Drug Coated Balloons Versus Drug‐Eluting Stents in Patients With De Novo Coronary Artery Disease

ABSTRACT

Background

In patients with de novo coronary artery disease (CAD), percutaneous coronary intervention (PCI) with drug eluting stent (DES) implantation is the gold standard therapy. New evidence is available with drug coated balloons (DCB) angioplasty. The objective of this meta‐analysis was to compare the efficacy and safety of DCB angioplasty versus DES‐based PCI for patients with de novo CAD.

Methods

A systematic search of PubMed, Scopus, Web of Science, and EMBASE databases was performed from January 1, 2006 up until November 1, 2024. Inclusion criteria were random treatment assignment; PCI with DCB versus DES; patients with de novo CAD; clinical follow‐up available for ≥1 year; and studies published in a peer‐reviewed journal.

Trial‐level incidence rate ratios (IRR) with 95% confidence intervals (CI) were pooled by random‐effects models with inverse variance weighting. The primary outcome was major adverse cardiac events (MACE). Secondary outcomes included cardiac death, all‐cause death, and target lesion revascularization.

Results

A total of nine clinical trials comprising of 4284 patients were included. The risk of the primary compositive endpoint of MACE was comparable with DCB versus DES (IRR 0.95, 95% CI 0.61 to 1.48). Secondary endpoints including cardiac death (IRR 1.49, 95% CI 0.96 to 2.33), all‐cause death (IRR 1.18, 95% CI 0.80 to 1.74), and target lesion revascularization (IRR 1.17, 95% CI 0.68 to 2.01) were comparable though CIs around treatment effects were wide.

Conclusions

DCB angioplasty for de novo CAD was associated with similar medium‐term clinical outcomes compared with DES. Additional evidence is warranted in view of numerical trends in important secondary endpoints.

Trial Registration

The analysis was registered at PROSPERO (CRD42024581222).

1. Introduction

Drug‐eluting stent (DES) implantation has been the standard approach for the percutaneous treatment of de novo coronary artery disease (CAD) in recent decades. Stenting with DES has demonstrated superior efficacy in comparison to bare metal stents, primarily driven by a reduced risk of target lesion revascularization (TLR) [1, 2]. Stent‐based percutaneous coronary intervention (PCI) leaves a permanent metallic implant on the vessel wall, which carries a life‐long risk of device‐related complications, primarily in‐stent restenosis (ISR) and stent thrombosis [3, 4, 5]. Rates of restenosis have been reducing with newer generation DES, however even with contemporary devices and implantation techniques rates of 2% to 3% per year of ISR are seen [6, 7].

Drug‐coated balloons (DCBs) are standard angioplasty balloon catheters with a matrix coating applied to the surface of the balloon. In 2004, the first preclinical studies of paclitaxel‐coated balloon catheters demonstrated effective inhibition of restenosis in a porcine coronary model [8]. Shortly thereafter a pilot randomized trial showed a favorable efficacy and safety profile of a paclitaxel‐coated balloon catheters in patients with ISR [9]. The matrix coating typically comprises of two elements: a lipophilic active, antiproliferative drug and a spacer or excipient. Treatment with a DCB involves the transfer of the antiproliferative drug into the vessel wall during transient balloon inflation at the level of a coronary lesion [10, 11].

Angioplasty with DCB has been investigated as an alternative treatment strategy to stenting with DES in a number of different clinical settings [12]. While several randomized trials have established the role of DCB catheters for the treatment of ISR [13], until recently, only a small number of randomized trials compared DCB angioplasty with DES‐based PCI for the treatment of de novo CAD and those available were often underpowered for clinical endpoints [14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25]. The REC‐CAGEFREE I trial recently is the largest trial to date comparing DCB angioplasty with DES‐PCI for any‐type de novo CAD and recently reported primary results [23]. Against this background, we conducted an updated comprehensive systematic review and meta‐analysis of randomized trials comparing DCB angioplasty with DES implantation for the treatment of de novo CAD.

2. Methods

We conducted a frequentist pair‐wise meta‐analysis in accordance with PRISMA and Cochrane Collaboration recommendations [26, 27, 28]. This meta‐analysis is registered with PROSPERO (CRD42024581222). As this was a meta‐analysis of trial level data, no ethical approval was required.

2.1. Eligibility Criteria

Studies fulfilling all the following inclusion criteria were included in our meta‐analysis: randomly assigned interventional strategy; treatment of de novo CAD presenting as either acute coronary syndrome (ACS) or chronic coronary syndrome (CCS); comparison of DCB angioplasty alone versus DES alone; follow‐up; ≥1 year; trials published in a peer‐reviewed journal. Studies with limited follow‐up (follow‐up <1 year), comparing DCB with bare‐metal stent (BMS), or employing hybrid strategies (e.g., DCB angioplasty followed by systematic stenting with bare metal stent vs. DES) were not considered.

2.2. Literature Search, Data Extraction, and Feasibility Assessment

A systematic search of PubMed, Scopus, Web of Science, and EMBASE databases for all reports of randomized clinical trials of angioplasty with DCB versus stenting with DES was performed from January 1, 2006 up until November 1, 2024. A detailed description of the search and methods for data extraction is provided in Supporting Information S2: Table 1. Two reviewers initially screened each record independently. Then a total of four reviewers screened each report retrieved to ensure studies met all inclusion and exclusion criteria. Data was extracted from original manuscripts by two researchers working independently then records were compared to ensure accuracy. Inclusion and exclusion criteria for individual trials listed in Supporting Information S2: Table 2.

2.3. Outcomes

The prespecified primary outcomes of interest was major adverse cardiac event (MACE). Although the components of MACE varied slightly across trials, this generally comprised the composite of cardiac death, myocardial infarction, or TLR. The individual trial‐level definitions of MACE are detailed in Supporting Information S2: Table 3 and the dual antiplatelet protocols in Supporting Information S2: Table 4. The secondary endpoints at the maximum available follow‐up included each individual component of the primary endpoint (i.e., cardiac death, myocardial infarction, and TLR), as well as all‐cause death, target lesion thrombosis, and target vessel revascularization.

2.4. Statistical Analyses

Fixed‐effect and random‐effects models with inverse variance weighting using trial‐level incidence rate ratios (IRRs) based on person‐years and 95% confidence interval (CI) were employed to compute summary estimates. The distribution of incidence rates between groups and the relative weight of each trial was presented by forest plots. Between‐trial heterogeneity was explored by variance τ² and I² statistics [29, 30]. I² values < 25% defines low heterogeneity, 25–50% moderate heterogeneity, and >50% high heterogeneity [29, 30]. The influence of individual trials on summary estimates was defined by leave‐one‐out analysis. All analyses were performed with R 4.4.1.

2.5. Bias Assessment

Trial‐level qualitative assessment was performed using the Risk of Bias RoB 2 tool [31]. We assessed reliability of our results for each outcome according to the GRADE system [28].

3. Results

After duplicates were removed and independent searches merged, a total of 994 reports were identified (Supporting Information S2: Table 1). A flow diagram illustrating the search and selection process is shown in Supporting Information S1: Figure 1. After screening at title and abstract level, 181 reports were identified for full‐text review. After full‐text assessment, a total of nine randomized trials met all the inclusion and exclusion criteria and were included in the primary analysis [16, 17, 18, 19, 20, 21, 22, 23].

The key features of study design and the main trial characteristics are summarized in Table 1. Inclusion and exclusion criteria of each trial are detailed in Supporting Information S2: Table 2. Five of the included trials focused primarily on de novo small vessel disease, three of the trials were almost exclusively patients presenting with ACS, whereas REC‐CAGEFREE‐1 included all‐comer population with a spectrum of lesion‐types and clinical presentations. There were 3 randomized trials excluded as clinical follow‐up available was less than 1 year, among these 3 trials excluded was PICCOLETO 1 as only 9‐month follow‐up was available [14, 24, 25]. Both the DCB and DES devices used in PICCOLETO 1 were early generation devices and differed greatly from the DCB and DES devices used in subsequent trials included in this meta‐analysis and so including PICCOLETO 1 would have added further heterogeneity to the pooled analysis. There were two randomized trials excluded as BMSs were used as the comparator arm [38, 39], and one randomized trial was excluded as both DES and BMS were used in the comparator arm [40].

Table 1.

Main characteristics of the included trials.

Trial (year, reference)	Patients randomized (DCB vs. DES)	Center (n)	Region	Vessel size	Enrollment period	Design	Primary endpoint	Maximum available follow‐up (months)	Registration
BELLO (2012 [18], 2015 [32])	90 versus 92	15	Italy	<2.8 mm	Mar 2010 to Mar 2012	Non‐inferiority	All‐cause death, myocardial infarction, or target vessel revascularization.	36	NCT01086579
BASKET‐SMALL 2 (2018 [16], 2020 [33])	382 versus 376	14	Germany, Switzerland, Austria	≥2.0 to <3.0 mm	Apr 2012 to Feb 2017	Non‐inferiority	Cardiac death, non‐fatal myocardial infarction, or target‐vessel revascularization.	36	NCT01574534
RESTORE SVD (2018 [17], 2020 [34])	116 versus 114	12	China	≥2.25 −2.75 mm	Aug 2016 to Jun 2017	Non‐inferiority	Cardiac death, target vessel myocardial infarction, or ischemia‐driven target lesion revascularisation.	24	NCT02946307
REVELATION (2019 [19], 2021 [35], 2024 [36])	60 versus 60	1	Netherlands	Any	Oct 2014‐Nov 2017	Non‐inferiority	Death, myocardial infarction, target lesion revascularization, stent thrombosis or major bleeding.	60	NCT02219802
PICCOLETO II (2020 [15], 2023 [37])	114 vs 118	5	Italy, Spain	≥2.25−2.75 mm	May 2015 to May 2018	Non‐inferiority	Cardiac death, myocardial infarction, target lesion revascularization.	36	NCT03899818
Hao et al. [21]	38 vs 42	1	China	2.5‐4.0 mm	Jan 2018 to Dec 2019	Non‐inferiority	Cardiac death, target vessel myocardial infarction or target lesion revascularisation.	12	Not available
Yu et al. [22]	84 vs 79	1	China	2.25−4.0 mm	July 2017 to July 2018	Non‐inferiority	Late lumen loss (LLL). Secondary endpoint was MACE consisting of Cardiac death, non‐fatal myocardial infarction, target lesion revascularization, and target vessel revascularization.	12	Not available
Dissolve SVD (2024 [20])	129 vs 118	10	China	≥2.25 to ≤2.75 mm	Apr 2018 to Mar 2019	Non‐inferiority	Cardiac death, target vessel myocardial infarction, or ischemia‐driven target lesion revascularisation.	12	NCT03376646
REC‐CAGEFREE‐1 (2024 [23])	1133 versus 1139	43	China	Any	Feb 2021 to May 2022	Non‐inferiority	Cardiac death, target vessel myocardial infarction, clinically and physiologically indicated target lesion revascularization.	24	NCT04561739

^a Registration number in www.clinicaltrial.gov.database.

Baseline patient and lesion characteristics are summarized in Table 2. The population enrolled were largely male (76%) and the average age was 62 years across the studies. The DCB and DES devices used across the trials are detailed in Table 2. All nine studies used a different DCB (with the exception of one device that was used in two trials) but all DCB catheters used in the included studies were paclitaxel‐coated.

Table 2.

Patient and device characteristics.

Trial	Interventions N	DCB type	Stent type	Mean age mean ± SD	Men N, (%)	DM N, (%)	Prior MI N, (%)	ACS N, (%)	RVD, mm mean ± SD	Bailout stenting N, (%)
BELLO (2012 [18])	DCB (90)	IN.PACT falcon	Taxus liberte (paclitaxel)	64.8 ± 8.5	72 (80)	39 (43.3)	46 (51.1)	22 (24.4)	2.15 ± 0.27	18 (20%)
	DES (92)			66.4 ± 9.0	71 (77.2)	35 (38)	33 (35.9)	20 (21.7)	2.26 ± 0.24
BASKET‐SMALL 2 (2018 [16])	DCB (382)	SeQuent please	Taxus element (paclitaxel) initially then xience (everolimus)	67.2 ± 10.3	295 (77)	122 (32)	160 (42)	112 (30)	NR	19 (5.1%)
	DES (376)			68.4 ± 10.3	262 (70)	130 (35)	133 (35)	102 (27)	NR
RESTORE SVD China (2018 [17])	DCB (116)	Restore	Resolute integrity (zotarolimus)	60.1 ± 10.5	66.4 (77)	39.7 (46)	22.4 (26)	69.0 (80)	2.42 ± 0.15	6 (5.2%)
	DES (114)			60.5 ± 10.8	77.2 (88)	42.1 (48)	24.6 (28)	71.1 (81)	2.42 ± 0.18
REVELATION (2019 [19])	DCB (60)	Pantera lux	Orsiro (sirolimus) or xience (everolimus)	57.4 ± 9.2	52 (87)	8 (13)	NR	60 (100)	NR	11 (18%)
	DES (60)			57.3 ± 8.3	52 (87)	4 (7)	NR	60 (100)	NR
PICCOLETO II (2020 [15])	DCB (118)	Elutax SV/emperor	Xience	64 (48–80)	83 (70.3)	45 (38.0)	45 (38)	37 (31.4)	2.23 ± 0.4	8 (6.8%)
	DES (114)		(Everolimus)	66 (50–82)	87 (76.9)	40 (35.4)	34 (30)	32 (21.1)	2.18 ± 0.4
Hao et al. [21]	DCB (38)	Yinyi biotech bingo	Not reported	59.00 ± 11.00	30, (78)	10 (28)	NR	38 (100%)	2.5– 4.0 mm	4 (9.5%)
	DES (42)			56.00 ± 11.00	35 (82)	15 (35)	NR	42 (100%)	2.5−4.0 mm
Yu et al. [22]	DCB (84)	Sequent please	Resolute integrity (zotarolimus) or xience (everolimus) or firehawk (rapamycin)	62.6 ± 8.8	62 (73.8)	16 (19.0)	NR	76 (91.5)	2.77 (2.50 to 3.25)	2 (2.4%)
	DES (79)			64.0 ± 10.5	56 (70.9)	23 (29.1)	NR	69 (87.3)	3.01 (2.65 to 3.39)
Dissolve SVD (2024 [20])	DCB (129)	Dissolve DCB	Resolute integrity (zotarolimus)	60.2 ± 9.5	94 (72.9%)	46 (35.7%)	33 (25.6%)	82 (63.6%)	2.20 ± 0.26	5 (3.9%)
	DES (118)			60.1 ± 9.3	82 (69.5%)	45 (38.1%)	27 (22.9%)	77 (65.3%)	2.21 ± 0.24
REC‐CAGEFREE I (2024 [23])	DCB (1133)	Swide DCB	Firehawk (rapamycin)	61.5 ± 10.3	67.9%	24.9%	7.1%	55.6%	2.74 ± 0.51	106 (9.4%)
	DES (1139)			61.2 ± 10.5	70.7%	29.7%	9.2%	55%	2.82 ± 0.51

Abbreviations: ACS, acute coronary syndrome; DCB, drug coated balloon; DES, drug eluting stent; DM, diabetes mellitus; MI, myocardial infarction; RVD, reference vessel diameter; SD, standard deviation.

The treatment effects for primary and secondary endpoints of interest for angioplasty with DCB versus stenting with DES in de novo CAD are summarized in Figure 1.

Figure 1.

Summary of treatment effects for primary and secondary endpoints of interest for angioplasty with DCB versus stenting with DES in de novo coronary artery disease. CI, confidence interval; DCB, drug coated balloon; DES, drug eluting stent; IRR, incidence rate ratio. [Color figure can be viewed at wileyonlinelibrary.com]

The analysis of the primary composite outcome of MACE showed that the annualized incidence of MACE was comparable between DCB angioplasty and DES‐based PCI (IRR 0.96, 95% CI 0.61−1.48) (Figure 2A). The analysis of cardiac (IRR 1.49, 95% CI 0.96−2.33) (Figure 2B) and all‐cause mortality (IRR 1.18, 95% CI 0.80‐1.74) (Supporting Information S1: Figure 2A) following PCI with DCB or DES showed non‐significant differences, though a numerical trend toward higher cardiac death was seen in the DCB arm. Rates of myocardial infarction (IRR 0.78, 95% CI 0.54 to 1.13) (Figure 2C), TLR (IRR 1.17, 95% CI 0.68−2.01) (Figure 2D), TVR (IRR 1.31, 95% CI 0.73−2.34) (Supporting Information S1: Figure 2B) and target lesion thrombosis (IRR 0.51, 95% CI 0.17−1.54) (Supporting Information S1: Figure 2C) did not differ significantly between treatment groups.

Figure 2.

Treatment effect of angioplasty with DCB versus stenting with DES in de novo disease. (A) Major adverse cardiac events. The analysis of the primary composite outcome of MACE showed that the annualized incidence of MACE was comparable between DCB‐ and DES‐based PCI (IRR 0.96, 95% CI 0.61−1.48). (B) Cardiac death. The analysis of cardiac death (IRR 1.49, 95% CI 0.96−2.33) following PCI with DCB or DES showed non‐significant differences, though a numerical trend toward higher cardiac death was seen in the DCB arm. (C) Myocardial infarction. Rates of myocardial infarction (IRR 0.78, 95% CI 0.54 to 1.13) following PCI with DCB or DES did not differ significantly between treatment groups. (D) Target lesion revascularization. Rates of TLR (IRR 1.17, 95% CI 0.68−2.01) following PCI with DCB or DES did not differ significantly between treatment groups. CI, confidence interval; DCB, drug coated balloon; DES, drug‐eluting stent; IRR, incidence rate ratio; Total, refers to total population size.

Results for MACE (Figure 3) and the main secondary endpoints of interest (Supporting Information S1: Figure 3A−3F) were consistent in a leave‐one‐out analysis. A sensitivity analysis for MACE showed broadly consistent treatment effects in the trials enrolling only patients with small vessel disease (and the small vessel cohort of REC‐CAGEFREE‐1) and those enrolling only patients with ACS (Figure 4). The large vessel component of REC‐CAGEFREE‐1 was an outlier and did not show comparable results between DCB and DES, (IRR of 3.06, 95% CI 1.72−5.39), with DES favored over DCB in this cohort of patients. A sensitivity analysis for the individual components of MACE also showed broadly consistent treatment effects (Supporting Information S1: Figure 4A−4C).

Figure 3.

Influence of leave‐one‐out analysis on treatment effect of angioplasty with DCB versus stenting with DES for major adverse cardiac events. Results were consistent in our leave‐one‐out analysis on treatment effect of angioplasty with DCB versus stenting with DES for MACE. CI, confidence interval; DCB, drug coated balloon; DES, drug‐eluting stent; IRR, incidence rate ratio.

Figure 4.

Sensitivity analysis for major adverse cardiac events. This sensitivity analysis for MACE shows broadly consistent treatment effects in the trials enrolling only patients with small vessel disease (and the small vessel cohort of REC‐CAGEFREE‐1) and those enrolling only patients with ACS. The large vessel component of REC‐CAGEFREE‐1 was an outlier and did not show comparable results. CI, confidence interval; DCB, drug coated balloon; DES, drug‐eluting stent; IRR, incidence rate ratio; Total, number of patients in each arm.

3.1. Qualitative Review

Qualitative assessment of trials showed overall a low risk of bias (Supporting Information S1: Figure 5). According to GRADE, evidence quality was high for the primary endpoint, and moderate or high for the secondary endpoints (Supporting Information S1: Table 5).

4. Discussion

The main finding of our meta‐analysis is that in patients undergoing intervention for de novo coronary disease outcomes are broadly comparable between DCB angioplasty and DES‐ based PCI in terms of MACE. This observation should be interpreted with caution as heterogeneity was high across trials. Key secondary endpoints of interest were also comparable in both groups though a numerical trend was observed for higher cardiac death in the DCB group with low heterogeneity across trials but low event numbers and broad CIs.

The findings of this analysis lend some support to the concept of angioplasty with DCB as a valuable strategy in selected patients undergoing intervention for de novo coronary stenosis. Importantly, however, in terms of MACE heterogeneity was high for the primary endpoint analysis. This may be explained by differences between the individual included trials: seven of the nine trials were small, and trials enrolled patients over a broad historical time frame, including patients treated with different DCB devices—in fact only one of DCB catheter types was studied in more than one trial. It is worth noting that DCB devices do not always perform comparably, even though all DCB catheters used in the included trials contained paclitaxel, the DCB catheters used remain highly heterogenous and grouping these together confers an additional degree of caution when interpretating our results. Details of the device characteristics used in each trial can be found in Table 2. Further data from well‐designed, appropriately powered randomized trials will be required to better define the role of DCB angioplasty in patients with de novo disease.

An additional source of heterogeneity may be the variation in clinical presentation and types of treated lesions in the included trials. In this respect, almost 2/3 of included patients were enrolled in trials focusing on small vessel disease. In addition, just under 20% of patients were enrolled in trials including patients only presenting with ACS. Nevertheless, sensitivity analysis for MACE showed broadly consistent treatment effects in the trials of small vessel disease and those enrolling only patients with ACS.

A recently published individual patient data meta‐analysis including de novo small vessel (<3 mm) disease patients combined 3 of the small vessel trials included in our meta‐analysis and found that DCB angioplasty led to a reduction in MACE (0.67 HR, 95% CI 0.47−0.96) when compared to DES [41]. A similar trend was observed in our sensitivity analysis for MACE in the small vessel (<3 mm) group (0.87 IRR, 95% CI 0.58−1.31) which consisted of trial level data from 6* trials comparing DCB versus DES in small vessel disease.

There is limited randomized data published to date investigating the role of DCB angioplasty versus DES in large vessel de novo CAD. A recent meta‐analysis focusing on this topic included randomized and non‐randomized data, found comparable clinical outcomes between DCB angioplasty and DES [42]. However, this meta‐analysis did not include REC‐CAGEFREE‐1 cohort which demonstrated worse outcomes with DCB angioplasty, which was particularly evident amongst the non‐small vessel (≥3 mm) cohort in that trial.

In relation to secondary endpoints, rates of myocardial infarction and TLR were generally comparable between patients treated with a DCB angioplasty versus a DES‐based strategy. However, the risk of cardiac death was numerically higher with DCB through medium term follow up (IRR = 1.49, 95% CI: 0.96 to 2.33). While the signal of a higher rate of cardiac death with DCB is of concern, the heterogeneity for this observation is low, and the effect should be interpreted cautiously in view of the relatively low number of cardiac deaths across all included trials and the absence of a signal of difference in all‐cause mortality (IRR 1.18, 95% CI 0.80 to 1.74). The higher cardiac death signal seen in this meta‐analysis is also at odds with a large meta‐analysis of 26 randomized trials including 4,590 patients published in 2020 comparing paclitaxel coated balloons versus stents in coronary in‐stent restenosis or de novo lesions where all‐cause mortality and cardiac death was significantly lower at 3 years follow‐up in the DCB arm [43]. Moreover, the observation in our meta‐analysis is not supported by concordant directional differences in myocardial infarction, which tended, in fact, to be lower with DCB, although the contribution of procedural and spontaneous myocardial infarction to the overall rate remains unclear. Moreover, the magnitude of the effect is less when the REC‐CAGEFREE‐I trial is excluded. Finally although the signal of increased cardiac death is somewhat at odds with prior analyses [43] it warrants further investigation, bearing in mind the previous signal seen with paclitaxel coated ballons for the treatment of peripheral vascular disease. Further large scale randomized controlled trials (RCTs) investigating DCB angioplasty versus DES based PCI in de novo CAD are ongoing and further data will help to determine if this is a real signal [44].

The largest included trial was the recently published REC‐CAGEFREE‐I trial. This study is notable for including a broad mix of clinical presentations and vessel size. The leave‐one‐out analysis findings highlight the importance of this trial on the overall analysis. It is the largest trial by a considerable margin, as REC‐CAGEFREE‐1 contributed just over half the total sample size in terms of patient numbers to the overall analysis. Without this trial, numerical trends favoring DCB over DES would have been more apparent in terms of the endpoints MACE, TLR, and myocardial infarction. To counter this however, the trend toward higher cardiac death in the DCB arm would have been less pronounced excluding REC‐CAGEFREE‐1, with the IRR for cardiac death favoring DES being 1.49 (95% CI 0.96 to 2.33) in the overall analysis dropping to 1.22 (95% CI 0.67 to 2.25) when omitting REC‐CAGEFREE‐1.

Certain patient cohorts were systematically excluded from trials evaluating DCB angioplasty versus DES based PCI in de novo CAD. Left main disease was an exclusion criterion in all RCT's to date. In addition, heavily calcified or chronically occluded lesions were also usually excluded from these trials. Finally interventions for in‐stent restenosis were not considered for this analysis [13].

5. Limitations

This study has some limitations that should be considered when interpreting the results. First, the absence of individual patient data limits the analysis to the summary estimates provided from the original publications. For this reason, there were some differences in the components of MACE across trials that could not be overcome. Nevertheless, heterogeneity in definitions were limited and did not preclude a meaningful analysis. Second, the maximum follow‐up length varied across trials. However, we accounted for this aspect by computing incidence rate per person‐years between treatment groups. Third, relevant lesion‐level characteristics and procedural aspects were not uniformly collected across trials, and the recommended dual antiplatelet therapy duration varied between trials and, in some cases, the DCB and DES arms. However, even with access to individual patient data, these limitations could not be fully addressed. Fourth, the DCB and DES devices investigated differed across trials with a possible influence on the comparison between treatments. Also, the heterogeneity in treatment effect is higher across the DCB devices than the DES investigated in the included trials. Fifth, only patients deemed suitable for either DCB or DES based revascularisation strategies were included, and they often mandated residual stenosis post lesion preparation to be ≤30% before randomization in these trials so generalizability of our findings to a true all‐comers population remains unclear. Finally, since all available trials conducted to date have excluded left main disease, heavily calcified lesions, and chronic totally occluded (CTO) lesions, the results may not be generalizable to patients presenting with these complex CAD patterns.

6. Conclusion

A comprehensive assessment of the available evidence from randomized trials of patients with de novo CAD shows that medium‐term incidences of major cardiac outcomes between DCB angioplasty and DES are comparable. Additional evidence is warranted however in view of high heterogeneity observed across trials as well as the numerical trends in important secondary endpoints including cardiac death.

Conflicts of Interest

Professor Robert A. Byrne does not accept direct or personnel payments from the medical device or pharmaceutical industry, he does report research funding to the institution from Abbott Vascular, Biosensors, Boston Scientific and Translumina, without impact on personal remuneration. The other authors declare no conflicts of interest.

Supporting information

supporting materials.

supporting materials.

supporting materials.