and PAPi change was observed (P=0.006), suggesting an iNO-mediated improvement in RV function.

Conclusions

In patients with cardiogenic shock requiring ECPELLA support, adjunctive iNO therapy was associated with greater improvement in PAPi at the time of VA-ECMO weaning. These findings suggest a potential role for iNO in optimizing hemodynamics during the transition from ECPELLA to Impella-only support.

Central Figure.

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is widely used for mechanical circulatory support in patients with cardiogenic shock. Recent reviews have also highlighted the evolving role of VA-ECMO in acute cardiovascular care.¹ However, VA-ECMO increases left ventricular (LV) afterload, which may delay myocardial recovery. To unload the LV, intra-aortic balloon pumping (IABP) has traditionally been used, but since its reimbursement in Japan in 2016, the Impella® device (ABIOMED) has been increasingly used as a percutaneous LV assist device in this setting. Impella functions as a direct LV vent by actively pumping blood from the LV into the ascending aorta using an axial-flow pump. Recent data have also suggested that Impella may provide effective LV unloading during VA-ECMO support.²

Editorial p 1488

For patients with biventricular failure, VA-ECMO and Impella are often used in combination, commonly referred to as ECPELLA. This dual-device strategy has been associated with improved outcomes compared to VA-ECMO alone.^3,4 However, a large-scale analysis using Japan’s DPC database reported no significant difference in in-hospital mortality rates between patients treated with ECPELLA and those treated with VA-ECMO plus IABP.⁵ More recent registry data from Japan have further characterized the short-term outcomes of patients with acute myocardial infarction (AMI) complicated by cardiogenic shock who were treated with ECPELLA.⁶ Prognosis may vary due to differences in underlying etiology, organ dysfunction (e.g., renal failure), and disease severity, indicating that optimization of ECPELLA management remains a clinical challenge.⁷ In addition, recent registry findings suggest that differences in shock phenotype and support strategy may influence clinical outcomes in patients treated with Impella.⁸

Successful weaning from VA-ECMO typically requires improved right ventricular (RV) function. Recent studies have also emphasized the importance of invasive hemodynamic assessment when determining readiness for VA-ECMO explantation.⁹ However, persistent RV dysfunction can impair LV preload, leading to septal shift and Impella low-flow alarms due to insufficient LV filling.¹⁰ In such cases, inhaled nitric oxide (iNO) may be beneficial, as it selectively reduces pulmonary vascular resistance and RV afterload without affecting systemic vascular resistance and is expected to improve right heart failure.^11,12 Additionally, iNO improves oxygenation by enhancing ventilation–perfusion matching via selective vasodilation near well-ventilated alveoli.¹³

Although iNO is approved in Japan (INOflo®, Mallinckrodt) for adult patients with pulmonary hypertension after cardiovascular surgery, data on its use in adults with cardiogenic shock are limited. AMI complicated by cardiogenic shock remains associated with high mortality rates despite advances in reperfusion therapy and pharmacological management.¹⁴ Moreover, few studies have assessed the hemodynamic impact of iNO in patients receiving ECPELLA.

In this study, we aimed to evaluate the impact of iNO on RV function assessed by PAPi during ECPELLA support, focusing on its bedside clinical utility at the time of VA-ECMO weaning.

Methods

Study Population

This retrospective study included 66 consecutive patients with cardiogenic shock who received ECPELLA support at Fujita Health University Hospital between September 2017 and June 2023. Clinical data were extracted from medical records to assess the association between iNO therapy and clinical outcomes in these patients. At this institution, iNO therapy is uniformly administered at a dose of 20 ppm. Based on prior studies, a pulmonary artery pulsatility index (PAPi) <0.9 is considered indicative of impaired RV function.¹⁵ Accordingly, iNO therapy was initiated when the PAPi fell below this threshold. To mitigate the risk of north–south syndrome, a condition in which deoxygenated blood from the native heart perfuses the brain due to impaired pulmonary oxygenation, iNO was initiated only after confirming adequate pulmonary gas exchange.¹⁶ Patients with severely impaired alveolar gas exchange were not treated with iNO.

Among the 66 patients, 12 were excluded: 7 died within 24 h due to circulatory collapse despite ECPELLA initiation, and 5 were successfully weaned from mechanical circulatory support within 24 h, making evaluation of treatment response difficult.

The remaining patients were divided into 2 groups based on the presence or absence of concomitant iNO therapy, and outcomes were compared accordingly (Figure 1A).

Figure 1.

(A) Flow diagram of the patient selection in the analytic cohort. (B) Process flow diagram. ECPELLA, combination therapy of VA-ECMO (venoarterial extracorporeal membrane oxygenation) and Impella; iNO, inhaled nitric oxide; MCS, mechanical circulatory support; VAD, ventricular assist device.

Evaluation of RV

RV function was evaluated using the PAPi, with hemodynamic measurements obtained at 2 time points: (1) upon return to the coronary care unit (CCU) after ECPELLA initiation, and (2) immediately after VA-ECMO weaning. At these respective time points, the mean pulmonary artery pressure (mPAP) and PAPi were recorded and defined as Pre-PAP, Pre-PAPi and Post-PAP, Post-PAPi (Figure 1B).

To assess improvement in RV function, we calculated both the absolute and relative changes in PAPi over time. The absolute change (∆PAPi) was defined as Post-PAPi minus Pre-PAPi (∆PAPi = Post-PAPi − Pre-PAPi). The relative change in PAPi was calculated as ∆PAPi divided by Pre-PAPi (rate of change in PAPi = ∆PAPi/Pre-PAPi).

iNO was administered using the INOflo® device (Mallinckrodt Pharma Co., Ltd., Tokyo, Japan). Hemodynamic measurements were performed using a Swan–Ganz thermodilution catheter (Edwards Lifesciences, Chiba, Japan).

The study protocol was approved by the Institutional Review Board of Fujita Health University (IRB No. HM23-236).

Statistical Analysis

Continuous variables were tested for normality using the Shapiro–Wilk test. Normally distributed variables are presented as mean±standard deviation (SD) and were compared using Student’s t-test. Non-normally distributed variables are summarized as median (interquartile range [IQR]) and were compared using the Mann–Whitney U test, as appropriate. Categorical variables were compared using the χ² test or Fisher’s exact test.

To evaluate changes in RV function over time (pre- and post-ECPELLA), a two-way repeated-measures analysis of variance (ANOVA) was performed to assess the interaction between time and iNO treatment. Variables analyzed using repeated-measures ANOVA are presented as mean±SD in accordance with model assumptions.

To compare the magnitude and rate of change in PAPi between groups, analysis of covariance (ANCOVA) was performed, adjusting for baseline variables that differed significantly between groups.

Propensity scores were calculated using logistic regression including clinically relevant covariates and variables with significant baseline differences. One-to-one nearest-neighbor matching without replacement was applied.

Correlations between baseline PAPi and ∆PAPi were assessed using Pearson correlation analysis.

A two-sided P value <0.05 was considered statistically significant. All statistical analyses were performed using JMP version 15 (SAS Institute Inc., Cary, NC, USA).

Results

Baseline Clinical Patient Characteristics

A total of 54 consecutive patients were divided into 2 groups: 25 patients in the iNO group and 29 in the non-iNO group (crude cohort). Table 1 summarizes the crude cohort’s baseline characteristics, including age, sex, comorbidities, smoking history, primary diagnosis, hemodynamic parameters, and laboratory values at the time of admission to the CCU.

Table 1.

Clinical Baseline Patient Characteristics and Hemodynamic Parameters in the iNO and Non-iNO Groups at the First Time Point (ECPELLA Established)

Patient characteristics	Crude cohort			Matched cohort
	Non-iNO group (n=29)	iNO group (n=25)	P value	Non-iNO group (n=14)	iNO group (n=14)	P value
Age (years)	62.6±17.4	61.6±18.2	0.83	71.5 (65.3–78)	66 (50.8–74)	0.87
Male sex (%)	22 (75.9%)	19 (76%)	0.99	10 (71.4%)	10 (71.4%)	1
Comorbidities
Hypertension (%)	19 (65.5%)	8 (32.0%)	0.05>	6 (42.9%)	7 (50.0%)	0.27
Hyperlipidemia (%)	11 (37.9%)	4 (16.0%)	0.07	4 (28.6%)	3 (21.4%)	0.66
Diabetes (%)	11 (37.9%)	8 (32.0%)	0.64	5 (35.7%)	5 (35.7%)	1
Previous coronary artery revascularization (%)	9 (31.0%)	2 (8.0%)	0.05>	2 (14.3%)	0 (0%)	1
Previous stroke (%)	0 (0%)	2 (8.0%)	0.12	0 (0%)	2 (14.3%)	0.14
Previous peripheral artery disease (%)	0 (0%)	3 (12.0%)	0.05	0 (0%)	3 (12.0%)	0.06
Maintenance hemodialysis (%)	1 (3.5%)	4 (16.0%)	0.17	0 (0%)	3 (21.4%)	0.06
Current smoker (%)	7 (24.1%)	5 (20%)	0.72	2 (14.3%)	3 (21.4%)	0.62
Out-of-hospital cardiac arrest (%)	8 (28%)	10 (40%)	0.33	5 (35.7%)	6 (42.9%)	0.7
Primary disease			0.36			0.75
Ischemic heart disease (%)	21 (72%)	14 (56.0%)		10 (71.4%)	10 (71.4%)
Fulminant myocarditis (%)	6 (21%)	5 (20.0%)		3 (21.4%)	1 (7.1%)
Lethal arrhythmia (%)	1 (3.5%)	3 (12.0%)		1 (7.1%)	2 (13.3%)
Other (%)	1 (3.5%)	3 (12.0%)		0 (0%)	1 (12.0%)
Hemodynamic parameters
Pre-PAPi	1.00 (0.60–1.50)	0.60 (0.55–0.85)	0.01	0.70 (0.38–1.33)	0.75 (0.55–0.93)	0.72
Pre-mean PAP (mmHg)	15.0 (10.0–21.0)	15.0 (12.0–19.0)	0.65	14.0 (10.8–19.0)	12.5 (10.0–21.5)	0.8
Laboratory data at CCU admission
Renal function, creatinine (mg/dL)	1.20 (0.85–1.99)	1.20 (0.83–2.35)	0.91	1.04 (0.79–1.21)	1.38 (0.87–2.91)	0.98
Total bilirubin (mg/dL)	0.7 (0.5–1.0)	0.7 (0.4–1.2)	0.92	0.9 (0.4–1.0)	0.6 (0.5–1.3)	0.09
NT-proBNP (pg/mL)	3,833 (724–13,818)	5,106 (690–19,276)	0.78	3,484 (183–6,294)	6,770 (703–18,707)	0.33

Values are mean±SD; n (%), or median (IQR). Crude cohort=Data before adjustment by propensity-score matching. Matched cohort=Data after adjustment using propensity-score matching. CCU, coronary care unit; iNO, inhaled nitric oxide; IQR, interquartile range; MCS, mechanical circulatory support.

Regarding comorbidities, the prevalence of hypertension was significantly higher in the non-iNO group than in the iNO group. Additionally, the proportion of patients with a history of coronary artery revascularization was significantly greater in the non-iNO group.

Among the hemodynamic parameters, Pre-PAPi at the time of ECPELLA initiation was significantly lower in the iNO group. Although the mean PAP (Pre-mean PAP) tended to be lower in the iNO group, the difference was not statistically significant (Table 1, crude cohort).

Hemodynamic Outcomes and Clinical Course

The clinical course, survival outcomes, hemodynamic parameters, and device-related complications are summarized in the Table 2 (crude cohort). There were no significant differences between the iNO and non-iNO groups regarding Post-PAP values or VA-ECMO duration.

Table 2.

Hemodynamic Parameters at Second Time Point (VA-ECMO Weaning): Complications, Clinical Course, and Survival Rate in the iNO and Non-iNO Groups

Clinical course and complications	Crude cohort			Matched cohort
	Non-iNO group (n=29)	iNO group (n=25)	P value	Non-iNO group (n=14)	iNO group (n=14)	P value
Hemodynamic parameters at the time of VA-ECMO weaning
Post-PAPi	1.60 (1.05–2.10)	1.80 (1.25–2.30)	0.18	1.40 (0.50–2.05)	1.95 (1.63–3.00)	0.05>
Post-mean PAP (mmHg)	20 (16.0–25.5)	20 (16.5–23.0)	0.93	18.5 (14.8–20.3)	19.5 (16.8–21.2)	0.31
ΔPAPi	0.60 (0.20–1.00)	1.30 (0.70–1.60)	0.01>	0.60 (0.18–1.05)	1.35 (0.80–1.85)	0.01>
Rate of change in PAPi	0.56 (0.41–1.09)	1.67 (0.88–2.86)	0.01>	1.00 (0.46–1.26)	1.92 (0.89–3.89)	0.01>
Complications
Hypoxic encephalopathy	4 (13.8%)	6 (24.0%)	0.33	3 (21.4%)	2 (14.3%)	0.62
Requiring CRRT (%)	11 (37.9%)	8 (32.0%)	0.64	5 (35.7%)	4 (28.6%)	0.68
Major bleeding (%)	14 (48.3%)	13 (52.0%)	0.78	6 (42.9%)	7 (50%)	0.7
Stroke (%)	3 (10.3%)	3 (10.3%)	0.84	1 (7.1%)	1 (7.1%)	1
Gastrointestinal bleeding (%)	6 (20.7%)	3 (12.0%)	0.4	5 (35.7%)	1 (7.1%)	0.06
Intraoral bleeding (%)	2 (6.9%)	3 (12.0%)	0.51	1 (7.1%)	2 (14.2%)	0.53
Limb ischemic (%)	1 (3.5%)	2 (8.0%)	0.46	1 (7.1%)	1 (7.1%)	1
Sepsis (%)	4 (13.8%)	3 (12%)	0.84	1 (7.1%)	3 (21.4%)	0.28
Rhabdomyolysis (%)	2 (6.9%)	1 (4.0%)	0.64	2 (14.2%)	0 (0%)	0.14
Clinical course and survival rate
VA-ECMO duration (days)	5 (4.0–11.5)	6 (3.3–10.0)	0.69	5 (3.5–8)	4.5 (3.3–7.3)	0.85
Survival rate on VA-ECMO withdrawal (%)	21 (72.4%)	21 (84%)	0.3	8 (57.1%)	12 (85.7%)	0.09
30-day survival rate (%)	19 (65.5%)	16 (64%)	0.91	8 (57.1%)	9 (64.3%)	0.69
Survival discharge rate (%)	17 (58.6%)	13 (52%)	0.62	8 (57.1%)	8 (57.1%)	1
Length of hospital stay among survivors (days)	71 (39.5–128)	41 (33–80.5)	0.17	40 (30.3–76.3)	64 (39.5–127)	0.4

Values are mean±SD, n (%), or median (IQR). Crude cohort=Data before adjustment by propensity-score matching. Matched cohort=Data after adjustment using propensity-score matching. ΔPAPi = Post-PAPi − Pre-PAPi. Rate of change in the PAPi = ΔPAPi (Post-PAPi − Pre-PAPi) / Pre-PAPi. CRRT, continuous renal replacement therapy; IQR, interquartile range; PAPi, pulmonary artery pulsatility index; VA-ECMO, venoarterial extracorporeal membrane oxygenation.

Similarly, survival outcomes, including successful VA-ECMO weaning, 30-day survival, and survival to discharge, did not differ significantly between groups. Although the Post-PAPi tended to be higher in the iNO group, the difference was not statistically significant.

However, the Pre-PAPi was significantly lower in the iNO group compared to the non-iNO group (1.00 [0.60–1.50] vs. 0.70 [0.55–0.85], P=0.01), suggesting a more impaired baseline RV function in patients who received iNO therapy.

Propensity-Score Matching and Comparative Analysis

To account for significant baseline differences between groups, including hypertension, history of coronary revascularization, and Pre-PAPi, we performed propensity-score matching using logistic regression. Post-matching comparisons of patient backgrounds are summarized in Table 1 (matched cohort), and clinical outcomes, hemodynamic parameters, and mechanical support-related complications are shown in Table 2 (matched cohort).

Changes in RV function (PAPi) before and after ECPELLA treatment were compared within and between groups using matched data (Figure 2). In the iNO group, Post-PAPi was significantly higher than Pre-PAPi (0.76±0.08 vs. 2.32±0.33, P=0.006). A similar, though smaller, increase was observed in the non-iNO group (0.82±0.13 vs. 1.37±0.22, P=0.001). Pre-PAPi did not significantly differ between the iNO and non-iNO groups (0.76±0.09 vs. 0.82±0.13, P=0.74), whereas Post-PAPi was significantly higher in the iNO group (2.32±0.33 vs. 1.37±0.22, P=0.016).

Figure 2.

Changes in pulmonary artery pulsatility index (PAPi) over time for the matched cohort data. Values are mean±SD. (1) Comparison between Pre-PAPi and Post-PAPi in the iNO groups. (2) Comparison of Pre-PAPi and Post-PAPi in the non-iNO groups. (3) Comparison of Pre-PAPi between the iNO and non-iNO groups. (4) Comparison of Post-PAPi between the iNO and non-iNO groups. iNO, inhaled nitric oxide.

The two-way repeated-measures ANOVA revealed a significant interaction between time (pre/post-ECPELLA) and iNO treatment for RV function (P for interaction=0.024).

In the matched cohort, both the absolute (∆PAPi) and relative changes in PAPi were significantly greater in the iNO group than in the non-iNO group (∆PAPi: 1.35 vs. 0.6, P=0.002; relative change: 1.92 vs. 1.00, P=0.009; Figure 3A,B).

Figure 3.

(A) ∆PAPi of the matched cohort data (∆PAPi=Post-PAPi−Pre-PAPi, median (IQR)). (B) Rate of change in PAPi of the matched cohort (∆PAPi (Post-PAPi−Pre-PAPi)/Pre-PAPi, median (IQR)). IQR, interquartile range; PAPi, pulmonary artery pulsatility index.

Survival outcomes, including VA-ECMO weaning, 30-day survival, and survival to discharge, showed no significant differences between groups before or after matching (Table 2). These findings suggest that although iNO therapy did not affect survival, it was associated with a greater improvement in RV function when used in conjunction with ECPELLA support.

Discussion

In this study we evaluated the therapeutic effects of iNO on RV function in patients with cardiogenic shock requiring ECPELLA (Impella® plus VA-ECMO). iNO is known to selectively dilate the pulmonary vasculature, thereby reducing PVR and RV afterload, with minimal systemic effects owing to its preferential action in ventilated lung regions.¹⁷ iNO has been also evaluated in mechanically supported patients, including a randomized trial after LVAD implantation.¹⁸ Because our RV assessment relied primarily on PAPi, our results indicated an association between iNO use and greater improvement in PAPi at the time of VA-ECMO weaning, rather than definitive improvement in overall RV function. However, improvement in PAPi during this phase may be clinically meaningful, as adequate RV output and LV preload are essential to maintain Impella flow during the transition from ECPELLA to Impella-only support.

Prior studies have demonstrated that iNO improves RV function and hemodynamics in patients with pulmonary hypertension secondary to elevated PVR, such as post-cardiovascular surgery or acute respiratory distress syndrome. However, in patients with preserved RV function, iNO typically reduces pulmonary pressures without improving cardiac output, suggesting its effects are limited to those with RV dysfunction.^19–22

A recent single-center study investigated the short-term physiologic effects of iNO in patients receiving ECPELLA support and reported improvements in RV indices, including echocardiographic parameters. Although that study provided important mechanistic insight into acute hemodynamic responses to iNO, it did not include a contemporaneous non-iNO control group and primarily evaluated immediate changes after iNO initiation. In contrast, our study incorporated a matched non-iNO cohort using propensity-score analysis and focused on clinically standardized time points, particularly the VA-ECMO weaning phase. Therefore, our findings complement prior physiologic investigations by providing comparative, real-world data regarding PAPi dynamics during routine ECPELLA management.²³ Notably, our cohort included patients with more advanced RV dysfunction than those reported in prior studies. In the matched cohort, the median baseline PAPi was approximately 0.7, reflecting severe impairment of RV–pulmonary circulation coupling. This contrasts with previously reported cohorts with relatively higher baseline PAPi values, suggesting that our study represents a more clinically unstable population encountered in real-world ECPELLA management. Therefore, our findings extend prior physiologic observations to patients with more advanced RV dysfunction, particularly during the critical phase of VA-ECMO weaning.²⁴

Elevated PVR often arises from left-sided heart failure, where chronic increases in left atrial pressure contribute to pulmonary venous hypertension and secondary remodeling of the pulmonary vasculature.^25,26 These changes exacerbate RV afterload and contribute to RV failure, particularly in advanced cardiogenic shock.^27,28 In patients supported with ECPELLA, such mechanisms, alongside cardiac arrest, chest trauma, or arrhythmia-induced instability, may contribute to RV dysfunction. Targeted reduction of PVR in these settings is critical for optimizing RV performance and facilitating timely VA-ECMO weaning.

VA-ECMO weaning often involves transitioning to Impella-only support, where adequate LV preload is essential for maintaining device flow and preventing suction or hemolysis. Elevated PVR can impair RV output and thereby compromise LV filling, limiting the ability to wean from ECMO. By selectively reducing PVR without systemic vasodilation, iNO may support RV output and LV filling during this critical transition phase.^29,30

In our study, propensity-score matching was performed to adjust for baseline imbalances, and two-way repeated-measures ANOVA confirmed a significant interaction between iNO administration and PAPi improvement. The observed effect was most pronounced in patients with severe baseline RV dysfunction (PAPi ≤0.9), suggesting that iNO may be particularly beneficial in this high-risk subgroup.

Future studies should aim to validate these findings in larger prospective cohorts and determine whether incorporating iNO into standardized RV management and ECMO weaning algorithms improves clinical outcomes.

Study Limitations

This was a single-center retrospective study, which inherently limits causal inference and generalizability.

Second, RV function was primarily assessed using PAPi, and comprehensive echocardiographic or pulmonary hemodynamic parameters were not uniformly available. Therefore, our findings should be interpreted as reflecting improvement in PAPi rather than definitive improvement in overall RV function.

Third, Swan–Ganz measurements were obtained at predefined clinical time points (post-ECPELLA initiation and post-VA-ECMO weaning) and were not time-locked to iNO initiation; therefore, mechanistic interpretation of pulmonary vasodilation (including PVR changes) is limited. In addition, pulmonary artery wedge pressure was not available in all patients, precluding robust PVR calculations.

Fourth, worsening pulmonary congestion after iNO initiation was not systematically adjudicated in this retrospective dataset.

Because iNO therapy started in July 2021 at our institution, most patients in the non-iNO group were treated before June 2021, which may have introduced temporal bias. Furthermore, the sample size was relatively small, and clinical outcomes such as VA-ECMO duration and survival are influenced by multiple factors (e.g., bleeding, hypoxic encephalopathy, and infection) in addition to cardiac function. Future multicenter prospective studies are warranted.

Conclusions

In patients with cardiogenic shock requiring ECPELLA support, adjunctive iNO therapy was associated with greater improvement in PAPi at the time of VA-ECMO weaning, suggesting a potential role in optimizing hemodynamics during the transition to Impella-only support.

Disclosures

H.I. is a member of Circulation Reports’ Editorial Team. The other authors declare no conflicts of interest.

Data Availability

The datasets generated or analyzed during the current study are not publicly available due to patient privacy and ethical restrictions. De-identified data may be made available by the corresponding author upon reasonable request and with approval from the institutional review board of Fujita Health University.