Antithrombotic Therapy After ICH
Survivors of intracerebral hemorrhage (ICH) face a fundamental therapeutic dilemma: many have compelling indications for antithrombotic therapy — atrial fibrillation (AF), coronary artery disease, prior stents, venous thromboembolism — yet these drugs may increase the risk of recurrent ICH, a frequently devastating and often fatal event. For decades, clinicians made these decisions in an evidence vacuum. The past five years have produced a series of landmark RCTs — RESTART, SoSTART, APACHE-AF, and most recently PRESTIGE-AF (2025) — that are beginning to clarify the risk-benefit balance. Simultaneously, left atrial appendage occlusion (LAAO) has emerged as a non-pharmacologic alternative for AF-related stroke prevention in patients for whom anticoagulation is deemed too risky.
🔹 Bottom Line: Antithrombotics After ICH
- Antiplatelets after ICH: RESTART showed that restarting antiplatelet therapy does NOT increase recurrent ICH (aHR 0.51 at 2 years, aHR 0.87 at extended follow-up). May be reasonable in patients with clear indications (AHA Class 2b).
- Anticoagulation in AF + ICH: PRESTIGE-AF (2025) — the first phase 3 trial — showed DOACs dramatically reduce ischemic stroke (NNT 13/year) but substantially increase recurrent ICH (NNH 24/year). Net clinical benefit trended toward DOAC (aHR 0.69).
- Lobar ICH / CAA: Anticoagulation is particularly risky — ENRICH-AF halted edoxaban in lobar ICH due to unacceptable recurrence. Avoid OAC if possible; consider LAAO.
- LAAO: Observational data shows effective stroke prevention with reduced bleeding in post-ICH patients. No RCT data yet (STROKECLOSE ongoing). Best option for AF + lobar ICH or high recurrence risk.
- Timing: Consider resuming antithrombotics ≥4 weeks post-ICH (AHA Class 2b). Optimal timing remains uncertain — balance urgency of ischemic prevention against recurrence risk.
1. Antiplatelets After ICH
The question of whether to resume antiplatelet therapy after ICH was addressed by the RESTART trial and its extended follow-up — currently the only RCT evidence on this topic.
1.1 RESTART Trial (2019)
RESTART (REstart or STop Antithrombotics Randomised Trial) enrolled 537 patients with spontaneous ICH who had been on antiplatelet therapy before the event and randomized them to restart or avoid antiplatelet therapy at a median of 76 days post-ICH. At 2-year follow-up, restarting antiplatelets was associated with a non-significant trend toward reduced recurrent ICH (adjusted HR 0.51; 95% CI 0.25–1.03; p=0.06) — the opposite of what most clinicians expected. There was no increase in ischemic events in the avoidance group, likely because of the relatively low annual ischemic event rate in this population.
1.2 Extended Follow-Up (2021)
At a median of 3.4 years, the initial protective signal attenuated. Recurrent ICH occurred in 8.2% of the antiplatelet group vs. 9.3% of the avoidance group (adjusted HR 0.87; 95% CI 0.49–1.55; p=0.64). The key conclusion: antiplatelet therapy does not increase ICH recurrence risk, and can be safely resumed in patients with clear vascular indications.
1.3 Observational Evidence
A systematic review and meta-analysis of observational studies in patients with any intracranial hemorrhage confirmed that antiplatelet resumption was associated with lower ischemic major adverse cardiovascular events (RR 0.61; 95% CI 0.48–0.79) with no increase in hemorrhagic events (RR 0.84; 95% CI 0.47–1.51).
🔹 Clinical Relevance: Antiplatelets After ICH
- AHA 2022: Starting antiplatelet therapy after ICH may be reasonable for patients with a clear indication (Class 2b, LOE B-R).
- This is distinct from the PATCH trial: Platelet transfusion in the acute setting is harmful (Class 3: Harm). Chronic oral antiplatelet use and acute platelet transfusion are completely different situations.
- Practical approach: Resume antiplatelets if clear cardiovascular indication (prior ACS, recent stent, established atherosclerotic disease), typically ≥2–4 weeks post-ICH when hematoma is stable. No specific antiplatelet agent has been shown to be superior; aspirin remains the default.
- Subgroups: RESTART did not identify harmful effects from antiplatelet therapy in patients with lobar ICH or cerebral microbleeds, though subgroups were small.
2. Resuming Anticoagulation After ICH in AF Patients
Approximately 15–20% of ICH patients have atrial fibrillation. Without anticoagulation, AF confers an annual ischemic stroke risk of 4–10% (depending on CHA₂DS₂-VASc score), while ICH recurrence risk is approximately 2–5% per year for deep ICH and 7–15% per year for lobar ICH. The challenge: OAC reduces ischemic stroke by ~65% but may increase the recurrence of the very event that made clinicians reluctant to prescribe it.
Four RCTs have now addressed this question with progressively larger sample sizes. Their results tell a consistent story: DOACs reduce ischemic events but increase hemorrhagic events, with the net balance likely favoring anticoagulation in most patients — except those with lobar ICH or CAA.
2.1 SoSTART (2021)
The Start or Stop Anticoagulants Randomised Trial enrolled 203 patients in the UK with spontaneous intracranial hemorrhage and AF, randomizing them to start (95% DOACs) or avoid OAC at a median of 115 days post-ICH. Over median 1.2-year follow-up, recurrent intracranial hemorrhage occurred in 8% of the OAC group vs. 4% of the no-OAC group — failing to demonstrate noninferiority (adjusted HR 2.42; 95% CI 0.72–8.09). Critically, 7 of 8 recurrent ICH events in the OAC arm were fatal, compared with 0 of 4 in the no-OAC arm. Major vascular events were numerically lower with OAC, but the trial was underpowered for efficacy.
2.2 APACHE-AF (2021)
The Apixaban After Anticoagulation-Associated Intracerebral Haemorrhage in Patients With Atrial Fibrillation trial randomized 101 patients (median 45 days post-ICH) to apixaban 5 mg twice daily or control (antiplatelet therapy or no antithrombotic). After median 1.9-year follow-up, the primary endpoint of stroke or vascular death was similar between groups (12.6% apixaban vs. 11.9% control; HR 1.05; 95% CI 0.48–2.31). ICH occurred in 8% of the apixaban group vs. 2% of control, while all major vascular events favored apixaban (12% vs. 22%). Like SoSTART, the trial was underpowered for definitive conclusions.
2.3 PRESTIGE-AF
PRESTIGE-AF, published in the Lancet in 2025, was the first completed phase 3 trial addressing this question. This PROBE design trial at 75 hospitals in six European countries randomized 319 ICH survivors with AF (mRS ≤4) to a DOAC or no anticoagulation. Patients were enrolled 14 days to 1 year after the index ICH.
| Outcome | DOAC Group | No Anticoagulation | Effect |
|---|---|---|---|
| Ischemic stroke (per 100 patient-years) | 0.83 | 8.60 | HR 0.05 (p<0.0001). NNT = 13/year |
| Recurrent ICH (per 100 patient-years) | 5.00 | 0.82 | HR 10.89. NNH = 24/year |
| All stroke & systemic embolism | 5.35 | 9.51 | Trend favoring DOAC |
| Net clinical benefit (composite) | — | — | aHR 0.69 (95% CI 0.33–1.40) |
| All-cause mortality | 10% | 13% | Similar (NS) |
PRESTIGE-AF confirmed what smaller trials suggested: DOACs are highly effective at preventing ischemic stroke in post-ICH AF patients (95% relative risk reduction, NNT 13/year), but this benefit is substantially offset by increased recurrent ICH (NNH 24/year). The net clinical benefit — incorporating all strokes, systemic embolism, MI, cardiovascular death, and major bleeding — trended toward DOAC (aHR 0.69) but was not statistically significant due to the competing effects.
2.4 Pooled Evidence
An individual patient data meta-analysis (2024, Lancet Neurol) pooling 412 patients from SoSTART, APACHE-AF, NASPAF-ICH, and ELDERCARE-AF found that OAC reduced ischemic MACE (HR 0.27; 95% CI 0.13–0.56) without a statistically significant increase in hemorrhagic MACE (HR 1.80; 95% CI 0.77–4.21). The composite of stroke or cardiovascular death favored OAC (HR 0.68) but was not significant. A 2025 meta-analysis pooling all 3 post-ICH RCTs (623 patients) confirmed the consistent pattern: ischemic benefit, hemorrhagic risk, uncertain net balance.
🔴 Lobar ICH & Cerebral Amyloid Angiopathy: Extreme Caution With OAC
- Lobar ICH carries 2–3× higher recurrence risk than deep ICH, especially with CAA markers (cortical microbleeds, superficial siderosis).
- The ENRICH-AF trial (edoxaban vs. no OAC in ICH+AF) halted enrollment of lobar ICH patients after a safety analysis revealed unacceptably high recurrent hemorrhagic stroke in the edoxaban arm. This is the strongest signal yet that OAC is particularly dangerous in lobar ICH.
- PRESTIGE-AF was stratified by location but was too small for definitive subgroup analysis. The increased recurrent ICH risk appeared present in both lobar and non-lobar subgroups.
- For patients with lobar ICH/probable CAA and AF, avoid OAC if possible and consider LAAO.
3. Left Atrial Appendage Occlusion (LAAO)
Because most AF-related thromboemboli originate from the left atrial appendage (LAA), mechanical closure of this structure offers stroke prevention without the hemorrhagic risk of chronic anticoagulation — making it particularly attractive for ICH survivors. Two device platforms are commercially available: the Watchman/Watchman FLX/FLX Pro (Boston Scientific) and the Amplatzer Amulet (Abbott).
3.1 Evidence in Post-ICH Patients
No RCTs of LAAO specifically in ICH patients have been completed. However, accumulating observational evidence is encouraging:
- NCDR LAAO Registry (2025): Among 133,947 LAAO patients, 15,428 (11.5%) had a history of prior ICH. Patients with prior ICH had higher periprocedural neurologic complications, but overall event rates for ischemic stroke and ICH during follow-up were well below those expected without OAC.
- Observational meta-analysis (2025): 19 studies, 1,671 patients with prior ICH. LAAO was successfully implanted in 99.3%. Using CHA₂DS₂-VASc-predicted risk as comparator, LAAO reduced ischemic stroke by 72% (RR 0.28; 95% CI 0.21–0.39) and bleeding events by 39% (RR 0.61; 95% CI 0.44–0.84). During follow-up: ICH 1.9%, ischemic stroke/TIA 2.9%, all-cause mortality 3.3%.
- Nordic propensity-matched study: 147 pairs of AF+ICH patients treated with Amplatzer Amulet vs. medical therapy showed 81% relative risk reduction in the composite of ischemic stroke, major bleeding, and death. Ischemic stroke reduced by 65%, ICH reduced by 71%.
- LAA-CAA Cohort: LAAO was feasible and safe specifically in patients with cerebral amyloid angiopathy and AF, offering a non-pharmacologic option for this highest-risk population.
3.2 Post-Procedural Antithrombotic Considerations
A critical caveat: LAAO does not eliminate the need for all antithrombotic therapy. Standard post-procedural regimens include:
- Watchman FLX: DAPT (aspirin + clopidogrel) for 1–3 months → aspirin monotherapy for 6 months → can discontinue all antithrombotics (per FDA label). Some centers use shorter regimens.
- Amplatzer Amulet: Aspirin for ≥6 months, with or without clopidogrel for 45 days.
- For ICH patients: Modified protocols with single antiplatelet therapy from the start are increasingly used in clinical practice, though not formally validated. Short-term antiplatelet exposure is generally considered acceptable given RESTART evidence.
3.3 Guidelines & Ongoing Trials
The 2025 SCAI/HRS Guidelines suggest LAAO for patients with non-valvular AF and contraindications to long-term OAC (conditional recommendation). ICH history qualifies as such a contraindication. Two ongoing RCTs will provide definitive evidence:
- STROKECLOSE: Amplatzer Amulet vs. medical therapy in AF+ICH patients with CHA₂DS₂-VASc ≥2 (PROBE design, 2:1 randomization). Primary composite endpoint: stroke, systemic embolism, life-threatening/major bleeding, and all-cause mortality. This trial will be practice-changing.
- CLEARANCE: LAAO vs. medical therapy in ICH survivors with AF.
🔹 Clinical Relevance: Who Is the Best LAAO Candidate?
- Ideal candidate: AF + prior ICH (especially lobar/CAA-related) + high CHA₂DS₂-VASc (≥3) + high HAS-BLED where OAC carries unacceptable recurrence risk.
- Less ideal: Patients with non-lobar deep ICH and high CHA₂DS₂-VASc may still benefit more from DOAC, given the more favorable risk-benefit balance in this subgroup.
- Multidisciplinary decision: This is the quintessential neurologist-cardiologist joint decision. Shared decision-making with the patient is essential.
- Practical barrier: LAAO requires cardiology referral, procedural availability, and TEE/ICE guidance. Not all centers offer this, and wait times can be prolonged — during which the patient remains unprotected.
4. Timing of Antithrombotic Resumption
The optimal timing for resuming antithrombotic therapy after ICH remains one of the most uncertain areas. Current evidence provides only broad guidance:
| Source | Antiplatelets | Anticoagulants |
|---|---|---|
| AHA 2022 ICH Guidelines | May consider after acute phase (weeks) — Class 2b | ≥4 weeks for AF patients — Class 2b, LOE B-NR |
| ESC 2024 AF Guidelines | — | Consider restarting OAC ≥4 weeks post-ICH |
| 2025 THANZ DOAC Guidelines | — | 1–6 weeks, depending on individual risk (GRADE 2C) |
| Observational evidence | Median ~76 days (RESTART) | Optimal window: 10–30 weeks (Majeed 2010); median 45–115 days across trials |
| Clinical practice surveys | — | 36.6% restart >30 days, 24.2% at 15–30 days, 16.3% at 10–14 days |
The early period (<2 weeks) carries the highest re-bleeding risk, while delayed resumption (>3 months) exposes AF patients to prolonged ischemic risk. A pragmatic approach is to repeat CT at 2–4 weeks to confirm hematoma stability, then initiate therapy based on the individual risk-benefit assessment.
5. Clinical Decision Framework
| Clinical Scenario | Recurrence Risk | Recommended Strategy |
|---|---|---|
| Deep ICH + AF + high CHA₂DS₂-VASc (≥4) | Low-moderate (2–5%/yr) | DOAC resumption at 4–6 weeks preferred (PRESTIGE-AF supports net benefit). NNT 13 > NNH 24. |
| Deep ICH + AF + moderate CHA₂DS₂-VASc (2–3) | Low-moderate | DOAC reasonable; consider LAAO if additional bleeding risk factors. |
| Lobar ICH + AF (no CAA markers) | Moderate (5–8%/yr) | Shared decision: DOAC vs. LAAO. Carefully weigh CHA₂DS₂-VASc against recurrence risk. LAAO increasingly preferred. |
| Lobar ICH + AF + CAA markers (CMBs, cSS) | High (7–15%+/yr) | Avoid OAC. LAAO strongly preferred. If LAAO not available, antiplatelet monotherapy may be considered. |
| Any ICH + vascular indication (no AF) | Variable | Resume antiplatelet therapy at 2–4 weeks if clear indication (RESTART evidence). Single agent preferred. |
| Any ICH + mechanical heart valve | Variable | Anticoagulation generally necessary. Timing and intensity require case-by-case multidisciplinary discussion. |
🔴 Key Principles
- Never use VKAs when DOACs are an option — DOACs confer ~50% lower intracranial hemorrhage risk.
- Avoid concomitant antiplatelet + anticoagulant therapy whenever possible — secondary analyses of DOAC trials identified dual therapy as a major modifiable risk factor for ICH.
- Repeat neuroimaging (CT or MRI) to confirm hematoma stability before resuming antithrombotics.
- Document the shared decision-making process — this is a high-stakes, preference-sensitive decision.
6. Ongoing & Future Trials
| Trial | Comparison | Population | Key Feature |
|---|---|---|---|
| ENRICH-AF | Edoxaban vs. no OAC | ICH + AF (non-lobar only, after lobar arm halted) | Largest OAC-after-ICH trial; lobar ICH halted for safety |
| ASPIRE | DOAC vs. no antithrombotic | ICH + AF | Enrolling; will contribute to IPD meta-analysis |
| A3ICH | Apixaban vs. no OAC | ICH + AF | Parallel trial to ENRICH-AF |
| STATICH | Start vs. avoid antithrombotic | ICH with any antithrombotic indication | Includes antiplatelet and anticoagulant arms; IPD meta-analysis planned |
| STROKECLOSE | LAAO (Amulet) vs. medical therapy | AF + prior ICH | First RCT of LAAO specifically in ICH population |
| CLEARANCE | LAAO vs. medical therapy | AF + ICH | Parallel LAAO trial |
7. Trial Comparison Table
| Trial | Year | N | Comparison | Median Time Post-ICH | Ischemic Events | Recurrent ICH | Key Finding |
|---|---|---|---|---|---|---|---|
| RESTART | 2019 | 537 | Antiplatelet vs. avoid | 76 days | Similar | 4% vs. 9% (aHR 0.51, p=0.06) | Antiplatelets safe; possible protective effect. Extended f/u: aHR 0.87 (NS) |
| SoSTART | 2021 | 203 | OAC (95% DOAC) vs. avoid | 115 days | Numerically ↓ with OAC | 8% vs. 4% (aHR 2.42, NS) | Failed noninferiority. 7/8 OAC recurrent ICH fatal. |
| APACHE-AF | 2021 | 101 | Apixaban vs. no OAC | 45 days | 12.6% vs. 11.9% (NS) | 8% vs. 2% | Neutral. Underpowered. All major vascular events favored apixaban (12% vs. 22%). |
| PRESTIGE-AF | 2025 | 319 | DOAC vs. no anticoagulation | 49 days | 0.83 vs. 8.60 /100 pt-yr (HR 0.05) | 5.00 vs. 0.82 /100 pt-yr (HR 10.89) | First phase 3 trial. NNT 13/yr ischemic stroke; NNH 24/yr ICH. Net clinical benefit trended DOAC (aHR 0.69). |
| IPD Meta-analysis | 2024 | 412 | OAC vs. no OAC (4 RCTs pooled) | Variable | 4% vs. 19% (HR 0.27) | 7% vs. 5% (HR 1.80, NS) | OAC ↓ ischemic MACE; hemorrhagic MACE not significantly ↑. Stroke/CV death: HR 0.68 (NS). |
References
- RESTART Collaboration. Effects of antiplatelet therapy after stroke due to intracerebral haemorrhage (RESTART): a randomised, open-label trial. Lancet. 2019;393(10191):2613–2623.
- Al-Shahi Salman R, et al. Effects of antiplatelet therapy after stroke caused by intracerebral hemorrhage: extended follow-up of RESTART. JAMA Neurol. 2021;78(10):1179–1186.
- SoSTART Collaboration. Effects of oral anticoagulation for atrial fibrillation after spontaneous intracranial haemorrhage in the UK (SoSTART). Lancet Neurol. 2021;20(10):842–853.
- Schreuder FHBM, et al. Apixaban versus no anticoagulation after anticoagulation-associated intracerebral haemorrhage in patients with atrial fibrillation in the Netherlands (APACHE-AF). Lancet Neurol. 2021;20(11):907–916.
- Veltkamp R, et al. Direct oral anticoagulants versus no anticoagulation for the prevention of stroke in survivors of intracerebral haemorrhage with atrial fibrillation (PRESTIGE-AF). Lancet. 2025;405(10482):927–936.
- Al-Shahi Salman R, et al. Oral anticoagulation for atrial fibrillation after cerebral haemorrhage: individual patient data meta-analysis of randomised controlled trials. Lancet Neurol. 2024;23(4):404–417.
- Mansour M, et al. Left atrial appendage occlusion in patients with prior intracranial hemorrhage. JACC Clin Electrophysiol. 2025;11(12):2729–2741.
- Greenberg SM, et al. 2022 Guideline for the Management of Patients With Spontaneous ICH. Stroke. 2022;53(7):e282–e361.
- 2025 SCAI/HRS Clinical Practice Guidelines on Transcatheter Left Atrial Appendage Occlusion. J Soc Cardiovasc Angiogr Intervent. 2025.