Predicting Hemodynamic Changes During Intra-Aortic Balloon Pump Support With a Longitudinal Evaluation

Abstract

The use of intra-aortic balloon pump (IABP) has decreased in recent years due to negative outcome studies in cardiogenic shock complicating acute myocardial infarction, despite its favorable adverse-event profile. Acute hemodynamic response studies have identified potential super-responders with immediate improvements in cardiac index (CI) in heart failure patients. This single-center retrospective study aimed to predict CI and mean arterial pressure (MAP) changes throughout the entire duration of IABP support. The study analyzed 336 patients who received IABP between 2016 and 2022. Linear mixed-effect regression models were used to predict CI and MAP improvement during IABP support. The results showed that CI and MAP increases during the first days of support, and changes during IABP support varied with time and were associated with baseline parameters. Longitudinal CI change was associated with body surface area, baseline CI, baseline pulmonary artery pulsatility index, baseline need for pressors, and diabetes. Longitudinal MAP change was associated with baseline MAP, baseline heart rate, need for pressors, or inotropes. The study recommends considering these parameters when deciding if IABP is the most appropriate form of support for a specific patient. Further prospective studies are needed to validate the findings.

Despite recent advances in cardiovascular therapies, cardiogenic shock (CS) mortality rates remain between 40% and 60%.¹ When CS deteriorates, the use of temporary mechanical circulatory support (tMCS) devices becomes essential for patient survival.^2,3 Temporary mechanical circulatory support devices currently utilized in clinical practice differ in terms of the level of support they can provide, ease of insertion, management, adverse event profile, and risk of hemocompatibility events.⁴ The intra-aortic balloon pump (IABP) is considered the tMCS device with the most favorable adverse event profile,⁵ but after the intra-aortic balloon pump in cardiogenic trial failed to show an improvement in cardiovascular mortality in patients with myocardial infarction and CS,⁶ ST elevation myocardial infarction guidelines have downgraded its indication.^7,8 On the other hand, IABP is still widely utilized to treat chronic heart failure (HF) patients with CS, where previous studies^9,10 on acute response have shown that IABP could be particularly effective in a subgroup of “super-responder” patients, in whom a robust acute hemodynamic improvement is observed. Although IABPs have been utilized for decades,¹¹ the understanding of which patient characteristics are associated with a positive hemodynamic response is still limited and, at the moment, it is not possible to determine a priori (i.e., before inserting the IABP) if and to what degree a given patient will respond. Thus, it is essential to develop predictive algorithms to improve patient selection and estimate the magnitude of IABP effect in individual patients. Accordingly, this study aimed to quantitatively assess the hemodynamic improvement (defined as the cardiac index [CI] and mean arterial pressure [MAP] increase from the baseline value before the IABP placement) during the entire duration of IABP support and identify which a priori patient characteristics are associated with these improvements.

Materials and Methods

Study Population

A retrospective single-center review of all adult patients who received an IABP between February 2016 and August 2022 was conducted. We excluded patients in whom IABP was placed solely as a temporary support during an interventional cardiology procedure and removed at the end of the case, or patients with missing CI assessment after the IABP insertion. Patients were followed until the IABP was removed due to either patient recovery, transition to other MCS support, heart transplantation, or death. The Albert Einstein College of Medicine institutional review boards approved this study.

Data Collection and Outcome of Interest

Demographics, baseline hemodynamics, laboratory values, IABP pressure values, medication use, and past medical history were extracted from our electronic medical record system using a proprietary software (Clinical Looking Glass Clinical Analytics—Streamline Health, Atlanta, GA). Longitudinal variation of CI values from baseline CI (ΔCI_t) and MAP values from baseline MAP (ΔMAP_t) while on IABP support were the outcome of interest. We utilized Fick’s formula (Supplemental Material, http://links.lww.com/ASAIO/B72) to calculate each CI value. To be consistent with previous literature,⁹ we defined super-responders as patients who exhibit a response (increase of CI or MAP) in the top quartile.

Exposure and Confounders

Definitions and formulas utilized to calculate the hemodynamic variables are presented in the Supplemental Material (http://links.lww.com/ASAIO/B72). We used clinical and hemodynamic variables to develop the predictive model for hemodynamic changes during IABP support (variables are listed in the Supplemental Material, http://links.lww.com/ASAIO/B72; Table 1; and Supplemental Table 1, http://links.lww.com/ASAIO/B72). CS was defined as the presence of all the following criteria: systolic blood pressure (BP) less than 90 mmHg or utilization of supportive medications to maintain a systolic BP and CI ≤ 2.2 L/min/m² with concomitant wedge pressure ≥ 15 mmHg^6,12 without inotropes.

Table 1.

Baseline Characteristics of Study Participants at the Moment of IABP Insertion and Results of Multiple Imputation

Variables	Before MI				After MI
	Mean	SD	N	% Missing	Mean	SE	95% Confidence Interval
Demographics
Age (years)	62.4	13.7	336	0	62.4	0.75	61.0–63.9
Male (%)	64.0		336	0	64.0		58.8–69.1
BMI (kg/m2)	28	6.5	336	0	28	0.35	27.3–28.7
BSA (m2)	1.9	0.3	336	0	1.9	0.02	1.9–1.9
Parameters at presentation
ACS (%)	46.7		336	0	46.7		41.4–52.1
Classic cardiogenic shock (%)	29.7		336	0	29.7		24.8–34.7
Reduced cardiac index* (%)	23.8		336	0	23.8		19.2–28.3
SBP (mmHg)	114.8	21.1	322	4.2	114.9	1.19	112.5–117.2
DBP (mmHg)	70.9	13.4	322	4.2	70.8	0.74	69.3–72.3
MAP (mmHg)	85.5	14.8	322	4.2	85.5	0.83	83.9–87.1
Oxygen saturation (%)	97† (94–100)		322	4.2	98†	0.29	97.7–98.2
Heart rate (bpm)	98.6	20.4	329	2.1	98.5	1.13	96.3–100.7
CVP (mmHg)	14.6	7.6	209	37.8	12.9	0.48	12.0–13.9
PAS (mmHg)	49.3	15.2	240	28.6	48.9	0.85	47.3–50.6
PAD (mmHg)	27.1	9.2	239	28.9	27.1	0.51	26.1–28.1
PAM (mmHg)	34.5	10.5	239	28.9	34.4	0.58	33.3–35.5
PCWP (mmHg)	25.6	10.2	197	41.4	25.8	0.59	24.6–27.0
Cardiac index (L/min/m2)	1.9	0.6	237	29.5	1.9	0.04	1.8–2.0
Stroke volume (ml)	36.6	16.1	237	29.5	39.5	1.13	37.3–41.7
PVR (wood)	2.6	1.5	184	45.2	2.6	0.11	2.3–2.8
SVR (dyne s/cm5)	1824.6	789.9	197	41.4	1789.6	46.9	1697.3–1882.0
TPG (mmHg)	9.1	5.0	194	42.3	8.6	0.35	7.9–9.3
PAPi	1.59† (1.07–2.33)		206	38.7	1.77†	0.41	1.62–1.93
Pressor use before IABP (%)	28.6		336	0	28.6		23.7–33.4
Inotrope use before IABP (%)	33.3		336	0	33.3		28.3–38.4
Past medical history
Ischemic cardiomyopathy (%)	58.9		336	0	58.9		53.6–64.2
Diabetes mellitus (%)	48.5		336	0	48.5		43.1–53.9
Hypertension (%)	52.1		336	0	52.1		46.7–57.5
CKD (%)	44.3		336	0	44.3		39.0–49.7
Heart failure (%)	77.4		336	0	77.4		72.9–81.9
Ejection fraction (%)	28† (20–42)		325	3.3	25†		24.3–25.7

^*Patients with CI < 2.2 L/min/m² but not meeting other shock criteria.

^†Median.

ACS, acute coronary syndrome; BMI, body mass index; bpm, beats per minute; BSA, body surface area; CI, cardiac index; CKD, chronic kidney disease; CVP, central venous pressure; DBP, diastolic blood pressure; IABP, intra-aortic balloon pump; MAP, mean arterial pressure; MI, multiple imputation; PAD, pulmonary artery diastolic pressure; PAM, pulmonary artery mean pressure; PAPi, pulmonary artery pulsatility index; PAS, pulmonary artery systolic pressure; PCWP, pulmonary capillary wedge pressure; PVR, pulmonary vascular resistance; SBP, systolic blood pressure; SD, standard deviation; SE, standard error; SVR, systemic vascular resistance; TPG, trans-pulmonary gradient.

Statistical Methods

The continuous data are presented as mean ± standard deviation (SD) or median (25–75 interquartile range), while categorical data are presented as a frequency count and percentages. The continuous scaled variables were assessed for normality. The associations between pre-IABP insertion variables and the trajectory of CI (ΔCI_t) and MAP (ΔMAP_t) over time during IABP support were evaluated using linear mixed-effects models (LMEMs) as described in the Supplemental Material (http://links.lww.com/ASAIO/B72). If not otherwise specified, a p value of less than 0.05 was set as a threshold for statistical significance.

Missing Data

We observed missing information in patient characteristics. Percentages of missing data are presented in Table 1. Missing data for baseline covariates were imputed as described in the Supplemental Material (http://links.lww.com/ASAIO/B72). Statistical analysis was performed with Stata (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC).

Results

Study Participants

Of a total of 552 patients who received an IABP in the study period, 197 patients were excluded from the analysis because IABP was placed as temporary support during an interventional cardiology procedure, and 19 patients were removed because they did not have any CI value assessment after IABP implantation. Three hundred thirty-six patients with 2,196 time-point CI assessments were included in the study. The baseline characteristics of the patients are presented in Table 1. The mean (±SD) age was 62.4 ± 13.7 years, 215 (64.0%) were male, and body surface area (BSA) was 1.9 ± 0.29 m². At the time of IABP insertion, despite the majority of the patients had a CI less than 2.2 L/m/m² (180, 53.8%), only a third of the patients formally met the all the three classic hemodynamic CS criteria (100; 29.7%) and nearly half (157; 46.7%) had an acute coronary syndrome (ACS). The mean CI before IABP insertion was 1.9 ± 0.6 L/min/m² and mean MAP was 86 ± 15 mmHg. Limited to patients meeting all three classic hemodynamic shock criteria only, the mean baseline CI was 1.6 ± 0.3 L/min/m², mean baseline MAP was 84 ± 15 mmHg, and the wedge pressure was 28.3 ± 8.8 mmHg. In the full cohort, 94 (28.6%) patients were on vasopressors and 109 (33.3%) patients were on inotropes at the time of IABP insertion. At presentation, patients had on average increased filling pressures on both the right (central venous pressure [CVP] 14.6 ± 7.6 mmHg) and left side (wedge pressure 25.6 ± 10.2 mmHg), a reduced Pulmonary Artery Pulsatility index (PAPi) with a median value of 1.59 (1.07–2.33), and significantly increased systemic vascular resistance, with a mean of 1825 ± 790 dyne s/cm⁵. Ejection fraction (EF) was overall severely reduced, with a median value of 28% (20–42%). Two hundred sixty (77.4%) patients had a previous history of HF, 198 (58.9%) ischemic cardiomyopathy, 175 (52.1%) arterial hypertension, 163 (48.5%) diabetes mellitus (DM), and 149 (44.3%) chronic kidney disease. The median duration of IABP support was 2.9 days (1.8–4.9 days). There was a trend difference in the duration of support between patients experiencing CS due to ACS and those due to HF decompensation (2.3 days [1.7–4.1 days] vs. 4.6 days [1.9–5.1 days]; p = 0.06). Patients had a median of 5 CI (2–9 CI) assessments during IABP support. The most utilized balloon sizes were 50 ml (49.6%) and 40 ml (46.8%); a small percentage of patients received a 25 ml (2.9%) or 30 ml (0.7%) balloon. Two hundred (59.5%) patients received inotropes and 203 (60.4%) received vasopressors during IABP support.

Acute Changes in CI and MAP After IABP Insertion

Two hundred ninety-three (87.2%) patients had at least one CI and MAP measurement obtained in the first 24 hours. The acute (within the first 24 hours during IABP support) changes of CI and MAP following IABP insertion are presented in Figure 1, A and B, respectively. Data were collected on average 7.3 ± 6.4 hours after IABP insertion. In the first hours post-IABP insertion, CI increased on average by 0.44 ± 0.82 L/min/m² while MAP decreased by 3.4 ± 19.4 mmHg. In shock patients, CI increased by 0.65 ± 0.74 L/min/m² while MAP decreased by 3.8 ± 19.9 mmHg. Univariable and multivariable regression analyses to predict the acute change in CI and MAP are reported in Supplemental Tables 2 and 3 (http://links.lww.com/ASAIO/B72). In brief, variables independently associated with acute CI change were the presence of ACS (β = –0.25; p = 0.017), baseline CI (β = –0.66; p < 0.001), baseline CVP (β = –0.02; p = 0.003), baseline wedge (β = 0.01; p = 0.041), EF (β = 0.01; p = 0.003), and use of pressor during IABP support (β = 0.24; p = 0.019). Baseline MAP (β = –0.82; p < 0.001), baseline heart rate (HR) (β = –0.18; p = 0.001), and pressor use during IABP support (β = –4.39; p = 0.026) were instead independently associated with acute MAP change.

Figure 1.

Acute changes in cardiac index and mean arterial pressure following IABP insertion. A: Acute change in cardiac index (L/min/m²) after IABP insertion. Each vertical line represents a single patient. The red dotted line represents the top quartile (super-responders). B: Acute change in mean arterial pressure (mmHg) after IABP insertion. Each vertical line represents a single patient. The red dotted line represents the top quartile (super-responders). IABP, intra-aortic balloon pump.

Longitudinal Changes in CI and MAP During IABP Support

The longitudinal changes of CI and MAP throughout the entire IABP support duration are summarized in Figure 2, A and B. In the study cohort, CI rapidly improved during the first two days. Following this steep increase, CI continued to rise in the model at lower rate until it stabilized between days 7 and 8, slowly declining thereafter. On average, MAP change remained negative in the model until day 5, reached its maximum increase ~ day 8, and then slowly declined thereafter.

Figure 2.

Longitudinal changes in cardiac index and mean arterial pressure during IABP support. A: Adjusted predictions with 95% confidence interval of mean longitudinal increase in cardiac index during IABP support. B: Adjusted predictions with 95% confidence interval of mean longitudinal increase in mean arterial pressure during IABP support. IABP, intra-aortic balloon pump.

Association Between the Longitudinal Change of CI and Covariates

Univariable and multivariable LMEM model estimates are presented in Supplemental Table 4 (http://links.lww.com/ASAIO/B72) and Table 2. The final LMEM to predict the change of CI over time (ΔCI_t) included the duration of IABP support (time and time²), age, sex, BSA, baseline CI, baseline PAPi, history of DM, inotrope use, and pressor use. Based on the fit of our model, we were able to create the following formula to estimate ΔCI_t.

Table 2.

Final Multivariable Model to Predict CI Change Over Time

Variables	Coefficient	SE	p	95% Confidence Interval
Support duration (per 1 day increase)	0.116	0.027	<0.001	0.063–0.169
Support duration2 (per 1 day2 increase)	–0.008	0.002	<0.001	–0.012 to –0.003
BSA (per 1 m2 increase)	–0.399	0.127	0.002	–0.649 to –0.15
Baseline cardiac index (per 1 L/min/m2 increase)	–0.748	0.06	<0.001	–0.867 to –0.629
Baseline PAPi (per 1 unit increase)	0.019	0.008	0.026	0.002–0.035
Diabetes mellitus	0.204	0.067	0.002	0.072–0.336
Pressor use during IABP	0.144	0.054	0.007	0.04–0.249
Constant	2.657	0.316	<0.001	2.039–3.276

BSA, body surface area; CI, cardiac index; IABP, intra-aortic balloon pump; PAPi, pulmonary artery pulsatility index; SE, standard error.

$${\displaystyle \begin{array}{c}{\displaystyle \begin{array}{c}{\displaystyle \mathrm{\Delta }\mathrm{C}{\mathrm{I}}_t\left(\frac{\frac{\mathrm{l}}{\mathrm{m}\mathrm{i}\mathrm{n}}}{{\mathrm{m}}^2}\right)=2.7+0.12\times t-0.01\times t^2-0.40}\\ {\displaystyle \times \mathrm{B}\mathrm{S}\mathrm{A}-0.75\times \mathrm{C}{\mathrm{I}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}+0.02\times \mathrm{P}\mathrm{A}\mathrm{P}{\mathrm{i}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}}\\ {\displaystyle +0.2\left(\mathrm{i}\mathrm{f}\mathrm{D}\mathrm{M}\right)+0.14\left(\mathrm{i}\mathrm{f}\mathrm{v}\mathrm{a}\mathrm{s}\mathrm{o}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{s}\right)}\end{array}}\end{array}}$$

Where t represents the duration of IABP support (in days), CI_baseline and PAPi_baseline the CI and the PAPi values before the IABP insertion, respectively (given the small value of the age coefficient, it was omitted from the formula). According to our model, the increase of CI over time followed parabolic behavior, with its peak after ~ 8 days of support ($\frac{d\mathrm{\Delta }\mathrm{C}{\mathrm{I}}_t}{dt}=0$ for t = 7.7 days). The increase of CI was negatively affected by BSA (β = –0.40; p = 0.002) and baseline CI before IABP insertion (β = –0.75; p < 0.001). PAPi value before IABP insertion (β = 0.02; p = 0.026), history of DM (β = 0.20; p = 0.002), and vasopressor use (β = 0.14; p = 0.007) had a positive independent association with the post-IABP CI change.

Association Between the Change of MAP Over Time and Covariates

Univariable and multivariable LMEM model estimates are presented in Supplemental Table 5 (http://links.lww.com/ASAIO/B72) and Table 3. The final LMEM to predict the change of MAP over time (ΔMAP_t) included the duration of IABP support (splines), age, sex, BSA, baseline MAP, baseline HR, inotrope use, and pressor use. Based on the fit of our model, we were able to create the following formula to estimate ΔMAP_t.

Table 3.

Final Multivariable Model to Predict MAP Change Over Time

Variables	Coefficient	SE	p	95% Confidence Interval
Support duration, before 8 days (per 1 day increase)	1.136	0.285	<0.001	0.577–1.695
Support duration, after 8 days (per 1 day increase)	–2.175	0.513	<0.001	–3.180 to –1.169
Baseline MAP (per 1 mmHg increase)	–0.843	0.040	<0.001	–0.921 to –0.765
Baseline heart rate (per 1 bpm increase)	–0.086	0.030	0.005	–0.145 to –0.026
Pressor use during IABP	–3.986	0.867	<0.001	–5.685 to –2.286
Inotrope use during IABP	–2.209	0.935	0.018	–4.042 to –0.376
Constant	81.537	5.578	<0.001	70.604–92.470

IABP, intra-aortic balloon pump; MAP, mean arterial pressure; SE, standard error.

$${\displaystyle \begin{array}{c}{\displaystyle \begin{array}{c}{\displaystyle \mathrm{\Delta }\mathrm{M}\mathrm{A}{\mathrm{P}}_t\left(\mathrm{m}\mathrm{m}\mathrm{H}\mathrm{g}\right)=82+1.14\times t^{\ast }-2.18\times t_{+}^{\ast }-0.84}\\ {\displaystyle \times \mathrm{M}\mathrm{A}{\mathrm{P}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}-0.09\times \mathrm{H}{\mathrm{R}}_{\mathrm{b}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{e}}-3.99\left(\mathrm{i}\mathrm{f}\mathrm{o}\mathrm{n}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{s}\right)}\\ {\displaystyle -2.21\left(\mathrm{i}\mathrm{f}\mathrm{o}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{p}\mathrm{e}\mathrm{s}\right)-1.39\left(\mathrm{i}\mathrm{f}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e}\right)}\end{array}}\end{array}}$$

Where t^* represents the duration of IABP support (in days), t₊^* is the duration of IABP after 8 days, MAP_baseline and HR_baseline the MAP and the HR values before the IABP insertion, (given the small value of age coefficient, it was omitted from the formula). The increase of MAP was negatively affected by baseline MAP before IABP insertion (β = –0.84; p < 0.001), HR value before IABP insertion (β = –0.09; p = 0.005), need for vasopressors (β = –3.99; p < 0.001), or inotropes (β = –2.21; p = 0.018) during support.

Discussion

In this study, we analyzed the temporal evolution of hemodynamics during the entire period of IABP support using a linear mixed-effect regression models. We aimed to identify patient characteristics associated with longitudinal improvement in CI and MAP. The main findings of our analysis are the following:

1. The hemodynamic response to IABP is variable and CI and MAP continued to evolve in our cohort throughout the duration of IABP support;

2. Low baseline CI, low BSA, higher baseline PAPi, the presence of DM, and the need for vasopressors are associated with a more robust increase in CI during the IABP support; and

3. Reduced baseline MAP, low baseline HR, and absence of vasopressor/inotrope need are associated with a more significant increase in MAP during IABP support.

Unlike the intra-aortic balloon pump in cardiogenic (SHOCK II) trial, our investigation is not an outcome study, but rather a physiology analysis of the interaction between the human body and the IABP. Our results are the most comprehensive to date supporting the notion of “IABP super responders,” a phenomenon well-known to practicing clinicians and evidenced by the ongoing clinically successful IABP use in many centers.⁹ Despite this, many cardiac catheterization operators have switched to the a priori deployment of what appear to be the—on average—more powerful devices, such as Impella. The Impella continuous-flow pumps provide the additional advantage of generating flow in the absence of effective cardiac contraction. Despite these clear advantages, the adverse event profile of Impella devices is inferior to that of IABP.⁵

After the SHOCK II trial failed to show a mortality improvement in patients with acute myocardial infarction complicated by CS,⁶ investigations have focused on the use of IABP outside of acute ischemic cardiomyopathy etiology.^9,13,14 These studies, analyzing the change in CI cross-sectionally (single time point) showed—on average—an acute increase in CI in the order of 0.2–0.5 L/min/m² in the first hours after the IABP placement. They also showed a very heterogenous response to IABP in decompensated HF patients, suggesting the existence of a “super-responder” patient phenotype¹⁰ that could benefit the most from IABP utilization. Similarly, we also observed a subgroup of acute “super-responders” in whom CI and/or MAP significantly improved in the first hours following the IABP insertion. However, given the complexities of shock physiology including for example dynamic changes of systemic vascular resistance after or in the acute phase of CS complicating myocardial infarction,¹⁵ we decided to investigate the change of CI and MAP over time during the “entire” duration of IABP support, using LMEMs. These models, specifically designed to characterize the change in the response of interest over time,¹⁶ unveiled a dynamic behavior of CI and MAP during IABP support. Indeed, we observed that the hemodynamic response (both in term of CI and MAP increase) continued to improve in the first days of IABP support, peaking in our model after 7–8 days. Considering that the level of IABP support did not change, this dynamic evolution could be explained by the resolution of the SIRS and vascular dysfunction in the days following IABP insertion well described in the setting of ST elevation myocardial infarction.¹⁵

Acute Response

In our initial analysis limited to the acute (within hours from IABP insertion) response, we identify the presence of ACS, high baseline CI, and high baseline CVP had a negative effect on the acute CI increase, while high baseline wedge, high EF, and the use of pressor during IABP support were associated to a more prominent acute increase of CI.

Patients with higher baseline MAP, baseline HR, and pressor use during IABP support experienced a reduced (or negative) change of MAP in the first hours post IABP.

Identifying Longitudinal “Super-Responders”

As following step, we examined which baseline characteristics were associated with the longitudinal sustained hemodynamic response to IABP. Regarding absolute CI improvement during IABP support, we found that the “baseline” value of CI was independently and inversely associated with the improvement of CI during IABP support. Baran et al.¹⁷ previously suggested that patients with high baseline CI were less likely to respond to IABP therapy; our data corroborates and further provides a quantification of this effect. Second, we found that patients with a low BSA experienced a more significative increase of CI, suggesting that IABP efficacy might be blunted in obese patients; this might simply be a function of relatively larger balloon to body size ratio. Third, we observed a linear relationship between baseline PAPi and the magnitude of CI improvement. IABPs (like the majority of other tMCS devices) are designed to support the left ventricle function, but they operate in a more complex system that also involves the right ventricle (RV), the pulmonary artery vasculature, and the aorta. As the left ventricle (and consequently the IABP) blood flow relies on the volume of blood received from the RV, it is reasonable to think that the performance of RV might impact the IABP effect. Our analysis showed that baseline PAPi was independently associated to an additional increase in CI, with a factor of ~ 0.02 L/min/m² per each point increase of baseline PAPi. Although this number may seem relatively small, it translates into a not negligible ~ 0.1 L/min/m² absolute difference in CI (or a cardiac output difference of 200 ml/min if BSA is 2.0), when comparing—for example—a patient with a depressed RV function (PAPi < 1) to a patient with robust RV function (PAPi = 5). This difference in CI increase based on PAPi may also explain why a better 30-day mortality is seen in IABP patients with higher baseline PAPi.⁹ Moreover, Malick et al.¹⁸ suggested that the superior IABP response observed in the acute-on-chronic decompensated HF patients compared ACS patients could be explained by a higher baseline PAPi in the acute-on-chronic decompensated HF group, while another study—limited by the very small sample size—showed a trend-association between PAPi and IABP response.¹⁹ Fourth, the presence of DM was associated to an additional CI increase of 0.2 L/min/m², suggesting that DM patients might benefit more than non-DM patients from IABP therapy. One might speculate that this finding could be explained by the increased aortic stiffness typical of DM patients.^20–22 Physiologically speaking, it is reasonable to imagine that the vacuum effect (afterload reduction) generated by the IABP inflation/deflation is more prominent if the balloon is placed in a stiffer vessel, where the energy delivered by the inflation/deflation is primarily converted into a pressure change, rather than in a more compliant aorta, where the pressure changed is converted into elastic energy by the vessel distension. Lastly, we found that IABP was more effective in patients who required a concomitant vasopressor support, with an additional CI increase of 0.14 L/min/m² associated with vasoactive medication use.

Limitations and Strengths

A few limitations should be acknowledged: first, given the retrospective nature of the study and the acuity of clinical condition, a significant number of patients had echocardiographic data acquired after initiation of IABP support. Consequently, we could not incorporate variables pertaining to the severity of mitral valve regurgitation, which could be influenced by afterload reduction. Second, although our models included the binary presence of vasopressors and inotropes, we could not account for changes in medication dosage due to insufficient data availability Third, due to the study’s retrospective nature, we used Fick’s formula to compute CI, which could have diminished result accuracy. Nevertheless, this approach aligns with current clinical practice and previous research. Additionally, alternative methods such as thermodilution pose challenges in patients experiencing CS, where valvular regurgitation is likely present. Fourth, this study is based on mathematical and statistical assumptions, simplifications, and projections; the interpretation of results should always be paired with clinical assessment. In particular, although the predicted temporal trends were similar to the real observed data and the robustness of the LMEM, interpretation of these results in case of prolonged support duration should be done with caution, given the decreasing number of participants. Fifth, since this study has a single-arm design, these findings are applicable under the assumption that any physiological changes observed can be solely attributed to the utilization of an IABP. Lastly, a validation study, potentially in a prospective fashion, is necessary to confirm these initial findings.

In addition to the novelty of longitudinal analysis of CI and MAP, other elements of this study are noteworthy: First, we decided to include all the patients who received an IABP, whether in shock or not and regardless of the IABP indication (acute on chronic HF, ACS with CS, or coronary perfusion support). This heterogeneity of patients allowed us to analyze a wide range of CI and MAP changes and generalize the results beyond a specific subgroup. Our investigation is a physiological study, aimed at identifying which variables were associated with a successful longitudinal hemodynamic response in order to improve our understanding of the hemodynamic effects of IABP support. Interestingly, in our final longitudinal models, indications for IABP did not have a significant association with CI, MAP change, or their interactions. It is important to highlight that while the presence of ACS negatively influenced the acute improvement of CI, this effect was no longer observed in the longitudinal analysis. Second, unlike previous studies where patients without a complete set of baseline variables were excluded, potentially creating a selection bias for sicker patients, we decide to mitigate this issue by utilizing our entire population and imputing the missing baseline data. As a result, and to the best of our knowledge, our analysis has the largest patient population for this type of study. Third, although our analysis is a mathematical model based on statistical assumptions prone to simplifications and generalization, we were able to numerically predict the change in CI and MAP, differently from other studies where arbitrary cutoffs to define “response” were studied.

Conclusions

CI and MAP changes during IABP support are not constant but vary with time and are associated with baseline parameters. These parameters should be considered when deciding if IABP is the most appropriate form of support for a specific patient. Further prospective studies are needed to validate our findings.

Acknowledgments

The authors thank Kathy Vandervoort and Kasra Jabbary Moghaddam for their assistance with the data extraction.