Risk of Recurrent ICH After OAC Resumption
In non-anticoagulated patients with a history of ICH, the 1-year risk of recurrent ICH ranges from 0 to 8.6% [
12]. In patients resuming OAC, this number ranges from 2.5 to 8% [
12]. For example, in one multicenter study (
n = 267), warfarin resumption was associated with an annual ICH recurrence rate of 2.56% [
13]. Whether OAC resumption per se truly increases the risk of recurrent ICH is still debatable, given the many associated comorbidities related to ICH recurrence.
The majority of studies demonstrate that OAC resumption did not increase the risk of recurrent ICH [
14,
15••,
16‐
19] (see Table
1). In one retrospective cohort study (
n = 160), recurrent ICH occurred with higher frequency after OAC resumption, but this was statistically nonsignificant compared with patients who did not resume OAC (7.6 vs. 3.7%,
p = 0.48) [
14]. Also, OAC resumption did not increase the risk of ICH in a Danish nationwide cohort study [
15••]. In another large observational cohort (
n = 2415), warfarin resumption after incident ICH was associated with similar risk of recurrent ICH to non-resumption of warfarin [
17••]. These results were confirmed by a recent systematic review and meta-analysis, including eight studies and 5306 ICH patients, in which reinstatement of OAC had a similar risk of recurrent ICH to non-OAC restarters [
18].
Table 1
Studies evaluating risk of hemorrhage and thromboembolism after ICH
| Population-based cohort | 6369 | N/A | N/A | N/A | N/A | N/A | N/A | N/A | 0.90 (0.44–1.82) | 0.58 (0.35–0.97) | 0.59 (0.43–0.82) |
| Retrospective cohort | 160 | 14 | 7.6 | 3.7 | 18.5 | 3.7 | 12.3 | 31.1 | 0.47 (0.10–2.30) | 0.28 (0.06–1.27) | 0.76 (0.30–1.89) |
| Nationwide cohort | 1752 | 34 | 8.0 | 5.3 | 9.7 | 8.6 | 10.4 | 19.1 | 0.91 (0.56–1.49) | 0.59 (0.33–1.03) | 0.55 (0.37–0.82) |
| Nationwide cohort | 719 | 31 | 3.9 | 5.2 | 8.2 | 3.9 | 15.0 | 37.5 | N/A | N/A | 0.26 (0.13–0.53) |
| Nationwide cohort | 2415 | 31 | 5.8 | 3.3 | 19.6 | 5.3 | 8. 9 | 35.5 | 1.31 (0.68–2.50) | 0.43 (0.21–0.86) | 0.51 (0.37–0.71) |
| Meta-analysis | 5306 | N/A | N/A | N/A | N/A | 7.8 | N/A | N/A | 1.01 (0.58–1.77) | 0.34 (0.25–0.45) | N/A |
| Nationwide cohort | 244 | N/A | N/A | 2.0 | 3.0 | N/A | 6.0 | 8.0 | N/A | 0.19 (0.06–0.60) | 0.17 (0.06–0.45) |
| Retrospective | 528 | 117 | 1.4 | 2.4 | 1.4 | 0 | 8.3 | 4.8 | N/A | 0.19 (0.08–0.47) | N/A |
Nevertheless, some contradictory results exist. For example, a retrospective study (
n = 428) showed that OAC restarting increased the risk of major bleeding (5.5 vs. 3.1 per 100 patient-years,
p = 0.024), and recurrent ICH was observed only in patients with OAC use [
20]. Different study designs and selection biases may explain the contradictory results. In many studies, only the patients with “less severe” ICH, that is ICH with smaller volume of hemorrhage and mild functional changes, could have received OAC resumption, hence leading to a lower recurrent ICH risk [
21,
22].
For clinical practice, different patient profiles may lead to a varying risk of ICH recurrence, and therefore, individualized evaluation is critical [
23]. We suggest that risk factors for recurrent ICH should be considered before deciding OAC resumption [
24,
25]. Currently, there are some well-identified risk factors for recurrent ICH (Table
2). For example, the location of ICH is a significant risk factor. Lobar hemorrhage has a higher ICH recurrent rate compared with hemorrhage in a deep cortical location (22 vs. 4% for cumulative 2-year rate,
p = 0.007) [
26].
Table 2
Clinical risk factors of recurrent ICH and thromboembolism
Risk factors for recurrent ICH | Large area ICH, ICH history, lobar ICH location, cerebral microbleeds, amyloid angiopathy, arteriovenous malformation, cerebral aneurysm, lacunar infarcts, leukoaraiosis, Asian population | Alcohol, tobacco, anemia, hepatic disease, high risk of fall |
Risk factors for both ICH and thromboembolism | Elderly, coagulopathy, previous IS, malignancy | Hypertension, diabetes, kidney dysfunction, labile INR |
Risk factors of thromboembolism | AF, HF, vascular disease, mechanical heart valve, VTE history, female sex, recent surgery | Decreased ambulation |
In postoperative (neurosurgery for ICH) patients with spontaneous ICH, diabetes mellitus (odds ratio [OR], 2.72; 95% CI, 1.01–7.35) has been related to recurrent ICH [
27]. In other studies, patients with hepatic C virus infection had increased risk of ICH (HR, 1.60; 95% CI, 1.24–2.06) [
28], as has severe hypertension (systolic blood pressure ≥ 160 mmHg or diastolic blood pressure ≥ 110 mmHg), which was associated with a sixfold increased risk of ICH [
29]. In addition, patients with recent intracranial microbleeds had substantial risk of incident ICH [
30]. Leukoaraiosis was also related with high risk of significant ICH (relative risk [RR], 1.65; 95% CI, 1.26–2.16) [
31]. For these two risk factors, brain imaging evidence provided by computed tomography (CT) or magnetic resonance imaging (MRI) is required.
Ethnicity is also a major risk factor for ICH. Asian populations had higher risk of ICH in the major trials of the non-VKA oral anticoagulants (NOACs) [
32‐
35]. For example, in the RE-LY (Randomized Evaluation of Long-term Anticoagulant Therapy) trial, Asian populations had higher risk of ICH in both of the warfarin (1.10 vs. 0.71%/year) and dabigatran (0.45 vs. 0.29%/year) arms compared with non-Asian populations [
35]. Further, the East Asian population had over twofold risk of ICH compared with non-East Asian ones in the ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial (warfarin, 1.88 vs. 0.67%/year; apixaban, 0.67 vs. 0.30%/year) [
33].
Modifiable bleeding risk factors should be addressed at every opportunity: for example, uncontrolled blood pressure and concomitant use of aspirin or non-steroidal anti-inflammatory drugs. An approach based on modifiable bleeding risk factors alone, however, is an inferior strategy to using a formal bleeding risk score to assess bleeding risk [
9,
10,
36].
Several bleeding risk scores have been proposed (Tables
3 and
4), including the mOBRI score (modified Outpatient Bleeding Risk Index) [
37], the HEMORR
2HAGES score (Hepatic or Renal Disease, Ethanol Abuse, Malignancy, Older Age, Reduced Platelet Count or Function, Re-Bleeding Risk, Hypertension, Anemia, Genetic Factors, Excessive Fall Risk, Stroke) [
38], the Shireman score [
39], the HAS-BLED score (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) [
40], the ATRIA score (Anticoagulation and Risk Factors In Atrial Fibrillation) [
45], and the ORBIT score (national Outcomes Registry for Better Informed Treatment of Atrial Fibrillation) [
46]. Of the various bleeding risk scores, the HAS-BLED score has been validated to predict ICH and recurrent ICH after first spontaneous ICH [
10,
47,
48]. For example, in patients with spontaneous ICH, the risk of ICH recurrence increased with the HAS-BLED score, ranging from 1.37% per patient-year for a score of 1 to 2.90% per patient-year for a score of 4 [
48].
Table 3
Risk scores for evaluating anticoagulated individual’s risk of bleeding
| 0.78 (N/A) | Age ≥ 65, previous stroke, gastrointestinal bleed, ≥ 1 of the following comorbidities (recent MI, hematocrit < 30%, creatinine > 1.5 mg/ml, diabetes) | Low risk: 0; intermediate risk: 1–2; high risk: ≥ 3 |
| 0.67 (N/A) | Prior bleed (2 points), hepatic or renal disease, ethanol abuse, malignancy, age ≥ 75, reduced platelet count or function, uncontrolled hypertension, anemia, genetic factor, excessive fall risk, stroke | Low risk: 0–1; intermediate risk: 2–3; high risk: ≥ 4 |
| 0.63 (N/A) | Age ≥ 70 (0.49 points), female (0.32 points), remote bleed (0.58 points), recent bleed (0.62 points), alcohol/drug abuse (0.71 points), diabetes (0.27 points), anemia (0.86 points), antiplatelet (0.32 points) | Low risk: 0–1.07; intermediate risk: 1.07–2.19; high risk: ≥ 2.19 |
| 0.72 (0.64–0.79) | Uncontrolled systolic blood pressure, abnormal renal/liver function, stroke, bleeding history, labile international normalized ratio, age ≥ 65, drug, concomitant alcohol | Low risk: 0–1; intermediate risk: 2; high risk: ≥ 3 |
| 0.74 (0.70–0.78) | Anemia (3 points), severe renal disease (eGFR < 30 ml/min or dialysis-dependent) (3 points), age ≥ 75 (2 points), previous bleed, hypertension | Low risk: 0–3; intermediate risk: 4; high risk: ≥5 |
| 0.69 (0.63–0.74) | Age ≥ 74, insufficient kidney function (eGFR < 60 ml/min), antiplatelet, bleeding history (2 points), anemia (2 points), abnormal hemoglobin (< 13 mg/dL for males and < 12 mg/dL for females) (2 points) | Low risk: 0–2; intermediate risk: 3; high risk: ≥ 4 |
Table 4
Risk scores for evaluating individuals risk of thromboembolism
| AF patients | 0.82 (0.80–0.84) | Congestive heart failure, hypertension, age ≥ 65, diabetes, IS/TIA/SE (2 points) | Low risk: 0–1; intermediate risk: 2–3; high risk: ≥ 4 |
| AF patients | 0.61 (0.51–0.70) | Congestive HF, hypertension, age ≥ 75 (2 points), diabetes, IS/TIA/SE (2 points), vascular disease, age 65–74, female gender | Low risk: 0; intermediate risk: 1; high risk: ≥ 2 |
Modified Wells score [ 43] | DVT/PE | N/A | Active cancer, immobilization, recent bedridden, tenderness along the deep venous system, entire leg swollen, calf swelling, pitting edema, collateral superficial veins | Low risk: 0; intermediate risk: 1–2; high risk: ≥ 3 |
Revised Geneva score [ 44] | DVT/PE | 0.79 (0.76–0.81) | Age 60–79, age ≥ 80 (2 points), previous PE/DVT (2 points), recent surgery (3 points), pulse rate > 100/min, PaCO2 < 4.8 kPa (2 points), PaCO2: 4.8–5.19 kPa, PaO2 < 6.5 kPa (4 points), PaO2: 6.5–7.99 kPa (3 points), PaO2: 8.9–9.49 kPa (2 points), PaO2: 9.5–10.99 kPa, pleatlike atelectasis, elevation of a hemidiaphragm on chest X-ray film | Low risk: 0–4; intermediate risk: 5–8; high risk: ≥ 9 |
Although some patients have significant risk factors or a high bleeding risk score, these need not be considered absolute contraindications. Indeed, a high bleeding risk score should be used to identify the patients at risk for more careful review and early follow-up after OAC resumption, and not used as an excuse to withhold OAC [
49], because during OAC cessation the patients face a higher risk of thromboembolism, which can increase mortality [
50].
Risk of Thromboembolism During OAC Cessation
OAC resumption is important for the patients at high risk of thromboembolism, such as those with prosthetic mechanical valve, high risk of PE, and AF patients with a high CHA
2DS
2-VASc score (congestive HF, hypertension, age ≥ 75 years, type 2 diabetes, previous stroke/TIA/thromboembolism, vascular disease, age 65~74 years, and gender category), i.e., CHA
2DS
2-VASc score ≥ 4. In a recent systematic review and meta-analysis of restarting OAC after ICH, AF is the most common reason for anticoagulation (34.7–77.8%), followed by prosthetic heart valve (2.6–27.8%), venous thromboembolism (7.9–20.8%), and previous IS (3.7–71.8%) [
18]. In other studies, the most common reason for OAC resumption after ICH was mechanical heart valve (39–68%) [
14,
16]. Undoubtedly, the prolonged cessation of OAC after ICH would expose these high-risk patients to a greater risk of thromboembolism.
For example, the risk of major thromboembolism among patients (
n = 13,000) with prosthetic heart valves and not on OAC was 4.0 per 100 patient-years (95% CI, 2.9–5.2) [
51]. Mitral valve prosthesis is associated with a fivefold higher incidence of valve thrombosis and 1.5 times greater incidence of thromboembolism [
52]. The combination of double mechanical prosthesis was related to an even higher risk of thromboembolism (91%) [
12]. Also, the risk of DVT was 2 to 15% in patients with recent ICH [
53]. PE occurs in 1 to 5% of recent ICH patients in which anticoagulation was stopped [
12]. Given the high risk of recurrent venous thromboembolism, these patients may require antithrombotic therapy despite the risk of recurrent ICH.
OAC resumption can reduce the risk of thromboembolism. For example, a retrospective study demonstrated that OAC in AF patients after ICH was associated with a significantly reduced incidence of thromboembolism (RR, 0.19; 95% CI, 0.11–0.54) [
20]. In a nationwide cohort (
n = 2415), warfarin resumption had a lower rate of IS and SE in AF patients with hemorrhagic stroke (HR, 0.49; 95% CI, 0.24–1.02) [
17••]. In a Danish cohort (
n = 1725), the patients with OAC resumption were associated with 41 to 54% lower risk of IS/SE and all-cause mortality, compared with no OAC treatment [
15••]. These results were also confirmed in the Danish nationwide registries (
n = 6369) (HR, 0.58; 95% CI, 0.35–0.97) [
1•], another Danish study (HR, 0.55; 95%, 0.39–0.78) [
15••] and in a German cohort (5.2 vs. 15% per 100 patient-years,
p < 0.001) [
16]. In a recent meta-analysis about OAC resumption after ICH, reinitiation of OAC resulted in a significantly lower risk of thromboembolic complications (HR, 0.34; 95% CI, 0.25–0.45) [
18]. Also, the MUCH-Italy study (Multicentre Study on Cerebral Hemorrhage in Italy) showed an 81% reduced risk of thromboembolism among patients restarted OAC after ICH [
19].
Despite the high risk of thromboembolism and the efficiency of anticoagulation, a large proportion of ICH survivors often do not resume OAC [
1•,
15••,
54,
55]. Although some prior studies suggest that OAC could be discontinued safely for a certain period without significant high risk of thromboembolism, the duration of this period is less certain [
41]. For example, in the REVERSE-AD study (Reversal Effects of Idarucizumab on Active Dabigatran), the majority of the thrombotic events occurred in patients who had not resumed OAC in the first 30 days after ICH [
42]. Thus, although immediate OAC resumption is not appropriate for most patients, the duration of withholding OAC should be carefully considered and balanced against the risk of recurrent thromboembolism.
There are some well-identified risk factors for thromboembolism (see Table
2), and several risk scoring systems have been proposed to evaluate the individual’s risk of thromboembolism, such as the CHADS
2 [
43] and CHA
2DS
2-VASc scores in AF [
44], and the modified Wells score [
56] and the revised Geneva score in venous thromboembolism [
57] (Table
4). Identifying these risk factors and risk score would be useful to predict the individuals’ risk of thromboembolism.