Bifur-EP-PICO7
2026-03-21
MID_CRIT = 10 # mortality, stent thrombosis
MID = 20 # everything else
# label.e\t= Label for experimental group. label.c\t= Label for control group.
# label.left = Graph label on left side of forest plot. label.right\t= Graph
# label on right side of forest plot.AI-generated code
AI-generated code
Summary of Findings (Inconsitency and Imprecision Assessments)
## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(common)
## Kim 2021 1.0569 [0.4538; 2.4617] 62.9
## Jin 2019 0.0929 [0.0053; 1.6320] 37.1
##
## Number of studies: k = 2
## Number of observations: o = 249 (o.e = 126, o.c = 123)
## Number of events: e = 24
##
## RR 95%-CI z p-value
## Common effect model 0.6993 [0.3270; 1.4955] -0.92 0.3564
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 1.7936 [0.0000; >100.0000]; tau = 1.3392 [0.0000; >10.0000]
## I^2 = 60.7% [0.0%; 90.8%]; H = 1.59 [1.00; 3.31]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.54 1 0.1107
##
## Details of meta-analysis methods:
## - Mantel-Haenszel method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## - Continuity correction of 0.5 in studies with zero cell frequencies (applied twice to sample sizes, like RevMan 5)
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using common/fixed effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(random)
## Kim 2021 0.9512 [0.1373; 6.5882] 12.7
## Niemelä 2011 0.7532 [0.1704; 3.3290] 21.5
## Pan 2011 2.4194 [0.4785; 12.2314] 18.1
## Jin 2019 0.3407 [0.0143; 8.1456] 4.7
## Kim 2015 2.5662 [0.8226; 8.0053] 36.7
## Watanabe 2021 0.9828 [0.0630; 15.3360] 6.3
##
## Number of studies: k = 6
## Number of observations: o = 1391 (o.e = 697, o.c = 694)
## Number of events: e = 35
##
## RR 95%-CI z p-value
## Random effects model 1.4719 [0.7388; 2.9325] 1.10 0.2716
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0 [0.0000; 2.2017]; tau = 0 [0.0000; 1.4838]
## I^2 = 0.0% [0.0%; 74.6%]; H = 1.00 [1.00; 1.99]
##
## Test of heterogeneity:
## Q d.f. p-value
## 3.15 5 0.6762
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## - Continuity correction of 0.5 in studies with zero cell frequencies (applied twice to sample sizes, like RevMan 5)
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using random effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(random)
## Niemelä 2011 3.0126 [0.1233; 73.5805] 18.6
## Pan 2011 1.4516 [0.2469; 8.5350] 60.6
## Kim 2015 5.1316 [0.2484; 106.0116] 20.7
## Watanabe 2021 NA 0.0
##
## Number of studies: k = 3
## Number of observations: o = 1142 (o.e = 571, o.c = 571)
## Number of events: e = 8
##
## RR 95%-CI z p-value
## Random effects model 2.1612 [0.5441; 8.5847] 1.10 0.2735
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0 [0.0000; 14.3556]; tau = 0 [0.0000; 3.7889]
## I^2 = 0.0% [0.0%; 89.6%]; H = 1.00 [1.00; 3.10]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.55 2 0.7601
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## - Continuity correction of 0.5 in studies with zero cell frequencies (applied twice to sample sizes, like RevMan 5)
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using random effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(random)
## Kim 2021 6.6627 [0.3497; 126.9313] 6.7
## Niemelä 2011 0.3347 [0.0351; 3.1951] 11.0
## Pan 2011 2.9032 [0.3062; 27.5233] 11.1
## Jin 2019 2.0455 [0.1923; 21.7548] 10.1
## Kim 2015 0.7106 [0.3130; 1.6132] 55.4
## Watanabe 2021 2.9492 [0.1226; 70.9192] 5.8
##
## Number of studies: k = 6
## Number of observations: o = 1391 (o.e = 697, o.c = 694)
## Number of events: e = 37
##
## RR 95%-CI z p-value
## Random effects model 1.0718 [0.4917; 2.3365] 0.17 0.8615
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0.1105 [0.0000; 5.6026]; tau = 0.3324 [0.0000; 2.3670]
## I^2 = 0.0% [0.0%; 74.6%]; H = 1.00 [1.00; 1.99]
##
## Test of heterogeneity:
## Q d.f. p-value
## 4.81 5 0.4389
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## - Continuity correction of 0.5 in studies with zero cell frequencies (applied twice to sample sizes, like RevMan 5)
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using random effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(random)
## Kim 2021 0.9512 [0.0605; 14.9455] 30.1
## Niemelä 2011 1.0042 [0.0632; 15.9616] 29.8
## Pan 2011 0.4839 [0.0445; 5.2665] 40.1
## Kim 2015 NA 0.0
## Watanabe 2021 NA 0.0
##
## Number of studies: k = 3
## Number of observations: o = 1302 (o.e = 653, o.c = 649)
## Number of events: e = 7
##
## RR 95%-CI z p-value
## Random effects model 0.7374 [0.1627; 3.3415] -0.40 0.6927
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0 [0.0000; 4.8841]; tau = 0 [0.0000; 2.2100]
## I^2 = 0.0% [0.0%; 89.6%]; H = 1.00 [1.00; 3.10]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.20 2 0.9047
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using random effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RR 95%-CI %W(random)
## Kim 2021 0.9512 [0.1979; 4.5721] 26.7
## Kim 2015 2.2583 [0.8037; 6.3452] 61.6
## Watanabe 2021 1.9655 [0.1833; 21.0777] 11.7
##
## Number of studies: k = 3
## Number of observations: o = 581 (o.e = 291, o.c = 290)
## Number of events: e = 25
##
## RR 95%-CI z p-value
## Random effects model 1.7641 [0.7840; 3.9696] 1.37 0.1701
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0 [0.0000; 7.9057]; tau = 0 [0.0000; 2.8117]
## I^2 = 0.0% [0.0%; 89.6%]; H = 1.00 [1.00; 3.10]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.82 2 0.6630
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Q-Profile method for confidence interval of tau^2 and tau
## - Calculation of I^2 based on Q
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using random effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## Number of observations: o = 115 (o.e = 58, o.c = 57)
## Number of events: e = 34
##
## RR 95%-CI z p-value
## Murasato 2022 0.6879 [0.3864; 1.2249] -1.27 0.2038
## [1] "entered RR path"
## [1] "Basing aboslute risk diff (RD) Using common/fixed effects model"## Warning: Argument 'Q.Cochrane' only considered for Mantel-Haenszel method in
## combination with DerSimonian-Laird estimator.## RD 95%-CI %W(random)
## Kim 2021 -0.0006 [-0.0351; 0.0338] 5.1
## Niemelä 2011 0.0000 [-0.0116; 0.0116] 44.6
## Pan 2011 -0.0086 [-0.0364; 0.0192] 7.8
## Kim 2015 0.0000 [-0.0127; 0.0127] 37.2
## Watanabe 2021 0.0000 [-0.0334; 0.0334] 5.4
##
## Number of studies: k = 5
## Number of observations: o = 1302 (o.e = 653, o.c = 649)
## Number of events: e = 7
##
## RD 95%-CI z p-value
## Random effects model -0.0007 [-0.0084; 0.0071] -0.18 0.8610
##
## Quantifying heterogeneity (with 95%-CIs):
## tau^2 = 0; tau = 0; I^2 = 0.0% [0.0%; 79.2%]; H = 1.00 [1.00; 2.19]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.34 4 0.9872
##
## Details of meta-analysis methods:
## - Inverse variance method
## - Restricted maximum-likelihood estimator for tau^2
## - Calculation of I^2 based on Q
## - Continuity correction of 0.5 in studies with zero cell frequencies (applied twice to sample sizes, like RevMan 5)
## [1] "entered RD path"
## [1] "Basing absolute risk diff (RD) Using random effects model"
## [1] "RD per 1000: -1 (95% CI -8 to 7)"Inconsistency
- Only 2 studies
- #374 Kim 2021 (SMART-STRATEGY II) - Non-true LM bifurcation lesions
- #838 Jin 2019 - True non-LM bifurcation
lesions
- GIST: Kim 2021 (#374) enrolled exclusively left main (LM), non-true bifurcation lesions (SB stenosis diameter <50%). As the authors note: “LM bifurcations have larger caliber vessels and a wider bifurcation angle than non-LM bifurcations. These characteristics could be protective factors for plaque and carina shift, a known mechanism of SB ostial lumen loss after MV stenting.” Because the side branch in this population has no significant baseline stenosis, a wide ostium, and a large-caliber vessel, it is inherently unlikely to occlude after main vessel stenting regardless of technique - “a significant SB compromise occurs in only 7% of cases after MV stenting.” KBI works by reopening the side branch ostium that stent struts cover and pushing the carina back toward its natural position, restoring flow to the side branch. Where occlusion is not a meaningful risk to begin with, KBI has little protective effect - consistent with Kim 2021’s finding that “the rate of procedure-related myocardial infarction were not significantly different between the 2 strategies” (11.5% no-KBI vs 12.2% KBI, p = 0.898).”
- GIST: Jin 2019 (#838) enrolled exclusively non-LM, true bifurcation
lesions (SB stenosis ≥50%, hemadostenosis ≥70%; LM explicitly excluded).
Here the side branch was already significantly narrowed at baseline, the
bifurcation angle was acute (<90°), and the vessels were smaller -
precisely the conditions where plaque and carina shift after main vessel
stenting are most likely to tip the side branch into complete occlusion.
This is confirmed by the observed SB occlusion rate of 15.6% in the
no-KBI arm. Those occlusions directly caused periprocedural MIs:
“perioperative MACEs in the JWT group included 5 cases of myocardial
infarction and 1 case of cardiac death.” In the KBI arm, zero side
branch occlusions occurred, and zero periprocedural MIs were observed
(RR ≈ 0.09, continuity-corrected due to zero events in KBI arm).
- The opposite direction point estimates reflects a population difference: KBI protects against periprocedural MI specifically through preventing side branch occlusion, and that mechanism works (only) where occlusion is a genuine risk. Kim 2021’s LM non-true bifurcation population is at the low end of that risk spectrum; Jin 2019’s non-LM true bifurcation population is at the high end. The pooled estimate (RR 0.70, 95% CI 0.33–1.50) averages across these two different contexts and should be interpreted with that caveat. Inconsistency is rated not serious given overlapping CIs and a biological explanation for the heterogeneity.
- TODO: Consider whether the overall estimate is meaningful for populations where true bifurcation and non-LM anatomy predominate.
| Feature | Kim 2021 (#374) - LM, non-true bifurcation | Jin 2019 (#838) - Non-LM, true bifurcation |
|---|---|---|
| Study / Trial name | Kim 2021 (SMART-STRATEGY II); #374 | Jin 2019 (BSKT vs JWT); #838 |
| Design | Prospective, randomised, open-label, multicentre RCT; 15 centres, Korea | Prospective, double-blinded, randomised controlled trial; single centre, China |
| Vessel type |
Left main (LM) bifurcation only “LM bifurcation lesion on coronary angiography with a reference diameter of the main branch (left anterior descending artery) and SB (left circumflex artery) ≥2.5 mm by visual estimation” |
Non-left main bifurcation only “non-left main coronary artery bifurcation lesions” (inclusion criterion) |
| Left main excluded? | No - LM is the entire study population |
Yes - explicitly excluded “The exclusion criteria were: …left main bifurcation lesions” |
| Bifurcation type (Medina) |
Non-true bifurcations (SB stenosis <50%) Medina: 1,0,0 (16.7%); 0,1,0 (29.5%); 1,1,0 (53.8%) - no third digit (no SB disease at baseline) “Patients were stratified according to whether they had a non-true bifurcation lesion (SB stenosis diameter <50%) or true bifurcation lesion (SB stenosis diameter ≥50%)” |
True bifurcations only (SB stenosis ≥50%) Medina: 1,1,1 (64.4%); 1,0,1 (13.3%); 0,1,1 (22.2%) - all have third digit = 1 (SB disease present) “true bifurcation lesions (defined as a vessel with ≥50% stricture); Medina classification 1.1.1, 1.0.1, or 0.1.1” |
| Side branch disease at baseline | None - SB stenosis <50% by definition (non-true bifurcation subgroup used for PICO 7) |
Significant - SB stenosis ≥50% by inclusion criterion; main
branch hemadostenosis ≥70% “true bifurcation lesions (hemadostenosis ≥70%)” Note: ‘hemadostenosis’ is a synonym for vascular stenosis, used in translated Chinese literature. It refers to the degree of narrowing of the target blood vessel (≥70% = severely obstructed main branch). |
| Bifurcation angle |
Wide - LM bifurcations characteristically have a wider angle “LM bifurcations have larger caliber vessels and a wider bifurcation angle than non-LM bifurcations. These characteristics could be protective factors for plaque and carina shift” |
Acute - inclusion criterion specifies bifurcation angle <90° “bifurcation angle <90°” (inclusion criterion) |
| Vessel calibre |
Large - LCX (SB) is a major epicardial artery “LM bifurcations have larger caliber vessels” |
Smaller - non-LM vessels (LAD/diagonal or LAD/LCX); SB reference diameter ~1.93–1.98 mm (Table 3) |
| KBI group (intervention) |
One-stent technique + mandatory final kissing balloon (FKB) “The aggressive strategy included MV stenting followed by mandatory FKB” |
BSKT (balloon-stent kissing technique): side branch balloon inflated
simultaneously with MV stent deployment - a form of KBI during
stenting “A standard coronary stent was advanced into the main-branch… meanwhile, a monorail balloon… was advanced into the side-branch vessel… the main-branch stent was then deployed to nominal pressure so that the side-branch balloon and main-branch stent ‘kissed’” |
| No-KBI group (comparator) |
One-stent technique alone, no FKB “The conservative strategy included MV stenting alone without FKB” |
JWT (jailed wire technique): wire placed in SB for protection but no
kissing balloon; stent deployed without simultaneous SB balloon
inflation “The main-branch stent was located and then released using a nominal pressure… If post-processing was not required then the guidewire was retracted followed by re-expansion of the main-branch stent balloon under high pressure” |
| SB occlusion rate after MV stenting (no-KBI arm) |
Low: ~9.3% had SB stenosis >50% post-MV stenting “A significant SB compromise occurs in only 7% of cases after MV stenting and in 17% when the SB has >50% stenosis… the observed rate of SB stenosis >50% after MV stenting without FKB was just 9.3%” |
High: 15.6% complete or partial SB occlusion “After main-branch stenting, occlusion of the side-branch occurred in 7 cases (15.6%) in the JWT group, of which 3 (6.7%) were completely occluded (TIMI grade 0) and 4 (8.9%) exhibited partial flow (TIMI grade 1 or 2). By contrast, there were no cases of side-branch occlusion in the BSKT group (P<.05 vs JWT group)” |
| Periprocedural MI: KBI arm | 10/82 (12.2%) | 0/44 (0%) |
| Periprocedural MI: No-KBI arm |
9/78 (11.5%) p = 0.898 (Table 3) “The rate of procedure-related myocardial infarction were not significantly different between the 2 strategies” |
5/45 (11.1%) - all driven by SB occlusion events “Perioperative MACEs in the JWT group included 5 cases of myocardial infarction and 1 case of cardiac death” |
| Direction of KBI effect on periprocedural MI | Null - KBI provided no detectable protection | Strongly protective - KBI (BSKT) prevented all SB occlusions and all periprocedural MIs |
| Approximate RR (periprocedural MI) | RR ≈ 1.06 (10/82 vs 9/78) | RR ≈ 0.09 (0/44 vs 5/45; continuity-corrected) |
| Key mechanistic quote | “LM bifurcations have larger caliber vessels and a wider bifurcation angle than non-LM bifurcations. These characteristics could be protective factors for plaque and carina shift, a known mechanism of SB ostial lumen loss after MV stenting… In LM bifurcations, a significant SB compromise occurs in only 7% of cases after MV stenting” | “The rate of perioperative MACEs was significantly higher (P<.05) in the JWT group (6 cases, 13.3%) than in the BSKT group (no cases). Perioperative MACEs in the JWT group included 5 cases of myocardial infarction and 1 case of cardiac death” |
| Abbreviations and notes: | ||
| LM = left main; SB = side branch; MV = main vessel; FKB = final kissing balloon; KBI = kissing balloon inflation; BSKT = balloon-stent kissing technique; JWT = jailed wire technique; TIMI = Thrombolysis In Myocardial Infarction flow grade. Hemadostenosis: synonym for vascular stenosis, from Greek haema (blood) + stenosis (narrowing); used in translated Chinese literature to denote the degree of vessel narrowing. Highlighted rows (yellow) are those most directly relevant to the mechanistic explanation for divergent periprocedural MI estimates. I² is unreliable with only 2 studies; inconsistency judgment is based primarily on visual CI overlap and clinical context. |
Risk of bias: Not Serious
D1: Randomization: #374 Kim 2021, #448 Pan 2011: Some concerns
- Allocation concealment NR
D2: Deviation from intended intervention: #448 Pan 2011: Some concerns
- not IIT as excluded people who got a 2nd stent from the analysis after randomization was done
- “Eleven patients (five from the KB group and six from the non-KB group) had a suboptimal immediate result at the SB and received a second stent at this vessel (Figure 1). These patients were excluded from the present analysis.”
- 11/244 (~4.5%) participants were excluded post-randomization, small and balanced exclusions make substantial impact unlikely.
D4: Measurement of the outcome: #374 Kim 2021: Some concerns
- NR if different people assessed the outcomes in each group. Likely not
- “Data regarding the primary and secondary end points were obtained through office visits or by telephone contact
D4 Outcome measurement: #58 Murasato 2022: High
- 4.5 PY on is it likely that
assessment of the outcome was influenced by knowledge of intervention
received?
Could only be an issue for TVR (not reported by this study) and ‘Unplanned additional procedures for high-risk stent failure’ (procedure-related adverse event)
No - some concerns
PY - High Risk
- “2.5 | Other technical considerations. To avoid fatal adverse events, additional treatment procedures, including re‐POT or additional KBT, were allowed only after SBD or KBT in cases of high‐risk stent failure with a stent malapposition of >400 µm from the vessel wall in any part of the stent, or serious lumen narrowing (>30% or <2.5 mm²) of the stent area in the distal MV ostium”
- The criteria for re-intervening was pre-defined and OCT-based (>400 µm malapposition or >30% lumen narrowing) based (objective)
- “The choice of additional treatment, re‐POT or additional KBT, was at the operators’ discretion”
- The type of additional procedure was discretionary, but the threshold for performing it was OCT-defined
- “since this randomized study followed an open-label design, the operators and patients were aware of the technique used, which may have introduced some bias or influence on the procedural results.”
- From Watanabe 2021 (same PROPOT trial, main paper): “Additional treatment was performed for 40.4% and 6.9% of patients in the POT and KBT groups, respectively (p<0.01).”
- Difference between the two reports (6.9% vs 25%) arises because Murasato counted re-KBT for proximal stent failure in the KBT group as an additional procedure, which the main paper did not: “In this study, recurrent KBT for proximal stent failure in the KBT group was counted as an additional procedure, which was not in the main study.”
D5 Selection of Reported Results : #679 Niemelä 2011, #448 Pan 2011, #838 Jin 2019, #93 Watanabe 2021 Some concerns
- Pre-specified analysis plan posted after study published and/or missing some reported outcomes
- Reported outcomes for all studies were standard PCI outcomes, so RoB in this domain did not greatly affect certainty of evidence
Evidence Profile: PICO 7
| Outcome | № studies / Design | Risk of bias | Inconsistency | Indirectness | Imprecision | Other | KBI | No KBI | Relative (95% CI) | Absolute (95% CI) | Certainty | Importance |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Periprocedural myocardial infarction (MI) (studies²: 1, 2) |
2 RCTs |
not serious | seriousa | not serious | very seriousb | none |
10/126 (7.9%) |
14/123 (11.4%) |
RR 0.70 (0.33 to 1.50) |
34 fewer per 1,000 (76 fewer to 57 more) |
⨁◯◯◯ Very lowa,b |
CRITICAL |
|
Target vessel revascularization (TVR) follow-up: mean 1 year (studies³: 2, 3, 4) |
3 RCTs |
not serious | not serious | not serious | seriousc | none |
16/291 (5.5%) |
9/290 (3.1%) |
RR 1.78 (0.80 to 3.95) |
24 more per 1,000 (6 fewer to 92 more) |
⨁⨁⨁◯ Moderatec |
CRITICAL |
|
Target lesion revascularization (TLR) follow-up: range 6 months to 1 year (studies⁶: 1, 2, 3, 4, 5, 6) |
6 RCTs |
not serious | not serious | not serious | seriousc | none |
21/697 (3.0%) |
14/694 (2.0%) |
RR 1.47 (0.77 to 2.96) |
9 more per 1,000 (5 fewer to 40 more) |
⨁⨁⨁◯ Moderatec |
CRITICAL |
|
All-cause mortality follow-up: range 6 months to 1 year (studies⁴: 3, 4, 5, 6) |
4 RCTs |
not serious | not serious | not serious | very seriousd | none |
6/571 (1.1%) |
2/571 (0.4%) |
RR 2.31 (0.60 to 8.82) |
5 more per 1,000 (1 fewer to 27 more) |
⨁⨁◯◯ Lowd |
CRITICAL |
|
Myocardial infarction (MI) follow-up: range 1 month to 1 year (studies⁶: 1, 2, 3, 4, 5, 6) |
6 RCTs |
not serious | not serious | not serious | seriousc | none |
19/697 (2.7%) |
18/694 (2.6%) |
RR 1.06 (0.57 to 1.96) |
2 more per 1,000 (11 fewer to 25 more) |
⨁⨁⨁◯ Moderatec |
CRITICAL |
|
Stent thrombosis (ST) follow-up: range 6 months to 1 year (studies⁵: 2, 3, 4, 5, 6) |
5 RCTs |
not serious | not serious | not serious | very seriouse | none |
3/653 (0.5%) |
4/649 (0.6%) |
RR 0.73 (0.16 to 3.24) |
2 fewer per 1,000 (11 fewer to 7 more)f |
⨁⨁◯◯ Lowe |
CRITICAL |
|
Procedure-related adverse events (studies¹: 7g) |
1 RCTs |
serioush | not serious | not serious | very seriousb,i | none |
14/57 (24.6%) |
20/52 (38.5%) |
RR 0.64 (0.36 to 1.13) |
138 fewer per 1,000 (246 fewer to 50 more) |
⨁◯◯◯ Very lowb,h,i |
CRITICAL |
| Abbreviations and study key: | ||||||||||||
|
CI: confidence interval; RR: risk ratio; KBI: kissing
balloon inflation; LM: left main; SB: side branch; MV: main vessel; OIS:
optimal information size; MID: minimally important difference; RCT:
randomised controlled trial; OCT: optical coherence tomography; SBD:
side branch dilation; POT: proximal optimisation technique; ITT:
intention-to-treat. Study key: 1 = Jin 2019 (#838); 2 = Kim 2021 (#374); 3 = Watanabe 2021 (#93); 4 = Kim 2015 (#531); 5 = Niemelä 2011 (#679); 6 = Pan 2011 (#448); 7 = Murasato 2022 (#58). |
||||||||||||
| Explanations: | ||||||||||||
| a I² = 61%, CIs overlap (Jin 2019 CI is wide enough to include Kim 2021 CI), but only 2 studies. Kim 2021 includes ONLY non-true LM bifurcations; Jin 2019 includes ONLY true non-LM bifurcations. The opposite-direction point estimates reflect a population difference: KBI protects against periprocedural MI specifically through preventing side branch occlusion, and that mechanism operates only where occlusion is a genuine risk. Kim 2021’s LM non-true bifurcation population sits at the low end of that risk spectrum (“LM bifurcations have larger caliber vessels and a wider bifurcation angle than non-LM bifurcations. These characteristics could be protective factors for plaque and carina shift, a known mechanism of SB ostial lumen loss after MV stenting” — Kim 2021); Jin 2019’s non-LM true bifurcation population sits at the high end (“After main-branch stenting, occlusion of the side-branch occurred in 7 cases [15.6%] in the JWT group… By contrast, there were no cases of side-branch occlusion in the BSKT group” — Jin 2019). The pooled estimate (RR 0.70, 95% CI 0.33–1.50) averages across these two mechanistically distinct contexts and should be interpreted with that caveat. Inconsistency is rated not serious given overlapping CIs and a biological explanation for the heterogeneity. | ||||||||||||
| b 95% CI includes both clinically meaningful benefit and harm (MID = −20 and +20 per 1,000). No optimal information size concerns. Crosses 2 thresholds → very serious. | ||||||||||||
| c 95% CI includes only clinically meaningful harm (upper bound exceeds MID = +20 per 1,000). No optimal information size concerns. Crosses 1 threshold → serious. | ||||||||||||
| d 95% CI includes only clinically meaningful harm (upper bound exceeds MID = +10 per 1,000 for catastrophic outcomes). Optimal information size not met (only 8 total events; threshold = 10 for catastrophic outcomes). Crosses 1 threshold + OIS not met → very serious. | ||||||||||||
| e 95% CI includes only clinically meaningful benefit (lower bound exceeds MID = −10 per 1,000 for catastrophic outcomes). Optimal information size not met (only 7 total events; threshold = 10 for catastrophic outcomes). Crosses 1 threshold + OIS not met → very serious. | ||||||||||||
| f Auto-calculation based on baseline risk presented a skewed 95% CI; replaced with manual calculation of risk difference (RD) in R. | ||||||||||||
| g Watanabe 2021 stated that analyses were performed on an intention-to-treat (ITT) basis. The randomised analysis population was 58 (KBT) and 57 (POT). The published clinical outcomes table reports a modified ITT population of 52 and 51, reflecting ~10% loss to 1-year follow-up. The randomised denominators were used in this meta-analysis to comply with the stated ITT principle. Given the low event rates across all clinical outcomes in this trial, the impact of this choice on the pooled estimates is minimal. | ||||||||||||
| h Domain 4 — outcome measurement. Open-label design; no blinding of operators. The outcome (unplanned additional procedure for correction of high-risk stent failure) is a subjective operator decision in the context of known treatment assignment. Trigger criteria were OCT-defined (“additional treatment procedures… were allowed only after SBD or KBT in cases of high-risk stent failure with a stent malapposition of >400 µm from the vessel wall in any part of the stent, or serious lumen narrowing [>30% or <2.5 mm²] of the stent area in the distal MV ostium” — Murasato 2022), but the choice and decision to intervene remained at operator discretion in an open-label setting. The authors note: “since this randomized study followed an open-label design, the operators and patients were aware of the technique used, which may have introduced some bias or influence on the procedural results” (Murasato 2022). | ||||||||||||
| i Evidence based on a single study (Murasato 2022); estimate is inherently imprecise regardless of CI width. |
References
- Jin Z, et al. Medicine. 2019;98(20):e15633.
- Kim J, et al. Rev Esp Cardiol. 2021;74(8):691–699.
- Watanabe Y, et al. EuroIntervention. 2021;17:747–756.
- Kim YH, et al. JACC Cardiovasc Interv.
2015;8(4):550–560.
- Niemelä M, et al. Circulation. 2011;123:79–86.
- Pan M, et al. Am J Cardiol. 2011;107:1460–1465.
- Murasato Y, et al. Catheter Cardiovasc Interv.
2022;99:1047–1058.