Sirolimus-Coated Versus Paclitaxel-Coated Balloons for Coronary Artery Disease: A Grade Assessed Systematic Review and Meta-Analysis

Abstract

Despite advances in percutaneous coronary procedures, in-stent restenosis remains a significant challenge. Although sirolimus- and paclitaxel-coated balloons are promising alternatives, their comparative safety and efficacy remain uncertain. PubMed, Embase, and Cochrane databases were searched using relevant keywords from inception until August 2025. A total of 11 studies (7 randomized controlled trials and 4 observational cohort studies) were included, comprising 3633 participants overall. The primary outcomes assessed were target lesion revascularization and target lesion failure. Meanwhile, the Secondary outcomes included stent thrombosis, all-cause mortality, myocardial infarction, major adverse cardiovascular events, survival, binary restenosis, and angiographic endpoints (acute gain, diameter stenosis, in-segment late lumen loss, in-lesion late lumen loss, and in-segment minimal lumen diameter). Interstudy heterogeneity was assessed using I² and X² statistics (I²>50% = significant heterogeneity). Interstudy heterogeneity was low for most outcomes, including all primary clinical endpoints, with moderate heterogeneity observed only for select angiographic measures (notably in-segment late lumen loss and diameter stenosis). Statistical analysis was conducted using R software and RStudio (version 4.4.2), with a P value of < 0.05 indicating statistical significance. This meta-analysis examined studies that compared paclitaxel-coated balloon (PCB) versus standard balloon [sirolimus-coated balloon (SCB)] angioplasty. Regarding primary outcomes, there were no notable variations in target lesion failure [risk ratio (RR), 1.08, 95% CI, 0.90–1.29, P = 0.36] or target lesion revascularization (RR, 1.16, 95% CI, 0.98–1.37, P = 0.08). With all aggregated estimates being nonsignificant, secondary outcomes such as stent thrombosis, all-cause mortality, myocardial infarction, major adverse cardiovascular events, and survival were similar between groups. Angiographic endpoints revealed no discernible variations in late lumen loss (in-lesion and in-segment), acute gain, or diameter stenosis. Nonetheless, the SCB group’s minimal lumen diameter was significantly smaller than that of the PCB group (MD, −0.08 mm, 95% CI, −0.14 to −0.01, P = 0.02). In treating coronary in-stent restenosis, SCB and PCBs show similar overall safety and effectiveness; lesion-specific angiographic variations indicate that customized selection may improve patient outcomes.

INTRODUCTION

Despite the advent of more recent drug-eluting stents (DESs), restenosis following coronary stent implantation is still a clinically significant issue. If in-stent restenosis (ISR) occurs, clinical results are negatively impacted. Introduction: Even with the advent of more advanced DESs, restenosis following coronary stent implantation is still a clinically significant situation. If ISR occurs, clinical results are negatively impacted.¹ There are 2 endovascular therapy approaches for ISR that have been found to be clinically relevant: using a drug-coated balloon (DCB) or implanting another DES.² For patients with coronary artery disease (CAD), the most popular revascularization technique is percutaneous coronary intervention (PCI) combined with the insertion of a DES.³ Nonetheless, universal DES implantation holds limitations related to permanent vessel caging, contributing to persistent accrual of adverse events overtime⁴ By providing homogeneous antiproliferative drug delivery with no foreign body implantation, DCBs offer a cage-free option preserving normal vessel anatomy and function.^5,6 DCB are alternative therapeutic modalities that enable rapid and effective transfer of antiproliferative drugs into the vessel wall following brief balloon inflation. DCB may offer advantages over DES by avoiding the implantation of permanent metal scaffolds in the vessel while still providing antiproliferative effects.^7,8 In recent years, sirolimus and its analogues have become established as the agents of choice for stent-based local drug delivery. In contrast, clinical evidence from randomized controlled trials (RCTs) of DCBs, except for a single study, exists only for paclitaxel-coated balloons (PCB).⁹ Paclitaxel has advantageous cell-specific actions and binds permanently to the microtubes, giving it exceptional persistence in vascular cells. On the other hand, sirolimus and its derivatives bind reversibly to the FK506-binding protein and combine with the rapamycin target in mammals to inhibit cell-cycle progression at the G1-S phase interface.¹⁰

The objective of this research is to conduct a systematic review and meta-analysis of previous studies to assess the safety and effectiveness of balloon angioplasty coated with sirolimus against paclitaxel in the treatment of CAD.

METHODS

Study Design and Protocol Registration

This systematic review adhered to the guidelines set by the Cochrane Collaboration¹¹ and the Preferred Reporting Items for Systematic Reviews and Meta-Analysis framework.¹² It encompassed the study design, stepwise implementation, analysis, and presentation of findings. Additionally, the study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under registration number CRD420251134405.

Search Strategy and Databases

An electronic search of PubMed, Embase, and Cochrane Library was conducted, covering all available entries from their inception to August 2025, without any language restrictions. The following keywords were used: “Sirolimus Coated Balloon,” “Paclitaxel Coated Balloon,” “Percutaneous Coronary Intervention,” “ Acute Coronary Syndrome,” and “Angioplasty.”

Study Selection and Eligibility Criteria

All studies identified through the online search were imported into the Rayyan software for screening, and duplicate records were removed. The remaining studies were initially screened based on their titles and abstracts. Full-text articles were retrieved for further assessment if either reviewer found the abstract potentially relevant. Two independent reviewers (H.N. and U.A.) evaluated the eligibility of each study according to predefined inclusion criteria. Any disagreements were resolved through discussion and consultation with a third reviewer (F.S.). Studies were included if they met the following inclusion criteria: (1) RCTs or observational studies on human patients undergoing PCI, including lesions with no, single, or multiple preexisting coronary stent layers, and (2) studies comparing clinical outcomes and/or angiographic outcomes after sirolimus-coated balloon (SCB) only versus PCB-only PCI for either ISR or native de novo coronary lesions. If there were multiple reports from the same RCTs or registries, the one with the longest available follow-up duration or the most information was used.

Data Extraction and Outcomes

Two authors (Z.R. and Z.I.) extracted data from the included studies into an Excel sheet using a prepiloted form. Baseline data included (age, gender, hypertension, diabetes, dyslipidemia, and acute coronary syndrome presentation). Outcomes were categorized into angiographic, primary, and secondary outcomes. The angiographic outcomes include acute gain, binary restenosis, diameter stenosis after intervention, in-lesion late lumen loss, in-segment late lumen loss, and in-segment minimal lumen diameter. The primary outcomes of this study included target lesion failure (TLF) and target lesion revascularization (TLR). TLF was typically defined as a composite clinical endpoint that included cardiac death, target vessel myocardial infarction (MI), and TLR. The secondary outcomes comprised stent thrombosis, major adverse cardiovascular events (MACE), MI, and death. See Supplemental Table S1, Supplemental Digital Content, https://links.lww.com/HPC/A292 for a detailed definition of outcomes.

Quality Assessment

The quality of the included studies was assessed using appropriate tools based on their study design. For RCTs, the Revised Cochrane Risk of Bias Tool for Randomized Trials (RoB 2)¹³ was used, evaluating bias across 5 domains: randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Each study’s overall risk of bias was categorized as low, some concerns, or high risk.

We assessed the quality of observational studies using the Newcastle–Ottawa Quality Assessment Form for Observational Studies,¹⁴ which includes 3 domains: selection, comparability, and outcome. We applied thresholds to convert the Newcastle–Ottawa scores to Agency for Healthcare Research and Quality standards (good, fair, and poor). The assessment was independently performed by 2 reviewers (I.S. and A.A.), with disagreements resolved by discussion or consultation with a third reviewer (A.A.). The selection domain was rated with a maximum of 4 stars, the comparability domain with a maximum of 2 stars, and the outcome domain with a maximum of 3 stars. Studies scoring 7–9 stars were rated as “low risk of bias,” studies scoring 5–6 stars were rated as “some concerns,” and studies scoring less than 5 stars were rated as “high risk of bias,” ensuring a comprehensive evaluation.

Certainty of Evidence

The Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) tool was employed by 2 independent authors (A.I.K. and U.A.) using the GRADEpro Guideline Development Tool¹⁵ to evaluate the level of certainty of the evidence in this meta-analysis, with categorizations ranging from high to very low.¹⁶ Any disagreements were discussed and resolved through consensus.

Statistical Analysis

Statistical analysis was conducted using R software and R Studio (version 4.4.2, 2024.10.31 + 87279; R Core Team, Vienna, Austria), applying DerSimonian and Laird’s random-effects model to compute pooled estimates with 95% confidence intervals (CIs).¹⁷ The results were visualized in forest plots. Binary outcomes were analyzed using risk ratios (RRs), while continuous outcomes were assessed with mean differences (MDs), both presented in forest plots. Heterogeneity was evaluated using the Cochrane Q χ² test and the I² statistic, with P values <0.10 and I² >50% indicating significant heterogeneity.¹⁸ To test the robustness of the pooled estimates, a leave-one-out analysis was performed, sequentially removing each study and reanalyzing the data, ensuring that no individual study unduly impacted the overall effect estimates.

Reconstructed Time-To-Event Analysis

In the reconstructed time-to-event data analysis, we reconstructed individual patient data (IPD) from the published Kaplan–Meier curves of eligible studies utilizing the curve approach.¹⁹ We adopted the 2-stage approach described by Liu et al,²⁰ adopting the IPD from KM R package. First, we extracted raw data coordinates (survival time and survival probability) of each arm of the included Kaplan–Meier curves. Then, IPD were reconstructed based on the raw data coordinates and the number of patients at risk at the reported time points. We also calculate the accuracy of our reconstructed individual data for reproducibility. Finally, we merged the reconstructed time-to-event data of all individual studies in a merged data set. The proportional hazards assumption was assessed visually and statistically using the Grambsch–Therneau test and Schoenfeld residual plots. We calculated flexible parametric survival models to estimate hazard ratios (HRs) with 95% CI. And using the R package survRM2, we analyzed the variation in restricted mean survival times over time.²¹ A Jackknife resampling analysis was performed on the overall population by systematically excluding 1 study at a time. P < 0.05 was considered statistically significant. Conventional pairwise meta-analyses were performed under random and fixed effects assumptions. Publication bias was analyzed graphically via funnel plot and with Egger’s²² and Begg’s²³ regression test.

Trial Sequential Analysis

Trial sequential analysis was performed for primary outcomes to assess the robustness of our results regarding type 1 and type 2 errors.²⁴ Required information size (RIS) was calculated considering a type 1 error of 5% and a power of 80%. Z-score was calculated to examine the benefit, harm, or futility of SCB compared with PCB in patients with CAD.²⁵

RESULTS

Study Selection

The initial database and register search identified a total of 1348 records. After removing 128 duplicates, 1220 records remained for screening. Of these, 1167 records were excluded based on title and abstract review, leaving 53 full-text reports for eligibility assessment. All full-text articles were successfully obtained; therefore, no reports were excluded due to failure to retrieve the full text. Following a detailed review, 42 studies were excluded (22 due to wrong study design, 12 due to wrong intervention, and 8 due to wrong outcomes). Ultimately, 11 studies met the inclusion criteria and were included in this systematic review (Fig. 1).

Figure 1.

Prisma flowchart. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis.

Study and Patient Characteristics

Eleven studies were ultimately included, comprising 7 RCTs^9,10,26–30 and 4 cohort studies,^31–34 conducted in Europe, Asia, and Malaysia between 2015 and 2025. Overall, these studies enrolled 3633 participants, of whom 2189 were treated with SCB and 1444 with PCB. Sample sizes varied substantially, ranging from 50–186 patients in smaller RCTs to over 1000 lesions in larger multicenter cohort studies. Across studies, the mean age of patients was between 59 and 70 years, with the majority being male (73%–90%). Procedurally, the mean DCB diameter ranged from 2.7 to 3.3 mm, with balloon lengths typically 22–36 mm. Inflation times were consistent across trials at approximately 55–60 seconds, with mean pressures of 9–14 atm. Follow-up duration of all clinical outcomes, that is, target vessel revascularization, TLF, stent thrombosis, all-cause mortality, MI, MACE, and binary restenosis was predominantly 12 months, with some studies reporting angiographic outcomes between 6 and 12 months. Detailed baseline characteristics, including study, patient, DCB, and angiographic characteristics, are presented in Tables 1 and 2.

TABLE 1.

Study Characteristics

Author, Year	Recruitment Period	Country	Type of Study	PCB Type	SCB Type	Use and Type of Intravascular Imaging	Lesion Type	Sample Size		No. of Lesions		Follow-Up Duration (Months)		Adjustment for Possible Cofounders and or Multivariable Analysis
								PCB	SCB	PCB	SCB	Angiography	Clinical
Ahmad 202210	Jun 2019–Feb 2020	Malaysia	RCT	SeQuent Please Neo	SeQuent	Not used	Coronary de novo stenotic lesions (≥70% stenosis or intermediate [50%–70%] with positive ischemia evidence)	35	35	36	37	6	12	NR
Ali 20199	Dec 2015–Jan 2017	Malaysia	RCT	SeQuent Please Neo	SeQuent	Not used	Drug-eluting stent, in-stent-restenosis	25	25	25	26	6	12	NR
Briguori 202331	NR	Italy	Cohort	Restore	Devoir	Not used	Coronary in-stent restenosis	186	186	NR	NR	NR	12	NR
Cortese 202132	NR	Italy, Europe, and Asia	Cohort	Elutax SV/ Emperor	MagicTouchTM	Not used	Broad spectrum including native lesions and in-stent restenosis (ISR)	290	290	NR	NR	NR	12	Multivariable logistic regression
Cuculi 202333	NR	Europe	Cohort	NR	NR	Not used	De novo lesions and in-stent restenosis	139	220	NR	NR	NR	12	Unadjusted descriptive comparison
Han 202326	NR	China	RCT	SeQuent Please Neo	SeQuent	Not used	Drug-eluting stent, in-stent-restenosis	128	130	NR	NR	9	12	NR
Leone 202534	Jan 2018–Dec 2023	Italy	Cohort	- Prevail (paclitaxel + urea) - IN.PACT (paclitaxel + urea) - SeQuent Please NEO (paclitaxel + iopromide) - RESTORE (paclitaxel + shellac)	- Magic Touch (phospholipid nanocarrier coating) - Selution (biodegradable polymer microspheres)	Used but not routinely	De Novo lesions and in-stent restenosis	330	990	408	1161	NR	12	Multivariable Cox regression
Liu 202527	Oct 2020–Jul 2021	China	RCT	SeQuent Please NEO, 3 µg/mm²	SeQuent SCB, 4 µg/mm²	Not used	In-stent restenosis	128	130	141	149	9	12	NR
Ninomiya 202328	NR	Europe	RCT	SeQuent Please Neo	MagicTouchTM	Optical coherence tomography	De novo lesions in small coronary vessels	60	61	63	66	6	12	NR
Pleva 202529	Jan 2019–May 2022	Czech Republic	RCT	SeQuent Please Neo	MagicTouchTM	Not used	Coronary in-stent restenosis	73	72	79	79	12	12	NR
Scheller 202230	Dec 2015–Jan 2017 (Malaysia) Jul 2017-Feb 2020 (Europe)	Malaysia and Europe	RCT	SeQuent Please Neo	SeQuent	Not used	Drug-eluting stent, in-stent restenosis	51	50	52	52	6	12	NR

NR, not reported; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; SCB, sirolimus-coated balloon.

TABLE 2.

Patient, Drug-Coated Balloon and Angiographic Characteristics

Patient Characteristics	Ahmad 202210		Ali 20199		Briguori 202331		Cortese 202132		Cuculi 202333		Han 202326		Leone 202534		Liu 202527		Ninomiya 202328		Pleva 202529		Scheller 202230
	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB	PCB	SCB
Age (years) Mean (SD)	59 (12)	60 (11)	59 (13)	62 (12)	68	67	67	66	66 (10)	67 (11)	64	64	68 (10)	69 (10)	64 (8.5)	64 (9.3)	67 (9.9)	70 (8.4)	68 (10.6)	70 (9.6)	63 (12)	67 (12)
Male (%)	86	74	76	88	77	75	75	77	NR	NR	76	73	85	87	76	73	83	90	81	78	76	86
Hypertension (%)	66	69	92	96	92	90	69	74	NR	NR	70	67	76	79	70	67	73	84	NR	NR	94	98
Diabetes (%)	49	54	76	72	55	56	35	45	25	33	41	34	30	33	41	34	25	25	36	44	59	54
Dyslipidemia (%)	57	51	84	92	90	86	61	67	NR	NR	14	18	82	77	14	18	82	82	NR	NR	84	88
Prior MI (%)	46	34	36	32	25	21	43	48	NR	NR	6	5	29	37	6	5	33	28	69	67	55	56
Prior CABG (%)	3	0	16	4	NR	NR	NR	NR	NR	NR	NR	NR	9	8	NR	NR	0	3	10	14	10	4
Smoker (%)	17	20	8	16	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	12	26	52	47	12	22
LVEF Mean (SD)	52 (11)	57 (16)	56 (14)	57 (7.4)	NR	NR	NR	NR	NR	NR	NR	NR	53 (7)	53 (8)	61 (7.5)	59 (8)	NR	NR	NR	NR	56 (12)	52 (10)
Insulin (%)	9	9	32	39	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	8	3	NR	NR	20	48
DCB procedural characteristics
DCB diameter (mm) Mean (SD)	NR	NR	2.9 (0.4)	2.9 (0.4)	2.8 (0.4)	2.7 (0.6)	2.8 (0.5)	2.8 (0.6)	3 (0.7)	2.7 (0.5)	NR	NR	2.8 (0.6)	2.9 (0.5)	3.0 (0.4)	3.0 (0.4)	2.3 (0.3)	2.5 (0.3)	3.23 (1.32)	3.32 (1.13)	3.0 (0.6)	3.0 (0.4)
DCB length (mm) Mean (SD)	NR	NR	27 (11)	26 (7.9)	NR	NR	NR	NR	NR	NR	NR	NR	31 (18)	34 (19)	23 (6.2)	23 (6.2)	22.8 (6.2)	23 (5.7)	32 (42)	36 (42)	NR	NR
Balloon inflation time (s) Mean (SD)	58 (11)	57 (8)	59.3 (6.1)	59 (7.1)	58 (6)	58 (5)	56 (30)	58 (13)	NR	NR	NR	NR	NR	NR	57.2 (13)	59 (13)	70 (38)	74 (20)	NR	NR	50 (14)	56 (16)
Inflation pressure (atm) Mean (SD)	13.8 (2.8)	13.5 (3.1)	10.3 (2.8)	11.6 (3.2)	13 (5)	12 (3)	11 (4)	11 (4)	9.7 (5.2)	7 (4)	NR	NR	11 (3)	12 (3)	9.4 (2.8)	9.3 (2.7)	9 (3)	10 (3)	18 (12)	17 (11)	8 (2)	10 (3)
Angiographic characteristics
Pre-PCI MLD (mm) Mean (SD)	0.90 (0.38)	0.96 (0.43)	0.80 (0.52)	0.81 (0.35)	0.79 (0.50)	0.78 (0.49)	NR	NR	NR	NR	NR	NR	NR	NR	0.94 (0.36)	0.86 (0.41)	0.92 (0.22)	0.98 (0.32)	0.94 (1.58)	0.72 (1.21)	0.80 (0.44)	0.82 (0.40)
Pre-PCI RFD (mm) Mean (SD)	2.81 (0.59)	2.72 (0.40)	2.42 (0.54)	2.53 (0.53)	2.55 (0.62)	2.55 (0.62)	NR	NR	NR	NR	NR	NR	NR	NR	2.52 (0.47)	2.47 (0.43)	2 (0.38)	2.15 (0.41)	2.77 (1.57)	2.72 (1.33)	2.58 (0.51)	2.66 (0.51)
Pre-PCI lesion length (mm) Mean (SD)	26.7 (8.04)	23.8 (7.32)	13.3 (7.18)	14.3 (7.83)	NR	NR	NR	NR	NR	NR	NR	NR	47 (32)	48 (30)	12.7 (6.44)	13.4 (7.51)	11.4 (5.32)	10.5 (6.23)	23 (36)	24.5 (37.7)	13.8 (8.20)	13.7 (7.16)
In-Lesion Pre-PCI diameter stenosis	63.6 (13.2)	61.1 (15)	69.3 (19.6)	67.4 (13.5)	79.0 (15)	78.0 (17)	NR	NR	NR	NR	NR	NR	NR	NR	62.8 (11.8)	65.2 (16.1)	53.7 (11.4)	53 (12.9)	70 (49)	76 (37)	70.4 (16.3)	62.8 (24.4)

CABG indicates coronary artery bypass grafting; DCB, drug-coated balloon; LVEF, left ventricular ejection fraction; MI, myocardial infarction; MLD, minimal lumen diameter; NR, not reported; PCB, paclitaxel-coated balloon; PCI, percutaneous coronary intervention; RFD, reference diameter; SCB, sirolimus-coated balloon.

Risk of Bias Assessment

Randomized Controlled Trials

The risk of bias assessment for RCTs was conducted using the Cochrane RoB 2.0 tool across 5 domains. Overall, the majority of studies were judged to be at low risk of bias. Six out of 7 studies demonstrated low risk across all domains, while 1 study (Scheller et al³⁰) showed some concerns in the domain related to deviations from intended interventions, leading to an overall judgment of “some concerns.” Domain-specific analysis revealed consistently low risk of bias for randomization, missing outcome data, measurement of outcomes, and selective reporting across all included studies. The only area with minor concerns was bias due to deviations from intended interventions, where a small proportion of studies were flagged. Taken together, the evidence base was considered to be of generally high methodological quality with minimal threats to validity (Supplemental Figure S1 and S2, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Observational Studies

The quality of the included observational cohort studies was assessed using the Newcastle–Ottawa Scale (NOS). Scores ranged from 4 to 9 stars, reflecting variability in methodological rigor. Leone et al³⁴ achieved the maximum score of 9, indicating high-quality design with strong selection, comparability, and outcome assessment. Briguori et al³¹ and Cortese et al³² both scored 7, suggesting good quality with minor limitations. In contrast, Cuculi et al³³ scored 4, reflecting concerns particularly in comparability and outcome domains. Overall, most cohort studies demonstrated acceptable methodological quality, even though some variability was noted across domains (Supplemental Table S2, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Certainty of Evidence

The GRADE assessment comparing SCBs versus PCBs for CAD found no significant differences across major clinical outcomes, including TLF, revascularization, MACE, death, MI, stent thrombosis, and binary restenosis, with most evidence rated as moderate certainty. Angiographic outcomes such as acute gain, diameter stenosis, and late lumen loss were also comparable, with certainty ranging from low to high. Overall, the results suggest that both devices demonstrate broadly similar efficacy and safety, though wide CIs and limited precision in some outcomes reduce confidence in the estimates. Detailed assessment is present in Supplemental Table S3, Supplemental Digital Content, https://links.lww.com/HPC/A292.

Outcomes

Primary Outcomes

Target Lesion Revascularization

The pooled analysis of 6 RCTs, including 1328 patients treated with SCB and 666 with PCB, showed no significant difference at 12 months (RR, 1.19, 95% CI, 0.97–1.47, P = 0.076, I² = 0%). The observational cohort (Cortese et al³²; 290 SCB vs. 290 PCB) also found no significant effect (RR, 1.04, 95% CI, 0.60–1.81) (Fig. 2). When combined, a total of 1618 SCB and 956 PCB patients were analyzed, and the overall effect remained nonsignificant (RR, 1.16, 95% CI, 0.98–1.37, P = 0.08, I² = 0%) with no subgroup difference (P = 0.65). Sensitivity analysis further confirmed robustness: omitting individual studies produced pooled RRs ranging from 1.10 to 1.20, with P values between 0.06 and 0.28, and I² consistently 0%, indicating stable results across all analyses (Supplemental Figure S3, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 2.

Forest plot of target vessel revascularization. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; RR, risk ratio; SCB, sirolimus-coated balloon.

Target Lesion Failure

In the RCT subgroup (7 studies; 503 SCB vs. 499 PCB patients), the pooled RR was 1.11 (95% CI, 0.84–1.46, P = 0.45) with no heterogeneity (I² = 0%, P = 0.82). In the observational subgroup (4 studies; 1686 SCB vs. 945 PCB patients), the pooled RR was 1.06 (95% CI, 0.68–1.64, P = 0.81) with low, nonsignificant heterogeneity (I² = 12.2%, P = 0.33). When all 11 studies (2189 SCB vs. 1444 PCB patients) were pooled, the overall effect remained nonsignificant (RR, 1.08, 95% CI, 0.90–1.29, P = 0.36, I² = 0%) (Fig. 3). Leave-one-out analysis confirmed the robustness of these findings, as omission of individual studies did not materially change the pooled estimates (RR range, 1.01–1.12), and heterogeneity remained negligible. Overall, the results indicate comparable outcomes between SCB and PCB, with no evidence of superiority for either strategy (Supplemental Figure S4, Supplemental Digital Content, https://links.lww.com/HPC/A292). These results indicate that PCB and SCB have comparable efficacy in preventing both TLR and TLF.

Figure 3.

Forest plot of target lesion failure. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; RR, risk ratio; SCB, sirolimus-coated balloon.

Secondary Outcomes

Stent Thrombosis

This analysis included 4 RCTs (Ali et al⁹, Scheller et al³⁰, Pleva et al³³, Ninomiya et al²⁸). Across these studies, there were 3 events out of 208 patients in the SCB group and 3 events out of 209 patients in the PCB group. The pooled RR was 1.03 (95% CI, 0.18–6.05), indicating no significant difference between SCB and PCB in terms of stent thrombosis. The wide CI reflects the low number of events and limited statistical power. Heterogeneity was absent (I² = 0%), and the test for overall effect was nonsignificant (P = 0.960), showing that the data do not favor either treatment arm (Fig. 4). Leave-one-out sensitivity analysis confirmed the robustness of the findings, with RRs ranging from 0.69 to 1.40 and all P-values remaining nonsignificant (0.64–0.74), indicating that no single study disproportionately influenced the overall result (Supplemental Figure S5, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 4.

Forest plot of stent thrombosis. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon.

All-Cause Mortality

The RCT subgroup (6 trials; 373 SCB vs. 371 PCB patients) yielded a pooled RR of 0.37 (95% CI, 0.10–1.46) with no heterogeneity (I² = 0%), suggesting a trend toward fewer events in the SCB group but without statistical significance. In the observational subgroup (2 studies; 1280 SCB vs. 620 PCB patients), the pooled RR was 1.40 (95% CI, 0.46–4.26) with mild heterogeneity (I² = 21.2%), showing no clear difference. When all studies were combined (31/1653 SCB vs. 17/991 PCB events), the pooled RR was 0.99 (95% CI, 0.39–2.52), again with no significant difference, although subgroup analysis indicated possible differences between RCTs and cohorts (P = 0.0385) (Fig. 5). Leave-one-out sensitivity analysis confirmed robustness as sequential omission of studies did not materially change the effect estimate (RR range, 0.56–1.13) and heterogeneity remained negligible to low (Supplemental Figure S6, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 5.

Forest plot of all-cause mortality. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; RR, risk ratio; SCB, sirolimus-coated balloon.

Myocardial Infarction

For MI, observational cohorts including 1280 SCB and 620 PCB patients showed a pooled RR of 1.43 (95% CI, 0.38–5.40; I² = 0%, P = 0.60), which was not statistically significant. In RCTs with 313 SCB and 311 PCB patients, the pooled RR was 1.44 (95% CI, 0.45–4.56; I² = 0%, P = 0.53), also nonsignificant. When all studies were combined, the overall RR was 1.43 (95% CI, 0.71–2.89; I² = 0%, P = 0.32), showing no significant difference between SCB and PCB (Fig. 6). Sensitivity analysis confirmed the stability of this result, with pooled RRs ranging from 1.38 to 1.54 and all but one P value above 0.10, indicating that no individual study substantially altered the overall conclusion (Supplemental Figure S7, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 6.

Forest plot of myocardial infarction. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; RR, risk ratio; SCB, sirolimus-coated balloon.

Major Adverse Cardiovascular Events

In the RCT subgroup (6 trials, 373 SCB vs. 371 PCB), the pooled RR was 1.03 (95% CI, 0.68–1.55) with negligible heterogeneity (I² = 0%), indicating no significant difference between SCB and PCB. In contrast, the observational cohort subgroup (2 studies, 1280 SCB vs. 620 PCB) showed a higher but highly uncertain risk with SCB (RR, 1.76, 95% CI, 0.16–19.60) and moderate heterogeneity (I² = 61%). When all studies were combined, the pooled RR was 1.07 (95% CI, 0.77–1.48) with low heterogeneity (I² = 0%) and no significant overall or subgroup differences, suggesting consistent findings across study designs (Fig. 7). The leave-one-out sensitivity analysis shows consistent results across studies, with pooled RR estimates ranging from 1.03 to 1.13 and all CIs crossing 1, indicating no significant difference between SCB vs PCB. The overall RR of 1.07 (95% CI, 0.77–1.48, P = 0.65) confirms the robustness and lack of statistical significance (Supplemental Figure S8, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 7.

Forest plot of major adverse cardiovascular events. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RCT, randomized controlled trials; RR, risk ratio; SCB, sirolimus-coated balloon.

Binary Restenosis

This forest plot compares the outcomes of SCB versus PCB across 5 studies. The pooled analysis included 341 patients in the SCB group and 343 in the PCB group, with 75 and 54 events, respectively. The overall pooled RR was 1.39 (95% CI, 0.83–2.31), suggesting a trend toward higher event rates in the SCB group compared with PCB, but this did not reach statistical significance (P = 0.149). Heterogeneity across studies was low (I² = 18%, P = 0.300), supporting consistency in findings (Fig. 8). The sensitivity analysis shows that omitting any single study does not significantly alter the pooled RR, which remains nonsignificant (overall RR, 1.39, 95% CI, 0.83–2.31, P = 0.15), indicating stable and consistent findings between SCB and PCB (Supplemental Figure S9, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 8.

Forest plot of binary restenosis. CI indicates confidence interval; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon.

Angiographic Outcomes

Acute Gain

This forest plot shows the pooled analysis of acute gain between SCB and PCB across 4 RCTs involving a total of 176 participants in the SCB group and 169 in the PCB group. The MD in acute gain was very small and nonsignificant in all trials, ranging from −0.19 (Ali et al⁹) to +0.06 (Ninomiya et al²⁸), with all 95% CIs crossing zero. The overall pooled effect demonstrated an MD of −0.01 (95% CI, −0.11 to 0.08; P = 0.76), indicating no significant difference between SCB and PCB. Heterogeneity was negligible (I² = 0%), suggesting consistency across studies (Fig. 9). The sensitivity analysis showed minimal variation in MD, with all CIs crossing zero. The overall MD was −0.01 (95% CI, −0.11 to 0.08, P = 0.76), indicating no significant difference between SCB and PCB and confirming the robustness of the findings (Supplemental Figure S10, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 9.

Forest plot of acute gain. CI indicates confidence interval; MD, mean difference; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon; SD, standard deviation.

Diameter Stenosis

Pooled analysis of 7 RCTs, involving a total of 516 participants in the SCB group and 509 in the PCB group, was analyzed. Analysis showed no statistically significant overall effect, with a pooled MD of 2.08 (95% CI, −1.14 to 5.30, P = 0.21). Moderate heterogeneity is present (I² = 39.7%, P = 0.1265), but not enough to invalidate the findings. Individual study results vary, yet none show a significant difference, indicating no clear advantage of SCB over PCB (Fig. 10). The leave-one-out sensitivity analysis yielded consistent, nonsignificant results across all study omissions, with an overall MD of 2.08 (95% CI, −1.14 to 5.30, P = 0.21), indicating no meaningful difference between SCB and PCB (Supplemental Figure S11, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 10.

Forest plot of diameter stenosis. CI indicates confidence interval; MD, mean difference; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon; SD, standard deviation.

In-Segment Late Lumen Loss

Pooled analysis of 6 RCTs, involving a total of 404 participants in the SCB group and 389 in the PCB group, was analyzed. The overall MD is 0.07 (95% CI, −0.01 to 0.15), which is not statistically significant (P = 0.10). While some individual studies, like Ahmad et al¹⁰ and Ninomiya et al²⁸, show significant positive effects favoring SCB, others show negligible or negative differences. The heterogeneity is moderate (I² = 58.5%, P = 0.0343), indicating variability among study results that may affect the reliability of the pooled estimate (Fig. 11). The leave-one-out analysis reveals stable results across all omissions, with the overall MD of 0.07 (95% CI, −0.01 to 0.15, P = 0.10), suggesting no statistically significant difference between SCB and PCB (Supplemental Figure S12, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 11.

Forest plot of in-segment late lumen loss. CI indicates confidence interval; MD, mean difference; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon; SD, standard deviation.

In-Lesion Late Lumen Loss

The second meta-analysis of 4 studies, involving a total of 227 participants in the SCB group and 233 in the PCB group, was analyzed. The overall MD is even smaller at 0.03 (95% CI, −0.02 to 0.09), with no statistical significance (P = 0.27). Heterogeneity is minimal (I² = 0%, P = 0.6654), suggesting consistent findings across studies. However, the lack of significant effect and narrow CI imply that SCB does not offer a meaningful advantage over PCB in this dataset (Fig. 12). The leave-one-out analysis showed consistent, nonsignificant results across all study omissions, with an overall MD of 0.03 (95% CI, −0.02 to 0.09, P = 0.27), indicating no meaningful difference between SCB and PCB (Supplemental Figure S13, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 12.

Forest plot of in-lesion late lumen loss. CI indicates confidence interval; MD, mean difference; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon; SD, standard deviation.

In-Segment Minimal Lumen Diameter (mm)

Seven RCTs, involving a total of 516 participants in the SCB group and 509 in the PCB group, were analysed. The overall MD is −0.08 (95% CI, −0.14 to −0.01), which is statistically significant (P = 0.02), indicating a slight but meaningful lower baseline value in the SCB group compared with PCB. Heterogeneity is absent (I² = 0%, P = 0.8543), suggesting consistent results across studies. While individual effects vary slightly, the pooled data support a small but significant baseline difference favoring PCB. This consistency across a large sample size strengthens the reliability of the finding (Fig. 13). The leave-one-out analysis consistently shows a significant MD favoring SCB, with an overall MD of −0.08 (95% CI, −0.14 to −0.01, P = 0.02), indicating a robust and statistically significant effect (Supplemental Figure S14, Supplemental Digital Content, https://links.lww.com/HPC/A292).

Figure 13.

Forest plot of in-segment minimal lumen diameter. CI indicates confidence interval; MD, mean difference; PCB, paclitaxel-coated balloon; RR, risk ratio; SCB, sirolimus-coated balloon; SD, standard deviation.

Publication Bias

Publication bias was assessed visually using funnel plots and statistically using Begg’s and Egger’s regression tests across all clinical and angiographic outcomes (Supplemental Figures S15–S26, Supplemental Digital Content, https://links.lww.com/HPC/A292). Overall, the majority of outcomes demonstrated symmetrical funnel plots with nonsignificant Begg’s and Egger’s tests, indicating no evidence of small-study effects or selective publication. This consistency supports the robustness of the pooled estimates for key clinical endpoints, including TLR, TLF, stent thrombosis, MI, MACE, and most angiographic measures.

For TLR, TLF, stent thrombosis, MI, MACE, binary restenosis, diameter stenosis, late lumen loss (in-segment and in-lesion), and in-segment minimal lumen diameter, funnel plots were visually symmetric and statistical tests were uniformly nonsignificant, suggesting a low risk of publication bias (Supplemental Figures S15–S17, S19–S21, S23–S26, Supplemental Digital Content, https://links.lww.com/HPC/A292).

In contrast, all-cause mortality and acute gain demonstrated some evidence of potential small-study effects. For mortality, funnel plot asymmetry was observed, and Egger’s test reached statistical significance, while Begg’s test was borderline, indicating that publication bias cannot be excluded for this outcome (Supplemental Figure S18, Supplemental Digital Content, https://links.lww.com/HPC/A292). Similarly, acute gain showed visual asymmetry with a significant Egger’s test despite a nonsignificant Begg’s test, suggesting possible small-study effects (Supplemental Figure S22, Supplemental Digital Content, https://links.lww.com/HPC/A292). Accordingly, the pooled estimates for these outcomes should be interpreted with caution.

Taken together, the overall risk of publication bias across outcomes was low, with isolated signals in select endpoints that did not materially alter the main conclusions of the meta-analysis.

Reconstructed Time-to-event Analysis for Target Lesion Failure

Four studies (2 RCTs^27,29 and 2 observational cohort^33,34) report Kaplan–Meier curve data for a total of 2316 lesions in patients treated with sirolimus (1576 lesions) and PCBs (740 lesions), with a median follow-up of about 12 months, with quantile (Q) 1 followed up to 7 months and Q3 up to 12 months. At 1 year, the pooled survival analyses of reconstructed time-to-event IPD showed no significant difference in TLF in patients treated with PCB and SCB for the treatment of CAD (HR, 1.06; 95% CI, 0.81–1.39; P = 0.657; Supplemental Figure S27 A, Supplemental Digital Content, https://links.lww.com/HPC/A292). There was no visual evidence of violation of the proportional hazards assumption, and in the plot of Schoenfeld residuals against time, also in the (Supplemental Figure S27 D, Supplemental Digital Content, https://links.lww.com/HPC/A292) Grambsch–Therneau test (Supplemental Figure S27 D, Supplemental Digital Content, https://links.lww.com/HPC/A292) was not statistically significant (P = 0.054), indicating time-invariant effect of HR. Supplemental Figure S27 B, Supplemental Digital Content, hazard ratiohazard ratiohttps://links.lww.com/HPC/A292, depicts the analysis of time-varying HR for overall survival based on flexible parametric survival models with B-splines, which also showed that HR was nonsignificant and constant over the follow-up of 1 year. Supplemental Figure S27 E, Supplemental Digital Content, https://links.lww.com/HPC/A292, presents the difference in mean event-free survival time [difference in survival time when time of follow-up is restricted to 1 year (restricted mean survival time)]. The curve over zero is stable almost the entire follow-up, showing no significant difference in event-free survival time with SCB and PCB. The jackknife leave-one-out sensitivity analysis shows that our results are robust and are not dependent on a single study Supplemental Figure S29, Supplemental Digital Content, https://links.lww.com/HPC/A292. Conventional meta-analysis of HR also shows no-significant difference in the TLF free survival time by SCB or PCB in patients treated with CAD Supplemental Figure S27 C, Supplemental Digital Content, https://links.lww.com/HPC/A292. Funnel plot showing no publication bias (Supplemental Figure S28, Supplemental Digital Content, https://links.lww.com/HPC/A292) and P value of 0.5 for Egger’s and Begg’s regression test (Supplemental Figure S28, Supplemental Digital Content, https://links.lww.com/HPC/A292), validating no publication bias for this outcome.

Trial Sequential Analysis

The cumulative Z-score for TLF (Supplemental Figure S30, Supplemental Digital Content, https://links.lww.com/HPC/A292) and TLR (Supplemental Figure S31, Supplemental Digital Content, https://links.lww.com/HPC/A292) did not reach RIS nor crossed the conventional trial sequential analysis boundary, indicating further large sample size trials are necessary to obtain firm evidence.

DISCUSSION

This meta-analysis of 11 studies, including RCTs and observational cohorts, reveals no significant differences between SCB and PCBs regarding the primary outcomes of TLR and TLF. Similarly, secondary outcomes, including stent thrombosis, all-cause mortality, MI, MACE, binary restenosis, and survival, demonstrated comparable efficacy. Angiographic results did not show a substantial advantage for either device, with the exception of a minor but statistically significant difference in minimal lumen diameter, which favored PCB. Overall, our findings suggest that SCB and PCB are equally safe and effective in contemporary cardiac treatments.

DCBs have emerged as a notable alternative to DESs in the treatment of CAD, particularly for ISR and small-vessel disease.^35–37 PCBs were the initial substantially utilized devices, and numerous significant studies conducted over the past decade have demonstrated their efficacy. SCBs have been introduced more recently. They may be more effective and safer in preventing restenosis, as sirolimus functions by inhibiting cellular proliferation rather than inducing cell death.^34,38 Sirolimus produces reversible cell-cycle arrest at the G1-S checkpoint, whereas paclitaxel causes cytotoxicity through irreversible microtubule stabilization. Nonetheless, material directly contrasting SCB with PCB is scarce and fragmentary. Preclinical and animal-model data extrapolated from stent-based platforms suggest more uniform neointimal suppression and possibly improved vascular healing with sirolimus; however, balloon-based delivery differs significantly in drug residence time, tissue uptake, and excipient technology, which may attenuate these theoretical advantages.^38,39

Our pooled analysis revealed no significant difference between SCB and PCB for TLR (RR, 1.16, 95% CI, 0.98–1.37, P = 0.08), with results consistent across RCTs and cohort studies. Prior meta-analyses (eg, Liyis et al³⁹) predominantly focused on the comparison between PCBs and DESs or bare-metal stents, consistently evidencing the superiority of PCB over angioplasty alone. However, direct comparisons with SCB were largely absent.⁴⁰ A recent meta-analysis by Shin et al⁴¹ suggested that SCB may not be inferior to PCB for ISR; nevertheless, the CIs were extensive. Our research expands upon this by compiling the most extensive direct comparison dataset to date, demonstrating parity rather than advantage of either balloon type.

No difference was observed in TLF, aligning with previous studies that indicated similar rates of lesion failure in small-vessel disease. Prior research focusing on PCB has demonstrated reduced TLF relative to conventional angioplasty; however, SCB was not investigated.^42,43 Our findings confirm SCB as a feasible alternative, exhibiting similar endurance to PCB across multiple lesion categories. The absence of variability among trials bolsters the validity of this claim, and the low interstudy heterogeneity observed for this outcome strengthens the robustness of this conclusion.

In 4 RCTs, the incidence of stent thrombosis was minimal and statistically equivalent across all groups (RR, 1.03). This corresponds with prior studies focused on PCBs that suggest a minimal risk of thrombosis, especially when DCBs are employed without stent placement.⁴⁴ Previous evidence regarding SCB was limited to small, single-center studies (Ahmad et al¹⁰), demonstrating enhanced safety. Our work, which integrates both methodologies, demonstrates that thrombosis is rather uncommon, regardless of the application of drug coating, which reinforces their safety profile.

The comprehensive pooled analysis indicated no disparity in mortality between SCB and PCB. Subgroup analysis indicated divergence: RCTs revealed a nonsignificant trend favoring SCB, whereas cohort data suggested the contrary. This difference is similar to what was found in an original study, where registry-based analysis showed that SCB had inferior results in high-risk real-world populations.⁴⁵ On the other hand, smaller RCTs, such as the one by Ninomiya et al,²⁸ found positive but not statistically significant trends in mortality with SCB.²⁸ Evidence of publication bias in mortality outcomes necessitates cautious interpretation. Publication bias identified in mortality outcomes affects interpretation. Consequently, our findings underscore the necessity for sufficiently powered randomized studies to elucidate mortality disparities, if present.

The rates of MI were comparable across SCB and PCB (RR, 1.43, 95% CI, 0.71–2.89), consistent across both research designs. Previous PCB investigations (Scheller et al⁴⁶) indicated decreased MI in comparison to angioplasty; however, these studies utilized nondrug comparators. Our analysis provides new evidence that SCB yields outcomes similar to PCB without increasing ischemia risk. Because MI occurrences happen so seldom, we require bigger groups of people to better set noninferiority margins.

Our findings indicated no difference in MACE, corroborating the conclusions of Vlieger et al⁴⁷, who reported similar MACE rates between SCB and PCB in real-world small-vessel populations. The consistency of our findings across RCTs and cohorts reinforces the conclusion that SCB performs comparably to PCB; nevertheless, long-term consequences extending beyond 12 months are yet inadequately investigated.

Binary restenosis exhibited a nonsignificant trend indicating elevated rates in SCB-treated lesions. Prior evidence on PCBs, including the Intracoronary Stenting and Angiographic Results: Drug Eluting Stents for In-Stent Restenosis 3 trial (2013), indicated that PCBs outperformed simple balloons and were not inferior to DES, dramatically decreasing the need for repeat revascularization at the 10-year follow-up.⁴⁸ Initial SCB trials (Ahmad et al¹⁰) indicated potential antirestenotic benefits; however, our synthesis reveals no statistically significant advantage to PCB. This is different from the biological reasons that support sirolimus, which may suggest that existing drug delivery technology may limit the clinical translation of its theoretical benefits.

Our aggregated survival analysis revealed no distinction between SCB and PCB (HR, 1.06). Survival has seldom been a principal endpoint in previous PCB trials, which concentrated on angiographic surrogates. Several investigations indicated parity among devices, aligning with our results. These results indicate that survival advantages are improbable to differ among coatings within the current follow-up durations.⁴⁹

No significant difference in angiographic outcomes was seen across groups, consistent with the findings of Ali et al.⁹ and Ninomiya et al.²⁸ Previous meta-analyses of PCB consistently demonstrated a greater acute gain compared with angioplasty alone; nevertheless, our data confirm that SCB and PCB are equivalent in this aspect.⁵⁰ The pooled analysis indicated no difference (MD, 2.08), corroborating the findings of Ahmad et al¹⁰ and Pleva et al.²⁹ PCB has exhibited superiority over angioplasty in DES-ISR lesions; our results affirm parity between coatings. Neither in-segment nor in-lesion late lumen loss exhibited significant differences; however, moderate heterogeneity was observed in the in-segment findings. Previous research, including Cortese et al³², demonstrated PCB’s superiority over DES in ISR; however, our aggregated results indicate that SCB achieves similar late luminal effects.³² A minor yet statistically significant difference favoring PCB (MD, −0.08, P = 0.02). This differs from Ahmad et al¹⁰, who documented enhanced lumen diameter with SCB in small-vessel disease. The consistency of this small signal across pooled analyses suggests a reproducible but clinically modest effect, potentially related to differences in drug-transfer kinetics or balloon-platform characteristics rather than true pharmacologic superiority.

Several included trials employed intravascular imaging modalities such as intravascular ultrasound or optical coherence tomography to guide lesion preparation and assess procedural success; however, imaging use was not standardized across all studies. This variability may partially explain the heterogeneity observed in angiographic endpoints and highlights the importance of imaging-guided protocols in future comparative trials.⁴¹

Strengths and Limitations

This meta-analysis is robust due to its inclusion of diverse RCTs and observational research across various populations and lesion types. This is the most extensive aggregated dataset that directly contrasts SCB with PCB. Comprehensive sensitivity analyses and assessments for publication bias enhance the credibility of the results, while limited heterogeneity across most outcomes strengthens confidence in the findings. Nevertheless, limitations must be acknowledged. The event rates for critical outcomes such as stent thrombosis and mortality were low, hence diminishing the statistical power. The follow-up period was generally short (6–12 months), hindering assessments of long-term outcomes. Third, variations in patient selection and the types of lesions examined generate heterogeneity that statistical analyses may not consistently reveal. Fourth, evidence of potential publication bias in mortality outcomes necessitates cautious interpretation. The slight yet significant variation in minimal lumen diameter indicates potential performance discrepancies related to the device that warrant further investigation.

Future Recommendations

Future research should prioritize comprehensive, multicenter RCTs that directly compare subcutaneous (SCB) and percutaneous (PCB) methods, with extended follow-up, ideally surpassing 3 years, to assess the durability of results. Trials should incorporate standardized intravascular imaging guidance, stratify outcomes by lesion subsets (eg, ISR, small-vessel disease, de novo lesions), and include clinically meaningful endpoints such as TLF and repeat revascularization rather than angiographic surrogates alone. Clinical trials should classify outcomes based on patient comorbidities, such as diabetes, which may influence therapy efficacy. Insights from the peripheral vascular literature, where DCBs have been used extensively for femoropopliteal ISR, suggest that late durability and long-term safety signals are critical and should inform future coronary trial design. Ultimately, standardized angiographic and clinical objectives would facilitate the comparison of future trials and meta-analyses.

CONCLUSIONS

This meta-analysis demonstrates that SCB and PCBs yield comparable results in clinical, angiographic, and survival outcomes. Despite a little difference in minimal lumen diameter favoring PCB, no consistent advantage was demonstrated for either method. These findings suggest that SCB could serve as a viable alternative to PCB in contemporary applications. Nonetheless, owing to limitations in event rates and follow-up, further extensive randomized studies are essential to confirm long-term comparative effectiveness and to ascertain whether specific patient subgroups may have larger advantages from one device type over another.

DISCLOSURES

Nothing to declare.