The Outcomes of Surgical Pulmonary Embolectomy for Pulmonary Embolism: A Meta-Analysis

Objectives: The purpose of this study is to assess the efficacy, short- and long-term cardiovascular and non-cardiovascular mortalities and postoperative morbidities of surgical pulmonary embolectomy (SPE) for patients with massive or submassive pulmonary embolism. Methods: A comprehensive literature review was performed to identify articles reporting SPE for pulmonary embolism. The outcomes included in-hospital and long-term mortality in addition to postoperative morbidities. The random effect inverse variance method was used. Cumulative meta-analysis, leave-one-out sensitivity analysis, subgroup analysis and meta-regression were performed. Results: Among the 1949 searched studies in our systematic literature search, 78 studies met our inclusion criteria, including 6859 cases. The mean age ranged from 42 to 65 years. The percentage of males ranged from 25.6% to 86.7%. The median rate of preoperative cardiac arrest was 27.6%. The percentage of contraindications to preoperative systemic thrombolysis was 30.4%. The preoperative systemic thrombolysis use was 11.5%. The in-hospital mortality was estimated to be 21.96% (95% CI: 19.21–24.98); in-hospital mortality from direct cardiovascular causes was estimated to be 16.05% (95% CI: 12.95–19.73). With a weighted median follow-up of 3.05 years, the late cardiovascular and non-cardiovascular mortality incidence rates were 0.39 and 0.90 per person-year, respectively. The incidence of pulmonary bleeding, gastrointestinal bleeding, surgical site bleeding, non-surgical site bleeding and wound complications was 0.62%, 4.70%, 4.84%, 5.80% and 7.2%, respectively. Cumulative meta-analysis showed a decline in hospital mortality for SPE from 42.86% in 1965 to 20.56% in 2024. Meta-regression revealed that the publication year and male sex were associated with lower in-hospital mortality, while preoperative cardiac arrest, the need for inotropes or vasopressors and preoperative mechanical ventilation were associated with higher in-hospital mortality. Conclusions: This study demonstrates acceptable perioperative mortality rates and late cardiovascular and non-cardiovascular mortality in patients who undergo SPE for massive or submassive pulmonary embolism.


Introduction
Venous thromboembolisms, such as pulmonary embolisms (PEs), are the third most common cardiovascular (CV) syndrome, with increasing incidence in the aging population [1,2].PE has been reported in the literature to have a high mortality rate [3].The clinical presentation of PE is often non-specific and can range from incidental findings on a computed tomographic chest scan with no clinical symptoms to patients presenting with hemodynamic instability, defined as individuals who are hypotensive needing pressor support and develop end organ hypoperfusion, and sudden death.Nevertheless, presentation in extremes accounts for only 5% of PE cases.
The primary mode of treatment for acute PE is anticoagulation (1).According to the European Society of Cardiology guidelines, surgical pulmonary embolectomy (SPE) for the treatment of PE should be reserved for individuals who deteriorate hemodynamically while being on rescue thrombolytic therapies, for those with contraindications for thrombolytic therapies or for failed catheter-directed thrombolysis [1,2].SPE usually included the performance of cardiopulmonary bypass (CPB), and the literature has demonstrated varying outcomes following surgical intervention [3].Therefore, we performed a systematic review and meta-analysis to assess the efficacy and short-and long-term CV and non-CV mortalities for patients that present with PE.

Materials and Methods
This meta-analysis was performed in concordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [4] and AMSTAR (A MeaSurement Tool to Assess systemic Reviews) Guideline.

Search Strategy
On 14 March 2024, the PubMed and Scopus databases were systematically searched for publications on SPE.The search terms in subject headings and main keywords included the following: "Pulmonary Embolectomy", "surgical embolectomy", "surgical pulmonary embolectomy", "surgical intervention", and "pulmonary embolism".This review was registered with the PROSPERO register of systematic reviews (ID: 542752).There was no individual patient involvement in this study; as such, research ethics board approval was not required.

Study Selection and Inclusion Criteria
Two investigators (HS, SA) independently performed data extraction.Database searches were conducted, and article de-duplication and screening were performed by these two reviewers.A third independent reviewer (MR) confirmed the adequacy of the studies based on the predefined inclusion and exclusion criteria.Articles were included if they were in full-text English on human subjects that included five or more patients with reported CV or non-CS mortality or morbidity outcomes following SPE.We included studies with the largest sample size and the most comprehensive follow-up period for each outcome of cumulative or longitudinal results in more than one publication.Studies were excluded if they were in a non-English language, did not include SPE, did not specify the number or proportion of mortality or morbidity or had a small case series with less than 5 patients.
The full article text of the screened studies was retrieved for the second round of eligibility screening.Prior meta-analyses and systematic reviews were searched to confirm the inclusion of all eligible studies (i.e., backward snowballing).A PRISMA flow diagram illustrating the study selection process is available in the Supplementary Materials (Supplementary Figure S1).The Newcastle-Ottawa scale (NOS) for assessing the quality of Cohort Studies was used for the critical appraisal of eligible studies.Studies with scores of six or more were included [5].

Clinical Outcomes/Definitions
The primary outcome of interest was SPE hospital mortality.Secondary outcomes included CV and non-CV mortality, postoperative pulmonary bleeding, gastrointestinal (GI) bleeding, surgical site bleeding, non-surgical site bleeding and wound complication.
Subgroup analysis for the primary outcome was conducted based on continents.

Data Extraction and Statistical Analysis
Extracted variables included the following: study name, publication year, study design, mean age, percentage of males, mean follow-up in years, percentage of individuals with a contraindication to systemic thrombolytic therapy, percent of preoperative cardiac arrest, preoperative mechanical ventilation, percent of individuals that underwent CPB or extracorporeal membrane oxygenation (ECMO) support, percent of right ventricular (RV) dysfunction, the need for inotropes or vasopressors, systemic thrombolysis, use of myocardial protective techniques and aortic cross-clamping.
Measurement data were reported as the mean and standard deviation or as the median and interquartile range based on the reported studies.The proportion per 100 observations with a 95% confidence interval (95% CI) was calculated for each binary outcome.For late mortality following SPE, the incidence rate with a constant event rate was used to account for different follow-up times of the various studies with the total number of events observed within the treatment group out of the total person-year of the follow-up.
Meta-regression was used to assess the effect of publication year, sex, systemic thrombolysis, contraindication to systemic thrombolytics, preoperative cardiac arrest, inotrope or vasopressor use, preoperative mechanical ventilation, use of CPB, myocardial protective techniques, use of intraoperative hypothermia and aortic cross-clamping percent on hospital mortality after SPE.Heterogeneity among the included studies was assessed using the Cochran Q statistic and the I 2 test.For the primary outcome, if heterogeneity was significant (I 2 > 75%), a leave-one-out sensitivity analysis was performed.The publication bias was assessed using a funnel plot and Egger's regression test.We used a random effect model (inverse variance method) for the entire analysis.The hypothesis testing for equivalence was set at a two-tailed value of 0.05.Analyses were performed using R (version 4.3.3R Project for Statistical Computing), using the following statistical packages: "meta" and "metafor" within RStudio (2023.12.1+402 "Ocean Storm" Release for windows; Postit: Boston, MA, USA).

Results
Among the 1949 searched studies in our systematic literature search, 78 studies met our inclusion criteria including 6859 cases that underwent an SPE intervention .A PRISMA flowchart is shown in Supplementary Figure S1.
The criteria of all included studies are presented in Table 1.The mean age of included patients ranged from 42 to 65 years.The percentage of males ranged from 25.6% to 86.7%.Preoperative cardiac arrest was reported in 57 studies and ranged from 0% to 87.2% of operations with a median preoperative cardiac arrest of 27.6%.The percentage of contraindications to preoperative systemic thrombolysis was reported by 29 studies with a median percent of 30.4% (interquartile range 20.00-45.50) in these studies.The preoperative systemic thrombolysis percent was reported by 35 studies with a median percent of 11.5% (interquartile range 3. 65-25.30) in these studies.The use of CPB appeared to be nearly universal (median 100% (IQR: 100-100)).The criteria of the included studies are shown in Table 1.A quality assessment of all studies was conducted according to the Newcastle-Ottawa scale (NOS) criteria, as shown in Supplementary Table S2.

Efficacy Outcomes
Point estimates for hospital and late mortality outcomes are reported in Figure 1A and Supplementary Figure S7.Hospital mortality was reported by all 75 studies involving 6779 cases.The hospital mortality was estimated to be 21.96% (95% CI: 19.21-24.98)(Figure 1A).The CV hospital mortality was reported in 53 studies and was estimated to be 16.05% (95% CI: 12.95-19.73).The non-CV hospital mortality was reported in 35 studies and was estimated to be 8.32% (95% CI: 6.22-11.06).

Safety Outcomes
Point estimates for pulmonary bleeding, gastrointestinal bleeding, surgical site bleeding, non-surgical site bleeding and wound complications are reported in Figure 1B.
There were 15 cases of GI bleeding reported, and most of them were due to abdominal surgical operations.Clarke et al.ʹs 1986 study reported that 10 patients had abdominal surgery for malignant tumor resection, and 4 of them had GI bleeding.Cases of GI bleeding and cerebral strokes were contraindicated for thrombolytics and anticoagulants.

Sensitivity and Subgroup Analyses and Meta-Regression
There is high heterogeneity in hospital mortality with an I2 of 73%.To explore such reasons for heterogeneity, we performed a leave-one-out analysis that showed the robustness of the obtained estimate for hospital mortality (Supplementary Figure S4).Additionally, cumulative meta-analysis showed a decline in hospital mortality for SPE from 42.86%

Safety Outcomes
Point estimates for pulmonary bleeding, gastrointestinal bleeding, surgical site bleeding, non-surgical site bleeding and wound complications are reported in Figure 1B.
There were 15 cases of GI bleeding reported, and most of them were due to abdominal surgical operations.Clarke et al.'s 1986 study reported that 10 patients had abdominal surgery for malignant tumor resection, and 4 of them had GI bleeding.Cases of GI bleeding and cerebral strokes were contraindicated for thrombolytics and anticoagulants.

Sensitivity and Subgroup Analyses and Meta-Regression
There is high heterogeneity in hospital mortality with an I2 of 73%.To explore such reasons for heterogeneity, we performed a leave-one-out analysis that showed the robustness of the obtained estimate for hospital mortality (Supplementary Figure S4).Additionally, cumulative meta-analysis showed a decline in hospital mortality for SPE from 42.86% in 1965 to 20.56% in 2024.

Discussion
This meta-analysis and systematic review examined the efficacy of SPE, as well as shortand long-term outcomes including CV and non-CV mortality in 78 studies, which included 6859 cases that underwent an SPE.The analysis demonstrated a hospital mortality rate of approximately 22%, with a CV mortality rate of 16%.Additionally, there were long-term CV and non-CV mortality rates of 39 per 100 person-year and 90 deaths per 100 person-year, respectively.The median preoperative cardiac arrest rate was approximately 28%, with the use of CPB universally in patients that underwent SPE.
The in-hospital mortality rates, as well as CV and non-CV mortality rates, reported in this study are similar to the reported mortality rates by Karla et al. who reported an in-hospital mortality rate of 26.3% [4].A study was conducted by Kilic et al. using a weighted nationwide inpatient sample, which included 1050 participating institutions in 44 states and identified 2709 patients that underwent an SPE for a PE.In this study, they reported an in-hospital mortality rate of 27.2% and identified that the comorbidity index and black race were independently associated with inpatient mortality following SPE [84].
A retrospective study performed by Hartman et al. reported a 30-day mortality rate of 4.2% for all comers but illustrated that patients that were unstable had a higher 30-day mortality rate of 12.5% compared to stable patients who had a 30-day mortality rate of 1.4% [6].Studies have also shown that mortality rates are higher following cardiac arrest, which could explain the reported in-hospital mortality rate of 27% in this study, given that 28% of patients that underwent SPE had preoperative cardiac arrest.Stein and colleagues reported an operative mortality rate of 59% in patients who had preoperative cardiac arrest compared to a rate of 20% in patients who did not have preoperative cardiac arrest [85].
Furthermore, we found in this study that there is a decline in the in-hospital mortality rate following SPE.It decreased from 42.86% in 1965 to 20.56% in 2024.Studies have previously shown this reduction in mortality over time [86].This trend is likely due to improvements in the diagnosis of PE, the stabilization of the patient and early intervention.There is also likely a significant selection bias at work, as the dramatic improvement in catheter-based interventions has offered many patients embolectomy in the absence of surgery.This lack of randomization is a major confounder of such a retrospective metaanalysis.This review supports the concept that in appropriately selected patients, surgical embolectomy may be performed safely and with a good outcome; it does not argue against the utility of popular catheter-based techniques/approaches that have rapidly evolved from catheter-directed thrombolysis to ultrasound-augmented thrombolysis and to multiple generations of percutaneous thrombectomy devices.
Among the included studies, there was an apparent trend toward higher inotrope/vasopressor use with RV dysfunction, in studies that reported both variables, but this was statistically insignificant (p-for trend = 0.347).The hospital mortality was mainly due to cardiovascular comorbidities which included the need for inotropes or vasopressors, preoperative mechanical ventilation, shock and cardiac arrest.The weighted median follow-up was 3.05 years.Late mortality causes included both CV and non-CV causes.Cardiovascular comorbidities such as hypertension and heart failure and non-CV causes such as malignant neoplasms are the most common causes for late mortality.
Finally, this study has limitations that include the lack of demographic data such as race in the majority of included studies, since previous studies have shown an association between race and in-hospital mortality following SPE.Specifically, the black and African American race was associated with higher mortality rates compared to white Americans [84].Additionally, hemodynamic information was not present in a reasonable number of the included studies.It would have been interesting to observe if there were differences in hospital mortality following SPE in stable and unstable patients or to understand the baseline presentation of the patient and why that contributed to a hospital mortality rate of approximately 27%; however, we were able to identify some predictors of mortality such as an earlier era of surgery, prior cardiac arrest, need for preoperative mechanical ventilation and the need for vasopressors or inotropes.There is a discernible lack of data on the institution of ECMO among included patients.There is a need to evaluate other late outcomes, such as the rate of development of chronic pulmonary hypertension in patients who undergo SPE for acute PE.

Conclusions
In conclusion, this meta-analysis and systematic review demonstrates acceptable perioperative mortality rates and late CV and non-CV mortality in patients who undergo SPE for massive or submassive PE.There is a noticeably reduced mortality rate with more recent studies using SPE.

JFigure 1 .
Figure 1.Forest plot of (A) hospital mortality (The * refers to the different subgroups of hospital mortality) and (B) hospital morbidity.

Figure 1 .
Figure 1.Forest plot of (A) hospital mortality (The * refers to the different subgroups of hospital mortality) and (B) hospital morbidity.

Figure 2 .
Figure 2. Bubble plots of meta-regression of (A) publication year, (B) percentage of males, (C) preoperative cardiac arrest and (D) preoperative mechanical ventilation on hospital mortality outcome.

Figure 2 .
Figure 2. Bubble plots of meta-regression of (A) publication year, (B) percentage of males, (C) preoperative cardiac arrest and (D) preoperative mechanical ventilation on hospital mortality outcome.

Table 1 .
The criteria of the included studies.

Table 2 .
Meta regression of hospital mortality.

Table 2 .
Meta regression of hospital mortality.