Immediate versus staged PCI in STEMI and multivessel disease: a systematic review and meta-analysis

INTRODUCTION

Mitral regurgitation (MR) is one of the most prevalent valvular heart diseases, affecting 2% of the global population, with incidence increasing significantly with age.1-4 For patients with severe symptomatic MR who are at high or prohibitive surgical risk, mitral transcatheter edge-to-edge repair (M-TEER) has become an established alternative to surgery.5 The COAPT trial demonstrated that M-TEER significantly reduces heart failure-related hospitalizations and mortality in patients with severe secondary MR receiving optimized guideline-directed optimal medical therapy.6 As M-TEER adoption has expanded, procedures are increasingly performed across hospitals with different levels of experience, structural heart infrastructure, imaging support, and referral patterns.7 In other cardiovascular interventions, an inverse relationship between procedural volume and adverse outcomes has been well described, with high-volume centers demonstrating lower mortality and fewer complications.8-11 Whether this relationship extends to M-TEER, in which procedural complexity and anatomical selection play critical roles, remains unknown. We therefore examined the association between hospital procedural volume and post-M-TEER outcomes, including mortality, readmission rates, and length of stay, using a nationally representative database. Understanding these relationships is essential to improve standardization of care across centers.

METHODS

Data source

We utilized the National Readmission Database (NRD) from January 2014 to December 2019 to conduct a retrospective analysis. The NRD is part of the Healthcare Cost and Utilization Project and is sponsored by the Agency for Healthcare Research and Quality of the United States.12 Covering multiple US states, it represents approximately 60% of the total US population and hospitalizations. The NRD has a well-established track record and has been widely used in studies of numerous diseases and cardiovascular procedures. Because the NRD contains no protected health information or personally identifiable data, this study was deemed exempt from Institutional Review Board approval. All analyses adhered to the Healthcare Cost and Utilization Project Data Use Agreement.

Study population

We queried the NRD using international code of diseases ICD-9 and ICD-10 codes to identify all hospitalizations for adults aged 18 years or older who underwent M-TEER between 2014–2019 (tables S1 and S2). Patient demographics, comorbidities, and hospital characteristics were extracted using established Healthcare Cost and Utilization Project coding definitions. For the 30- and 90-day analyses, patients discharged during months without complete follow-up availability were excluded.

The NRD does not contain echocardiographic information such as MR mechanism, functional vs degenerative MR, MR severity, leaflet morphology, coaptation depth, or postprocedural MR reduction. Accordingly, differentiation between functional and degenerative MR was not possible. Given procedural adoption patterns during the study period, the cohort likely reflects predominantly functional MR; however, this limitation affects the generalizability of the findings to M-TEER for degenerative MR.

Hospital volume stratification

Annualized hospital procedural volume was calculated as the total number of M-TEER procedures performed at each hospital from 1 January through 31 December of each year from 2014 to 2019. Hospitals were categorized into 3 groups according to the distribution of annual procedural volumes, consistent with previous volume-outcome analyses in structural heart interventions.11 Hospital-volume categories were defined using quartile-based annual procedural-volume cutoffs, with the middle 2 quartiles combined to ensure an adequate sample size and to capture clinically meaningful gradients in institutional experience: low-volume hospitals were defined as quartile 1, with < 13 M-TEER procedures per year; medium-volume hospitals as quartiles 2 and 3, with 14 to 54 procedures per year; and high-volume hospitals as quartile 4, with ≥ 55 procedures per year.

Endpoints

The primary endpoints were in-hospital events, including mortality, stroke, acute kidney injury, vascular complications, cardiogenic shock, and cardiac arrest, as well as in-hospital procedures such as implantation of multiple clips during the same index procedure, repeat clip implantation after the index procedure during the same admission, surgical mitral valve replacement (SMVR), and use of mechanical circulatory support. Secondary endpoints included 30-day readmission and 30-day inpatient mortality, defined as death during the index admission or any subsequent hospitalization within 30 days. A composite adverse outcome was defined as in-hospital mortality, repeat M-TEER during the index hospitalization, or SMVR.

Statistical analysis

We evaluated the association between procedural volume and clinical outcomes by categorizing annualized procedural volume into low-, medium-, and high-volume groups. Baseline comorbidities, hospital characteristics, and clinical outcomes were compared across volume groups using the chi-square test for categorical variables and the Kruskal–Wallis H test for continuous variables. To examine the relationship between procedural volume, as a categorical variable, and endpoints, we used multivariable logistic regression.

A set of clinically significant variables was selected a priori on the basis of presumed impact and clinical importance and was incorporated as covariates into the regression models. Multivariable models were adjusted for age, sex, median household income quartile, primary payer/insurance type, hospital bed size, and Elixhauser comorbidity components.

To examine nonlinear relationships between annual hospital M-TEER volume and the composite endpoint, we used models incorporating restricted cubic splines. These models assessed the association between procedural volume, treated as a continuous variable, and the weight-adjusted composite endpoint. The relationship was visualized by plotting the adjusted odds ratio (aOR) and 95%CI for the composite endpoint on the y-axis against procedural volume on the x-axis.

All statistical analyses were performed using STATA 18 (StataCorp LLC, United States); P values < .05 were considered statistically significant.

RESULTS

Admission characteristics

Of the 33 980 patients who underwent M-TEER between 2014-2019, 1054 (3.1%), 11 734 (34.5%), and 21 192 (62.4%) were treated at low-, medium-, and high-volume hospitals respectively. The number of hospitals performing M-TEER increased substantially during the study period, as shown in figure 1A, with growth observed across all volume categories.

Figure 1. A: trends in United States hospitals performing mitral transcatheter edge-to-edge repair; B: annual distribution of procedural volumes per center.

Figure 1B illustrates the annual distribution of M-TEER procedural volumes across US centers from 2014 through 2019. During this period, procedural uptake increased, and the distribution widened, reflecting greater heterogeneity in practice. A bimodal pattern emerged, with a dense lower peak representing low-volume centers, defined as < 13 M-TEER procedures per year, and a sparse upper tail representing high-volume centers, defined as ≥ 55 M-TEER procedures per year.

Regarding demographics, there was a marginal difference in the proportion of female patients across hospital-volume groups (43.4%, 46.4%, and 46.2% in the low-, medium-, and high-volume hospitals, respectively; however, this difference was not statistically significant (P = .46). However, age increased significantly from low- to high-volume hospitals (75.1 ± 12.1; 77.3 ± 10.9; and 77.7 ± 10.2 years, respectively, P < .01). Payment sources varied substantially across hospital-volume groups, with different proportions of Medicare, Medicaid, private insurance, self-pay, and other payment categories (P < .01). Regarding hospital characteristics, the proportion of teaching hospitals increased from low- to high-volume hospitals (85.0%, 88.7%, 91.6%, respectively: P = .16).

When comorbidities were examined, notable differences were observed across tertiles. For example, the prevalence of coronary artery disease increased substantially from low- to high-volume hospitals: 62.3%, 64.5%, and 68.3%, respectively; P < .01. Similarly, liver disease differed significantly across groups, with the highest prevalence observed in high-volume hospitals: 10.9%, 8.8%, and 12.3%, respectively; P = .02. Furthermore, other comorbidities, including atrial fibrillation, chronic kidney disease, chronic obstructive pulmonary disease, diabetes mellitus, congestive heart failure, hyperlipidemia, hypertension, prior coronary artery bypass grafting, prior percutaneous coronary intervention, peripheral vascular disease, and smoking demonstrate varying prevalence across hospital- volume groups, as shown in table 1.

Table 1. Baseline characteristics

Short term outcomes

In-hospital and 30-day outcomes

Unadjusted in-hospital and 30-/90-day outcomes according to hospital volume are summarized in table 2. A detailed analysis of in-hospital and 30-day outcomes, adjusted for baseline characteristics and comorbidities, is provided in table 3. Regarding in-hospital mortality, no statistically significant differences were observed between low- or medium-volume hospitals and high-volume hospitals, as shown in figure 2. However, 30-day mortality was significantly higher in both low-volume hospitals (aOR, 1.65; 95%CI, 1.05–2.61; P = .03) and medium-volume hospitals (aOR, 1.31; 95%CI, 1.02–1.68; P = .03) compared with high-volume centers.

Table 2. Clinical outcomes

Table 3. Adjusted associations by hospital M-TEER volume

Figure 2. Procedural volume and clinical outcomes of mitral transcatheter edge-to-edge repair (M-TEER). 95%CI, 95% confidence interval; aOR, adjusted odds ratio; MCS, mechanical circulatory support; SMVR, surgical mitral valve replacement.

In addition, patients treated at low-volume hospitals had significantly higher 30-day readmission rates (aOR, 1.27; 95%CI, 1.00–1.61; P = .048), whereas medium-volume hospitals did not differ significantly in readmission risk. Importantly, there were no significant differences in repeat M-TEER during the same admission or within 90 days, and no difference was observed in the incidence of SMVR across hospital-volume groups. A visual summary of these adjusted outcomes, including procedural and safety endpoints, is provided in figure 3.

Figure 3. Central illustration. In-patient and 30/90-day outcomes of mitral transcatheter edge-to-edge repair procedures. aOR, adjusted odds ratio; MCS, mechanical circulatory support; SMVR, surgical mitral valve replacement.

Procedural volume

Notably, patients treated at low-volume hospitals had lower odds of receiving multiple clips during the same procedure (aOR, 0.56; 95%CI, 0.22-1.44; P = .23), whereas those treated at medium-volume hospitals had higher odds of receiving multiple clips (aOR, 2.05; 95%CI, 1.28-3.29; P < .01) compared with high-volume centers. Moreover, low-volume hospitals showed lower odds of vascular complications (aOR, 0.59; 95%CI, 0.35-1.00; P = .049) compared with high-volume centers. Acute kidney injury, although not significantly different between low- and high-volume hospitals, was significantly less frequent in medium-volume hospitals (aOR, 0.77; 95%CI, 0.64-0.93; P < .01). The use of mechanical circulatory support was less likely in medium-volume hospitals (aOR, 0.66; 95%CI, 0.46-0.95; P = .026) compared with high-volume hospitals. In-hospital stroke, in-hospital cardiac arrest, and the incidence of SMVR did not differed significantly across low-, medium-, and high-volume hospitals. Finally, the restricted cubic spline analysis model demonstrated a significant nonlinear association between annual M-TEER procedural volume and the odds of the composite adverse outcome. The odds of the adverse outcome were highest at lower procedural volumes and declined steeply up to approximately 12 to 13 procedures per year, after which the association plateaued near the reference value. Formal testing demonstrated both a significant overall association between procedural volume and the outcome (P < .001 for overall association) and significant nonlinearity (P < .01 for nonlinearity; figure 2).

Length of stay

During the study period, the length of stay for M-TEER procedures decreased, with the greatest reduction observed in low-volume centers. Despite this overall decrease, low-volume centers consistently had the longest length of stay throughout the study period, as shown in figure 2.

DISCUSSION

In our study, we used the NRD to perform a national analysis of the volume-outcome relationship in M-TEER. Among 33,980 M-TEER procedures performed between 2014 and 2019, we observed that high-volume centers accounted for 63% of the overall volume and that adjusted 30-day mortality and readmission were significantly higher in low-volume centers. Furthermore, length of stay was consistently longest in low-volume hospitals.

Approximately two-thirds of cases were performed in high-volume hospitals, whereas 3.1% were performed in low-volume hospitals. This notable variation in case numbers across M-TEER centers may be attributable to established referral networks, leading to a distribution of cases toward somce centers rather than others.13 A similar finding was reported in the 2019 study by Chhatriwalla et al., which examined institutional experience and M-TEER outcomes and noted that, owing to the referral system, several years would be required for low-volume institutions to achieve the number of cases required to reach the experience threshold.13

Of note, the NRD does not capture echocardiographic parameters such as MR mechanism, leaflet morphology, or postprocedural MR reduction. Because M-TEER during this period was predominantly performed for functional MR, the cohort likely reflects mostly functional MR cases; however, the inability to stratify by MR etiology limits assessment of how anatomic complexity influenced outcomes.

Low-volume centers demonstrated higher adjusted 30-day mortality, although in-hospital mortality remained low across all groups. This finding is consistent with data from a Japanese registry, reporting in-hospital mortality of < 5% among patients undergoing M-TEER for secondary mitral regurgitation.14

A strength of this study is the evaluation of outcomes beyond inpatient mortality, including readmission and length of stay. These metrics suggest that differences in procedural volume may influence recovery and postdischarge care. Our findings, together with previous work by Chhatriwalla et al., reinforce the inverse volume–outcomes relationship observed in other structural heart procedures.13

Institutional resources and infrastructure at higher-volume centers may play a crucial role in influencing outcomes. High-volume centers are more likely to have established protocols, heart teams, and specialized facilities dedicated to transcatheter procedures.9,13,15 These resources may support a more streamlined care process and facilitate comprehensive patient selection, which could contribute to the observed differences in outcomes. Identifying and standardizing best practices from high-volume centers could be instrumental in improving outcomes in low-volume settings.

Furthermore, the disparities observed in readmission rates and length of stay between low- and high-volume hospitals raise questions about differences in postoperative care and coordination. Low-volume centers may face challenges in managing post-discharge issues or ensuring consistent follow-up, which could contribute to higher readmission rates and longer hospitalizations. The finding that in-hospital complication rates were similar whereas 30-day outcomes differed suggests that variations in postdischarge care, including follow-up coordination, access to heart failure services, and timely recognition of clinical deterioration, may play a role. The inverse volume-outcome association observed in our study aligns with previous research demonstrating similar patterns for various cardiac procedures, such as transcatheter aortic valve replacement, percutaneous coronary intervention, and surgical mitral valve repair, as well as noncardiac procedures, including surgery for heptopancretobiliary disease, colon cancer, and endocrinopathies.16-21

A key factor influencing outcomes between low- and high-volume centers is the variation in operator experience and expertise. High-volume centers are more likely to have a team of operators who routinely perform M-TEER procedures, thereby accumulating experience and refining their skills. In contrast, low-volume centers may lack the same level of exposure and experience. This may contribute to the higher rates of procedural complications and adverse events observed in lower-volume centers. Although operator volume is known to independently influence M-TEER outcomes, operator identifiers are not available in the NRD; thus, we were unable to adjust for operator-level procedural experience, representing an important source of unmeasured confounding. Our dataset does not allow us to directly assess whether higher-volume centers achieve greater M-TEER success, because specific measures of mitral regurgitation improvement are unavailable. Instead, we relied on indirect indicators, such as conversion to surgery, readmission, and repeat procedures, which are also influenced by differences in patient selection across centers. For instance, low-volume centers may treat lower-risk or less complex patients, whereas high-volume centers may treat patients with more complex anatomy, thereby influencing immediate procedural decisions and complication rates.

The need for multiple clips during a single TEER procedure and the occurrence of repeat M-TEER provide significant insights into the complexity and effectiveness of mitral valve repair strategies. Our analysis showed that medium-volume hospitals had higher odds of using multiple clips (aOR, 2.05; 95%CI, 1.28-3.29; P < .01) compared with high-volume centers. This pattern may reflect differences in procedural strategy, anatomical complexity, or operator learning curves rather than procedural inefficiency. Without echocardiographic variables, the precise mechanism contributing to this observation cannot be determined. The need for multiple clips often arises from the challenge of achieving adequate mitral valve coaptation, particularly in complex cases in which a single clip may not be suffice. It is difficult to determine whether patient selection, anatomical complexity, or procedural experience contributed to this difference. However, the rate of repeat M-TEER during the same admission and within 90 days did not show differ significantly between hospital-volume groups, suggesting that although initial M-TEER procedures may be technically demanding, the need for repeat interventions may not vary substantially according to hospital experience.

With the increase in M-TEER centers nationally, determining which patients should undergo M-TEER or SMVR can become more challenging, particularly among patients who may be eligible for both procedures. SMVR is a surgical procedure associated with an increased risk of complications and adverse events, and M-TEER for functional MR with reduced left ventricular ejection fraction has previously been shown to have lower perioperative complications and mortality than SMVR.22 Feldman et al. and Stone et al. also demonstrated the safety and efficacy profile of the MitraClip system (Abbott Vascular, Unites States) for M-TEER.6,23 However, just as with clinical experience has contributed to improved SMVR outcomes and success, the proficiency and experience of the medical facility and operator are important factors associated with M-TEER outcomes.9,13,24-28

Recent advances in M-TEER with MitraClip have expanded treatment options for patients undergoing mitral valve repair. However, our study, as well as previous studies, suggests that referral of patients to higher-volume centers with more experienced operators is associated with improved outcomes.9,13 To achieve optimal and consistent patient care and outcomes across hospitals with different procedural volumes, assessment and standardization training and procedural standards for M-TEER with MitraClip may be necessary.

Although differences may exist between high- and low-volume centers, certain outcomes showed no significant disparities. In-hospital stroke, in-hospital cardiac arrest, and the incidence of SMVR remained consistent across all volume groups. Notably, vascular complications outcomes favored low-volume centers compared with high-volume centers. In addition, medium-volume hospitals showed a significant reduction in acute kidney injury compared with both low and high-volume hospitals. Differences in mechanical circulatory support use across volume groups likely reflect institutional resource availability and differing thresholds for mechanical support rather than inherent procedural differences, because high-volume centers may treat a greater proportion of hemodynamically complex patients.

Although it may be assumed that factors like acute kidney injury and vascular complications in high-volume centers could be associated with increased trainee activity, leading to higher rates of vascular access complications and greater contrast use,29 previous studies have shown that only fluoroscopy time tends to increase with trainee involvement, whereas vascular access complications remain comparable when trainees are supervised.29-31 Interestingly, although high-volume centers had a higher proportion of teaching hospitals, this difference was not statistically significant and was accounted for in our adjusted analysis. This finding suggests that the observed outcomes in high-volume centers cannot solely be attributed to trainee involvement. Moreover, higher rates of mechanical circulatory support and permanent pacemaker placement in high-volume centers likely reflect the greater availability of these devices and the treatment of more complex patients undergoing M-TEER.

Taken together, these findings highlight the complex interplay between institutional volume, operator experience, patient selection, and periprocedural practices, which cannot be fully disentangled using administrative data alone.

Real-world application of findings

The disparities in M-TEER procedure outcomes observed in our study prompt further exploration of the potential mechanisms underlying the volume-outcome relationship. Factors such as operator experience, institutional resources, and adherence to best practices may contribute to these differences. Investigation of these aspects is crucial to better understand underlying mechanisms and to develop targeted interventions aimed at improving outcomes across hospitals with different procedural volumes. This understanding is imperative for guiding future research directions and improving outcomes in low-volume centers.

The implications of our findings extend to health care policy and resource allocation. If low-volume centers consistently show inferior outcomes, policymakers may need to consider consolidation of selected procedures at higher-volume centers or provide additional support to improve low-volume institutions. However, such decisions must balance patient access, geographic distribution, and overall health care system capacity. As transcatheter interventions continue to evolve, ongoing monitoring of the volume-outcome relationship is essential. Longitudinal studies tracking outcomes, including technological advances and changes in practice, will offer a more comprehensive understanding. Collaborative efforts among professional societies, regulatory bodies, and health care providers are crucial to developing evidence-based guidelines for clinical practice and reducing disparities across hospital-volume groups.

Limitations

Our study has several limitations. First, as an observational analysis of NRD administrative data, it cannot establish causality, and residual confounding may persist. Second, the absence of data on postdischarge clinic visits or specialist follow-up may have influenced readmission mortality rates after the index hospitalization. Third, the NRD excludes interstate hospitalizations and lacks patient data linkage across years, which may affect readmission estimates. In addition, the NRD does not include echocardiographic information such as MR mechanism, functional vs degenerative MR, MR severity, leaflet morphology, coaptation depth, pulmonary pressures, or postprocedural MR reduction. As a result, we were unable to stratify outcomes by MR etiology or evaluate anatomical procedural success, both of which are important determinants of M-TEER complexity. This limitation may influence interpretation of the volume–outcome relationships.

Fourth, the results primarily reflect outcomes related to M-TEER for functional MR, because functional MR M-TEER was adopted before degenerative MR M-TEER during the study period. Consequently, the findings may not be directly applicable to M-TEER procedures for degenerative MR. However, this focus on functional MR may have contributed to sample homogeneity, thereby enhancing the consistency and comparability of the results. Another important limitation is that the NRD does not provide operator-level identifiers. Therefore, we could not adjust for operator procedural experience, which previous studies have shown to independently influence M-TEER outcomes. This unmeasured factor may contribute to differences observed across hospital-volume strata. Fifth, the use of ICD codes for administrative data analysis introduces the possibility of errors. Notably, while coding changed from ICD-9 to ICD-10 in 2015, this transition is unlikely to have substantially affected the proportion of readmissions. In addition, the inclusion of more states to the NRD enhances national representation without substantially affecting readmission proportions. The NRD does not capture readmissions across states; however, this is likely inconsequential because very few patients would be expected to be readmitted in a state other than that in which the procedure was performed. Sixth, 30-day mortality captured in the NRD reflects inpatient mortality only, because the database does not include out-of-hospital deaths. This may underestimate true 30-day mortality, particularly in centers with shorter lengths of stay and earlier discharge practices. Seventh, patient selection bias represents an important concern: high-volume centers may treat patients with more complex anatomy and a higher comorbidity burden, whereas low-volume centers may manage more selected, lower-risk cases. This differential selection could influence the observed volume–outcome association and cannot be fully addressed without anatomical and echocardiographic data. Eighth, the study period, from 2014 to 2019, predates the broader expansion of M-TEER to degenerative MR and reflects an era of earlier-generation devices and evolving learning curves. The volume thresholds and outcome patterns observed may therefore not be directly applicable to contemporary practice, in which device technology, operator training pathways, and procedural indications have evolved substantially. Ninth, our analysis examined multiple outcomes across 3 hospital-volume strata, raising the possibility of multiple comparisons. Thirty-day inpatient mortality and 30-day readmission were prespecified as the primary endpoints, whereas procedural variables, in-hospital safety events, 90-day outcomes, costs, and discharge disposition were considered exploratory and interpreted descriptively. We did not apply a formal multiplicity adjustment, such as Bonferroni or Benjamini–Hochberg correction, which is consistent with the hypothesis-generating nature of observational volume–outcome studies, in which effect size, confidence interval precision, biological plausibility, and consistency with previous literature are central to interpretation. However, some exploratory associations may not remain statistically significant after formal correction and should not be interpreted as confirmatory. Validation in registries with operator-level and echocardiographic data, ideally using prespecified outcome hierarchies, is needed before any specific association can be considered established.

Finally, certain anatomic and clinical features that influence M-TEER outcomes, such as left ventricular size, valvular anatomy, leaflet thickness, and overall complexity, cannot be captured in the NRD. Differential patient selection at low- and high-volume centers may therefore contribute to residual confounding. The absence of race data in the NRD also limits assessment of racial disparities in readmission. Furthermore, the NRD captures data at the hospitalization level and lacks individual patient-level details, such as specific procedural information and long-term follow-up.

CONCLUSIONS

The number of centers performing M-TEER increased substantially during the study period, with procedural adoption extending across hospitals with varying procedural volumes. However, an inverse relationship was observed between procedural volume and 30-day outcomes, with higher mortality and readmission rates at low-volume hospitals than at high-volume centers. Although detailed anatomical and operator-level information was not available in this dataset, the consistency of these findings with previous work highlights the importance of institutional experience and coordinated periprocedural care. Continued efforts to standardize protocols, optimize training, and support developing M-TEER programs may help reduce disparities and improve outcomes across hospital volumes.

FUNDING

No funding was received for this study.

ETHICAL CONSIDERATIONS

This study used deidentified HCUP-NRD data and was deemed exempt from institutional review board review; informed consent was not required because no individual patient data are reported. All analyses complied with the HCUP Data Use Agreement. With respect to SAGER guidance, sex was included in the analysis where available; gender identity is not captured in the NRD, which limits assessment of gender-related bias.

STATEMENT ON THE USE OF ARTIFICIAL INTELLIGENCE

No generative artificial intelligence tools were used to create, analyze, or interpret the data. Artificial intelligence was used only for limited language editing of selected sections; all scientific content, analyses, conclusions, and final approval are the sole responsibility of the authors.

AUTHORS’ CONTRIBUTIONS

Concept and design: F. Ghanem, L. Alhuneafat, O. Obeidat, and A. Sharma. Acquisition, analysis, or interpretation of data: all authors. Drafting of the manuscript: F. Ghanem, L. Alhuneafat, O. Obeidat, A. Elhamdani, and T. Tarawneh. Critical revision of the manuscript for important intellectual content: A. Naser, A. Asassfeh, R.M. Hatttab, A. Jabri, A. Al Turk, S. Gurevich, P. Villablanca, and A. Sharma. Statistical analysis: F. Ghanem and L. Alhuneafat. Supervision: A. Sharma and P. Villablanca. All authors read and approved the final manuscript.

CONFLICTS OF INTEREST

None declared.

SUPPLEMENTARY DATA

REFERENCES

INTRODUCTION

Aortic stenosis (AS) is one of the most common valvular heart diseases, particularly among older adults. In younger patients, bicuspid aortic valve (BAV) is a major underlying cause of AS. In developed countries, BAV is the most common cause of AS in individuals younger than 70 years, affects approximately 0.5%-2% of the population and is more frequent in men.1,2

The anatomical characteristics of BAV pose specific challenges that differ from those associated with tricuspid aortic valve disease. Abnormal leaflet structure, characterized by asymmetric morphology and frequent calcified raphe, predisposes the valve to accelerated calcification and fibrosis, often leading to clinically significant valvular dysfunction at a younger age.3 In addition, BAV is commonly associated with aortic root dilatation and an increased risk of aortic dissection, requiring a comprehensive assessment of the entire aortic root during treatment planning.4

Transcatheter aortic valve implantation (TAVI) has emerged as a minimally invasive alternative to surgical aortic valve replacement in patients with severe AS, with expanding indications that now include younger, low-risk populations.5,6 However, the anatomical complexity of BAV poses specific challenges for TAVI that differ substantially from those of typical tricuspid anatomy. Asymmetric leaflets, eccentric calcification patterns, and calcified raphe may influence procedural success and long-term outcomes.7,8 These anatomical considerations may affect the choice between self- expanding valves (SEVs) and balloon-expandable valves (BEVs), as each technology has distinct characteristics that may interact differently with complex BAV morphology.

Despite the increasing use of both valve technologies in patients with BAV, dedicated comparative studies evaluating the specific outcomes of SEVs vs BEVs in this challenging population remain scarce. Historically, patients with BAV have often been excluded from pivotal randomized controlled trials (RCTs) comparing TAVI with surgical aortic valve replacement because of a perceived higher risk of procedural complications. Most available evidence comes from small observational cohorts or subgroup analyses of broader TAVI registries, limiting the ability to draw definitive conclusions regarding optimal device selection.

Sá et al. previously conducted a meta-analysis of 8 observational studies (n = 1080 patients) and found no significant differences in procedural, 30-day, or 1-year mortality between valve types. However, BEVs were associated with a significantly higher risk of annular rupture than SEVs Although no statistically significant difference in paravalvular leak (PVL) was observed overall, a subgroup analysis of newer-generation devices suggested lower PVL rates with BEVs.9 Since then, several additional studies have been published, expanding the evidence base to 12 studies and more than 2000 patients. Therefore, this meta-analysis aims to provide a comprehensive and updated evaluation of the comparative safety and efficacy of SEVs vs BEVs in patients with BAV, incorporating the most recent evidence, a larger patient cohort and newer- generation devices.

METHODS

This systematic review and meta-analysis was performed in full compliance with the principles outlined in the Cochrane Collaboration and the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.10 The prespecified research protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) under protocol No. CRD42025634772.

Eligibility criteria

Inclusion in this meta-analysis was restricted to studies that met all the following eligibility criteria: a) randomized controlled trials and observational studies; b) studies comparing the use of SEVs and BEVs in TAVI in patients with BAV; c) inclusion of adult patients (≥ 18 years); and d) a follow-up period of at least 30 days. The exclusion criteria were the absence of a detailed evaluation of the BAV population and the absence of outcomes of interest.

Search strategy and data extraction

We systematically searched MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials on 2 March 2025, using the following search terms: “transcatheter aortic valve replacement”, “transcatheter aortic valve implantation”, “balloon-expandable”, “self-expandable”, “bicuspid aortic valve” (table S1). References from all included studies, previous systematic reviews, and meta-analyses were also manually searched for any additional eligible studies. Two authors (N.V. Sala da Silva and L. Marqueño da Cunha) independently extracted the data using predefined search criteria.

Endpoints

The clinical primary endpoints were procedural mortality, and all-cause and cardiovascular mortality at 30 days and 1 year. Secondary endpoints included stroke, annular rupture, need for a second valve, coronary obstruction, new pacemaker implantation, and moderate or severe PVL.

Sensitivity analysis

To assess the robustness of our results, we performed a leave-one-out sensitivity analysis. This approach involved sequentially excluding each study and recalculating the overall effect size for the remaining studies for each outcome.

Quality assessment

Quality assessment of observational studies was performed using the ROBINS-I tool, which assesses risk of bias in nonrandomized studies.11 Studies were rated as having low, moderate, serious or critical risk of bias across 7 domains: confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of the reported result. Two independent authors (P.F. Gomes Nicz and W.F. Gomes) performed the risk of bias assessment. Disagreements were resolved by consensus. Risk of bias was presented by a diagram using the ROBINS I tool (table S2).

Statistical analysis

Pooled odds ratios (OR) with 95% confidence intervals (95%CI) were calculated using the Mantel-Haenszel method with a random-effects model, which was selected given the binary nature of outcomes and the potential for sparse data in some subgroups. The I2 statistic was used to assess heterogeneity; P values < .10 and I2 > 25% were considered to indicate significant heterogeneity. A leave-one-out sensitivity analysis was conducted by sequentially omitting each individual study and recalculating the pooled estimates to assess the robustness of the results and identify any disproportionately influential studies. R statistical software version 4.3.2 was used for the statistical analysis.

RESULTS

Study selection and characteristics

This study analyzed 12 observational studies including 2013 patients, of whom 1099 received BEVs and 914 received SEVs (figure 1). The mean age of the participants was 77.15 years, and 60.17% were men. The mean STS score and EuroSCORE were 4.5% and 11.97%, respectively.

Figure 1. PRISMA flow diagram of study screening and selection.

Echocardiographic parameters showed a median aortic valve mean gradient of 49.06 mmHg and a median aortic valve area of 0.74 cm². BEVs used in the included studies the SAPIEN, SAPIEN XT, and SAPIEN 3/3Ultra valves (Edwards Lifesciences LLC, United States). SEVs included the Acurate Neo valve (Boston Scientific Ltd, United States), CoreValve/Evolut R/Pro valves (Medtronic Inc, United States), Portico valves (Abbott Structural Heart, United States), and Venus A-Valve (Venus MedTech, China).

The access routes for valve implantation were predominantly transfemoral (87%), followed by transapical (4%) and other approaches, approximately 9%.

The complete characteristics of the included studies and the baseline characteristics of the patients are shown in table 1.

Table 1. Baseline characteristics and study design of included studies comparing balloon-expandable and self-expandable valves in patients with bicuspid aortic valve undergoing TAVI

Pooled analysis of all studies

Mortality endpoints

Analysis of mortality outcomes revealed no statistically significant differences between BEV and SEV approaches across multiple time horizons. The pooled analysis showed no significant differences in 1-year all-cause mortality (10.5% vs 13.0%; OR; 0.84; 95%CI, 0.60-1.19; P = .331) or cardiovascular mortality (5.2% vs 5.5%; OR,1.07; 95%CI, 0.55-2.05; P = .851), with minimal heterogeneity in both analyses (I² = 0% and 2.5%, respectively) (figure 2).

Figure 2. Forest plots of 1-year all-cause and cardiovascular mortality: BEVs vs SEVs. No significant difference between groups was identified at 1 year for all-cause mortality (A) or cardiovascular mortality (B). Individual study results, ORs and 95%CI are shown. 95%CI, 95% confidence interval; BEV, balloon- expandable valve; MH, Mantel-Haenszel; OR, odds ratio; SEV, self-expandable valve. The bibliographical references cited in this figure correspond to Yoon et al.,8 Boiago et al.,14 Yousef et al.,19 Mangieri et al.,16 Hayashida et al.,22 Mylotte et al.,21 and Buono et al.12.

Short-term mortality outcomes followed a similar pattern, with 30-day all-cause mortality showing no significant difference in the largest pooled analysis, which included 12 studies comparing BEVs with SEVs (3.2% vs 3.7%; OR, 0.85; 95%CI, 0.51-1.42; P = .536; I² = 0.0%). Thirty-day cardiovascular death, analyzed across 10 studies, likewise showed no significant treatment effect (4.1% vs 4.9%; OR, 0.87; 95%CI, 0.50-1.51; P = .619; I² = 0.0%). Procedural mortality, examined in 9 studies, showed no significant difference between the interventions (1.2% vs 1.0%; OR, 1.17; 95%CI, 0.48-2.90; P = .728; I² = 0.0%) (figure 3).

Figure 3. Forest plots of procedural (A), and all-cause (B) and cardiovascular mortality at 30 days: BEV vs SEV. No significant between-group differences were observed in short-term mortality outcomes. Individual study results, odds ratios (OR), and 95% confidence intervals (95%CI) are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references mentioned in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Hayashida et al.,22 Kosek et al.,18 Mylotte et al.,21 Costopoulos et al.,20 Buono et al.12

Secondary endpoints

Annular rupture was more frequently associated with BEVs than with SEVs across 9 studies (1.7% vs 0.1%; OR, 3.65; 95%CI, 1.30-10.24; P = .014; I² = 0.0%) (figure 4).

Figure 4. Forest plot of annular rupture incidence: BEV vs SEV. The pooled analysis showed a higher incidence of annular rupture with BEV. Individual study results, odds ratios (OR), and 95% confidence intervals (95%CI) are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Hayashida et al.,22 Mylotte et al.,21 Costopoulos et al.,20 Buono et al.12

The need for second valve implantation was significantly lower with BEVs than with SEVs in an analysis of 8 studies (2.6% vs 5.1%; OR, 0.49; 95%CI, 0.26-0.93; P = .029), although moderate heterogeneity was observed (I² = 22.3%) (figure 5).

Figure 5. Forest plot comparing the need of a second valve between BEV and SEV. The pooled analysis indicates a lower need for a second valve with BEV. Individual study results, odds ratios (OR), and 95%CI are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Buono et al.12

Moderate-to-severe paravalvular leak was significantly less frequent with BEVs than with SEVs across 9 studies, (4.0% vs 10.1%; OR, 0.38; 95%CI, 0.18-0.80; P = .011). This outcome demonstrated substantial heterogeneity (I² = 63.5%) (figure 6).

Figure 6. Forest plot comparing the incidence of moderate or severe PVL between BEV and SEV. The pooled analysis shows a significantly higher risk with SEV. Individual study results, odds ratios (OR), and 95%CI are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Kosek et al.,19 Buono et al.12

New pacemaker implantation requirements differed significantly between treatment approaches, with BEV associated with lower odds than SEVs in a pooled analysis of 9 studies (12.9% vs 19.2%; OR, 0.64; 95%CI, 0.48-0.85; P = .002; I² = 0.0%) (figure 7).

Figure 7. Forest plot of new pacemaker implantation: BEV vs SEV. Forest plot of the need for new pacemaker implantation between BEV and SEV, demonstrating a significantly lower rate with BEV. Individual study results, odds ratios (OR), and 95%CI are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Mylotte et al.,21 Buono et al.12

Stroke incidence was assessed in 9 studies and showed no significant difference between BEV and SEV approaches (2.5% vs 2.7%; OR, 0.93; 95%CI, 0.52-1.67; P = .805; I² = 0.0%) (figure 8). Coronary obstruction, although analyzed in only 4 studies, also showed no significant difference between groups (1.1% vs 1.3%; OR, 1.10; 95%CI, 0.30-4.07; P = .888; I² = 0.0%) (figure 9).

Figure 8. Forest plot of stroke incidence: BEV vs SEV. The pooled analysis showed no significant between-group difference. Individual study results, odds ratios (OR), and 95%CI are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mangieri et al.,16 Hayashida et al.,22 Mylotte et al.,21 Costopoulos et al.,20 Buono et al.12

Figure 9. Forest plot of coronary obstruction incidence: BEV vs SEV. No significant between-group difference was observed in the occurrence of coronary obstruction. Individual study results, odds ratios (OR), and 95%CI are shown. BEV, balloon-expandable valve; SEV, self-expandable valve. The bibliographical references included in this figure correspond to: Yoon et al.,8 Lee et al.,13 Boiago et al.,14 Yousef et al.,19 Jilaihawi et al.,17 Deutsch et al.,15 Mylotte et al.21

Sensitivity analysis

Leave-one-out sensitivity analyses were performed for all outcomes to evaluate the influence of individual studies on the pooled effect estimates (figures S1-S11).

For mortality outcomes, all leave-one-out iterations yielded nonsignificant results. The 1-year all-cause mortality analysis showed ORs ranging from 0.798 to 0.893 across iterations, with all confidence intervals crossing unity. Similarly, 30-day all-cause mortality, 1-year cardiovascular mortality, and 30-day cardiovascular mortality remained nonsignificant in all sensitivity iterations.

Moderate-to-severe paravalvular leak, the outcome with the highest heterogeneity (I² = 63.5%), retained statistical significance in most leave-one-out iterations, although the effect size varied (figure S1). The leave-one-out sensitivity analysis for the need for a second valve implantation showed a consistent direction of effect favoring BEVs across all iterations, with the pooled OR remaining < 1 in every scenario. However, statistical significance was not uniformly maintained: when Deutsch et al.,15 Mangieri et al.,16 Jilaihawi et al.,17 Yoon et al.,8 or Buono et al.12 were omitted, the confidence intervals marginally crossed unity. These findings indicate that, although the overall pooled result is statistically significant, it should be interpreted with caution, because it may be influenced by the specific composition of the included studies (figure S2). No individual study was identified as having a disproportionate influence on the statistical significance or direction of effect for the other outcomes (figures S3-S11).

Quality assessment

Upon reassessment of all 12 included studies using the ROBINS-I tool, no study was rated as having an overall low risk of bias rating. Eight studies were classified as having a moderate overall risk of bias: Yoon et al.,8 Mylotte et al.,21 Jilaihawi et al.,17 Mangieri et al.,16 Buono et al.,12 Boiago et al.,14 Deutsch et al.,15 and Lee et al.13 These studies showed mixed risk profiles across the 7 assessed domains, with most domains rated as low to moderate risk; however, residual confounding and concerns regarding participant selection concerns were the most frequently identified methodological limitations.

Four studies were classified as having a serious overall risk of bias: Kosek et al.,18 Hayashida et al.,22 Yousef et al.,19 and Costopoulos et al.20 These studies had significant methodological limitations across multiple domains, most notably bias due to confounding (D1) and selection of participants (D2).

The detailed results of the ROBINS-I assessment for each study across all 7 domains are shown in table S2.

DISCUSSION

This meta-analysis of 12 observational studies including more than 2000 patients represents the largest comparative assessment to date of self-expanding vs balloon-expandable valve platforms in patients with bicuspid aortic stenosis undergoing TAVI. Our findings show comparable short- and intermediate-term mortality outcomes between valve types, while identifying potential differences in procedural complications and valve performance.

No statistically significant differences were observed between BEVs and SEVs for any of the mortality endpoints evaluated, including 30-day all-cause mortality, cardiovascular mortality, procedural mortality, and 1-year all-cause and cardiovascular mortality. This finding is consistent with the existing literature. A study focusing on patients with BAV and small aortic annulus reported no significant differences in 1-year all-cause mortality or cardiac mortality between the 2 valve types.23 The 30-day and 1-year mortality rates were also comparable to those reported for TAVI in tricuspid aortic valve disease in contemporary studies24. Taken together, these consistent results suggest that, in terms of overall survival, both platforms offer comparable safety in the short- to intermediate-term in patients with BAV.

In contrast, our analysis identified a significantly higher risk of annular rupture with BEVs, a finding consistent with prior evidence. The AD-HOC registry, one of the most robust datasets included in this meta-analysis, reported annular rupture exclusively in the BEV group (0.7% vs 0.0% for SEVs).12 Excessive area-based oversizing > 20% has been identified as a key predictor of BEV-related rupture, with an 8-fold increased risk reported in one study.25 In contrast, SEVs are rarely associated with annular rupture unless aggressive balloon pre- or postdilatation is performed.26 The mechanism underlying this difference is often attributed to radial force and deployment mechanism: BEVs expand with a fixed balloon size, potentially exerting greater and more localized stress, whereas SEVs deploy gradually, allowing greater conformability to irregular anatomies such as BAV. Notably, Lee et al. reported no cases of annular rupture in either the BEV or SEV groups when utilizing Wei’s sizing method was used, suggesting that meticulous preprocedural planning can mitigate this severe complication regardless of valve type.13 Overall, these data suggest that, in patients with BAV, especially those with complex annular calcification or noncircular annuli, SEVs may offer a safer profile regarding annular integrity.

Regarding paravalvular leak, BEVs were associated with a significantly lower incidence of moderate-to-severe PVL than SEVs, a finding supported by the existing literature. The AD-HOC registry reported a markedly lower risk of moderate or greater paravalvular regurgitation with BEVs,12 and the BEAT Registry specifically highlighted that, in patients with large aortic annuli, SEVs were associated with substantially higher PVL rates.27 Furthermore, Sá et al. found that new-generation BEVs were associated with significantly less PVL than newer-generation SEVs.9 Although significant heterogeneity was observed for this outcome in our analysys, sensitivity analyses consistently favored BEVs. The observed heterogeneity is likely attributable to differences in BAV morphology, including Sievers classification and severity of raphe calcification, as well as variations in sizing strategies and the inclusion of different device generations. The clinical relevance of this finding should not be understated, because moderate-to-severe PVL has been consistently associated with higher rates of overall mortality, rehospitalization, and cardiovascular mortality.28 While contemporary SEV designs incorporate modifications aimed at minimizing paravalvular regurgitation, BEVs continue to demonstrate lower rates of moderate- to-severe PVL in the current comparative evidence base.

Moreover, BEVs were associated with a significantly lower need for second valve implantation, a finding that differs from some earlier meta-analyses in which this outcome did not reach statistical significance. The inclusion of more recent and larger datasets, particularly the AD-HOC registry, likely increased the statistical power to detect this difference.24 This result is consistent with an observational study showing second valve implantation in 9.3% of SEV patients treated with SEVs vs 0% of those treated with BEVs among patients with a dilated ascending aorta, in which BAV anatomy emerged as an independent predictor of device failure in multivariable analysis.29 The need for a second valve often arises from procedural complications, such as significant PVL, device malposition, or device failure, all of which may be compounded by the complex aortic anatomy inherent to BAV. These findings underscore the importance of tailored procedural strategies for optimizing TAVI outcomes in this population.

New permanent pacemaker implantation (PPI) was significantly less frequent with BEVs, which is consistent with previous data in both general TAVI populations and BAV-specific subgroups. A meta-analysis in patients with small aortic annulus found that BEVs were associated with a lower risk of PPI,20 and a recent meta-analysis of 16 studies comparing third-generation devices confirmed that BEV use was associated with a significantly reduced PPI risk.30 The higher PPI rates observed with SEVs are generally attributed to their radial force characteristics and tendency toward deeper implantation, which may compress the conduction system. Of note, specific BAV morphological features, such as Sievers type 1 classification, have also been independently associated with higher PPI risk, highlighting the multifactorial nature of this complication23.

No statistically significant differences were identified between valve types for stroke or coronary obstruction. The absence of a difference in stroke incidence is consistent with most contemporary literature in patients with BAV,31 although some analyses of specific subgroups, such as patients with small aortic annulus, have reported lower stroke risk with BEVs, suggesting that patient-level anatomical and clinical factors may modulate this risk.20 Coronary obstruction occurred at very low absolute rates in both groups, which is consistent with pooled estimates from the broader BAV-TAVI literature,32,33 and the comparable incidence between platforms is particularly reassuring for this potentially catastrophic complication.

Limitations

This meta-analysis provides valuable insights but is subject to several limitations inherent to the pooled observational data. First, the primary limitation is the reliance on observational studies, which introduces potential selection bias and unmeasured confounding. The absence of dedicated randomized clinical trials (RCTs) comparing BEVs and SEVs specifically in patients with BAV means that conclusions drawn from meta-analyses of observational data, although informative, cannot establish causality with the same certainty as RCTs. Second, heterogeneity across the included studies for some outcomes may reflect differences in patient characteristics, evolving procedural techniques and the inclusion of different generations of transcatheter heart valves. Third, although the meta-analysis focused on patients with BAV, the specific characteristics of the included BAV anatomies could not be explored and may not represent the full spectrum of BAV disease. Fourth, the inclusion of various device generations, including the Acurate neo/neo2 transcatheter heart valve, a platform no longer commercially available, represents an additional source of heterogeneity and may limit the generalizability of the findings to contemporary practice. Unfortunately, a subgroup analysis comparing device generations was not possible because individual patient-level data were not available from the included studies.

Future directions

The current meta-analysis, along with the existing literature, highlights several critical areas for future research to further refine valve selection in TAVI for bicuspid aortic stenosis. Although dedicated randomized controlled trials comparing SEVs and BEVs in patients remain absent from the literature, several registered trials may provide valuable insights into the performance of these valve types in the BAV population. The STAR trial (NCT02541877), a multicenter randomized study focusing on sizing strategies for type 0 bicuspid aortic stenosis using SEVs, may provide critical data on procedural outcomes, hemodynamic performanc, and optimal sizing approaches specific to BAV anatomy. The study of safety and efficacy profile of TAVI in intermediate-risk BAV patients (NCT03163329) and the HANGZHOU Solution trial in bicuspid aortic stenosis (NCT04722796) may contribute data on both valve types across different risk profiles and procedural approaches, although without direct comparison of these devices for definitive conclusions on valve selection.

CONCLUSIONS

This meta-analysis of 12 observational studies represents the largest comparison of SEVs and BEVs in patients with BAV undergoing TAVI. Both valve platforms showed equivalent mortality outcomes across all the time horizons. However, BEVs were associated with a higher risk of annular rupture but lower rates of moderate-to-severe paravalvular leak, reduced need for second valve implantation and fewer permanent pacemaker implantations than SEVs.

These findings suggest potential differences in complication profiles between valve types, though current evidence is limited by the nature and quality of available observational data. Further research with randomized designs and contemporary valve generations is needed to confirm these associations.

FUNDING

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

ETHICAL CONSIDERATIONS

This systematic review and meta-analysis used previously published data and did not involve primary data collection from human participants. Therefore, ethical approval and informed consent were not required. All included studies had received appropriate ethical approval from their respective institutions, as reported in the original publications. This study complies with the SAGER guidelines for reporting sex and gender data in research.

STATEMENT ON THE USE OF ARTIFICIAL INTELLIGENCE

Claude (Anthropic) was used during the preparation of this manuscript. It assisted the authors with language editing, grammar, and readability of the text. Furthermore, it was used to support rephrasing and connecting content within the introduction and discussion sections.

The tool was not used in the study design, literature searches, data collection, data extraction, statistical analyses, or interpretation of the results. All scientific content, references, and conclusions were produced by the authors. All output from the tool was reviewed and edited before inclusion in the manuscript, and the authors take full responsibility for the final content.

AUTHORS’ CONTRIBUTIONS

Conceptualization: L. Marqueño da Cunha, N.V. Sala da Silva, and W.F. Gomes. Methodology: L. Marqueño da Cunha, N.V. Sala da Silva, and W.F. Gomes. Literature search: L. Marqueño da Cunha, N.V. Sala da Silva, and W.F. Gomes. Data extraction: L. Marqueño da Cunha and N.V. Sala da Silva. Quality assessment: P.F. Gomes Nicz , W.F. Gomes. Statistical analysis: W.F. Gomes. Writing—original draft: L. Marqueño da Cunha, N.V. Sala da Silva, W.F. Gomes, P.F. Gomes Nicz, D.C. Nercolini, A. Dumsch de Aragon Ferreira, and C.A. Kenji Nakashima. Writing—review and editing: all authors; Supervision: W.F. Gomes. Final approval: all authors.

CONFLICTS OF INTEREST

W.F. Gomes has received proctorship honoraria from Meril Lifesciences and Abbott Cardiovascular.

ACKNOWLEDGEMENTS

The authors thank all investigators of the original studies included in this meta-analysis for their valuable contributions to the evidence base.

SUPPLEMENTARY DATA

REFERENCES

INTRODUCTION

Percutaneous coronary intervention (PCI) has progressively expanded to increasingly complex anatomical and clinical scenarios, including chronic total coronary occlusions (CTOs), long and diffuse disease, severe calcification, and tortuous coronary anatomy. In this setting, coronary microcatheters have become indispensable adjunctive devices, facilitating guidewire support and exchange, distal contrast injection, and device delivery.

Despite their widespread use, clinical evidence supporting the safety and performance of coronary microcatheters is largely derived from single-center experiences,1,2 retrospective series,3-5 or extrapolated from broader CTO registries.6-11 Under the current European Medical Device Regulation (MDR 2017/745), manufacturers and sponsors are required to generate post-market clinical follow-up (PMCF) evidence to continuously confirm the safety and clinical performance of CE-marked devices throughout their life cycle. Real-world data, collected prospectively in routine clinical practice, play a central role in fulfilling these regulatory requirements.

The Navitian coronary microcatheter (iVascular, Spain) is a single-lumen, over-the-wire compatible device designed to facilitate guidewire advancement, exchange, and distal infusion during PCI. While its use is well established in daily practice, prospective multinational PMCF evidence evaluating its performance in real-world complex PCI remains limited.

The present study was designed as a prospective, multicenter, multinational PMCF investigation to evaluate the safety and clinical performance of the Navitian coronary microcatheter in routine practice across Spain and Portugal. By focusing on highly complex lesions and contemporary PCI workflows, this study aims to provide clinically relevant and regulatorily robust evidence supporting the real-world use of the device under MDR requirements.

METHODS

Study design

We conducted a prospective, multicenter, multinational observational PMCF study conducted in routine clinical practice in Spain and Portugal (ClinicalTrials.gov ID NCT05292118). The study was designed and performed in full compliance with the European Medical Device Regulation (MDR 2017/745), ISO 14155:2020 for clinical investigations of medical devices, the Declaration of Helsinki, and applicable national regulatory requirements.

The study followed a single-arm design without protocol-mandated additional diagnostic or therapeutic procedures beyond standard clinical care. Device use was limited to its CE-marked indications, and all procedural decisions were left to the discretion of the treating operators.

Study population

Adult patients aged 18 years or older undergoing PCI in whom use of the Navitian coronary microcatheter was attempted, either alone or in combination with other coronary devices, were consecutively included. Eligibility was based on routine clinical practice and operator judgment.

Patients were excluded only if they explicitly declined participation or refused to provide written informed consent, in accordance with local ethical and regulatory requirements.

Participant centers

The study was conducted across 7 high-volume PCI centers in Spain and Portugal, reflecting a broad spectrum of contemporary European practice. This multinational design aimed to enhance the external validity and generalizability of the findings across different healthcare systems and procedural environments.

Device description

The Navitian coronary microcatheter is a single-lumen device compatible with 0.014 in guidewires, designed to provide guidewire support, facilitate wire exchange, and allow distal infusion of contrast or saline. The device features a hydrophilic distal coating to enhance trackability through tortuous and complex coronary anatomy and incorporates a rounded atraumatic distal tip to reduce the risk of arterial damage during advancement, as well as radiopaque markers to aid fluoroscopic visualization (figure 1). The microcatheter was used strictly according to its instructions for use.

Figure 1. Navitian coronary microcatheter. Schematic representation of the Navitian microcatheter (iVascular, Barcelona, Spain), a rapid-exchange coronary microcatheter designed for use in complex percutaneous coronary intervention. The device features a hydrophilic distal coating to optimize trackability through tortuous and calcified coronary anatomy, a rounded and atraumatic distal tip to reduce the risk of vessel injury during advancement, and radiopaque markers to aid fluoroscopic visualization. The shaft design is intended to provide an optimal balance of pushability, flexibility, and torque transmission for crossing chronic total coronary occlusions and other complex lesions.

Endpoints

The primary device effectiveness endpoint was successful crossing of the target lesion or occlusion with the microcatheter.

Procedural effectiveness was defined as final Thrombolysis in Myocardial Infarction grade-3 flow12 with < 30% residual percent diameter stenosis in the treated segment at the end of the procedure.

Device safety endpoints included the absence of device-related mechanical failures, defined as device rupture, kinking, or complicated device retrieval.

Clinical outcomes were assessed during in-hospital follow-up until discharge and included the occurrence of target lesion failure, defined as the composite of cardiac death, target-vessel myocardial infarction, or clinically driven target lesion revascularization.

Given the sequential dependency inherent to complex PCI and CTO techniques, device effectiveness was assessed only in lesions in which successful guidewire crossing was achieved. Microcatheter performance cannot be meaningfully evaluated in the absence of wire passage; therefore, analyses based on evaluable lesions represent standard practice in CTO and complex PCI studies.

Data collection and monitoring

Data were collected prospectively using a standardized electronic case report form. Remote data monitoring was performed to ensure data completeness and consistency. Cases involving device or procedural failures were subject to targeted review, including source document verification when deemed necessary.

Statistical analysis

Continuous variables are expressed as mean ± standard deviation or median [IQR], as appropriate. Categorical variables are expressed as absolute counts and percentages. The incidence rate of clinical and procedural events is expressed as cumulative incidence during the in-hospital follow-up period.

The planned sample size was based on a noninferiority framework using published benchmarks for device effectiveness,13-15 assuming a reference success rate of 85%, a noninferiority margin of 10%, an alpha level of 0.05, and 80% statistical power.

All analyses were performed using JMP statistical software (version 16; SAS Institute, Cary, NC, United States).

Causality assessment

Causality between adverse events and the study device was assessed based on temporal sequence, inspection of device integrity, procedural context, and operator adjudication. Events occurring after the use of downstream interventional devices, such as balloons or stents, and in the absence of microcatheter malfunction were not considered causally related to the Navitian microcatheter.

RESULTS

Study conduct and population

After approval by the reference ethics committee and local ethics committees, patient enrollment started on 30 September, 2022, and was completed on 11 July, 2024. A total of 102 patients treated across participant centers in Spain and Portugal were included. Overall, 115 coronary lesions were addressed using the Navitian coronary microcatheter.

In 12 patients, the device was used to treat > 1 lesion (11 patients with 2 lesions and 1 patient with 3). Baseline clinical characteristics reflected a population with high cardiovascular risk and complex coronary disease. Table 1 and table 2 summarize the baseline characteristics of patients and lesions.

Table 1. Baseline characteristics

Table 2. Baseline characteristics

The mean age was 68.7 ± 9.9 years, and 81.4% of patients were male. Relevant comorbidities included diabetes mellitus in 37.2%, chronic kidney disease in 12.7%, and prior myocardial infarction in 36.3%. Previous PCI had been performed in 45.1%, and 21.6% of patients were o chronic oral anticoagulation at the time of the index procedure. Multivessel coronary artery disease was present in 58.8% of patients. Clinical presentation included ST-segment elevation acute coronary syndrome in 8.8% and non–ST-segment elevation acute coronary syndrome in 33.3%.

Lesion and procedural characteristics

Lesion complexity was high. Among the 115 treated lesions, 57.4% were CTOs, 87.8% were classified as American Heart Association type B2/C,16 64.3% showed diffuse disease, and 47.0% exhibited moderate-to-severe calcification. The median lesion length was 30 mm [IQR, 20–40 mm], with a median baseline stenosis of 100%. Lesions with ostial involvement were treated in 16.5% and bifurcation lesions in 19.1%. High vessel tortuosity, considered a key determinant of microcatheter performance, was documented in 16.5% of lesions.

The Navitian microcatheter was used to facilitate guidewire support or advancement in 83.5% of cases, for guidewire exchange after initial wire crossing of very complex lesions in 21.7%, and for distal infusion of contrast, drugs or saline in 6.1%, with overlap of indications in some procedures. Additional support techniques, such as guiding catheter extensions or balloon anchoring, were required in 13.9% of cases.

Device effectiveness and procedural success

Overall, successful crossing of the target lesion or occlusion with the Navitian microcatheter was achieved in 100 of 115 lesions (87.0%). All failed device crossings occurred in CTOs. Among lesions in which guidewire crossing was successfully achieved (n = 105), device effectiveness for lesion or occlusion crossing was 95.2%. In 5 cases, the lesion was successfully wired but the Navitian device could not cross: in 3 cases, no device was able to cross despite multiple attempts; in only 2 cases did another device successfully crossed the lesion, another microcatheter in 1 case and a 1-mm balloon in the other.

All 5 lesions in which the Navitian microcatheter failed to cross were CTOs (100%) located in the right coronary artery (4 of 5) or the first diagonal branch (1 of 5), mostly American Heart Association type C lesions (80%) and de novo (80%). This subgroup showed a high proportion of diffuse disease (80%) and a median lesion length of 32.5 mm [IQR, 28.5–44.75 mm], with moderate- to-severe calcification in 40% and high tortuosity in 40%. In 4 of 5 procedures additional support techniques were required, including a guide catheter extension system in 3 cases, and no device-related technical problems, perforations, dissections or thromboses occurred in this subgroup. Given the small sample size (n = 5), no formal statistical comparison with successfully crossed lesions (n = 100) was performed, but these findings suggest that failures were concentrated in long, diffuse right-coronary CTOs rather than in cases driven by a single identifiable anatomical feature.

In all cases in which the Navitian microcatheter successfully crossed the lesion, procedural success was achieved in 100%, defined as final Thrombolysis in Myocardial Infarction grade-3 flow with < 30% residual percent diameter stenosis in the treated segment.

Device safety and procedural complications

No device-related mechanical failures were observed. Specifically, there were no cases of device rupture, kinking, or complicated device retrieval.

Procedural complications occurred in a limited number of cases and included 3 coronary perforations, all treated with covered stents, 1 coronary dissection of grade C or higher, and 1 probable acute coronary thrombosis. However, the report’s adjudication narrative is important: the 3 perforations occurred after subsequent stent implantation and were managed with covered stents; 1 fatal despite pericardiocentesis and surgery. The probable acute thrombosis occurred after complex left main/left anterior descending coronary artery stenting and drug-coated balloon use, making exclusive attribution to the microcatheter unlikely. The dissection occurred in a failed crossing context in which aggressive CTO wiring, such as with a Pilot 200 guidewire, was used; it had no clinical sequelae.

In-hospital clinical outcomes

The median length of stay was 1 day [IQR, 1–3 days]. Three adverse clinical events were recorded during in-hospital follow-up: 1 target-vessel myocardial infarction and 2 cardiac deaths, 1 following coronary perforation, and 1 following probable acute coronary thrombosis. In all cases, these events occurred after the use of other interventional devices and were not directly attributable to the Navitian microcatheter.

DISCUSSION

In this prospective, multicenter, multinational PMCF study conducted under the requirements of the European Medical Device Regulation (MDR 2017/745), the Navitian coronary microcatheter demonstrated high effectiveness and an excellent device-related safety profile in a real-world population characterized by very high anatomical and clinical complexity. More than half of the treated lesions were CTOs and nearly 90% were classified as American Heart Association type B2/C, closely reflecting contemporary complex PCI practice in high-volume European centers.

Device effectiveness in the context of complex PCI and CTO practice

The primary effectiveness endpoint—successful lesion or occlusion crossing—was achieved in 87% of the total population, and 95.2% of evaluable lesions in which the guidewire successfully crossed, exceeding the 85% benchmark used for PMCF sample size assumptions. Importantly, failure to cross the lesion with the microcatheter was frequently driven by the inability of the guidewire to cross the lesion rather than by microcatheter malfunction. This observation underscores the sequential dependency inherent to complex PCI and CTO interventions and supports the methodological decision to assess device effectiveness only in lesions where guidewire passage was achieved.

When contextualized with published microcatheter experience, these findings appear consistent with contemporary CTO practice (table S1). In the first clinical experience with the NHancer locking microcatheter, Wilson et al.13 reported that the device contributed to at least 1 major procedural step in 85.9% of CTO cases and was the only microcatheter required in 68.4% of successful interventions. While Navitian is not a locking microcatheter, the present PMCF results similarly show that Navitian frequently enabled guidewire support and exchange: it was used for guidewire support in 83.5%, guidewire exchange in 21.7%. Adjunctive backup support, including guide extension or anchor techniques, was required in 13.9% of cases, a figure compatible with escalation patterns described in complex CTO practice.

Sidik et al.15 citer former studies showing procedural success rates in the mid-70% to mid-80% range for CTO-PCI using specialized microcatheters, such as the Corsair series, underscoring that real-world performance depends heavily on lesion complexity, crossing strategy, whether antegrade/retrograde, operator technique, and adjunctive device use.

Registry-based evidence further supports the central role of microcatheters in modern CTO algorithms. Data from contemporary European CTO registries9 indicate overall CTO PCI technical success rates in the range of 85% to 90%, with microcatheter use considered standard practice for wire support, exchange, and escalation strategies. In this context, the performance observed with Navitian aligns with expected outcomes for contemporary coronary microcatheters used in highly complex interventions.

Safety profile and attribution of adverse events

Device safety endpoints were robust: no device fracture, kinking, or difficult withdrawal were reported despite the complexity of the lesions treated. This is a key PMCF reassurance signal for a microcatheter intended for demanding anatomies, particularly in settings in which repeated device exchanges and tortuosity or calcification can increase mechanical stress.

Although 5 procedure-level complications were observed—3 perforations, 1 dissection grade ≥ C, 1 probable acute thrombosis—all the events occurred after the use of additional coronary devices, including guidewires, balloons, and stents, and were not adjudicated as causally related to the microcatheter. These nuances should be explicitly stated in the manuscript because they align with contemporary complex PCI reality: complications often reflect the cumulative risk of the full procedural sequence (wiring, microcatheter manipulation, ballooning, atherectomy/lithotripsy when used, stenting, and optimization), rather than a single device.

Compared with published CTO microcatheter series,13-15 the observed complication profile appears consistent with the underlying risk associated with complex PCI. For example, in an early clinical experience with the NHancer microcatheter,13 Wilson et al. reported a low complication rate, with 1 case of tamponade following guidewire exit that was successfully managed with a covered stent and no reported device failures. Differences compared with the procedural complications observed with Navitian should be interpreted cautiously because a) the Navitian PMCF study included non-CTO complex disease in addition to CTO and reflects broader routine-practice indications, and b) event attribution in the Navitian study points to downstream therapy, particularly stenting, rather than microcatheter malfunction.

Clinical outcomes and regulatory relevance

Three adverse clinical events were recorded during in-hospital follow-up: 1 target-vessel myocardial infarction and 2 cardiac deaths, 1 following coronary perforation, and 1 following probable acute coronary thrombosis. In all cases, these events occurred after the use of other interventional devices and were not directly attributable to the Navitian microcatheter.

Although longer-term clinical outcomes were not assessed, the absence of early device-related safety signals and the favorable procedural success rates provide meaningful evidence supporting the performance of the device under intended-use conditions.

From a regulatory perspective, this study illustrates the value of prospective real-world PMCF investigations in fulfilling MDR 2017/745 requirements.17,18 By generating device-specific evidence in routine clinical practice across multiple centers and countries, such studies complement premarket evaluations and contribute to ongoing benefit–risk assessment throughout the device life cycle. The multinational design further enhances the generalizability of the findings across different European health care environments. Consistent with this approach, prospective multicenter clinical follow-up has recently been applied to other coronary devices for complex PCI, as exemplified by the first-in-man evaluation of the Naviscore scoring balloon for moderate-to-severe calcified lesions.19

Limitations

Some limitations inherent to pragmatic PMCF studies should be acknowledged, such as a) single-arm observational design without a concurrent comparator microcatheter, limiting causal inference; b) heterogeneity of lesion subsets, including CTO and non-CTO complex disease, which enhances generalizability but complicates direct comparison to CTO-only microcatheter series such as NHancer; c) event attribution constraints in complex PCI: complications may be multifactorial and temporally linked to downstream devices or procedural steps, as occurred in this cohort. In addition, angiographic outcomes were site reported and not adjudicated by an independent core laboratory, and formal independent event adjudication was not performed; d) short follow-up and limited to discharge, restricting assessment of longer-term target lesion failure components, particularly repeat revascularization. Clinical follow-up was limited to the in-hospital period. This timeframe reflects the primary endpoint of the present PMCF study, which was to confirm device safety and technical performance under intended-use conditions, in full compliance with MDR 2017/745 requirements, rather than to assess long-term clinical effectiveness; e) standardized definitions for some anatomical features of complex PCI were lacking. No prespecified definition was applied in this study for diffuse disease, significant calcification, or coronary tortuosity; operators classified these characteristics according to their usual angiographic criteria. Several definitions of long or diffuse disease, with lesion length thresholds ranging from 20 mm to ≥ 40 mm have been proposed in the literature, but none was adopted uniformly across centers. This may introduce interobserver variability and should be taken into consideration when interpreting the complexity descriptors of the cohort.

CONCLUSIONS

In a prospective, multicenter, multinational PMCF study conducted under the European MDR framework, the Navitian coronary microcatheter demonstrated high effectiveness, excellent device-related safety, and favorable early clinical outcomes in real-world complex PCI, including CTO interventions. These findings support its continued use in routine practice and provide regulatorily robust evidence of clinical performance across different European health care environments.

FUNDING

This investigator-initiated post-market clinical follow-up study was sponsored by Fundación EPIC, a nonprofit academic organization. An unrestricted grant from the manufacturer of the device, iVascular (Barcelona, Spain) supported the development of the study. However, iVascular had no role in study design, data collection, data analysis, interpretation of the results, or preparation of the manuscript.

ETHICAL CONSIDERATIONS

The study protocol was approved by the Ethics Committee of the Lead Site (Hospital Universitario de León) and by each participant site as required by national regulations. The investigation was conducted in full compliance with the principles outlined in the Declaration of Helsinki, ISO 14155:2020 for clinical investigation of medical devices, the EU Medical Device Regulation (MDR 2017/745), and applicable local laws and regulations for postmarket clinical follow-up studies. Written informed consent was obtained from all participants prior to enrollment. Sex and gender considerations were addressed in accordance with the SAGER (Sex and Gender Equity in Research) guidelines: participants were enrolled consecutively without selection based on sex, and sex-disaggregated baseline characteristics and outcomes are reported where appropriate. The study sample reflects the real-world distribution of patients referred for complex percutaneous coronary intervention during the enrollment period.

STATEMENT ON THE USE OF ARTIFICIAL INTELLIGENCE

During the preparation of this work the authors used ChatGPT to support language editing and manuscript structuring. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

AUTHORS’ CONTRIBUTIONS

A. Pérez de Prado conceived and designed the study, served as coordinating investigator, supervised data acquisition and analysis, and drafted and critically revised the manuscript. A. Rodrigues, M. Sabaté, I.J. Amat Santos, T. García Camarero, A. Gómez Menchero, and B. García del Blanco served as site principal investigators, contributed to the study design, and participated in patient enrollment, data acquisition, and clinical follow-up. M. López Benito, P. Braga, A. Regueiro, C. Cortés Villar, J. Roa Garrido, and B. Serra Creus participated in patient enrollment, data collection, procedural documentation, and clinical follow-up at the respective sites. J.M. de la Torre-Hernández contributed to the study design, data interpretation, and critical review of the manuscript. All authors reviewed, revised, and approved the final version of the manuscript and agreed to be accountable for all aspects of the work.

CONFLICTS OF INTEREST

J.M. de la Torre-Hernández is editor-in-chief of REC: Interventional Cardiology. A. Pérez de Prado is associate editor of REC: Interventional Cardiology; the journal’s editorial procedure to ensure impartial handling of the manuscript has been followed. A. Pérez de Prado, M. Sabaté, B. García del Blanco, and J.M. de la Torre-Hernández report consulting or speaker fees from iVascular outside the submitted work. The remaining authors declare no conflicts of interest directly related to the subject of this study. The sponsor of the study (Fundación EPIC) received an unrestricted grant from iVascular for the conduct of this postmarket clinical follow-up; iVascular had no role in the study design, collection, analysis or interpretation of data, manuscript preparation, or the decision to submit the manuscript for publication.

SUPPLEMENTARY DATA

REFERENCES

INTRODUCTION

Acute coronary syndrome, particularly ST-segment elevation myocardial infarction (STEMI), is the leading cause of out-of-hospital cardiac arrest.1 The main underlying mechanism is reversible ischemia, provided that reperfusion is achieved within the first few hours after STEMI.2 Approximately 1 in 20 patients with STEMI presents with cardiac arrest as the initial or early sign of myocardial infarction.3-5 Resuscitated cardiac arrest before primary angioplasty (RCABPA) is an important prognostic marker, and has been associated with 30-day mortality rates of 40% to 60%.3,6 Published studies differ regarding the long-term prognosis of patients who survive the in-hospital phase after RCABPA. Some registries have reported no long-term differences in outcomes,7,8 whereas others have found higher mortality, even after adjustment for the poorer baseline clinical profile of patients with RCABPA.9,10 However, all available studies consistently show that most events occur during the in-hospital phase or within the first 30 days after RCABPA.

“STEMI network” programs are designed to ensure rapid identification of STEMI and timely delivery of the most appropriate reperfusion strategy, preferably primary angioplasty.5 Studies evaluating the characteristics and prognosis of RCABPA have generally paid little attention to the specific point in the STEMI care pathway at which cardiac arrest occurs. Their inclusion criteria have typically included patients with RCABPA occurring outside the hospital setting, without specifying whether the cardiac arrest occurred before or after first medical contact or how close the patient was to resources for advanced cardiopulmonary resuscitation (CPR). In addition, the prognosis of RCABPA occurring in hospitals with CPR capability but without primary angioplasty capability has not been specifically analyzed. The poor short-term prognosis associated with RCABPA may be related to the clinical characteristics and consequences of the STEMI per se or to delays and limitations in CPR when cardiac arrest occurs outside the hospital setting.

Therefore, analyzing the characteristics and consequences of RCABPA based on the place of occurrence may help identify opportunities to improve the care and prognosis of patients with STEMI.

The aim of this study was to analyze short-term mortality in patients with RCABPA, with particular attention to the place where cardiac arrest occurred within the care pathway initiated by activation of the STEMI code after diagnosis.

METHODS

Design

This observational study used a historical cohort of consecutive patients admitted for primary angioplasty with a diagnosis of STEMI and an indication for reperfusion.

Study population

We included all patients who arrived at the cath lab of Hospital Clínico Universitario Virgen de la Arrixaca (El Palmar, Murcia, Spain) between 1 January, 2021, and 31 December, 2024 through the regional STEMI code program and with an indication for primary angioplasty. In patients with > 1 episode during the inclusion period, only the index episode was included. In 2023, the reference population covered by the regional STEMI code program for the study center was 1,132,310 inhabitants. Six non–PCI-capable hospitals referred patients to the study hospital, which served as the reference center. According to the STEMI code protocol, activation must occur at first medical contact, and patients should be transferred directly from the place of activation to the cath lab, without intermediate stops at other hospitals or emergency departments.

Variables

Data were obtained from the prospective patient registry of the cath lab at the reference hospital for primary angioplasty, where demographic characteristics, clinical presentation, procedural data, and follow-up information are systematically recorded. Missing data were completed by reviewing the regional electronic health record or by contacting patients or their relatives by telephone.

RCABPA was defined as cardiac arrest with return of spontaneous circulation after cardiopulmonary resuscitation maneuvers. Cardiac arrests occurring after arrival at the cath lab were excluded. Patients who died before or during transfer to the reference hospital were not included.

Patients were classified into 5 groups according to the location where RCABPA occurred: a) in-hospital RCABPA, occurring in referral hospitals before transfer or at the reference hospital prior to the arrival at the cath lab; b) RCABPA in a medicalized ambulance occurring in an ambulance with defibrillation capability; c) RCABPA in primary care, occurring in a health center or out-of-hospital emergency department with defibrillation capability; d) RCABPA outside the health care setting, occurring at home or in a public place without medical or paramedical personnel present; and e) no RCABPA, used as the control group.

All primary care centers and medicalized ambulances have the capacity to provide advanced CPR. We could not determine the exact number of patients with out-of-hospital RCABPA who may have exceptionally benefited from the proximity of a defibrillation team.

For comparisons by grouped location, groups a, b, and c were classified as RCABPA occurring in a health care setting, whereas groups b and c were classified as RCABPA occurring in an out-of-hospital health care setting.

Vital status at 30 days was obtained from the patients’ electronic health records or, when unavailable, by telephone contact.

The study was conducted in full compliance with the principles outlined in the Declaration of Helsinki. Episodes were collected retrospectively. The study was approved by the local ethics committee.

Statistical analysis

Quantitative variables are expressed as mean and standard deviation or as median and 25^th-75^th percentiles when they did not meet normality criteria, as assessed with the Shapiro-Wilk test. Quantitative variables that did not meet the normality criteria were compared using the nonparametric Mann-Whitney U test; normally distributed variables were compared using the Student t test for independent samples. Qualitative variables are expressed as absolute frequencies and percentages and were compared using the Pearson chi-square test. As specified in the tables and Results section, patients without RCABPA and those with in-hospital RCABPA were used as reference groups, as appropriate. Thirty-day mortality was assessed using Kaplan-Meier survival analysis. Survival curves were plotted for patients without RCABPA and for those with RCABPA at each individual and grouped location. Survival curves were compared using the log-rank test, with patients without RCABPA and those with in-hospital RCABPA used as reference groups. In all cases, P values < .05 were considered statistically significant.

To determine the contribution of RCABPA to 30-day mortality compared with the absence of RCABPA, as well as the contribution of RCABPA according to its location, logistic regression models were constructed. Models were adjusted for variables that were asymmetrically distributed across the study subgroups in the univariate analysis (P > .1) or that have been associated in the literature with higher mortality: age, diabetes, cardiogenic shock, performance of percutaneous coronary intervention (PCI), anterior infarct location, and delay from symptom onset to first medical contact. The absence of significant multicollinearity among the included variables was confirmed. The predictive performance of each model was assessed using the receiver operating characteristic curve, based on 30-day mortality predicted by the models and observed mortality.

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (IBM Corp., United States), and SigmaPlot for Windows, version 11.0 (Systat Software, Inc., United States).

RESULTS

Population description

During the 4-year study period, there were 2463 activations of the STEMI code program with an indication for primary angioplasty. A total of 119 cases (4.9%) were excluded because they had previously been referred for primary angioplasty, leaving a final sample of 2344 patients, 170 of whom (7.3%) experienced RCABPA: 40 (1.7%) in a hospital setting, 13 (0.6%) in primary care, 33 (1.4%) in an advanced life support ambulance, and 84 (3.6%) outside the health care setting. Eight aborted STEMI code activation due to patient death before arrival at the cath lab were recorded, including 3 deaths in an advanced life support ambulance. The final cause of death could not be determined. All these patients had cardiogenic shock at the time of activation and were not included in the study.

The initial rhythm of RCABPA was shockable in 158 cases (92.9%). Baseline characteristics and infarct presentation based on the presence of RCABPA are shown in table 1. Patients with RCABPA underwent PCI less frequently than those without RCABPA (76.5% vs 85.6%; P = .001). PCI was not performed in 2 cases of RCABPA because the patients died before the procedure could be initiated, or in 28 cases because no culprit coronary lesion was identified.

Table 1. Baseline, acute myocardial infarction presentation, and procedural characteristics according to the occurrence or absence of resuscitated cardiac arrest before primary angioplasty

Baseline characteristics and infarct presentation based the presence and location of RCABPA are shown in table 2.

Table 2. Baseline, acute myocardial infarction presentation, and procedural characteristics based on the location of resuscitated cardiac arrest before primary angioplasty

A total of 32 patients received mechanical circulatory support, representing 1.4% of the overall population and 17.6% of patients with cardiogenic shock. In this group, the 30-day mortality rate was 47%. Mechanical circulatory support was used in 10 patients with cardiogenic shock and RCABPA (21.2%; 30-day mortality, 70%) and in 22 patients with cardiogenic shock without RCABPA (16.3%; 30-day mortality, 36.4%).

Mortality based on the location of RCABPA

The 30-day all-cause and cardiovascular mortality rates are shown in table 3, and mortality grouped by RCABPA location is shown in figure 1. The 30-day all-cause mortality rate was significantly higher in patients with RCABPA than in those without RCABPA (31.2% vs 6.3%; P < .0005). Mortality increased progressively from in-hospital RCABPA to RCABPA occurring outside the health care setting, where it approached 50% (48.8%). Compared with in-hospital RCABPA, RCABPA occurring outside the hospital was associated with a significantly higher 30-day mortality rate or showed a clear trend toward a higher mortality rate (table 3 and figure 2).

Figure 1. Thirty-day mortality according to the location of resuscitated cardiac arrest before primary angioplasty (RCABPA).
^a Comparison with patients without RCABPA.
^b Comparison with patients with in-hospital RCABPA.

Table 3. Interventional procedure outcome and 30-day mortality based on the location of resuscitated cardiac arrest before primary angioplasty

Figure 2. Kaplan-Meier curves for 30-day survival and significance of the log-rank test, using patients without resuscitated cardiac arrest before primary angioplasty (RCABPA) as the reference group.

Patients with RCABPA in a nonhospital health care setting and those with out-of-hospital RCABPA had a significantly higher 30-day mortality rate compared with patients without RCABPA (19.6% vs 6.3%; P < .0005, and 38.5% vs 6.3%; P < .0005, respectively). The 30-day mortality rate among patients with RCABPA in a nonhospital health care setting was numerically much higher than that among patients with in-hospital RCABPA, although the difference did not reach statistical significance (19.6% vs 7.5%; P = .123). When out-of-hospital RCABPA was analyzed as a grouped category, the mortality rate was significantly higher than that observed for in-hospital RCABPA (38.5% vs 7.5%; P < .0005) (figure 2).

After adjustment in logistic regression models, RCABPA was independently associated with a higher 30-day mortality rate, regardless of the presence of cardiogenic shock and other variables associated with poorer prognosis (OR, 5.99; 95%CI, 3.78-9.51; P < .0005). This association was not observed for in-hospital cardiac arrest. Furthermore, the independent predictive value for mortality was observed for out-of-hospital RCABPA (OR, 9.15; 95%CI, 5.59-14.95; P < .0005) and for RCABPA occurring outside the health care setting (OR, 20.9; 95%CI, 11.40-38.30; P < .0005). When in-hospital RCABPA was used as the reference category, the independent predictive value was maintained for virtually all nonhospital RCABPA locations. There was no significant interaction between cardiogenic shock and RCABPA in any of the models. The areas under the receiver operating characteristic curves exceeded 85% for all models (table 4).

Table 4. Adjusted contribution of recovered cardiac arrest before primary angioplasty, based on location, to 30-day mortality. Logistic regression analysis with predictors of 30-day mortality

DISCUSSION

To our knowledge, this study analyzes the largest series of patients with RCABPA in the setting of STEMI based on the location where cardiac arrest occurred. The main findings were as follows: a) approximately half of RCABPA episodes in patients with STEMI treated within a STEMI code program occurred in a health care setting; b) although STEMI complicated by RCABPA before arrival at the cath lab was associated with a significantly higher 30-day mortality rate, prognosis varied based on RCABPA location; c) patients with in-hospital RCABPA did not have a higher crude or adjusted mortality rate compared with patients without RCABPA; and d) the prognostic differences observed based on RCABPA location may reflect differences in CPR availability and effectiveness, rather than differences in RCABPA per se. These findings suggest that improvements in the STEMI code care pathway could reduce mortality (figure 3).

Figure 3. Central illustration. Study design and main results. PCI, percutaneous coronary intervention; RCABPA, resuscitated cardiac arrest before primary angioplasty; STEMI, ST-segment elevation myocardial infarction.

Incidence of cardiac arrest before reperfusion in patients with STEMI

Although it has been estimated that approximately half of sudden deaths are of coronary origin, these cannot be definitely classified because they represent the first sign of ischemic heart disease and patients are not resuscitated.11,12 STEMI code programs have reported an incidence of RCABPA ranging from 2% to 10%,3,6,13-15 with variability depending on inclusion criteria and delays to reperfusion. In our series, the incidence of RCABPA was 7.3%, which is close to the highest reported rates. This may be explained by the comprehensive nature of the registry, its regional coverage, and the median interval from symptom onset to reperfusion, which exceeded 200 minutes. A Danish nationwide case-control study16 including 1901 patients with STEMI reported an 11.6% incidence of ventricular fibrillation before primary angioplasty; although 83% of these episodes occurred outside the hospital, the precise location was not specified.

We found no previous studies that analyzed prognosis based on the location of RCABPA within the STEMI care pathway. This limits comparisons with our series, except for episodes occurring during transfer in an advanced life support ambulance. The first trials showing the superiority of primary angioplasty over thrombolysis in patients requiring transfer for angioplasty reported ventricular fibrillation during ambulance transport in 1.4% of the patients from the DANAMI-2 trial17 and 0.7% of the patients from the PRAGUE-2 trial.18 These figures are similar to the 1.4% observed in our cohort. In an observational study of 7393 patients with myocardial infarction transferred to a tertiary referral center, 5.6% experienced cardiac arrest before hospital arrival.19 In our series, > 35% of RCABPA episodes occurred in the out-of-hospital health care setting, underscoring the importance of appropriate training and the availability of material and human resources to provide high-quality cardiac resuscitation in these settings.

Differences in the prognosis of cardiac arrest in patients with STEMI based on location

Despite its high mortality rate, out-of-hospital cardiac arrest of coronary origin has a better prognosis than cardiac arrest of noncoronary origin.20 Currently, most studies21 assessing the characteristics or prognosis of out-of-hospital RCABPA in patients with or without ST-segment elevation myocardial infarction, have defined it simply as cardiac arrest occurring outside the hospital, without considering the specific setting or whether cardiopulmonary resuscitation was immediately available. In the present study, prognosis differed significantly based on the location of out-of-hospital RCABPA. Mortality increased progressively from RCABPA occurring at the hospital or health care setting to RCABPA occurring in a nonhospital or outside the health care setting (7.5%, 19.6%, and 48.8%, respectively). These differences do not appear to be fully explained by the patients’ baseline clinical characteristics or infarct presentation, because the independent predictive value of RCABPA location for 30-day mortality persisted after adjustment in the different multivariate models. Notably, mortality among patients with in-hospital RCABPA was not significantly different from that observed in patients without RCABPA (7.5% vs 6.3%; P = .748; figure 2). By contrast, mortality was 3 times higher among those with RCABPA occurring in a nonhospital health care setting (19.6% vs 6.3%; P < .0005). In the adjusted model (table 4), RCABPA occurring in a nonhospital health care setting was significantly associated with a higher 30-day mortality rate compared with patients without RCABPA and those with in-hospital RCABPA (OR, 13; P = .044). A previous study analyzing the prognosis of in-hospital cardiac arrest in 40 670 patients with STEMI22 reported an in-hospital mortality rate of 53%, which is substantially higher than that observed in our cohort and significantly higher than that among patients without cardiac arrest. However, unlike in our study, cardiac arrests in that series occurred throughout hospitalization and therefore probably reflected, in many cases, more severe clinical deterioration, including potentially irreversible situations.

Despite the adjustments made, we cannot rule out that patients with RCABPA had location-related characteristics that influenced prognosis. However, our findings suggest that, although health care settings have defibrillation capability and advanced CPR, and although almost all cardiac arrests are caused by shockable rhythms, in-hospital RCABPA is treated more effectively than out-of-hospital RCABPA.

Although in the present study, most RCABPA episodes involved shockable rhythms, RCABPA was associated with poor prognosis regardless of cardiogenic shock. This suggests that most RCABPA episodes were not primarily related to an irreversible hemodynamic condition, but rather that prognosis may have been determined by the consequences of delayed rhythm reversal. This interpretation is supported by the absence of significantly worse prognosis when cardiac arrest occurred in the hospital setting.

The diversity observed in mortality between in-hospital RCABPA and RCABPA occurring in other locations suggests that prognosis may be more closely related to the quality and immediacy of CPR than to cardiac arrest per se. We consider this one of the most important findings of our study, because it identifies potential opportunities to improve survival among patients with RCABPA occurring in out-of-hospital health care setting. Future studies should further explore differences in the clinical presentation and treatment of patients within the health care system outside and inside the hospital to identify actionable areas for improvement.

Prognosis of RCABPA outside the health care setting

Although most studies report in-hospital mortality rates of 55% to 70% among survivors of out-of-hospital cardiac arrest,12 the precise location of arrest is usually not specified. Cause of cardiac arrest, initial rhythm, comorbidity, and socioeconomic status have all been associated with prognosis.23 Rapid initiation of resuscitation, even by the first witness,24,25 is a key determinant of survival.23,26 In our study, mortality among patients with RCABPA occurring outside the health care setting was 6 times higher than that among patients with in-hospital RCABPA and more than twice that among patients with RCABPA occurring in out-of-hospital health care settings. This difference persisted after adjustment for the main variables associated with prognosis. These findings support the importance of population-based campaigns aimed at improving recognition of cardiac arrest and promoting early defibrillation. Current CPR guidelines27 recommend facilitating access to defibrillators in public places and training the general population in CPR maneuvers.

Two studies have analyzed differences in all-cause mortality after out-of-hospital cardiac arrest based on the precise place of occurrence.28,29 In both studies, mortality was highest when cardiac arrest occurred at home, probably because of poorer access to defibrillation and CPR. In our series, the precise location of out-of-hospital RCABPA episodes could not be determined; therefore, differences within this group based on whether they occurred in a public or private place cannot be ruled out.

Limitations

The study design does not allow to determine precisely how many patients with cardiac arrest in the context of STEMI died before STEMI code activation. An undetermined number of patients may have died before the arrival of emergency teams, during first contact, or after unsuccessful resuscitation attempts. Although nonresuscitated cardiac arrest after activation but before arrival at the cath lab is exceptional, this may not be the case for patients who experience sudden death before contact with health care services, without successful circulatory restoration allowing transfer to the cath lab. This may introduce biases that are difficult to control. We believe this limitation is common to most registries of out-of-hospital cardiac arrest and reflects the real-world clinical practice setting of the study. The similarity between our RCABPA rates and those reported in previous studies on primary angioplasty or STEMI suggests that our findings are representative of the routine clinical practice.

Some differences in mortality according to RCABPA location may be explained by clinical characteristics that were not controlled for in the multivariate analyses. Information on the presence of witnesses, availability of a semiautomatic external defibrillator, and qualifications of the first rescuer was not available in our series; therefore, their influence on the observed results could not be assessed. Despite these unmeasured variables, we believe that the observed differences, particularly within the health care setting, suggest important opportunities for improvement in STEMI code programs. The high area under the receiver operating characteristic curve observed in the different models reduces the relevance of these potential uncontrolled variables.

CONCLUSIONS

Cardiac arrest in patients with STEMI before reperfusion occurs in 7.3% of cases referred for primary angioplasty, and approximately half of these events occurred outside the health care setting. Although RCABPA was associated with significantly higher mortality rates, prognosis varied based on the location of the cardiac arrest. In-hospital RCABPA may not adversely affect prognosis, whereas out-of-hospital RCABPA, particularly when occurring outside the health care setting, was associated with significantly higher mortality rates. Improvements in regional STEMI code programs focused on early medical care and out-of-hospital CPR could reduce mortality.

FUNDING

This study received no funding.

ETHICAL CONSIDERATIONS

This study was approved by the local ethics committee, with a waiver of informed consent because it was an anonymized retrospective study. The SAGER guidelines were followed with respect to possible sex/gender bias.

STATEMENT ON THE USE OF ARTIFICIAL INTELLIGENCE

Artificial intelligence was not used in the preparation of this work.

AUTHORS’ CONTRIBUTIONS

R. López-Palop contributed to the conception and design of the study, data acquisition, analysis, and interpretation, drafting of the original project and final manuscript, and final approval. P. Carrillo Sáez collaborated in data acquisition, analysis, and interpretation, and in drafting, reviewing, and approving the manuscript. R. López-Palop López collaborated in data acquisition, analysis, and interpretation, drafting of the initial project, and review and final approval of the manuscript. M.D. Vallés García collaborated in data acquisition, analysis, and interpretation. N. Fernández Villa collaborated in drafting, reviewing, and final approval of the manuscript. J.R. Gimeno Blanes, J.M. Durán Hernández, F.J. Lacunza Ruiz, J. García de Lara, J.A. Hurtado Martínez, A. Riquelme Pérez, and E. Pinar Bermúdez collaborated in data acquisition and in review and final approval of the manuscript. D. Pascual-Figal participated in reviewing, drafting, editing, and final approval of the manuscript.

CONFLICTS OF INTEREST

None declared.

REFERENCES