Comparative Effectiveness of Suction Thrombectomy vs. Catheter-Directed Thrombolysis in Intermediate-Risk Pulmonary Embolism

Abstract

OBJECTIVES:

Catheter-directed thrombolysis (CDT) has been shown to rapidly reverse hemodynamic and echocardiographic abnormalities seen in intermediate-risk pulmonary embolism (IRPE). Suction thrombectomy (ST) devices have emerged as alternative treatment modalities demonstrating immediate results, obviating the need for thrombolytics. Comparative data between the two methods are sparse.

METHODS:

We retrospectively reviewed interventions for IRPE (CDT or ST) at a multihospital healthcare system (2017–2022). IRPE was defined by evidence of right heart strain (RHS) on imaging (echocardiogram and/or CT-angiography) or elevated biomarkers (troponin or B-natriuretic peptide). Patients with high-risk PE (systolic blood pressure<90 mm Hg) or those who received systemic thrombolytics were excluded. The primary endpoint was a composite of 7-day all-cause mortality, intracranial bleeding, major bleeding, clinical deterioration and/or need for bailout therapy. Secondary outcomes included the primary endpoint individual components, intensive care unit length-of-stay (ICULOS), 30-day mortality and resolution of RHS at 3 months. Inverse probability of treatment weighting (IPTW) was used to adjust for baseline imbalances between groups, generating weighted odds ratios (wOR).

RESULTS:

332 patients were included with 152 CDT and 180 ST. IPTW successfully balanced baseline differences between groups (Table I). On univariable analysis, the primary outcome did not differ between groups (CDT 6.6% vs ST 12.8%; p=0.06), but ST was associated with increased 30-day mortality (CDT 1.3% vs ST 5.5%, p=0.039) and the need for bailout intervention (CDT 4.6% vs ST 11.1%, p=0.031). Major bleeding occurred in 3.3% CDT vs 2.2% ST (p=0.551). There were no intracranial bleeds. Post-IPTW analysis showed a significantly higher rate of the primary outcome in the ST group (wOR 4.4, 95% CI [1.27–15.3], p=0.02). There were no differences in 7-day mortality, 30-day mortality, major bleeding or intracranial bleeding. The need for bailout intervention was significantly higher in the ST group (wOR 3.7, 95% CI [1.04–13.4], p=0.044). The use of ST was significantly associated with resolution of RHS (wOR 3.46, 95% CI [1.32–9.11], p=0.012).

CONCLUSION:

ST is associated with significantly increased odds of the primary outcome in patients with IRPE when compared to CDT after IPTW. These results were mainly driven by the bailout intervention rate. There was no statistically significant reduction in major bleeding or ICULOS compared to CDT. RHS resolved more frequently in patients after ST suggesting there may be benefit to rapid thrombus removal in appropriately selected IPRE patients despite the increased need for bailout therapy. CDT should remain an integral part of the interventional armamentarium in IRPE.

Table of Contents Summary

After IPTW, ST was associated with a higher rate of the composite outcome (7-day all-cause mortality, major bleeding, clinical deterioration, need for bailout intervention) and resolution of right heart strain compared to CDT in this retrospective cohort of 332 patients. CDT should remain an integral part of the armamentarium in IRPE.

INTRODUCTION

An estimated 100,000 Americans die each year from PE, 1 in 4 of whom die suddenly (1). According to societal guidelines, PE can be risk-stratified by severity from low-risk, intermediate-risk (or submassive) to high-risk (or massive). Intermediate-risk PE (IRPE) is defined as PE without hemodynamic instability, but with evidence of right ventricular (RV) dysfunction (i.e. right heart strain [RHS]) on CT angiography (CTA) or echocardiogram combined with biomarker evidence of myocardial necrosis with elevated troponin and/or right-sided heart failure with elevated brain natriuretic peptide (BNP)(2).

IRPE has traditionally been treated with anticoagulation alone. However, decompensation risk still ranges from 2–18% (3–7). Emerging evidence suggests that invasive treatment of IRPE, with catheter-directed thrombolysis (CDT) or suction thrombectomy (ST), can reduce PE-related 5-year mortality up to 5 years and the incidence of chronic thromboembolic pulmonary hypertension (CTEPH) (8, 9).

However, choice of invasive therapy is unclear. CDT has been associated with reversal of hemodynamic and echocardiographic abnormalities but requires ICU monitoring and can be contraindicated in those with bleeding diathesis. Conversely, ST provides quicker results and obviates the need for thrombolytics. However, it requires much larger bore devices, which can potentially damage the cardiac, pulmonary and peripheral vasculature (10–13) and be associated with substantial blood loss. Overall, real-world comparative data are sparse and the optimal treatment strategy for IRPE patients currently remains poorly defined. Thus, the aim of this study was to compare short-term outcomes of CDT vs ST amongst patients with IRPE.

METHODS

Cohort

The study protocol was approved and exempted from informed consent by the Quality Review Board of the University of Pittsburgh Medical Center (STUDY22110154). We conducted a single-institution multihospital health network retrospective cohort study comparing CDT and ST in IRPE. We included adults > 18 years of age with a confirmed diagnosis of acute PE who underwent catheter-based intervention between 2017–2022. We chose to exclude interventions performed prior to 2017 given the paucity of available off-the-shelf devices at the time. Diagnosis was obtained via ICD-9 or 10 code and confirmed by CTA in all patients. Transfers within our healthcare network were only included once. We excluded patients with missing operative or imaging reports. Low-risk PE, high-risk PE and/or patients who received systemic thrombolytics were then excluded as well. High-risk PE was defined as systolic blood pressure (SBP) on admission < 90 mm Hg, need for vasopressor support prior to intervention and/or cardiac arrest or syncope on presentation. IRPE was defined by imaging evidence of RHS (echocardiogram and/or CTA) and/or elevated biomarkers (troponin or BNP) without any systemic hypotension (SBP > 90 mm Hg). RHS was confirmed by the presence of any of the following findings on echocardiogram: RV anomalies (dilatation, hypertrophy, systolic dysfunction (typically with apical sparing i.e. McConnell’s sign), hypokinesis), elevated RV pressures, interventricular septal flattening, paradoxical septal motion, right atrial enlargement, tricuspid regurgitation, dilated inferior vena cava, elevated pulmonary arterial pressures) or an RV/LV ratio as measured on CTA > 0.9. RV/LV ratio was obtained by measuring the transverse diameter of the RV and the transverse diameter of the LV, excluding the myocardium. Increased biomarkers (troponin, BNP) were defined by the highest high-sensitivity troponin T level in ng/dL and the highest BNP value in ng/dL within the first 24 hours after admission and/or PE diagnosis. Interventions were identified using the Current Procedural Terminology codes (codes 36013, 36014, 36015, 37212, 37201, 37213, 37214, 37187, 37188) and included CDT, ST and ultrasound-assisted CDT.

Variables of Interest and Outcomes

Baseline patient demographics, comorbidities, vital signs on admission, simplified Pulmonary Embolism Severity Index (sPESI) scores, laboratory results and periprocedural data were obtained through the electronic medical records (EMR) system. Primary outcome was defined as the rate of major adverse events (MAE), defined as all-cause mortality, intracranial bleeding, major bleeding, and clinical deterioration and/or need for bailout therapy (at discharge or 7 days, whichever is sooner), and analyzed as a composite outcome. Secondary outcomes included the individual components of the primary outcome, intensive care unit length-of-stay (ICULOS) in days, 30-day mortality and resolution of RHS at 3 months.

Major bleeding was defined by the International Society of Thrombosis and Haemostasis (ISTH): Fatal bleeding, symptomatic bleeding in a critical area or organ (intracranial, intraspinal, intraocular, retroperitoneal, intra-articular, pericardial or intramuscular causing compartment syndrome), bleeding with decrease in hemoglobin ≥ 2g/dL or bleeding requiring transfusion of ≥ 2 units of packed red blood cells were considered major bleeding events. Clinical deterioration was defined as the need for vasopressor support either intra-operatively or post-operatively for hemodynamic collapse/deterioration. Bailout therapy was defined as requiring additional catheter-directed intervention (CDT and/or ST), surgical embolectomy or extracorporeal membrane oxygenation (ECMO) initiation.

Treatment protocol

All patients were managed per our previously published institutional Pulmonary Embolism Response Team (PERT) team algorithm (8, 9, 14–19). IRPE patients were treated with CDI (CDT vs ST) at the discretion of the interventionalist, which included vascular surgeons, interventional cardiologists, and interventional radiologists. Access for CDT as well as tPA dosage were also left at the discretion of the interventionalist. Internal jugular venous access was utilized preferentially. CDT was performed using standard thrombolysis catheters (Cragg-McNamara [Medtronic, Dublin, Ireland] or Unifuse [AngioDynamics, Latham, NY]) or ultrasound-assisted thrombolysis (USAT) catheters (EKOS Endovascular System^™ [Boston Scientific, Malborough, MA]). Two catheters were placed in bilateral pulmonary arteries. tPA was typically infused at 0.5 mg/hr in each catheter for a total of 1 mg/hr. Duration of infusion ranged from 8–48 hours. ST was performed using FlowTriever^™ [Inari, Irvine, CA], Indigo^™ CAT-8/CAT-12/CAT-16 [Penumbra, Alameda, CA) or AngioVAC^™ [AngioDynamics, Latham, NY] thrombectomy catheters, based on interventionalist preference.

Statistical analysis

Categorical variables were reported as counts and percentages. Continuous variables were reported as means +/− standard deviation (SD) or medians with interquartile range (IQR), depending on the normality of distribution. Univariable comparisons between both groups were performed using Fisher’s exact and χ² tests as appropriate for categorical variables. Continuous variables were compared using t test or Wilcoxon rank sum test, depending on the normality of distribution. Outcome analysis was performed using inverse probability of treatment weighting (IPTW) to adjust for inherent differences between both groups. Pre-IPTW outcomes were evaluated using Fisher’s exact and χ² tests for binary outcomes, as appropriate. Post-IPTW outcomes were then evaluated with multivariable logistic regression, generating weighted odds ratios (wOR) with 95% confidence intervals (95% CI). For continuous outcome variables, linear regression was performed generating predicted means with 95% CI. This was performed by adjusting for all covariables with a standardized mean difference (SMD) of less than < 0.2 or variance < 0.5 or > 2 after IPTW, as those were considered inappropriately balanced by IPTW (20–23). Patients were weighted according to treatment type (CDT vs ST), age in years, gender, race, diabetes, hypertension, hyperlipidemia, cancer, coronary artery disease (CAD), congestive heart failure (CHF), non-coronary heart disease, chronic obstructive pulmonary disease (COPD), chronic kidney disease (CKD), end-stage renal disease (ESRD), stroke, hemorrhagic stroke, hypercoagulable disorders including history of VTE, antiphospholipid syndrome and heparin-induced thrombocytopenia (HIT), liver disease, pulmonary arterial disease, smoking status, anticoagulation on admission, recent surgery within 90 days and vital signs on admission.

Post-hoc analysis

To assess the effect of clinical deterioration and need for bailout therapy on the primary outcome, we performed two separate post-hoc analyses. We first re-examined the primary outcome after excluding clinical deterioration, thus including only all-cause mortality, intracranial bleeding, major bleeding, and need for bailout therapy (at discharge or 7 days, whichever is sooner). We then re-examined the primary outcome excluding both clinical deterioration and need for bailout therapy, only including all-cause mortality, intracranial bleeding and major bleeding.

A p-value of < 0.05 was considered statistically significant. Statistical analysis was performed using StataSE 17 (StataCorp, College Station, TX, April 2021).

RESULTS

7632 adults were diagnosed with acute PE at our institution between 2017–2022, of which 384 underwent CDI. After exclusion, 332 patients were included in our study (Figure 1) with 152 patients in the CDT group and 180 in the ST group. Patients were 47% female with a median age of 64 years (Table 1). Pulmonary hypertension was more frequent in the CDT group (39% vs 35%, p=0.007). 19% of patients in the ST group had undergone surgery within 90 days prior to their CDI (vs. 10% CDT, p=0.015). Overall, IPTW successfully balanced baseline differences between groups (Table 1). We adjusted for imbalanced variables (CHF, CAD, non-coronary cardiac disease, unknown smoking status, liver disease and anticoagulation on admission) in our post-IPTW analysis. There were no significant differences in pre-operative physiological characteristics on admission (Table 2) or baseline biomarkers (Table 3) between both treatment groups. Most ST interventions (92%) were performed using the FlowTriever^™ device (Table 4). As expected, estimated blood loss was greater in the ST group compared to CDT with a median of 100 cc (vs 0 cc CDT, p<0.001) (Table 4). Bailout intervention rates remained low overall, with only 3 patients requiring ECMO and 2 patients requiring surgical embolectomy in the entire cohort (Table 5).

Figure 1.

Flow diagram demonstrating identification of patients with intermediate-risk pulmonary embolism (IRPE) treated with catheter-directed intervention (CDT). DVT, Deep venous thrombosis; PE, pulmonary embolism; RHS, right heart strain; ST, suction thrombectomy; tPA, alteplase.

Table 1.

Baseline demographics

	PRE-IPTW				POST-IPTW
	CDT	ST	p-value	SMD**	CDT	ST	SMD**
BASELINE DEMOGRAPHICS	n=152	n=180			n=152	n=180
AGE (years)	62 (51–71)	64 (54–73)	0.12	0.199	63 (14)	63 (14)	0.01
FEMALE	76 (50)	79 (44)	0.27	−0.123	73 (48)	79 (44)	0.054
Race			0.44
WHITE	135 (89)	154 (86)		−0.098	127 (85)	156 (86)	0.046
BLACK	11 (7)	13 (7)		−0.001	18 (12)	13 (7)	−0.138
OTHER	6 (4)	13 (7)		0.143	6 (4)	13 (7)	−0.032
Hypertension	70 (47)	97 (53)	0.15	0.157	79 (52)	96 (53)	0.033
Hyperlipidemia	51 (34)	73 (40)	0.19	0.145	60 (39)	72 (40)	0.069
Diabetes	26 (17)	39 (21)	0.30	0.116	27 (18)	39 (21)	0.088
Cancer	18 (12)	32 (18)	0.13	0.168	29 (19)	32 (18)	−0.044
Congestive Heart Failure	6 (4)	15 (8)	0.10	0.183	5 (4)	15 (8)	0.195
Coronary Artery Disease	18 (12)	25 (14)	0.58	0.061	9 (6)	25 (14)	0.234
COPD	11 (7)	20 (11)	0.23	0.135	8 (5)	20 (11)	0.197
Pulmonary hypertension (including prior PE)	59 (39)	45 (25)	0.007	−0.246	36 (24)	45 (25)	0.03
Smoking status			0.061
Never	87 (57)	88 (49)		−0.167	86 (57)	88 (49)	−0.150
Former	43 (28)	59 (33)		0.098	48 (32)	59 (33)	0.039
Current	20 (13)	21 (12)		−0.008	15 (10)	21 (12)	0.021
Stroke	10 (7)	11 (6)	0.20	−0.019	10 (7)	19 (10)	0.129
Hemorrhagic	1 (1)	2 (1)	0.66	0.048	3 (2)	2 (1)	−0.06
Hypercoagulable disorders *	53 (35)	50 (27)	0.10	−0.153	34 (23)	49 (27)	0.081
Chronic kidney disease	7 (5)	11 (6)	0.55	0.067	7 (5)	11 (6)	0.058
Recent surgery (< 90 days)	15 (10)	35 (19)	0.015	0.273	37 (21)	35 (19)	0.001
Liver disease	2 (1)	5 (3)	0.36	0.103	10 (6)	5 (3)	0.081
Anticoagulation on admission	10 (7)	21 (12)	0.11	0.177	8 (5)	31 (17)	0.198

Pre-IPTW continuous variables reported as median +/− interquartile range (IQR)

Post-IPTW continuous variables reported as mean +/− standard deviation (SD)

Categorical variables reported as frequency (%)

^*Including history of venous thromboembolism (VTE)

^**Standardized mean difference

Standard mean differences (SMD) were reported in lieu of p-values as mandated by IPTW. SMD less than <0.2 were considered appropriately balanced by IPTW.

Table 2.

Pre-operative patient data

	PRE-IPTW				POST-IPTW
	CDT	ST	p-value	SMD**	CDT	ST	SMD**
	n=152	n=180			n=152	n=180
Heart rate (bpm)	104 (93–116)	104 (90–116)	0.55	−0.102	106 (18)	103 (17)	−0.126
Oxygen saturation (%)	95 (93–97)	95 (93–98)	0.94	0.000	95 (4)	95 (4)	−0.059
Respiratory rate (per min)	22 (19–26)	21 (18–26)	0.33	0.000	22 (6)	23 (5)	0.018
Oxygen requirements (L/min)	3 (2–5)	4 (2–6)	0.061	0.153
sPESI score > 0	120 (79)	136 (76)	0.46	−0.081

Pre-IPTW continuous variables reported as median +/− interquartile range (IQR)

Post-IPTW continuous variables reported as mean +/− standard deviation (SD)

Categorical variables reported as frequency (%)

^*Including history of venous thromboembolism (VTE)

^**Standardized mean difference

Standard mean differences (SMD) were reported in lieu of p-values as mandated by IPTW. SMD less than <0.2 were considered appropriately balanced by IPTW.

Table 3.

Pre-operative biomarkers

	CDT	ST	p-value
	n=152	n=180
BNP* (ng/dL)	373 (122–797)	438 (146–1061)	0.27
High-sensitivity troponin T (ng/dL)	30 (67)	37 (66)	0.95

Continuous variables reported as median +/− interquartile range (IQR)

Categorical variables reported as frequency (%)

^*BNP, B-natriuretic peptide

Table 4.

Intra-operative data

	CDT	ST	p-value
	n=152	n=180
Thrombectomy device			<0.001
Penumbra Indigo™	0 (0)	14 (8)
Inari FlowTriever™	3 (2)	165 (92)
AngioDynamics  AngioVAC™	0 (0)	1 (1)
None	149 (98)	0 (0)
Estimated blood loss (cc)	0 (0–0)	100 (0–350)	<0.001

Continuous variables reported as median +/− interquartile range (IQR)

Categorical variables reported as frequency (%)

Table 5.

Bailout interventions

	CDT	ST	p-value
	n=152	n=180
ECMO	1 (0.6)	2 (1.1)	1.000
CDT	0 (0)	3 (1.7)	0.253
ST	1 (0.6)	0 (0)	0.458
Surgical embolectomy	0 (0)	2 (1.1)	0.502

Continuous variables reported as median +/− interquartile range (IQR)

Categorical variables reported as frequency (%)

ECMO, Extracorporeal membrane oxygenation; CDT, catheter-directed thrombolysis; ST, suction thrombectomy

Pre-IPTW analysis

On univariable analysis, the primary outcome did not significantly differ between groups (CDT 6.6% vs ST 12.8%; p=0.06) (Table 6). However, ST was significantly associated with an increased rate of clinical deterioration and/or need for bailout therapy (CDT 4.6% vs ST 11.1%, p=0.031). Major bleeding occurred in 3.3% CDT vs 2.2% ST (p=0.551). There were no intracranial bleeds. The need for transfusion was similar between both groups. Median ICULOS was significant longer in the ST group (1.68 days CDT vs 2.09 days ST, p=0.0008) (Table 7), but there was no difference in the rate of discharge to home vs. facility between both groups (84.7% CDT vs 82.3% ST, p=0.450). There was no difference in 7-day mortality, but ST was associated with increased 30-day mortality (CDT 1.3% vs ST 5.5%, p=0.039). Resolution of RHS was seen in 63% of ST patients vs. 56% of CDT patients (p=0.34).

Table 6.

Outcomes (categorical)

	PRE-IPTW			POST-IPTW
	CDT	ST	p-value	weighted odds ratio (wOR)	95% CI	p-value
OUTCOMES	n=152	n=180
Primary outcome	10 (6.6)	23 (12.8)	0.06	4.4	1.3–15.3	0.02
Secondary outcomes
7-day mortality	1 (0.7)	4 (2.2)	0.38	2.95	0.3–26.2	0.332
30-day mortality	2 (1.3)	10 (5.5)	0.039	0.69	0.1–4.5	0.702
Major bleeding	5 (3.3)	4 (2.2)	0.551	1.86	0.3–12.2	0.517
Transfusion *	7 (4.6)	3 (1.8)	0.196	0.99	0.15–6.54	0.989
Clinical deterioration/need for bailout intervention	7 (4.6)	20 (11.1)	0.031	3.7	1.04–13.4	0.044
Discharge to home	127 (84.7)	133 (82.3)	0.654	3.04	1.27–7.3	0.013
Resolution of RHS** (n=198)	49 (56)	70 (63)	0.336	3.46	1.3–9.1	0.012

Categorical variables reported as frequency (%)

^*Transfusion of 1 or more units of packed red blood cells

^**RHS, right heart strain

Table 7.

Outcomes (continuous)

	PRE-IPTW			POST-IPTW
	CDT	ST	p-value	CDT	ST	p-value
OUTCOMES	n=152	n=180		n=152	n=180
ICULOS *	1.68 (1.00–2.51)	2.09 (1.44–3.18)	0.0008	2.09 (1.81–2.38)	2.78 (2.39–3.17)	<0.0001

Continuous variables reported as median +/− interquartile range (IQR)

^*ICULOS, ICU length-of-stay (reported in days)

Post-IPTW analysis

Pulmonary hypertension was well balanced between both groups following IPTW. Adjusted analysis showed a significantly higher rate of the primary outcome in the ST group (wOR 4.4, 95% CI [1.3–15.3], p=0.02). Of the components, there were no differences in 7-day mortality, 30-day mortality, major bleeding, intracranial bleeding or need for transfusion on weighted analysis. The rate of clinical deterioration/need for bailout therapy was significantly higher in the ST group (wOR 3.7, 95% CI [1.04–13.4], p=0.044). Predicted mean ICULOS remained significantly longer in the ST group (2.09 days CDT vs 2.78 days ST, p<0.0001). Conversely, the use of ST was significantly associated with discharge to home (wOR 3.04, 95% CI [1.27–7.3], p=0.013) and resolution of RHS (wOR 3.46, 95% CI [1.32–9.11], p=0.012).

Post-hoc analysis

Post-hoc analysis was performed on the primary outcome as a composite of MAE without including clinical deterioration and/or need for bailout therapy, as the composite primary outcome was largely driven by these two individual components, both on unweighted and weighted analysis.

First post-hoc analysis excluded both components (clinical deterioration and need for bailout therapy). Once again, the primary outcome did not differ between groups (CDT 3.3% vs ST 4.4%, p=0.8) on unweighted analysis. However, post-IPTW analysis did show a higher rate of the primary outcome in the ST group (wOR=3.96, 95% CI [0.9–17.8], p=0.07), but these results were no longer statistically significant.

Second post-hoc analysis was performed on the primary outcome, but in this instance, after excluding clinical deterioration alone. The primary outcome did not differ between groups (CDT 4.6% vs 6.1%, p=0.6) on unweighted analysis. However, post-IPTW analysis did show a significantly higher rate of the primary outcome in the ST group (wOR=4.5, 95% CI [1.3–15.9], p=0.02).

DISCUSSION

Our single-institution multihospital health network retrospective cohort study comparing CDT and ST in IRPE demonstrated that after adjusting for demographics, comorbidities, and clinical status on presentation, ST is associated with a significantly increased rate of MAE at 7 days. These findings are mainly driven by the increased rate of clinical deterioration and/or need for bailout therapy associated with ST. There was no difference in mortality or major bleeding between groups before or after weighting. There was no intracranial bleeding in either group. ICULOS in days was increased with ST, even after weighting. Despite increased MAE and slightly longer ICULOS, patients who underwent ST had higher rates of discharge to home and significantly improved rates of RHS resolution on 3-month follow-up.

A growing body of evidence suggests that CDT reduces long-term mortality presumably from decreased rates of CTEPH. Ultimately, this may have impacted therapeutic selection in our cohort, namely in patients with known pulmonary hypertension or history of prior PE, who more commonly underwent CDT. However, there are now several catheter-based ST devices available on the market, with the potential advantages of single-session thrombus clearance without thrombolytics. Most recently, Tsukagoshi’s series (24) found that ST was associated with decreased 30-day pulmonary hypertension and 5-year CTEPH rates compared to CDT. Existing industry data reports short-term mortality rates of 0.8–1% with ST, which are comparable to our mortality rates of 0.7–2.2% (10–12).

A 550-patient industry-sponsored prospective, multicenter, randomized controlled trial (RCT) was recently published comparing the Inari FlowTriever^™ device to CDT in acute IRPE (PEERLESS) (25, 26). Mortality rate at 7 days was 0% in the ST group and 0.4% in the CDT group (p=1.00). They did not find a mortality difference at 30 days either (0.4% ST vs 0.8% CDT, p=0.62). Existing real-world, retrospective data comparing ST to CDT, including the industry-sponsored REAL-PE (27) and a TriNetX Data Network (TriNetX, Cambridge, MA) analysis (24), commonly combine high-risk PE with IRPE, thus making results difficult to interpret. Unsurprisingly, mortality rates reported in these studies were higher and ranged from 3–5%. However, in line with our findings, they did not demonstrate any mortality difference between CDT and ST on univariable or adjusted analyses.

The authors of PEERLESS (25) opted for a hierarchical win ratio (WR) approach to analyze their data. The hierarchy was established as follows: 1) all-cause mortality, 2) intracranial hemorrhage, 3) major bleeding, 4) clinical deterioration and/or escalation to bailout therapy, and 5) postprocedural intensive care unit (ICU) admission and LOS. This statistical analytic technique uses paired subjects and establishes a “winner” by comparing ST vs CDT for each outcome sequentially, until one of the subjects experiences the outcome first, at which point they “lose”. The subsequent outcomes are thus not included in the WR analysis. While they did find a WR of 5.01 (p<0.001) in favor of ST, it is reasonable to assume that these findings are at least partially driven by ICU utilization as their 4-component win ratio (excluding postprocedural ICU admission and ICULOS) was no longer statistically significant (WR 1.24, p=0.30). The regulated structure of a randomized clinical trial may also have influenced ICU utilization more favorably than what would be expected in a real-world setting.

Existing industry data reports MAE rates with ST between 1.7 and 4.1%, while our MAE rates were much higher at 6.6%−12.8%. These findings can likely be explained by the real-world nature of our data, as opposed to an industry-sponsored clinical trial with strict inclusion and exclusion criteria. Roughly half of the patients from the existing FLARE and EXTRACT-PE trials had an sPESI score of 1 or greater, while almost 80% of our cohort did. We did demonstrate an increased rate of MAE in our study in the ST group after IPTW. These results are in line with another retrospective single-center trial by Feroze et al (28) which compared CDT vs ST in both high and intermediate risk-PE. ST was associated with increased absolute MAE rates compared to CDT (12% vs 19.6%, p=0.14), although their findings were not significant, possibly due to a smaller sample size.

Our post-hoc analyses demonstrate that the largest contributor to MAE was the increased need for post-operative vasopressor support and rescue catheter-based interventions in ST patients. Very few patients required escalation to ECMO or surgical embolectomy. 5 of the 7 cases requiring bailout intervention mentioned incomplete or unsuccessful thrombectomy with presence of chronic thrombus. This constitutes a known limitation of large-bore ST devices. Interestingly, in PEERLESS (25), their rate of clinical deterioration and/or need for bailout therapy was lower in the ST group (1.8% vs 5.4%, p=0.04). We hypothesize that our findings may be related to physiological and hemodynamic confounders within the ST population which were potentially unaccounted for in our IPTW analysis, given the retrospective nature of our data. In addition, PEERLESS’ strict exclusion criteria may have impacted the generalizability of their results, while our results may be more representative of real-world data. It is also plausible that a learning curve with ST devices may have contributed to an increased bailout rate within our cohort over time, which perhaps may be more pronounced than within the PEERLESS trial.

One theoretical benefit of ST over CDT involves the avoidance of thrombolytic medications, which have the potential to cause fatal bleeding. However, following the widespread adoption of low dose tPA protocols (29), there is now accumulating evidence that the bleeding risk associated with CDT is not as high as previously reported, with meta-analytic data showing thrombolytic therapy (including systemic administration) in IRPE was associated with a 9.24% major bleeding risk, compared to 3.42% with anticoagulation alone. Thrombolysis was also associated with a 1.46% risk of ICH vs 0.19% with anticoagulation alone (30). The KNOCOUT-PE registry (31) was a prospective, industry-sponsored cohort study of acute both high-risk and IRPE patients undergoing USAT using EKOS^™. Similarly, their reported major bleeding rate within 72 hours of intervention was 1.6% with a mean total tPA dose of 18.1 mg. In our study, we reported low overall major bleeding rates (3.3% CDT vs 2.2% ST, p=0.551) with no intracranial bleeding event with either treatment modality. This may be explained by our routine use of low-dose tPA protocols in CDT patients and rigorous patient selection. Despite the anticipated up-front blood loss with ST, particularly with older devices, we did not demonstrate any difference in transfusion requirements. Of note, the FlowSaver^™ device only became available in late 2021 and may not have significantly impacted overall EBL rates in our cohort. In PEERLESS (25), their major bleeding rates were higher than in our series at 6.9% in each group, with an access complication rate causing a hemoglobin drop > 2 g/dL or requiring transfusion of 2.9% in the ST group and 3.6% in the CDT group.

Somewhat unexpectedly, we found that ST patients had significantly longer ICULOS compared to CDT. This remained significant even after weighting. We believe this reflects the reality of real-world clinical practice, outside the constraints of a clinical trial. In our practice, most ST patients were already admitted to the medical ICU (MICU) or had already been assigned a MICU bed upon PERT activation and simply returned to their assigned bed for post-operative monitoring. Otherwise, increased ICULOS was mainly seen in patients with clinical deterioration, need for bailout therapy or with complications unrelated to their IRPE.

Existing data comparing bleeding rates amongst CDT and ST are mixed, and most stem from studies that included both intermediate and high-risk PE. The REAL-PE study (27) compared USAT using EKOS^™ to ST using FlowTriever^™ within the Truveta (Truveta Inc, Bellevue, WA) database. Their contemporary analysis showed more frequent major bleeding and ICH with ST, with no difference in all-cause in-hospital mortality. However, this study had limitations usually associated with large database analyses. Included patients were identified through a PE diagnosis in the electronic medical record within 30 days prior or up to 1 day following intervention which may capture patients who underwent CDI for other diagnoses such as deep venous thrombosis or acute arterial thrombosis. Moreover, patients were not stratified by PE severity due to database limitations. Major bleeding could only be assessed through surrogate data i.e. a decrease in hemoglobin regardless of transfusion, and may represent expected periprocedural blood loss from ST. Furthermore, this database could not account for systemic thrombolytics administered in the perioperative period, which may have confounded the results. Tsukagoshi et al (24) compared CDT and ST using the TriNetX Data Network using 1:1 propensity score matching. They were able to account for several bleeding risk factors such as prior gastrointestinal bleed (GIB) or ICH. They found ST to be associated with a better safety profile than CDT, with lower risks of ICH and GIB. However, once again, patients were not stratified by PE severity in this study due to limited data granularity within the database.

There is a growing body of evidence to suggest that CDT can reduce RHS more effectively than anticoagulation alone (11, 17, 29, 32). However, there remains a theoretical benefit of rapid thrombus removal using ST to quickly improve hemodynamics. Most recently, the APEX-AV trial found that the AlphaVac F18⁸⁵ System significantly decreased the RV/LV ratio at 48 hours (12, 33). PEERLESS observed improved RV function following ST (25). In our series, while we did not assess improvement in RV/LV ratio following intervention, we did find ST to be associated with decreased RHS on repeat echocardiogram at 3 months compared to CDT. As decreasing RHS is perhaps the most important premise behind CDI in PE thus far, these results may be worth the increased bailout intervention rate seen with ST, especially in patients on the higher-risk end of the IRPE spectrum. Given the reduction in RHS and the increased rate of discharge to home, these findings may suggest a net clinical benefit that could potentially offset the modestly higher need for bailout therapy observed with ST. Future PERT protocols should include standardized pre and post-treatment imaging in order to monitor both individual and system-wide outcomes.

Our study has a few limitations worth noting. Firstly, our study is based on a retrospective cohort obtained through an EMR query, subject to data missingness and lack of data precision due to the inherent limitations of EMR documentation. Selection bias remains a significant issue in any retrospective data analysis comparing PE intervention, as ST has become widely accepted as a more rapid therapeutic approach in patients perceived to be higher on the PE-risk spectrum. While we did perform IPTW analysis to address this bias, it is conceivable we were not able to adjust for all confounding factors and that this may ultimately have impacted our results. Despite the presence of a well-established PERT at our institution, laboratory data, echocardiographic data and CTA protocols were not uniformly adhered to which invariably led to data variability. In addition, interventional practices did differ between providers. Pulmonary artery pressure changes as well as transcatheter tPA dosages could not be collected as this data was not uniformly obtained and/or documented by all interventionalists. Estimated blood loss (EBL) was similarly variable as well. An institutional prospectively collected PE intervention/PERT registry would be ideal and is highly warranted. Lastly, as 92% of our ST cohort underwent intervention using the FlowTriever^™ device, our results may ultimately not be generalizable to other thrombectomy devices. Ultimately, an independent, non-industry-sponsored RCT comparing CDT vs ST may be needed to make definitive conclusions.

Conclusion

ST is associated with significantly increased odds of MAE in patients with IRPE when compared to CDT after IPTW. These results were mainly driven by the bailout intervention rate. There was no statistically significant reduction in major bleeding or ICULOS associated with the use of ST compared to CDT. Patients were discharged to home and saw their RHS resolve more frequently after ST, suggesting there may be benefit to rapid thrombus removal in appropriately selected IPRE patients despite the increased need for bailout therapy. Given the overall low bleeding rates and known clinical benefits, CDT should remain an integral part of the interventional armamentarium in IRPE.

ARTICLE HIGHLIGHTS.

Type of Research:

Multicenter single-institution retrospective analysis

Key Findings:

Catheter-directed thrombolysis (CDT) was compared to suction thrombectomy (ST) in intermediate-risk pulmonary embolism (IRPE) using inverse probability of treatment weighting (332 patients). ST was associated with a higher rate of the composite outcome (7-day all-cause mortality, major bleeding, clinical deterioration, need for bailout intervention) and resolution of right heart strain.

Take home Message:

ST is associated with significantly increased odds of major adverse events in patients with IRPE. These results were mainly driven by the bailout intervention rate. There may be benefit to rapid thrombus removal with ST in appropriately selected IPRE patients. CDT should remain an integral part of the interventional armamentarium in IRPE.