Two massive Swedish trials put paclitaxel-coated devices to the test in PAD - and the answer is the same for both
A vascular surgeon in Sweden sees two patients back-to-back. Both need endovascular revascularisation. The surgeon reaches for a paclitaxel-coated balloon. Should they?
Patient A: A 68-year-old man with severe intermittent claudication (Rutherford 3). He can't walk 100 metres without stopping. His ankle-brachial index is 0.55. He wants to get back to gardening.
Patient B: A 79-year-old woman with critical limb-threatening ischaemia (CLTI). She has rest pain and a non-healing ulcer on her right foot (Rutherford 5). Her ABI is 0.3. Without revascularisation, she may lose her leg.
Different patients. Different goals. Same question: does the drug-coating on the balloon matter?
SWEDEPAD - the Swedish Drug-Elution Trial in Peripheral Arterial Disease - was not one trial but a coordinated program of two parallel randomised controlled trials.
Rutherford 4-6
Primary endpoint:
Major amputation
Rutherford 1-3
Primary endpoint:
Quality of life (VascuQoL-6)
SWEDEPAD used a powerful but underappreciated design: the registry-based randomised controlled trial, embedding research into routine clinical care.
Swedvasc captures >95% of all vascular procedures in Sweden. By embedding randomisation directly into this registry, SWEDEPAD achieved:
Patients were randomised after successful guidewire crossing of the target lesion. This is a critical detail: it means only patients who could actually undergo balloon angioplasty were randomised.
Randomised
1,206 drug-coated vs 1,194 uncoated
ITT: 1,180 vs 1,175
Randomised
577 drug-coated vs 578 uncoated
ITT: 565 vs 571
Combined, SWEDEPAD enrolled approximately 3,500 patients across the two trials. No prior RCT of paclitaxel-coated devices in PAD comes close to this size. And unlike most prior trials, these were government-funded and independent from industry.
Understanding WHO was enrolled in each trial is essential for interpreting the results - especially the mortality data.
Both femoropopliteal + infrapopliteal: 23.8%. TASC distribution: ~17% A, 33% B, 30% C, 19% D.
Planned primary stenting 18.8% but actual 42.8% (many needed supplementary stenting). TASC: ~24% A, 40% B, 24% C, 12% D.
In 2,355 patients with CLTI, the question was simple: does a paclitaxel-coated device reduce major amputation?
HR 1.05 (95% CI 0.87-1.27), p=0.61
NO DIFFERENCE
amputations per
100 person-years
amputations per
100 person-years
Drug-coating reduced target vessel reintervention at 1 year: HR 0.81 (0.66-0.98), p=0.031. But by 5 years: HR 0.97 (0.83-1.14), p=0.71.
The cumulative reintervention curves converged by year 2. The uncoated group had more reinterventions early, but then the rate dropped as those lesions were treated. The drug-coated group's reintervention rate stayed constant.
Drug-coating delayed rather than prevented reinterventions. At 1 year there appeared to be a benefit, but it vanished as the uncoated group caught up. And crucially, this reintervention delay translated into zero improvement in amputation, survival, or quality of life.
In 1,136 claudicants, the primary question was about what patients care about most: can they walk better and live better?
Mean difference -0.02 (95% CI -0.66 to 0.62), p=0.96
NO DIFFERENCE
VascuQoL-6 improvement
72.9% improved at 1yr
VascuQoL-6 improvement
73.7% improved at 1yr
Both groups had massive QoL improvement - the procedure itself works. The coating adds nothing.
All-cause mortality at 5 years: HR 1.47 (95% CI 1.09-1.98), p=0.010
4.57 vs 3.28 deaths per 100 person-years
The mortality curves began to diverge at approximately 2 years, with the drug-coated group experiencing progressively more deaths. The separation persisted through the 5-year landmark.
Per-protocol analysis confirmed: HR 1.56 (1.15-2.13), p=0.0042
Over full follow-up (median 7.1 years): HR 1.18 (0.94-1.48), p=0.16
The mortality signal was strongest at 5 years but attenuated over the full follow-up period, suggesting a time-limited effect of paclitaxel exposure.
This is the central epidemiological question of the SWEDEPAD program. The same drug, the same device types, the same country - but a mortality signal in claudicants and not in CLTI. Why?
This is a classic epidemiological concept. Consider the background mortality rates:
deaths per 100 person-years
deaths per 100 person-years
If paclitaxel adds approximately 1 extra death per 100 person-years:
The CLTI cohort was TWICE as large (2,355 vs 1,136) and STILL couldn't detect it. The problem wasn't sample size - it was signal-to-noise ratio.
SWEDEPAD tells us drug-coating doesn't help. But the data also tell us something important about what DOES work for claudication patients.
A common worry with claudication is progression to limb loss. SWEDEPAD 2 puts hard numbers on this: the major amputation rate was only ~0.5-0.7 per 100 person-years over 5 years. That's about 1-2% at 5 years - and it stayed flat, with no late spike.
This is actually reassuring for the intervention strategy. These patients were revascularised and they kept their limbs. The late-amputation bogeyman didn't show up.
Both groups improved from a baseline VascuQoL-6 of ~10 to ~15 at 1 year. That's a 50% improvement in disease-specific quality of life. The procedure works - the drug coating just doesn't add anything on top of it.
This is an important distinction: SWEDEPAD does not argue against intervention for claudication. It argues against drug-coating the devices.
Both the ESVS 2024 and ESC 2024 guidelines recommend supervised exercise therapy FIRST for claudication, before considering endovascular intervention. But when patients fail or can't access supervised exercise, intervention is a reasonable next step.
Exercise therapy improves walking distance, QoL, and cardiovascular fitness. But in practice, access to structured supervised programs varies enormously - and many patients end up in the cath lab regardless.
The SWEDEPAD results land in the middle of a controversy that has been simmering for seven years. Here is the timeline.
Pooled data from RCTs suggested increased late mortality with paclitaxel-coated devices. Critically, 89% of included patients were claudicants, only 11% CLTI. This meta-analysis shook the vascular world.
The FDA issued safety communications. SWEDEPAD paused enrollment for 15 months while safety was assessed.
Published interim safety analysis showed no mortality difference at mean 2.49 years of follow-up. Enrollment resumed March 2020. But as we now know, the signal appears between years 2-5.
Industry-sponsored individual patient data meta-analysis found no increased mortality. But follow-up was limited.
Updated individual patient data meta-analysis: no increased mortality detected. But again, mix of CLTI and claudicants, limited long-term follow-up.
UK trial in CLTI: no differences with drug-coated devices. Consistent with SWEDEPAD 1's CLTI finding.
CLTI (SWEDEPAD 1): No mortality signal (HR 1.04, p=0.54).
Claudication (SWEDEPAD 2): Mortality signal at 5 years (HR 1.47, p=0.010).
The SWEDEPAD 2 mortality signal follows the same pattern as the 2018 Katsanos meta-analysis: divergence at approximately 2 years, persisting through 5 years. And remember, 89% of Katsanos patients were claudicants.
The 2020 SWEDEPAD interim analysis found nothing at 2.49 years - because the signal hadn't emerged yet. This is why long-term follow-up matters.
Most prior individual RCTs of drug-coated devices were:
SWEDEPAD 2 is the first adequately-powered individual trial with pre-specified mortality analysis and sufficient follow-up to detect this signal in a pure claudication population.
Every trial has trade-offs. Understanding them helps you weigh the evidence appropriately.
Apply what you've learned about both SWEDEPAD trials.
~3,500 patients across 2 parallel trials in 22 Swedish centres, with up to 10 years of follow-up.
HR 1.05 (0.87-1.27), p=0.61. Drug-coated devices did not save limbs.
Mean difference -0.02 (-0.66 to 0.62), p=0.96. Both groups improved from 10 to 15 on VascuQoL-6.
1-year TVR HR 0.81 (p=0.031) converged to HR 0.97 (p=0.71) at 5 years. Delayed, not prevented.
1-year TVR HR 0.86 (p=0.37), 5-year TVR HR 0.96 (p=0.71). Not even a temporary benefit.
HR 1.47 (1.09-1.98), p=0.010. Death rate 4.57 vs 3.28 per 100 person-years. Confirmed on per-protocol analysis (HR 1.56, p=0.0042).
HR 1.04 (0.92-1.17), p=0.54. Background mortality of ~15/100py makes a small excess undetectable.
~0.5-0.7 amputations per 100py with no late spike. QoL improved 50% (10→15) in both arms. The procedure helps; the coating doesn't.
Drug-coating didn't add benefit in either population. In claudication, intervention itself works well (QoL improved 50%), but the paclitaxel adds nothing except a concerning 5-year mortality signal. Consider carefully whether drug-coated devices are worth the trade-off when uncoated ones achieve the same outcomes.
Your hospital's vascular lab currently defaults to paclitaxel-coated balloons for all femoropopliteal interventions. Based on SWEDEPAD, what do you recommend?