Percutaneous Coronary Interventions for Calcified Lesions: A Systematic Review with Meta-analyses of Randomised Trials

Effective lesion preparation and stent optimisation influence long-term clinical outcomes after percutaneous coronary intervention (PCI).¹ Coronary calcifications complicate lesion preparation and stenting, and affect 16–31% of patients undergoing PCI.^2–4 Patients with severe calcifications have a lower procedural success rate and an increased risk of procedural complications, target vessel failure, and death, compared with patients with less calcification.^5,6 As a consequence, these patients are often excluded from randomised trials and present the treating clinicians with difficult choices regarding the most effective interventional strategy.

Various lesion preparation techniques exist for calcium modification, including specialised balloons (high-pressure, lithoplasty, scoring and cutting balloons) and debulking techniques (rotational atherectomy, orbital atherectomy and excimer laser).¹ Recent guidelines recommend calcium modification in patients with calcified coronary lesions, but the optimal technique remains unclear.⁷ Additionally, it is unclear whether calcium modification improves clinical outcomes compared with conventional lesion preparation. Furthermore, limited data exist regarding the comparative efficacy of second- or newer-generation drug-eluting stents (DESs) in this population.

While reviews assessing percutaneous interventions for calcified lesions have been published, they are limited by non-systematic methodology, inclusion of non-randomised trials, or lack of adherence to reporting guidelines.⁸ To date, no systematic review has comprehensively investigated all available percutaneous interventions for calcified coronary lesions.

The present objective was to systematically assess the effects of PCI strategies for calcified lesions, focusing on all-cause mortality, serious adverse events and other patient-centred outcomes.

Methods

The protocol for this systematic review has been published and registered at PROSPERO (CRD42021226034).⁸ Furthermore, we adhered to the PRISMA statement and the Cochrane Handbook for Systematic Reviews of Interventions.^9,10 Ethics approval was not required for this study.

Search Strategy

An information specialist searched the CENTRAL, MEDLINE, Embase, LILACS, and SCIE databases to identify relevant trials. We searched databases from inception to 8 May 2025. The search was supplemented by manually screening the reference lists of included trials, published reviews, and selected studies. We also searched clinical trial registers and websites of cardiology conferences. The search strategy is given in Supplementary Material Methods.

Eligibility Criteria

We included randomised trials assessing participants undergoing PCI on any native coronary artery de novo stenosis due to coronary artery disease with moderate or severe lesion calcification (any definition of calcification was accepted). For the intervention group, we included any conventional or specialised lesion preparation technique, any stent or scaffold implantation, or any stent optimisation technique. For the comparator group, any comparison was eligible. Conventional preparation was defined as interventions labelled ‘standard therapy’ or equivalent, mostly using standard balloons but in some cases this included the non-mandatory use of specialised balloons or atherectomy, at the operator’s discretion. In trials randomising participants to stents, we included only trials investigating second- or newer-generation DESs to reflect contemporary practice.

Data Collection

Two authors independently screened search results, extracted trial-level data and assessed the risk of bias using the Cochrane Risk of Bias tool (ROB 2).¹¹ Disagreements were solved through discussion or via consultation with a third author (JCJ).

Outcomes

Outcomes were extracted according to the protocol.⁸ The primary outcomes were all-cause mortality and serious adverse events as defined by the International Conference on Harmonisation – Good Clinical Practice (ICH-GCP).¹² If the trialists did not explicitly use the ICH-GCP definition or the term ‘serious adverse event’, we included the data if the criteria in the ICH-GCP definition were met. This assessment was done by two authors, with events presumed serious unless stated otherwise by the trialists. If the trialists did not report the proportion of participants with an event, we extracted the most prevalent event to avoid double-counting participants with multiple events. This systematic approach reflects standard procedures.^13,14 Secondary outcomes were quality of life, MI, stroke, coronary angiography, target vessel MI and target vessel revascularisation. Exploratory outcomes were cardiovascular mortality, failed or no stenting during the procedure, and study group cross-over. We evaluated outcomes at the maximum follow-up time available.

Statistical Analysis

Results were based on intention-to-treat data, if available. Meta-analyses were conducted according to the Cochrane Handbook for Systematic Reviews of Interventions and Jakobsen et al.^10,13 Random-effect (DerSimonian–Laird) and fixed-effect (Mantel–Haenszel) meta-analyses were conducted and the estimate with the highest p value was reported. Results are expressed as RR and 95% CI. Double-zero-event trials were excluded from the calculation of RR and p values, and a continuity correction of 0.5 was applied to single-zero-event trials to calculate the RR, which reflects standard practice.¹⁰ For single-trial comparisons, we used Fisher’s exact test to calculate p values.

Meta-analyses were accompanied by trial sequential analysis (TSA) to calculate the diversity-adjusted required information size (DARIS, i.e. the number of participants required to confirm or reject a prespecified RR reduction of 25% based on the observed proportion of patients with an outcome in the comparator group, an alpha of 3.3%, a beta of 10%, and the observed diversity as suggested by the trials in the meta-analysis, as detailed in the protocol) and the cumulative Z-curve’s breach of trial sequential monitoring boundaries for benefit, harm or futility.⁸ Because we assessed two primary outcomes, we considered p<0.03 as statistically significant.¹⁵ We assessed the certainty of the evidence of primary and secondary outcomes using the GRADE approach and created summary of findings tables.¹⁶ We considered the risks of bias, inconsistency, indirectness, imprecision and publication bias.¹⁶ Imprecision was determined by comparing the number of analysed participants to the DARIS. The statistical software R (version 4.2.2) and Trial Sequential Analysis Viewer (version 0.9.5.10) were used for analyses.¹⁷

Differences between the Protocol and the Review

We did not report the individual serious and non-serious adverse events or the exploratory outcomes for in-stent restenosis, any revascularisation, any physiological or imaging-derived measurement of improved myocardial perfusion, procedural success, stent delivery, procedural duration, fluoroscopy time, and contrast dose due to lack of relevant data. We did not report the exploratory outcomes for successful device crossing or bailout atherectomy due to inconsistent reporting and unclear definitions. Lastly, the stent subgroups of coating strategies were not prespecified.

Results

A systematic search, performed on 8 May 2025, identified 7,172 records after removal of duplicates. We included 31 unique trials randomising 8,453 participants (Figure 1). The list of excluded trials is given in Supplementary Table 1. Lesion preparation techniques were investigated in 24 trials (4,392 randomised participants) assessing 16 unique interventions across 17 different comparisons.^18–41 Stents or scaffolds were investigated in seven trials (4,061 randomised participants) assessing five unique interventions across five different comparisons.^42–48

Trial characteristics are listed in Table 1 and Supplementary Table 2. The median follow-up was 12 months from randomisation (IQR: 3.5–18.5 months). The severity of lesion calcification was severe in 16 trials (52%), moderate and severe in 14 trials (45%), and unclear in one trial (3%).^18–48 The calcification severity was defined using invasive angiography in 19 trials (61%), intravascular ultrasound in four trials (13%), optical coherence tomography in one trial (3%), combined modalities in four trials (13%), and there were no data on the diagnostic modality used to assess calcification severity in three trials (10%) (Supplementary Table 3).^18–48 Twenty-three trials (74%) required lesion calcification for inclusion and eight (26%) reported subgroup results from participants with calcification, but did not require it for inclusion.^18–48 In trials requiring lesion calcification for inclusion, intravascular imaging was used in 16 trials (70%), of which only intravascular ultrasound was used in seven (30%), only optical coherence tomography in seven (30%) trials, and both imaging techniques were used in two (9%).^{18–20,23–30,32,33,35,39–41} Four trials (13%) were assessed at overall low risk of bias (Supplementary Figure 1).^39–41,43 Eight trials (26%) were not at risk of for-profit bias (Supplementary Table 2).^{18,23,25,31,32,34,36,37} No trials reported data on quality of life or coronary angiography, and no trials randomised participants to stent optimisation techniques. Due to lack of data, it was not possible to conduct network meta-analyses or subgroup analyses.

Lesion Preparation Techniques

Orbital atherectomy was investigated in two trials (2,105 randomised participants), rotational atherectomy in 14 trials (1,530), scoring balloon in four trials (653), cutting balloon in seven trials (602), lithoplasty in six trials (431), excimer laser in two trials (243), and super high-pressure balloon in one trial (74) (Figure 2).^18–41 The primary outcome in these trials was based solely on clinical outcomes in three trials (13%), imaging in 11 trials (46%), procedural success in seven trials (29%), and undefined in three trials (13%) (Table 1).^18–41 Conventional preparation (definitions listed in Supplementary Table 4) was the comparator in nine trials (2,679 randomised participants) assessing orbital atherectomy (2005), rotational atherectomy (270), cutting balloon (192), lithoplasty (140) and excimer laser (72).^{26,32–35,37–40}

Results for each interventional comparison are listed in Table 2 and a network graph of comparisons is shown in Figure 2. We did not find evidence to support a benefit of any lesion preparation technique in the primary or secondary outcomes, and the certainty of the evidence was low or very low (Table 3). Except for orbital atherectomy versus conventional preparation for the outcome serious adverse events, TSA showed that all comparisons were insufficiently powered to confirm or reject an RR reduction of 25% of all-cause mortality, serious adverse events, or any secondary outcome (Supplementary Tables 7–23). Results of exploratory outcomes are given in Supplementary Material Results.

Stents and Scaffolds

Permanent polymer DES was investigated in seven trials (4,061 randomised participants), bioresorbable polymer DES in five trials (3,595), polymer-free DES in one trial (1,672), bioactive stent (Titan-2) in one trial (352), and bioresorbable scaffold in one trial (114) (Supplementary Figure 18).^42–48 The primary outcome in these trials was based solely on clinical outcomes in six trials (86%) and on imaging in one trial (14%).^42–48 All trials were subgroup analyses from larger trials that did not require calcification for inclusion.

Results for each interventional comparison are listed in Supplementary Table 5 and a network graph of comparisons is shown in Supplementary Figure 18. We did not find evidence to support a benefit of any stent or scaffold technique in the primary or secondary outcomes (Supplementary Tables 24–28). TSA showed that we could reject an RR difference of 25% in all-cause mortality for the comparisons polymer-free versus permanent polymer DES and bioresorbable polymer versus polymer-free DES (boundary for futility crossed, Supplementary Table 6). Furthermore, we could reject an RR difference of 25% in serious adverse events for all comparisons between bioresorbable polymer, permanent polymer, and polymer-free DES. Last, we could reject an RR difference of 25% in target vessel revascularisation for the comparison polymer-free versus permanent polymer DES. Meta-analysis and TSA for the remaining stent/scaffold comparisons were insufficiently powered to confirm or reject an RR difference of 25% of all-cause mortality, serious adverse events, or any secondary outcome (Supplementary Tables 24–28). Results of exploratory outcomes are given in Supplementary Material Results.

Discussion

PCI of calcified lesions is associated with increased procedural and long-term risks, but the optimal interventional techniques remain unclear. To clarify the current state of evidence, we systematically reviewed randomised trials comparing the effects of PCI techniques on all-cause mortality and serious adverse events in patients with calcified coronary lesions based on a predefined protocol.⁸ We identified 31 trials randomising 8,453 participants across 22 different interventional comparisons, and found no lesion preparation or stenting technique to be superior for all-cause mortality, serious adverse events, or other patient-centred outcomes. Most trials were assessed at high risk of bias and the certainty of the evidence was very low or low, emphasising the need for robust trials evaluating clinical outcomes. Existing evidence is underpowered, emphasising the clinical equipoise between strategies such as rotational atherectomy, lithoplasty, scoring balloons or polymer-coating stent strategies.

Challenges in Current Evidence

The field of calcium modification is rapidly expanding and lacks randomised evidence supporting the superiority of any lesion preparation or stenting strategy. Without clear evidence, decisions rely on procedural findings, operator experience, and resources. Expert statements provide guidance, but often rely on consensus or non-randomised data.¹ The number of trials for each interventional comparison was very limited, and TSA showed that most meta-analyses on lesion preparation techniques were underpowered, making it challenging to draw definitive conclusions. Furthermore, meta-analysis was not possible for several comparisons due to the absence of events in both groups.

The majority of trials were not powered for major adverse cardiovascular outcomes (MACE) and only a few trials assessed MACE as the primary outcome. Instead, many trials focused on surrogate outcomes such as stent expansion or procedural success. While these trials provide valuable knowledge and are more feasible than trials assessing clinical outcomes, their importance to the patient is less clear. A beneficial effect of lesion preparation techniques on clinical outcomes may exist, but we have not been able to demonstrate it with the available evidence.

Designing a trial with implications for daily clinical practice is challenging for several reasons. First, comparison of individual techniques is difficult given their complementary nature in daily practice. Operators may use multiple different techniques to optimise outcomes based on the individual patient factors. As reported in multiple trials assessing conventional preparation (Supplementary Material Results), the risk of study group cross-overs due to resistant calcium or balloon uncrossability is not negligible and adds complexity to the interpretation of the results.^{26,27,32,38–40} Therefore, it is attractive to test interventional strategies rather than individual techniques (e.g. upfront balloon-based versus an ablation-based strategy). A secondary randomisation in the case of uncrossability could compare the efficacy of rotational versus orbital atherectomy, which currently are considered equivalent in several treatment algorithms.^1,49–51 Another unexplored question is how to effectively modify deep calcium, for which the lithoplasty and atherectomy techniques are considered equal.^1,49–51

Second, although the prevalence of calcified lesions is expected to increase, the average enrolment rate of all of the included trials was low (a median of 0.9 patients randomised per month per enrolling site [IQR 0.5–1.8]; Supplementary Table 2) despite many trials allowing moderate calcification. Trials powered for MACE may require inclusion across centres and countries to achieve sufficient sample sizes. Given that severe calcification is conventionally diagnosed using invasive angiography, this presents a logistical challenge because eligible participants can rarely be identified before the procedure. The low inclusion rate is contrasted by the rapid development and introduction of new calcium modification techniques, which may render randomised trials less applicable when published. Adjusting the interventional protocol during the trial may be necessary to accommodate new techniques, but this adds complexity to the interpretation of the results.

Last, the heterogeneity in calcification in terms of location, extent and morphology may render a one-technique-fits-all approach less effective. The efficacy of interventional techniques for specific calcium morphologies is not well established and trial stratification on calcium morphology could be valuable.

Strengths

This review adhered to a predefined protocol and Cochrane standards.⁸ We used TSA to control for the risk of random errors and false-positive results that can occur in meta-analyses. TSA reflects the power of a meta-analysis to determine whether the available evidence is sufficient to draw firm conclusions. Furthermore, we used GRADE assessment and evaluated the certainty of the findings. Last, we included all available treatment options for a comprehensive assessment.

Limitations

First, the distinction between serious and non-serious adverse events was difficult in cases of inadequate reporting. However, we used a predefined standardised definition and discussed severity in cases of doubt, and all serious adverse events are listed in Supplementary Table 2 for clarity. Second, most trials were assessed at an overall high risk of bias, primarily because of the risk of bias due to selective outcome reporting or missing data. Third, the combined assessment of many comparisons may have caused problems with multiplicity. However, we adjusted the threshold of significance to account for this. Fourth, the protocol-defined RR reduction of 25% was relatively high and only a few meta-analyses, namely those based on large stent trials with long-term follow-up, were sufficiently powered to detect this difference for specific outcomes.^42–45,47 However, GRADE recommends an RR reduction of 20–30% to determine the optimal information size.¹⁶ Fifth, it was not possible to conduct the prespecified subgroup analyses (stent generations) and network meta-analyses due to inadequate data.⁸ Sixth, the trials used varying definitions of calcification, and more than half of the trials accepted moderate calcification. The efficacy of specialised calcium modification may be diluted in patients with moderate calcification. Lastly, none of the trials assessing stents or scaffolds required calcification for inclusion and, in some cases, severe calcification was an exclusion criterion.^44,46 This post hoc approach introduces a risk of selection bias because the calcification may not have been prospectively identified, potentially impacting outcomes.

Clinical Implications and Future Directions

Currently, there is clinical equipoise in important aspects of percutaneous intervention of calcified lesions. Promising clinical trials (ISAR-WAVE NCT06369142 and VICTORY NCT05346068, among others) are under way and may pave the way for refined guidelines. While the findings of this review may not have immediate clinical applicability, they could help shape the design of future randomised trials.

Conclusion

We did not find randomised evidence to support a reduction of all-cause mortality or serious adverse events with any lesion preparation technique for percutaneous treatment of moderately or severely calcified lesions. The certainty of the evidence was low or very low, and future high-quality randomised trials are needed to assess the efficacy and safety of lesion preparation techniques. For stenting strategies, we could reject with moderate certainty that the choice of polymer-coating strategy influenced all-cause mortality and serious adverse events.

Click here to view Supplementary Material.