The use of systemic thrombolysis in the setting of high-risk PE is supported by small-scale RCTs and observational studies published over the last few decades. A state-of-the-art systematic review by Marti et al published in 2014 included fifteen clinical studies amounting to 2,057 patients and compared systemic thrombolysis to anticoagulation alone across risk groups. The authors found that thrombolytic therapy reduced total in-hospital fatality (OR: 0.59, 95% CI: 0.36-0.96), PE-related fatality (OR: 0.29, 95% CI: 0.14-0.60), and PE recurrence (OR: 0.50, 95% CI: 0.27-0.94) among patients with acute PE. However, this benefit was at the cost of an increased incidence of major (OR: 2.91, 95% CI: 1.95-4.36) and fatal or intracranial hemorrhage (OR: 3.18, 95% CI: 1.25-8.11). In a subgroup analysis, there was a lower risk of major bleeding or fatal/intracranial hemorrhage in studies using alteplase compared with tenecteplase. Among intermediate-risk patients, these findings were driven by the results of the PEITHO trial, which showed that systemic thrombolysis with tenecteplase (vs anticoagulation alone) reduced the risk of developing hemodynamic decompensation but substantially increased the risk of major and intracranial bleeding within 7 days of randomization. In PEITHO, younger patients appeared to have a more favorable benefit-to-risk ratio compared to older patients.

The concept of reduced-dose thrombolysis has been previously explored. A meta-analysis from 2014 included five RCTs with a total of 440 patients; two RCTs compared half-dose thrombolysis to anticoagulation alone, while three RCTs compared it to full-dose thrombolysis. The study showed no difference between low-dose thrombolysis and heparin in terms of major bleeding events (OR: 0.73, 95% CI: 0.14-3.98), recurrent PE (OR: 0.13, 95% CI: 0.01-2.64), or all-cause in-hospital fatality (OR: 0.63, 95% CI: 0.12-3.34). However, in the comparison of different thrombolysis dosages, standard-dose rt-PA had a higher risk of major bleeding than low-dose rt-PA (OR: 0.33, 95% CI: 0.12-0.91). In this meta-analysis from 2014, there were no differences in recurrent PE (OR: 0.96, 95% CI: 0.30-3.04) and all-cause fatality (OR: 0.88, 95% CI: 0.23-3.37) between the two rt-PA dosage groups.

CBI is currently suggested for selected patients with high-risk or intermediate-high-risk PE, particularly when full-dose thrombolysis is ineffective or contraindicated. CBI is emerging in recent years as a rapidly evolving field in interventional medicine and has the potential of becoming a first-line therapy in patients at higher risk of decompensation in the near future. Guidelines and expert recommendations give emphasis on the limited evidence from clinical RCTs supporting their use primarily because of a lack of evidence concerning early and long-term clinical outcomes, and cost-effectiveness. ^,, However, despite the uncertainty of current data, real-world analyses demonstrate that the use of CBI is rising exponentially, partly due to its ease of use and favorable effectiveness as well as safety profile. ^,,

Two conceptual modalities, catheter-directed thrombolysis (CDT) and catheter-based thrombectomy (CBT), are currently used. Both modalities appear effective and safe in the management of severe PE, mostly in observational, single-arm studies and RCTs with surrogate markers for clinical outcomes.

In the KNOCKOUT-PE prospective registry, all-cause fatality at 30 days was 1.0% and major bleeding events within 72 hours occurred in 1.6% of 489 patients treated with ultrasound-assisted CDT (US-CDT, EKOS®, Boston Scientific). In the FLASH registry, all-cause fatality at 30 days was 0.8% and major bleeding events occurred in 1.4% of 800 patients treated with large-bore mechanical thrombectomy (FlowTriever®, Inari Medical). The FLAME study was a multicenter study that showed low rates of death and hemodynamic decompensation in selected patients with acute PE treated with the FlowTriever system. This benefit was achieved with a low incidence of major bleeding and device-related adverse events, suggesting that FlowTriever may be a safe and effective alternative to systemic thrombolysis in critically ill patients. In the RESCUE trial, the BASHIR catheter (Thrombolex), which employs a pharmaco-mechanical basket delivery mechanism of lytic into the pulmonary arteries, was evaluated in 109 patients with intermediate-risk acute PE, showing a significant decrease in the RV/LV ratio from 1.66 to 1.10, with no device-related major complications. The EXTRACT-PE trial was a multicenter, single-group study involving 119 patients with intermediate-risk PE who underwent thrombus aspiration using the Penumbra Indigo Aspiration System. The primary efficacy outcome showed a significant reduction in the RV/LV ratio by 27% reduction, with a 1.7% major adverse event rate. The APEX-AV trial described the use of the AlphaVac F18 ⁸⁵ thrombectomy system (AngioDynamics) in 122 patients with acute PE, demonstrating a 48-hour reduction in the RV/LV ratio of 29.1% and a clot burden reduction, evaluated by the Modified Miller Index, of 35.5%.

Observational data are conflicting regarding comparisons between the two modalities. An analysis of administrative data showed no difference in clinical outcomes between CDT and CBT, while others showed either superiority of CBT or CDT, and hence, no robust conclusions can be derived. The actual number of RCTs published so far on CBI remains low and concerns mainly small trials (total number of patients accounts to 825; of those, 550 from the recent PEERLESS trial ) that were underpowered to detect differences in hard clinical outcomes.

Three trials have compared CDT to anticoagulation alone. The first RCT (ULTIMA) was conducted in 2014 by Kucher et al and investigated the use of ultrasound-assisted (US)-CDT (EKOS®, Boston Scientific) against anticoagulation alone in 59 patients, finding that US-CDT reduced more effectively the RV/LV ratio and pulmonary artery pressures compared to anticoagulation alone. This trial was appropriately powered for effectiveness with a surrogate outcome. In a trial of 23 patients by Kroupa et al the primary efficacy outcome of RV function improvement, pulmonary pressure decrease, and thrombus burden reduction was achieved more frequently in the CBI group than in the standard-of-care group (7 of 12 patients vs 1 of 11 patients). The CANARY trial, an open-label randomized study, compared CDT (using the Cragg-McNamara Valved Infusion Catheter, Medtronic) with anticoagulation alone in patients with intermediate-high risk PE. CDT involved alteplase infusion (12 mg for unilateral PE, 24 mg for bilateral PE), whereas the control group received enoxaparin. The study was prematurely stopped due to the slow enrollment during the COVID-19 pandemic after inclusion of 94 of 288 planned patients. At 3 months, the primary outcome (RV/LV ratio >0.9) did not differ significantly between groups, but the CBI group had numerically a lower mean RV/LV ratio. There were three deaths at the 3-month follow-up (all in the anticoagulation only group), and only one major bleeding (in the CDT group). STORM-PE was the first study to assess computer-assisted vacuum thrombectomy (CAVT) showing superiority to anticoagulation alone in reducing RV/LV ratio within 48 hours in patients with intermediate-high risk PE.

Another three RCTs, have compared CBI to each other, or to other reperfusion therapies. The SUNSET sPE trial was the first to compare two different CBI modalities to each other, namely standard CDT (Uni-Fuse AngioDynamics catheter or Cragg-McNamara catheter) to US-CDT (EKOS Endovascular System). In this trial of 82 patients, both treatments significantly reduced obstruction index and thrombus load, with no significant difference between groups ( P =.77), whereas, standard CDT resulted in a greater reduction in RV/LV ratio compared to US-CDT ( P =.01). The trial was powered on the basis of an ambitious 50% reduction in RV/LV ratio with ultrasound augmentation, and the thrombolytic regimens lacked standardization in both arms, making generalizability uncertain. In a single-center noninferiority trial by Stortecky et al, 27 patients with intermediate-high or high-risk acute PE were randomized to undergo either US-CDT with the EKOS Endovascular System or surgical embolectomy. The trial was prematurely terminated due to slow enrolment and was likely underpowered, and failed to show noninferiority of US-CDT to surgical embolectomy for reduction of the RV/LV ratio at 72 hours, with a post hoc superiority analysis suggesting greater RV function improvement and thrombus burden reduction with surgical embolectomy. Lastly, the PEERLESS trial compared large-bore CBT to CDT (either standard or US-CDT with variable thrombolytic dosing) and included 550 patients with intermediate-risk PE. The similar clinical outcomes reported from observational studies were also captured in the PEERLESS trial, which showed no difference in hard clinical outcomes, such as in all-cause death and major bleeding, while a difference in favor of CBT was shown in regards to physician-adjudicated clinical deterioration and need for bail-out treatments, and intensive care unit (ICU) admission rates, though convention requires ICU admission for tPA administration in many centers.

As shown, currently, no large-scale, adequately powered RCT has been conducted to assess the impact of CBI on clinical outcomes. Data from large multicenter RCTs which will provide more robust data (and especially on hard clinical outcomes) for the use of CBI are ongoing ( Table 1 ).

It is of note that in high-risk PE, reperfusion strategies have not been adequately investigated in RCT settings. Even for systemic thrombolysis, which has a class I indication for use, only one trial of eight patients, four in each group, was performed exclusively in high-risk patients (and hence, the high level of recommendation in the ESC guidelines) and tested reperfusion against standard of care. In addition, only four of the fifteen trials summarized in the meta-analysis by Marti et al included high-risk PE as part of a mixed cohort of patients giving a sum total of 26 patients with high-risk PE randomized in clinical trials up to 2015. This highlights the lack of randomized evidence especially for this vulnerable population.

Only recently, a noninferiority trial (FORPE) randomized 310 high-risk patients to either nonimmunogenic staphylokinase or alteplase and showed low rates of the primary efficacy outcome of all-cause death within 7 days in both groups, 2% and 3% respectively. Staphylokinase had numerically slightly fewer major bleeds than alteplase (0% vs ∼3%).

Conducting a single, large-scale trial that incorporates all distinct reperfusion treatment strategies as separate arms is not feasible. This challenge highlights the critical importance of synthesizing the results from the upcoming RCTs upon their publication. The large RCTs designed to assess hard clinical outcomes—such as death, hemodynamic deterioration, and recurrence of PE—which are only just published, currently underway, or being designed will include approximately 2,000-2,500 patients. The most recent systematic review of all reperfusion therapies in PE included approximately 2,500 patients. As such, the volume of evidence in the field of reperfusion treatment of severe PE is expected to substantially increase with results from ongoing RCTs anticipated in 2026-2027.

A robust, predefined synthesis of the results of these trials with a standardized methodology is of imperative importance to bring this field forward. This living meta-analysis will not only provide insights into the impact of novel reperfusion treatments on early clinical outcomes but also allow precise estimation of their net benefit and pose the basis for cost-effectiveness analysis. Moreover, it will allow for pooled estimates of long-term and rare clinical outcomes. Clinicians may see conflicting rankings of reperfusion strategies across outcomes or patient groups, which should be interpreted in the context of trial design, patient risk, and outcome definitions rather than as definitive proof of one approach. The conclusions of this effort will influence the landscape of severe PE management and inform clinical guidelines and healthcare providers as new evidence emerges. The following outlines the protocol for this prospective, living systematic review, ensuring a structured and transparent approach to synthesizing evidence.

In this prospective meta-analysis, we will routinely (every six months) perform a systematic search of clinical trial registries (ie, clinicaltrials.org and EudraCT) and bibliographic databases, as well as searching for protocols with general searching engines (eg, Google Scholar), approaching relevant stakeholders, and publicizing the meta-analysis through protocol and websites. By implementing this process, ongoing studies will be identified as eligible for inclusion in the meta-analysis. Therefore, hypotheses and analysis strategies are specified before the results of individual studies are known.

In parallel, a systematic search of MEDLINE (via PubMed), the Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science will be routinely (every six months) performed through automatic query of databases to screen published research. A search string will be created for PubMed and modified accordingly for other search databases (Supplement 1). To complement our search, all references from selected studies will be retrieved and manually reviewed according to the snowball effect. No language restrictions will be posed. The prospective systematic review will be continuously updated with new data as they become available. In the final data analysis, we will include the final results of RCTs comparing reperfusion treatments against the standard of care (and/or each other) in adult patients with severe acute PE. Major changes in results will be communicated through publication in peer-reviewed journals, rather than preprint servers, to ensure accuracy and rigorous peer review.

Severe acute PE will be defined as high- or intermediate-high risk (according to the European Society of Cardiology criteria ), or as massive, or sub-massive PE, while the standard of care will be defined according to the 2019 European Society of Cardiology guidelines. Commonly, severe acute PE includes patients with hemodynamic instability (high-risk PE), or hemodynamically stable patients with presence of right ventricular dysfunction and/or abnormal troponin levels (intermediate-risk PE) with or without additional clinical criteria of severity. Nonrandomized interventional studies, crossover trials, and observational studies will be excluded.

The reperfusion strategies to be considered are full-dose systemic thrombolysis (FD-ST), half-dose systemic thrombolysis (HD-ST), standard catheter-directed thrombolysis (S-CDT), ultrasound-assisted catheter-directed thrombolysis (US-CDT), catheter-based thrombectomy including mechanical aspiration with or without clot fragmentation (CBT), and surgical embolectomy (SE). Thrombolytic agents to be considered will be recombinant tissue plasminogen activators (rt-PA: alteplase, tenecteplase). Studies with plasminogen activators (PA: streptokinase, urokinase, staphylokinase) will be excluded to ensure clinical and pharmacological homogeneity. If novel techniques or reperfusion strategies are developed, we will consider if they meet the criteria for inclusion and analysis.

The primary efficacy outcome will include all-cause death (or PE-related death, as provided in the original publication), PE recurrence, objectively assessed cardiorespiratory decompensation or collapse, and/or treatment failure, defined as deterioration or lack of improvement of (compromised) vital parameters.

The safety outcomes will include fatal bleeding, major bleeding (classified as Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries [GUSTO] severe bleeding, International Society on Thrombosis and Hemostasis [ISTH] major bleeding, or Bleeding Academic Research Consortium [BARC] types 3b, 3c, 5a, and 5b bleeding). Harmonization of clinical outcomes will be based on the definition of major bleeding adopted in the individual trials.

The secondary outcomes will include the individual components of the primary outcome, as well as additional outcomes, such as rehospitalization, length of stay, length of stay in the ICU, and diagnosis of chronic thromboembolic pulmonary hypertension (CTEPH). Functional outcomes will also be assessed, including persisting dyspnea, functional impairment (ie, postvenous thromboembolism functional scale), six-minute walk distance (6MWD), cardiopulmonary exercise testing (CPET) parameters, post-PE impairment/syndrome, and quality of life (QoL) measures, both generic and disease-specific. For all primary and secondary outcomes, measures of utility, such as the number needed to treat or to harm will be provided, as well as pooled estimates of absolute risks for individual treatment groups.

Given the varying timeframes for outcome evaluation, the primary efficacy outcomes and its components, as well as the safety outcomes will be pooled together, provided they fall within 30 days of randomization. Diagnosis of CTEPH and the functional outcomes will also be pooled together, as long as they are assessed at least three months after randomization and over a follow-up period of up to one year. For studies reporting outcomes at multiple time points (eg, both 7-day and 30-day assessments), the longer timeframe will be prioritized. Additionally, studies reporting in-hospital assessments will be assumed to reflect a 30-day evaluation.

Heterogeneous definitions of hemodynamic deterioration will be addressed through trial-level extraction and reporting of prespecified outcome definitions, without post hoc reclassification.

Our systematic review will be performed in respect to Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. All studies will be imported into a reference management software. All duplicates will be removed, and two reviewers will independently screen titles and abstracts and peruse full texts for eligible studies. A third review author will be consulted to resolve any discordance regarding study eligibility. All reasons for exclusion will be reported.

Two reviewers will independently extract data regarding study design and efficacy and safety outcomes on a predefined Excel spreadsheet. A pilot test will be performed before initiation to ensure coherence between the two review authors. Any disagreement will be resolved by consensus. Only aggregated data will be extracted from the eligible studies. Authors will be contacted in order to obtain any missing data relevant to the analysis.

Eligible studies will be evaluated for quality using the Cochrane collaboration risk-of bias tool for randomized controlled trials (RoB 2).

A sensitivity analysis will be performed excluding studies with low quality. We will evaluate the confidence in the results from network meta‐analysis (for primary outcome) using the CINeMA (Confidence in Network Meta‐Analysis) framework as proposed by Salanti and colleagues. CINeMA is an adaptation of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for network meta‐analysis.

The data synthesis will be separated in two different strategies. For both of the following strategies, the primary analysis will be limited to clinical trials powered for and designed to investigate hard clinical outcomes adjudicated by an independent committee, while a secondary analysis will be performed including all studies reporting on clinical outcomes, including those designed to study the impact of CBI on hemodynamic outcomes.

For the first data synthesis strategy, we will perform pairwise meta-analyses, using the Restricted Maximum Likelihood (REML) approach, and compare:

• (i)

  any reperfusion (systemic thrombolysis, half dose systemic thrombolysis, CDT, CBT) vs standard of care.

• (ii)

  half dose systemic thrombolysis vs standard of care.

• (iii)

  CBI vs standard of care.

• (iv)

  CDT vs standard of care.

• (v)

  CBT vs standard of care.

The effect measure will be the risk ratio (RR) or the mean difference (MD), as appropriate for binary and continuous outcomes, respectively, with the corresponding 95% confidence intervals (CI).

For the second data synthesis strategy, we will perform a frequentist random effects network meta-analysis to assess direct and indirect evidence across trials comparing reperfusion treatments not only to standard of care, but also to each other, therefore including studies in which two novel reperfusion strategies are assessed. The generalized DerSimonian-Laird estimator will be used to estimate the common between-study variance. The total amount of variation related to heterogeneity and not sampling error will be assessed with the I ² statistic as follows: <30% low, 30-60% moderate and >60% substantial heterogeneity. Prespecified subgroup and sensitivity analyses mentioned below, are planned to enhance interpretation in case of substantial heterogeneity.

Network graphs will be generated with node sizes proportional to the number of patients per group and edge widths proportional to the number of pairwise comparisons. The assumption of consistency in the network will be assessed using both local and global checks. For local checks, we will use the node-splitting method to evaluate differences between direct and indirect comparisons. The design-by-treatment interaction method will be used to simultaneously assess the consistency assumption globally across all comparisons in the network. P-scores will be used to produce the relative ranking of all interventions in the network. Overall network meta-analyses results will be summarized in league tables.

Prespecified analyses will be conducted separately based on acute PE severity (high-risk PE vs intermediate-risk PE) according to the different standard of care in the two groups. Subgroup analyses stratified by follow-up duration (≤7 days vs >7-30 days) to reduce misclassification bias will be performed where applicable. Given expected differences in the background anticoagulation strategies, a subgroup analysis will be performed according to the anticoagulation regimen received from the majority of patients at the trial level. A sensitivity analysis excluding studies that do not define eligibility criteria according to the baseline PE severity (eg, right ventricular dysfunction, and/or markers of laboratory or hemodynamic severity) will be performed.

For rare binary endpoints (defined as an event rate below 5%), we will conduct a sensitivity analysis using the penalized likelihood meta-analytical model. This approach is warranted because, in the presence of rare events, conventional random-effects methods such as REML and DerSimonian–Laird may produce biased and imprecise estimates. ^,, In contrast, the penalized likelihood approach has been shown to yield more robust effect estimates and allows inclusion of studies with zero events in one or both treatment arms.

All analyses will be performed using R and the meta and netmeta packages. A summary of the present protocol and study design paper has been registered to PROSPERO (registration number: CRD420251207053, International prospective register of systematic reviews; https://www.crd.york.ac.uk/PROSPERO/ ).