Critical review · Mar 20, 2026
SPACIOUS: sirolimus versus paclitaxel DCB for bifurcation side branches
Zhou Y, Hu Y, Zhao X, Chen Z, Li C, Ma L, et al. Sirolimus-coated versus paclitaxel-coated balloons for bifurcated coronary lesions in the side branch: the SPACIOUS trial. EuroIntervention. 2025;21:e307-e317.
doi:10.4244/EIJ-D-24-00742 · Multicenter randomized noninferiority trial (China, 14 centers; NCT04899583), n = 230
Summary
SPACIOUS randomized 230 patients 1:1 to a sirolimus DCB (SCB, Acotec) or a paclitaxel DCB (PCB, Bingo) for side branch dilation after main vessel stenting, in non-left main true bifurcations. The primary endpoint was side branch diameter stenosis (DS) by QCA at 9 months, with a noninferiority margin of 15 absolute points.
| Outcome | SCB | PCB | Difference or p |
|---|---|---|---|
| DS at 9 months | 30.5 ± 16.1% | 33.5 ± 16.2% | −2.94 (95% CI −7.62 to +1.74); p for NI < 0.01; p = 0.127 |
| Binary restenosis | 4.4% | 12.8% | p = 0.043 |
| Target lesion revascularization | 2.6% | 5.2% | p = 0.321 |
Median side branch reference diameter was about 1.9 mm.
Strengths
- Randomized and multicenter, with an objective angiographic primary endpoint and a prespecified noninferiority margin.
- Direct comparison of two DCBs with different drugs in a specific setting (side branch).
Methodological issues
- Reversed labels in Figure 3A. The figure assigns 30.5% to the control (PCB) and 33.5% to sirolimus, the opposite of the abstract and the central illustration. The direction matters for interpretation and needs clarification.
- Wide margin without clinical justification. With a mean PCB DS of about 33.5%, the 15-point margin allows the SCB to reach about 48.5% and still be declared noninferior. The paper does not anchor the margin on a clinical criterion.
- Binary restenosis sensitive to measurement error. QCA variability is ±0.10–0.15 mm. In a 1.9 mm vessel near the 50% threshold, that equals 5–8% uncertainty in DS, which reclassifies borderline cases without a biological cause. With an SD of about 16% in both groups, a large part of the distribution sits near the threshold. A small difference in means, plus slightly higher pre-DCB stenosis in the PCB arm, can shift the count. The p = 0.043 may be noise.
- Clinical events assessed by investigators. There was no independent blinded committee. For hard events this matters less. For target lesion revascularization it matters a great deal: an operator who knows which device was used may read a borderline lesion differently. The numerical TLR difference (5.2% vs 2.6%) should be read with that in mind.
- Comparator without a benchmark. Bingo is approved in China, but I am not aware of comparisons against benchmark PCBs.
Opinion
SPACIOUS proves that this SCB is angiographically noninferior to this PCB, in small non-left main side branches, within a wide margin. It does not prove clinical noninferiority, superiority in restenosis, or applicability beyond this device and this comparator.
It is a first step and should be read as one. I would have started somewhere else.