Percutaneous Coronary Intervention versus Optimal Medical Therapy in Patients with Chronic Total Occlusion: A Meta-Analysis

Background/Objectives: Chronic total occlusion (CTO) is a prevalent finding in patients with coronary artery disease and is associated with increased mortality. Prior reports on the efficacy of percutaneous coronary intervention (PCI) compared to optimal medical therapy (OMT) were controversial. Following the emergence of recently published new evidence, a meta-analysis is warranted. The current meta-analysis assessed the effects of PCI compared to OMT in the treatment of CTO. Methods: A structured literature search was performed. Randomized controlled trials (RCTs) and non-randomized controlled studies of interventions were eligible. The primary outcome was an accumulated composite of cardiac mortality, myocardial infarction and target vessel/lesion revascularization events. Results: Thirty-two studies reporting on 11260 patients were included. Of these, 5712 (50.7%) were assigned to the PCI and 5548 (49.3%) were allocated to the OMT group. The primary outcome occurred in 14.6% of the PCI and 20.1% of the OMT group (12 trials, OR 0.66, 95% CI 0.50 to 0.88, p = 0.005, I2 = 67%). Subgrouping demonstrated a consistent reduction in the primary outcome for the PCI group in RCTs (six trials, OR 0.58, 95% CI 0.33 to 0.99, p = 0.05). The primary outcome reduction was irrespective of the study design, and it was replicable in sensitivity and subgroup analyses. Advantages in other outcomes were rather related to statistical pooling effects and dominated by observational data. Conclusions: CTO-PCI was associated with improved patient-oriented primary outcome compared to OMT in a study-level meta-analysis. This composite outcome effect was mainly driven by target vessel treatment, but a significant reduction in mortality and myocardial infarction was observed, irrespectively. These findings have hypothesis-generating implications. Future RCTs with adequate statistical power are eagerly awaited.


Introduction
Chronic total occlusion (CTO) is a prevalent finding affecting 15 to 33% of patients with coronary artery disease undergoing coronary angiography [1][2][3][4][5].CTO itself is associated with increased mortality [4,6,7].The available evidence from mainly observational data comparing percutaneous coronary intervention (PCI) to optimal medical therapy (OMT) demonstrated symptom relief, rather than a consistent reduction in major cardiovascular outcomes [8,9].CTO-PCI is a complex coronary procedure: technical success rates depend on operators' experience and on trial design with about 85% in dedicated CTO registries, but only 64% in national reports [10,11].Successful CTO-PCI was associated with a more favorable outcome compared to failed PCI attempts in prior observational studies and a pooled data analysis [2,[12][13][14][15][16][17].The current European Society of Cardiology guidelines on myocardial revascularization or acute coronary syndrome (ACS) lack explicit and specific recommendations for the treatment of CTO despite a class IIa indication for revascularization in the case of refractory angina [8,18,19].In this context, CTO-PCI mainly relies on symptomatic rather than on prognostic indication, is still a subject of discussion and requires a careful decision-making process [9,20].
Additional studies, including follow-up reports of randomized-controlled trials (RCTs, EURO-CTO [21], COMET-CTO [22]), were recently published after the latest guideline reports [8,18].The results of these studies could potentially further inform treatment practice.Following the emergence of this new evidence, a meta-analysis is warranted, in order to further elucidate the efficacy of PCI compared to OMT in CTO treatment.

Methods
This meta-analysis was conducted using a protocol and reproducible plan for the literature search and synthesis, and it was reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines [23].The systematic literature search was performed using three databases, including Medline (via PubMed), Web of Science, and Cochrane Library.The search strategy for each database is provided in the Supplementary Materials.The search was performed on 27 November 2023.No restrictions on the publication date, language or study size were applied.After the exclusion of duplicates and screening of titles and abstracts according to the eligibility criteria, full texts of the remaining articles were assessed.Additional handsearching of screened references was performed.The screening of titles and abstracts, full text assessment and final study selection were independently performed by two reviewers (SM, KS).In case of any disagreement, this was resolved based on a consensus with one of the senior authors (SL/CA) at each step.
Controlled trials comparing PCI to OMT in patients with CTO were included.Randomized controlled trials (RCTs) and non-randomized controlled studies of interventions (NRSIs) were eligible.In NRSI, matched data (e.g., propensity score matching, PSM) were preferred over crude data.No restrictions were applied for specific interventional or conservative treatment strategies, follow-up duration or timing of the intervention itself.Double publications, case reports, case series without control groups, reviews, meta-analyses and conference abstracts were excluded.Studies grouping revascularization techniques (e.g., PCI and coronary artery bypass grafting (CABG)) or studies with solely CABG revascularization were excluded.
Data were independently extracted by two investigators (SMM, KS) using a standardized pre-specified data collection form.In case of any disagreement, this was resolved based on a consensus with one of the senior authors (SL/CA) at each step.Main study reports, as well as any Supplementary Materials and study protocols, were reviewed.Pre-specified data elements included the study design, patient baseline characteristics, intervention and follow-up data.In the case of published follow-up reports or several reports of the same cohorts, the longest available observational period was preferred, and event data were extracted accordingly.
The primary end point was constructed as the accumulated composite equivalent to major adverse cardiac events (MACE).The primary outcome was manually calculated through the extraction of total events of cardiac mortality, myocardial infarction (MI) and target vessel revascularization (TVR) or target lesion revascularization (TLR) from the included studies.Either TVR or TLR were selected depending on available data.The primary outcome was named the MACE equivalent.
As TVR and TLR might be clinically less impactful, we investigated a second constructed composite called Mortality&MI, which consequently included cardiac mortality and MI.Only trials eligible for primary outcome analysis were considered in the Mortal-ity&MI analysis to adjudicate on the robustness of the primary outcome measurement.The individual components of the primary outcome, stroke, major adverse cardiac and cerebrovascular events equivalent (MACCE, composite of primary outcome and stroke) and all-cause mortality were secondary efficacy endpoints.Peri-procedural MI was ad-hoc excluded from the analysis of MI given the uncertainty in the measurement and inequality in the distribution between the PCI and OMT group [24,25].TVR and TLR included PCI, but CABG was also considered during the observational period.
Risk of bias (RoB) at the study level was assessed using the Cochrane Collaborations risk-of-bias tool (RoB2, version 08/22/2019) for RCTs [26].NRSIs were assessed using the Cochrane Collaborations Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I, version 10/20/2016) tool [27].
The risk of bias assessment was independently performed by two investigators (SMM, KS).In the case of a discrepancy, a third independent investigator was consulted (SH).The risk of bias assessment was performed focusing on the primary outcome.
Random-effects meta-analyses were performed using the Mantel-Haenszel method for dichotomous event data.Pooled odds ratios (ORs) and 95% confidence intervals (CIs) are given for each analysis with a two-sided significance level of p < 0.05 (RevMan 5.3, Nordic Cochrane Centre, Cochrane Collaboration).The extent of heterogeneity was approximated by I 2 tests considering 0-40% as non-important, 30-60% as moderate, 50-90% as substantial and 75-100% as considerable heterogeneity.Pre-specified funnel plot analysis was performed for the primary outcome.
A sensitivity meta-analysis of the primary outcome was performed according to the risk of bias judgement.RCTs at "high" and NRSIs with a "serious" or "critical" risk of overall bias were excluded.Subgroup meta-analysis of the primary outcome was performed according to the study design (RCT vs. NRSI), intention-to-treat principle (ITT, exclusion of studies who assigned failed PCI to OMT), patient selection (exclusion of unmatched NRSI) and underlying disease (exclusion of studies considering patients with concomitant ACS).The latter was based on the hypothesis that ACS itself might impact the MACE rate, rather than the treatment of CTO in the non-infarct related artery.
We did not obtain ethical approval for this meta-analysis because we did not collect data from individual human subjects.

Study Selection
A total of 266 articles were identified by the literature search (see Figure 1, PRISMA flow chart, see Supplementary Materials).After removing duplicates, the titles and abstracts of 209 remaining articles were screened.Of these, 170 articles were excluded, which left 39 references for an assessment of full-text eligibility.Seven additional fulltexts were assessed for eligibility by handsearching [13,[28][29][30][31][32][33].Thirty-two publications reporting on nineteen patient cohorts were finally included in the quantitative analyses [4,13,21,22,.
The CTO definition varied within the trials.In the majority of studies, a thrombolysis of myocardial infarction (TIMI) flow 0 with a known or estimated duration of at least or greater than 3 months was required [4,21,29,31,34,41,43,46,47,51,53,54].The indication for CTO treatment was sparsely reported, but stabile angina pectoris combined with a proof of myocardial ischemia in the CTO territory was a recurring definition.Additionally, a few trials required myocardial viability in the CTO territory.Patients with prior CABG [4,32,41,43,46,47,51,54] or concomitant ACS [13,21,22,30,31,38,54] at the time of CTO treatment were excluded in a considerable number of cohorts.Notably, the EXPLORE trial only enrolled patients with ST-segment elevation MI, and Qin et al. exclusively included patients with ACS [32,34].In RCTs, all patients were randomly assigned to intervention.In all but one trial [30], each patient allocated to the PCI strategy was considered in the interventional group following the ITT principle.In NRSI, the treatment decision (PCI or OMT) was at the local heart team's discretion or followed a shared-decision making process between the treating physician and the patient.In six NRSI patients with unsuccessful PCI were allocated to the OMT arm for an efficacy analysis [13,33,41,42,46,53], and the remaining six NRSIs followed an ITT principle [4,32,38,47,51,54].

Risk of Bias Assessment
The RoB assessment of RCTs is summarized in Table 2a.Two trials [22,34] were judged to be at a low risk of bias.Two other trials [21,31] were associated with "some concerns" in the overall RoB assessment.In REVASC [31], imbalances in patient characteristics and slight deviations between pre-specified and reported outcomes resulted in this judgement.In EURO-CTO [21], deviations from intended interventions with a considerable cross-over rate, especially in the PCI arm, might have affected the outcome.This reflects real-world practice but indicates some sort of bias.Two RCTs [29,30] were judged to be at high risk of bias.DECISION [29] had a remarkable cross-over rate in both directions with 19.9% of patients assigned to OMT immediately crossing over to PCI after randomization.In IMPACTOR [30], patients with unsuccessful PCI and those with non-adherence to study medication were automatically excluded from the outcome analysis.These treatment strategies each indicate a substantial risk of bias.Two trials were prematurely stopped: one [21] due to limited funding (lacking 33% of targeted patient enrollment) and one [29] because of slower enrollment than expected.This indicates a source of "other bias", which is currently not included in the RoB2 tool, but should be acknowledged.The risk of bias assessment of NRSIs is summarized in Table 2b.Six [4,32,43,47,51,54] NRSIs were judged to be at a moderate risk of overall bias.Six [13,38,41,42,46,53] other NRSIs were judged to be at "serious", and one [33] was judged to be at a "critical risk" of overall bias.This judgement was primarily driven by severe imbalances between treatment groups in baseline characteristics (confounding domain).Precisely, patients in the revascularization or successful revascularization groups had more favorable baseline characteristics.In the study judged to be at a critical risk of bias, the selection into the study was very strongly related to the intervention and outcome, and the patients were extremely imbalanced in baseline characteristics without any adjusted analysis attempts [33].
Secondarily, the judgement was affected by the assignment of failed PCI patients to the OMT group (derivations from intended interventions)-which stands in contrast to higher quality trials following the ITT principle.These sources of bias might have reasonably influenced the measured outcome.

Patient Baseline Characteristics and Procedural Data
A total of 11,260 patients with coronary artery disease and at least one CTO were included in the analysis.Of these, 1890 (16.8%) were enrolled in RCTs and 9370 (83.2%) were included in NRSIs.Overall, 5712 (50.7%) were assigned to the PCI group and 5548 (49.3%) were allocated to the OMT group.The median/mean age ranged from 58.0 to 69.6 years, and 70.6 to 90.1% were male.An analysis of the cardiovascular risk profile demonstrated that 14.9 to 53.2% had diabetes, 6.5 to 81% had dyslipidemia, 39.9 to 91.6% had hypertension and 20.2 to 87.9% were ever smokers.
CTO lesions were predominantly located in native coronary arteries.The majority of CTO lesions were diagnosed in main vessels, including the left anterior artery descending (LAD), right coronary artery (RCA) and left circumflex artery (LCX).Only a negligible low number of patients with CTO of the left main stem or branches were considered in the studies included.RCA-CTO was documented in 40 to 78% of patients.Accordingly, RCA-CTO was the most prevalent affected vessel followed by LAD-CTO (10 to 44.8%) and LCX-CTO (10.2 to 41.7%).
The overall success rate of PCI varied from 60.2 to 100%.The mean fluoroscopy time ranged from 29.4 to 49.6 min.The contrast medium volume varied between 280 and 340.7 mL.Whenever reported, antegrade recanalization techniques (60.4 to 88%) were more commonly used than retrograde approaches, but these data were limited to RCTs.Detailed baseline, lesion and procedural characteristics are summarized in Table 1a,b.

MACE Equivalent
Twelve trials were included in the analysis of the constructed primary outcome.Overall, 5324 patients were considered.Of these, 2752 were treated with PCI and 2572 were managed using OMT.The event rate was 14.6% in the PCI and 20.1% in the OMT group (see Figure 2a, OR 0.66, 95% CI 0.50 to 0.88, p = 0.005, I 2 = 67%, favoring PCI, moderate heterogeneity).In detail, 296 cardiac deaths (32%), 236 MIs (26%) and 389 TVRs/TLRs (42%) were reported.In the subgroup of RCTs, 61% of events were caused by TVR/TLR, 24% of events were due to cardiac death and 15% of events were related to MI.
A funnel plot analysis of the primary outcome did indicate slight asymmetry on the visual plot analysis (see Figure S1).Detailed events of single composite domains are summarized in Table S1.

Mortality&MI
All twelve trials considering 5324 patients were included in the outcome analysis.The cumulative event rate was 7.7% in the PCI and 12.4% in the OMT group (see Figure S2, OR 0.60, 95% CI 0.47 to 0.76, p < 0.001, I 2 = 25%, favoring PCI, non-important heterogeneity).

Mortality&MI
All twelve trials considering 5324 patients were included in the outcome analysis.The cumulative event rate was 7.7% in the PCI and 12.4% in the OMT group (see Figure

Sensitivity Analysis of Primary Outcome
Eight trials remained for the sensitivity analysis according to the RoB assessment (see Section 3.3).The event rate was 17.2% in the PCI and 24.3% in the OMT group (see Figure S5, OR 0.66, 95% CI 0.49 to 0.90, p = 0.008, I 2 = 53%, moderate heterogeneity).The observed treatment effect favoring PCI was robust in the sensitivity analysis.

Subgroup Analyses of Primary Outcome
Subgrouping following the trial design demonstrated a reduction in the primary outcome for the PCI group in RCTs (see Figure 2a, OR 0.58, 95% CI 0.33 to 0.99, p = 0.05), but this effect was not present in the NRSI comparison (OR 0.70, 95% CI 0.49 to 1.02, p = 0.06).
The observed treatment effect of the primary outcome reduction favoring PCI in the overall analysis was robust in the subgroup following the ITT principle (see Figure S5, OR 0.69, 95% CI 0.53 to 0.90, p = 0.007) and in patients with chronic coronary syndrome (OR 0.73, 95% CI 0.54 to 0.99, p = 0.04).But there was little to no difference after excluding unmatched trials leaving NRSIs with PSM and RCTs for analysis (see Figure S5, OR 0.75, 95% CI 0.56 to 1.00, p = 0.05).

Cardiac Mortality
Seventeen trials were included in the analysis of cardiac mortality.Overall, 10,354 patients were analyzed.Of these, 5285 were treated with PCI and 5069 were managed using OMT.The event rate was 3.7% in the PCI and 9.9% in the OMT group (see Figure 2b, OR 0.41, 95% CI 0.34 to 0.49, p < 0.001, I 2 = 49%, favoring PCI, moderate heterogeneity).

Myocardial Infarction
Seventeen trials were included in the analysis of myocardial infarction.A total of 10,660 patients were assessed.Of these, 5409 were treated with PCI and 5251 were managed on OMT.The event rate was 2.9% in the PCI and 4.8% in the OMT group (see Figure 2c, OR 0.70, 95% CI 0.57 to 0.86, p < 0.001, I 2 = 5%, non-relevant heterogeneity).

TVR
Fourteen trials were included in the analysis of TVR.Overall, 5774 patients were considered.Of these, 2954 were treated with PCI and 2820 were managed on OMT.The event rate was 6.6% in the PCI and 7.1% in the OMT group (see Figure 2d, OR 0.80, 95% CI 0.48 to 1.34, p = 0.40, I 2 = 75%, substantial heterogeneity).

TLR
Four trials were included in the analysis of TLR.A total of 2083 patients were analyzed.Of these, 1108 were treated with PCI and 975 were managed on OMT.The event rate was 6.8% in the PCI and 6.5% in the OMT group (see Figure 2e, OR 0.75, 95% CI 0.19 to 2.99, p = 0.69, I 2 = 91%, considerable heterogeneity).

All-Cause Mortality
Sixteen trials were included in the analysis of all-cause mortality.Overall, 8916 patients were considered.Of these, 4543 were treated with PCI and 4373 were managed on OMT.The event rate was 8.7% in the PCI and 19.0% in the OMT group (see Figure 2f, OR 0.53, 95% CI 0.41 to 0.68, p < 0.001, I 2 = 63%, favoring PCI, moderate heterogeneity).

Stroke
Ten trials were included in the analysis of stroke.Overall, 4783 patients were considered.Of these, 2457 were treated with PCI and 2326 were managed on OMT.The event rate was 0.8% in the PCI and 1.5% in the OMT group (see Figure S3, OR 0.54, 95% CI 0.30 to 0.99, p = 0.05, I 2 = 0%, favoring PCI, non-important heterogeneity).

MACCE equivalent
Eight trials were included in the analysis of the MACCE equivalent.A total of 3167 patients were analyzed.Of these, 1675 were treated with PCI and 1492 were managed on OMT.The event rate was 11.5% in the PCI and 16.0% in the OMT group (see Figure S4, OR 0.62, 95% CI 0.42 to 0.92, p = 0.02, I 2 = 67%, favoring PCI, moderate heterogeneity).

Discussion
This meta-analysis was based on the most recent available long-term follow-up data and comprehensively assessed the effect of PCIs compared to OMT in patients with CTO.The main and novel findings of the pooled data analyses on the study-level were as follows:

‚
PCI was associated with a 34% lower likelihood of a constructed primary outcome MACE equivalent; this was consistent in RCTs and replicable after an adjustment for confounding factors in the ITT subgroup analysis and after the exclusion of studies considering patients with concomitant ACS.‚ Primary outcome reduction was mainly related to TVR or TLR, but Mortality&MI outcome analysis PCI was associated with a decreased event rate again.‚ PCIs were associated with a 47% decrease in all-cause mortality.‚ PCIs were associated with a 46% lower likelihood of stroke and a subsequent 38% decrease in the MACCE equivalent.

‚
There was little to no detectable difference in TLR and TVR rates between the interventions.
Pooled data analysis is the uncontested strength of a meta-analysis, but interpretation of the overall effects requires cautious revision.
In the present context, CTO-PCI was associated with the above-mentioned advantages.In contrast, prior meta-analyses showed an improved outcome following the CTO-PCI in pooled analyses or in NRSIs but not in the RCT subgroup [55][56][57].A funnel plot analysis indicated slight asymmetry, which could reflect publication bias and interstudy heterogeneity.Furthermore, this difference might be related to the recent long-term followup reports from EURO-CTO [21] and COMET-CTO [22].Mainly, potential differences in the primary outcome definition and measurement in prior meta-analyses might be contributing factors.For example, Simsek et al. [55] did not specify the MACE definition, and Abo-Ali et al. considered MACE according to the individual study definition resulting in heterogenous reports [56].During the extraction process, the authors of the present review identified various definitions of major adverse cardiac events used.In brief, in six RCTs we documented five different definitions: In the cardiovascular literature, the definition of MACE lacks standardization; this is a major limitation in the pooled data analysis of composite outcomes [58].Consequently, the authors defined the primary outcome MACE equivalent according to preexisting CTO publications in the post-hoc analysis [57].We decided to accumulate event data of the single domains and construct a reasonable MACE equivalent outcome instead of considering heterogenous composites from the included studies.In doing so, the pure event rate in the present accumulated MACE analysis might be overestimated compared to conventional time-to-event composite analyses.Repeated events in the same patients might be theoretically considered, but this reflects a patient-oriented approach in the outcome assessment and evaluation.On the study-level this seems to be the most reasonable approach, but does not imply the true event rate.On the one hand, a patient-level meta-analysis might be a future approach to overcome this inconsistency.On the other hand, we evaluated each single domain of the composite to adjust for confounding factors in the data analysis.
The primary outcome difference was mainly attributable to TVR and TLR (see Figure 2a).This observation was consistent in both the overall analysis and in the subgroup of RCTs.TVR and TLR are widely accepted cardiovascular outcomes [24,59], but standardization regarding indication and interpretation in trials comparing PCI to OMT is controversial.One might assume that in the comparator group, TVR/TLR was clinically driven in patients with refractory angina pectoris despite optimized medical management.On the contrary, in the PCI group, not only a symptomatic indication but also device-failure after the PCI might have contributed to repeated revascularization.As the authors were not able to adjust for this confounding factor, we designed the Mortalty&MI outcome.PCI was associated with a 40% lower likelihood of MI or cardiac death in the overall analysis, but this advantage was not replicable in the subgroup of RCTs.One should bear in mind that the event rate was remarkably lower in RCTs than in NRSIs (4.3% vs.15.2%) and might lack statistical power even in the pooled data analysis [55].
Finally, the reduction in the composite of cardiac mortality, MI and TVR/TLR following PCI was a consistent measurement and was sustained in the sensitivity and subgroup analysis.
Pooled data analysis showed an approximately 50% lower likelihood of all-cause and cardiac mortality following CTO-PCI, but this was not a robust finding in the subgroup of RCTs (see Figure 2b, see Figure 2f).Firstly, single RCTs did not have enough statistical power to detect mortality differences [55].Secondly, the advantageous mortality data were mostly related to events in NRSIs underlying especially selection bias.As previously acknowledged, patients included in CTO-RCTs have a more favorable risk profile and less complex lesions compared to real-world populations and registries [11].The latter are at an increased risk of procedural and long-term cardiovascular complications.Accordingly, the cardiac mortality rate was 7.7% in the NRSIs and 4.2% in RCTs in the present report.Finally, the follow-up period was longer in NRSI compared to RCTs.The potential effect of PCIs on survival data might be uncertain because RCTs had this shorter follow-up period (RCTS: 4 to 56 months, NRSI: 1.0 to 7.9 years) and were not adequately powered on an individual trial level to assess survival [55,57].Patients with CTO have an increased risk of ventricular arrhythmia [60], cardiogenic shock [61] and MI-as shown in the present analysis.CTO-PCI might optimize targeted artery and donor-vessel flow and subsequently improve overall hemodynamics and contribute to remodeling.These processes might reduce the number of events of cardiac death based on a reduction in MI, arrhythmia and shock.These hypotheses require a further work-up.
One might speculate whether PCI offers a time-dependent advantage in mortality or whether advantages in NRSIs are purely related to confounding factors.An increased follow-up of preexisting RCTs or new evidence might add further information.In this context, the authors want to point out that the present literature indicates considerably higher CTO-PCI success rates in dedicated registries by experienced operators [10,11,62,63].The trend of successful revascularization contributing to a better prognosis might only reliably appear if patients survive the risky and complex CTO-PCI [2,[12][13][14][15][16][17]64,65].The referral of CTO patients to experienced centers might be the preferred strategy to optimize interventional and clinical outcomes in this vulnerable cohort, but this requires prospective validation [20].
The expected results from the NOBLE CTO (NCT03392415) and ISCHEMIA CTO (NCT03563417, [66]) trial, each comparing OMT to PCI in the CTO treatment, might add further evidence to the research question of optimized treatment.NOBLE CTO will enroll 2000 patients, and study completion is expected in July 2030.The primary outcome is all-cause mortality at ≥6 months.ISCHEMIA CTO will enroll 1560 patients investigating differences in major adverse cerebral and cardiovascular events after 5 years.Completion is expected in November 2032.The results of these studies could potentially further inform treatment practice and are eagerly awaited.Until then, angina relief and quality of life improvement remain the leading indication for CTO-PCI [14,[18][19][20]67].

Limitations and Strengths
Confounders on the individual study level and interstudy heterogeneity were acknowledged.The intervention itself varied throughout the trials, and this restricts generalizability: the majority of patients were treated in dedicated CTO-PCI centers with specific interventional expertise, most trials were single-center reports, the treatment decision had substantial confounding potential in NRSIs and the indication for treatment varied.The single-blind design might have contributed to cross-over rates in RCTs with up to one-fifth in DECISION.Uncertainty of the treatment decision regarding TVR/TLR, especially in OMT groups in primary studies, are an inherent limitation.The heterogeneity in follow-up limits transferability, especially in the survival analysis.Both subgroup and sensitivity analyses were performed to adjust for these sources of selection, detection and performance bias.As mentioned before, the primary outcome construction might overrate the measured effect of CTO-PCI, but this follows a patient-oriented approach.
The extracted data were restricted to study-level reports, and the authors did not have access to patient-level data.The latter would have allowed for a more accurate analysis of patient-and procedural-related differences (e.g., collateralization vs. insufficient collateralization in CTO territory, calcification vs. no calcification, treatment regimen, stent type, intravascular imaging, antegrade vs. retrograde approach, lesion crossing technique, successful vs. failed PCI).Procedural complications were inconsistently reported and not analyzed in this review.Accordingly, type 4a myocardial infarction was excluded from the analysis of MI [68].This might have favored the outcome of the PCI group.The authors faced an unclear overlap of two patient cohorts [46,51], and even a detailed revision of the manuscripts and supplements did not result in clarification.Accordingly, only one trial [51] was considered in the primary outcome assessment to overcome this inconsistency.Moreover, the explicit combination of OMT and adherence to contemporary guidelines could not be assessed on the trial level.
The important strengths of this meta-analysis are the systematic description and discussion of bias, the demonstration of weaknesses and gaps in the evidence and the adjusted analysis of most recent available long-term data.

Conclusions
CTO-PCI was associated with an advantageous patient-oriented primary outcome-a composite of accumulated event rates of cardiac mortality, MI, TVR/TLR-compared to OMT in the study-level meta-analysis.This benefit was mainly driven by TVR/TLR.The primary outcome reduction was replicable in sensitivity and subgroup analysis of the RCTs or following the ITT principle.
Cardiac mortality, all-cause mortality, MI and stroke were decreased in the CTO-PCI group, but this was mainly driven by pooling effects dominated by observational data.Deaths were less frequently observed in RCTs, and the authors assume a lack of statistical power in assessment of this outcome.These results have hypothesis-generating implications.Future RCTs with adequate statistical power are eagerly awaited to clarify the optimal treatment strategy in CTO.

Impact on Daily Practice
CTO-PCI was superior to OMT in the pooled data analysis measuring a patientoriented composite outcome of cardiac mortality, myocardial infarction and revascularization of the target vessel.This hypothesis-generating finding arises from considerable bias and requires confirmation based on large scaled randomized controlled studies.Until then, these data might contribute to the decision-making process from the interventionalist's perspective.

Table 1 .
Characteristics of the included studies and patients.

Table 2 .
Risk of bias assessment of included studies.(a) Risk of bias assessment of randomized controlled trials; (b) risk of bias assessment of non-randomized controlled studies of interventions.