The central point: The PRAETORIAN, published in the New England Journal of Medicine (2020), is the first randomized clinical trial to directly compare the subcutaneous ICD (S-ICD) with the transvenous ICD (TV-ICD). In 849 patients analyzed (of 876 randomized), followed for a median of 49.1 months, S-ICD demonstrated noninferiority for the composite outcome of device-related complications and inappropriate shocks (HR 0.99; 95% CI 0.71-1.39; P=0.01 for noninferiority). The main conclusion: subcutaneous ICD is a safe alternative to transvenous ICD for patients who do not require cardiac pacing.
Why this comparison matters
The implantable cardioverter-defibrillator (ICD) is the standard treatment for prevention of sudden cardiac death in high-risk patients. Since its introduction, the device has used leads positioned inside the heart through veins (the transvenous route). Although effective, this approach carries risks inherent to venous access and direct contact with the myocardium: systemic infection, cardiac perforation, venous thrombosis, lead fracture and long-term vascular complications.
The subcutaneous ICD (S-ICD, short for subcutaneous implantable cardioverter-defibrillator) emerged as an alternative. The device is implanted entirely in subcutaneous tissue, with the lead positioned along the sternum, without entering the venous system or heart. The trade-off: S-ICD does not offer antitachycardia pacing (ATP, short for antitachycardia pacing) or bradycardia pacing.
Before PRAETORIAN, the choice between the two technologies was based on observational studies and registries. Evidence from a randomized trial with adequate statistical power and prolonged follow-up was lacking.
Key PRAETORIAN data
- 876 randomized patients (849 included in analysis: 426 S-ICD, 423 TV-ICD) at 39 centers in Europe and the USA
- Composite primary outcome: device-related complications + inappropriate shocks
- Cumulative incidence at 48 months: 15.1% (S-ICD) vs. 15.7% (TV-ICD)
- HR 0.99 (95% CI 0.71-1.39; P=0.01 for noninferiority)
- Median follow-up: 49.1 months
Study design
PRAETORIAN (Prospective, RAndomizEd comparison of subcuTaneOus and tRansvenous ImplANtable cardioverter-defibrillator therapy) was an international, multicenter, randomized noninferiority clinical trial. Recruitment took place between March 2011 and January 2017, with follow-up completed in December 2019.
Inclusion and exclusion criteria
Patients aged 18 or older with a class I or IIa ICD indication for primary or secondary prevention (ACC/AHA/HRS or ESC guidelines) were included. Exclusion criteria were prior ICD, unsuitability for S-ICD by QRS-T waveform analysis, an indication for bradycardia pacing or cardiac resynchronization therapy, and known ventricular tachycardia (VT) below 170 bpm refractory to medication or ablation.
Randomization and intervention
Randomization was 1:1, stratified by center, with variable block sizes (2 to 8). All S-ICDs were Cameron Health/Boston Scientific systems (EMBLEM). For the TV-ICD group, the manufacturer was chosen by the attending physician. Single-chamber ICDs were standard, with a dual-chamber option only when needed for arrhythmia discrimination.
Outcomes
The composite primary outcome included device-related complications (infection requiring explantation, pocket hematoma requiring drainage or transfusion, thrombosis, pneumothorax or hemothorax requiring intervention, cardiac perforation with tamponade, and lead repositioning or replacement) and inappropriate shocks (defined as shocks delivered for any rhythm other than ventricular tachycardia or ventricular fibrillation). The noninferiority margin was defined as an upper 95% confidence interval limit for the hazard ratio (HR) less than or equal to 1.45.
Main results
Primary outcome: noninferiority confirmed
The primary outcome occurred in 68 patients in the S-ICD group and 68 in the TV-ICD group. Kaplan-Meier cumulative incidence at 48 months was 15.1% versus 15.7%, respectively. The hazard ratio was 0.99 (95% CI 0.71-1.39), with P=0.01 for noninferiority. The noninferiority criterion was met because the upper CI limit (1.39) was below the prespecified margin of 1.45. The superiority analysis did not show a significant difference (P=0.95).
Components of the primary outcome
Device-related complications occurred in 31 patients (5.9%) in the S-ICD group versus 44 (9.8%) in the TV-ICD group (HR 0.69; 95% CI 0.44-1.09). The trend favored S-ICD but was not statistically significant. The complication profile differed between groups: in S-ICD, the most frequent complication was pocket hematoma requiring drainage or transfusion; in TV-ICD, lead-related complications (displacement or need for repositioning) and infections requiring explantation predominated (8 cases in TV-ICD versus 4 in S-ICD). Inappropriate shocks occurred in 41 patients (9.7%) in the S-ICD group versus 29 (7.3%) in the TV-ICD group (HR 1.43; 95% CI 0.89-2.30). This difference was also not statistically significant.
A relevant finding: causes of inappropriate shocks differed between groups. In S-ICD, 58.5% of shocks were due to oversensing of cardiac signals (excessive detection of signals such as the T wave). In TV-ICD, 93.1% were due to supraventricular arrhythmias.
Secondary outcomes
Appropriate shocks were more frequent in the S-ICD group: cumulative incidence of 19.2% versus 11.5% with TV-ICD (HR 1.52; 95% CI 1.08-2.12). This statistically significant difference reflects the inability of S-ICD to deliver ATP. In the TV-ICD group, antitachycardia pacing was used in 12.9% of patients and successfully terminated 55% of VT episodes, avoiding the need for a shock.
All-cause mortality was 83 patients in the S-ICD group versus 68 in the TV-ICD group (HR 1.23; 95% CI 0.89-1.70), without statistical significance. Sudden cardiac death was identical: 18 patients in each group.
Limitations and points of caution
The study was not blinded. The event adjudication committee knew each patient’s group. Device technology evolved during the six years of recruitment. The SMART Pass filter, designed to reduce oversensing in S-ICD, was unavailable or not activated in 78% of patients who received the first inappropriate shock in that group.
Operators had less experience with S-ICD, a newer technology at the time. Median follow-up of 49 months, although substantial, may be insufficient to capture chronic transvenous lead complications (fracture, insulation failure), which typically appear after 5 to 10 years.
The noninferiority margin of 1.45 is debatable. It means accepting up to 45% more relative risk with S-ICD as clinically equivalent. In addition, 27 patients were excluded before implantation and 38 were lost to follow-up.
Implications for clinical practice
PRAETORIAN established that the subcutaneous ICD is noninferior to the transvenous ICD for the composite outcome of complications and inappropriate shocks. The data nevertheless reveal a trade-off in event profiles: fewer device-related complications with S-ICD (5.9% vs. 9.8%), but more inappropriate shocks (9.7% vs. 7.3%) and appropriate shocks (19.2% vs. 11.5%).
The absence of ATP is the main clinical limitation of S-ICD. For patients with monomorphic VT amenable to pacing, the transvenous ICD offers the advantage of terminating arrhythmia without a shock. For patients without documented VT or with channelopathies (Brugada syndrome, catecholaminergic polymorphic VT), S-ICD eliminates vascular and endocardial risks without compromising protection against sudden death.
My view as an electrophysiologist: PRAETORIAN was a milestone because it validated with randomized evidence what clinical practice had already suggested: S-ICD is a real alternative to transvenous ICD, not merely a niche option. In my experience, patient selection remains decisive. For those who do not need pacing and do not have known slow VT, S-ICD should be considered as a first option, especially in young patients with long life expectancy, where the accumulated risk of transvenous lead complications far exceeds what was shown over 4 years of follow-up. Technology has advanced since PRAETORIAN, and newer discrimination filters have substantially reduced inappropriate shocks.
