This was a retrospective, single-center observational study that used data from medical records. This study included consecutive cases of ICH that occurred between April 2012, when factor Xa inhibitors were first introduced in Japan, and June 2022, when the reversal agent andexanet alfa became clinically available.

Inclusion criteria were as follows: patients who had been regularly taking one of the factor Xa inhibitors (rivaroxaban, apixaban, or edoxaban) for at least 1 month prior to the onset of hypertensive ICH, who developed ICH within 24 h of taking a factor Xa inhibitor and presented to Hitachi General Hospital within 24 h of their onset.

Exclusion criteria included cases with the concomitant use of antiplatelet agents, cases in which the primary lesion was intraventricular hemorrhage, and cases lacking follow-up CT imaging. Additionally, patients with vascular abnormalities capable of causing hemorrhage, such as aneurysms, arteriovenous malformations, or Moyamoya disease, were excluded.

Clinical outcomes were evaluated using the modified Rankin Scale (mRS) 3 months after the onset of ICH, with favorable outcomes defined as an mRS scores of 0–3.

Hematoma volumes were calculated using the ellipsoid formula: (length × width × height)/2, based on measurements obtained from CT images with a slice thickness of 2.5 mm (19).

HE was defined as an increase in hematoma volume of ≥33% or an absolute increase of ≥6 mL within 24 h of the initial CT scan (20).

The initial CT was, in principle, obtained within 30 min of hospital arrival.

Follow-up CT scans were generally obtained at 3, 12, and 24 h after the initial CT. Additional imaging was conducted if clinical deterioration was observed.

The CT morphology of the hematoma was assessed based on its margin configuration and Hounsfield unit (HU) characteristics. An irregular hematoma margin was defined as meeting at least one of the criteria for the island sign or the satellite sign. Heterogeneous hematoma density was defined as meeting at least one of the criteria for the blend sign or the black hole sign. The following imaging signs were defined as follows (7, 9–11):

Factor Xa inhibitors are administered in both standard and reduced dosing regimens, with specific dose reduction criteria defined for each agent based on creatinine clearance (CrCL) (21), age, body weight, and concomitant medications. In Japan, the approved standard and reduced daily doses of rivaroxaban are 15 mg and 10 mg once daily, respectively, which differ from the internationally recommended dosing guidelines. In contrast, the dosing regimens for apixaban and edoxaban in Japan align with global standards, with standard and reduced daily doses of 10 mg (5 mg twice daily) and 5 mg (2.5 mg twice daily) for apixaban, and 60 mg and 30 mg once daily for edoxaban, respectively.

Overdosing was defined as the administration of a standard dose in patients who met the criteria for dose reduction.

The following factors were analyzed for their relationship with HE:

Statistical analyses were performed using the median test for continuous variables and the chi-squared test or Fisher’s exact test for categorical variables, with a significance threshold set at p ≤ 0.05. Variables with a p-value ≤0.20 in the univariate analysis were entered into a multivariate logistic regression model. All statistical analyses were conducted using Stat Mate version 4.01 (ATMS Co., Ltd., Japan).

Ninety-two cases of spontaneous ICH occurring within 24 h of factor Xa inhibitor intake were identified during the study period. Of these, 24 cases were excluded for the following reasons: two in which intraventricular hemorrhage was the primary lesion, 12 cases involving the concomitant use of antiplatelet agents, and 10 cases lacking follow-up CT imaging. Among the latter, five patients underwent surgical intervention after initial CT, and five were in critical condition and unable to undergo repeat imaging.

We included 68 patients in this study. The median age was 77.5 years (IQR 70.8–85.0), and 39 (57.4%) were male. The factor Xa inhibitors and dosages were rivaroxaban in 32 patients (standard dose, 14; reduced dose, 18), apixaban in 15 (standard dose, 7; reduced dose, 8,), and edoxaban in 21 (standard dose, 8; reduced dose, 13). The indications for factor Xa inhibitor therapy were non-valvular atrial fibrillation, including paroxysmal cases, in 62 patients; deep vein thrombosis or pulmonary embolism in 14; and both conditions in 8.

The HE was observed in 18 of the 68 included cases (26.5%). Among these, 10 cases occurred within 6 h and 7 within 12 h after the initial CT. In one case, hematoma enlargement was confirmed at the 24-h evaluation.

Favorable outcomes at 3 months were achieved in 29 cases (58.0%) without HE and in 5 cases (27.8%) with HE. The presence of HE was significantly associated with worse functional outcomes (p = 0.028, χ² test) (Figure 2).

During the period of factor Xa inhibitor discontinuation following ICH onset, thromboembolic events occurred in 3 patients (4.4%), comprising one case each of pulmonary embolism, myocardial infarction, and cerebral venous sinus thrombosis. The outcomes were fatal in the pulmonary embolism and myocardial infarction cases, whereas the patient with cerebral venous sinus thrombosis had a mRS score of 4.

The median values or incidence of various patient, clinical, hematoma, and drug factors stratified by the presence or absence of HE are summarized in Table 1.

In univariate analysis, significant predictors of HE were hematoma morphology on CT (p = 0.009), an interval of ≤4 h from the last factor Xa inhibitor intake to the initial CT (p = 0.034), and edoxaban use (p = 0.041). Factors showing a trend toward association included larger initial hematoma volume (p = 0.054), higher sBP on admission (p = 0.066), factor Xa inhibitor overdose (p = 0.113), and lower platelet count (p = 0.169). In contrast, HT, HL, DM, Child–Pugh classification, sBP at 3 h, PT-INR, APTT, an interval of ≤4 h from the last factor Xa inhibitor intake to symptom onset, interval from symptom onset to initial CT, and factor Xa inhibitor dosage were not associated with HE.

Multivariate analysis of variables with p ≤ 0.20 in univariate analysis identified irregular margin and/or heterogeneity of the hematoma on non-contrast CT (p = 0.030) and an interval of ≤4 h from the last Xa inhibitor intake to the initial CT (p = 0.048) as independent predictors of HE. Factor Xa inhibitor overdose demonstrated a trend toward association with HE (p = 0.089), whereas the type of Xa inhibitor (p = 0.200), sBP on admission (p = 0.245), initial hematoma volume (p = 0.737), and platelet count (p = 0.766) were not predictive.

The presence of an irregular margin and/or heterogeneity of the hematoma on non-contrast CT associated with HE. These imaging features have been recognized in previous studies as predictors of HE in general ICH (5–11).

Heterogeneous attenuation on CT is considered to reflect incompletely clotted blood, indicating ongoing bleeding.

Areas within a hematoma with attenuation values below 30 Hounsfield units (HU) have been linked to a higher risk of HE. Fully clotted blood typically demonstrates values of approximately 100 HU, while partially clotted or active bleeding regions range from 30 to 50 HU (24). Irregular hematoma margins may indicate continued bleeding from ruptured perforating arteries at the hematoma edge.

Despite the predictive value of these features, HE still occurred in 22.2% of cases without these CT features. Blacquiere et al. (6) reported that density heterogeneity demonstrated a sensitivity of 44% and a specificity of 69%, whereas margin irregularity showed a sensitivity of 72% and a specificity of 46%. In contrast, Kim et al. (10) found that 38.8% of hematomas with density heterogeneity did not exhibit expansion, and 41.4% of those with margin irregularity also did not demonstrate growth. It should be noted that the risk of HE cannot be ruled out even in cases that do not exhibit these hematoma morphologies.

The “target sign” on contrast-enhanced CT is another established indicator of active bleeding; however, we were unable to assess this sign in the present study due to the limited number of patients who underwent contrast-enhanced imaging (11). Future studies incorporating this parameter may improve the prediction of HE.

In general ICH, a shorter interval from symptom onset to CT is associated with an increased risk of HE, which may be due to incomplete hemostasis early in the hemorrhagic process (8).

In the present study, we examined both the interval from factor Xa inhibitor intake to initial CT and the interval from symptom onset, only the former was significantly associated with HE.

Factor Xa inhibitors reach peak activity within 1–4 h of their administration and have varying half-lives: rivaroxaban (5–9 h), apixaban (10–14 h), and edoxaban (9–11 h) (15, 25–27). These pharmacological characteristics indicate that anticoagulant activity is higher within 4 h after administration. OD further increases the plasma concentration of factor Xa inhibitors. In addition, the longer half-life of edoxaban compared with other agents may explain its association with HE observed in the univariate analysis. Therefore, in order to predict HE in patients receiving factor Xa inhibitors, it would be ideal to evaluate the degree of anticoagulant activity in each patient. However, conventional coagulation assays have proven to be poor indicators of the pharmacological activity of factor Xa inhibitors (13–15).

While HE was identified as high-risk within 4 h of the last intake, 55.6% of HE cases occurred beyond this timeframe. Accordingly, it would be inappropriate to consider patients ineligible for reversal therapy solely on the basis that more than 4 h have elapsed since the last intake.

Godier et al. (14) reported that the anticoagulant effects of DOACs may persist for up to 72 h in patients with moderate renal impairment. Routine coagulation assays are insufficient to reliably quantify this effect. Due to the potential for ongoing anticoagulant activity, reversal therapy should be considered for all ICH cases occurring within 24 h of the last factor Xa inhibitor dose.

The benefits of reversal beyond the 24-h window remain unclear and, thus, warrant further investigation.

In general ICH, larger hematoma volumes have been associated with an increased risk of HE, which may be attributed incomplete hemostasis of the perforating artery leading to hematoma enlargement and mechanical stress at the hematoma margins promoting additional vascular rupture (5).

In the absence of anticoagulant effects, small hematomas typically do not enlarge because hemostasis is already complete. In contrast, in patients receiving factor Xa inhibitors, the suppression of coagulation increases the risk of both persistent bleeding and new vessel rupture, even in cases with smaller hematomas.

The interval from the last dose to ICH onset was not a predictor of HE. A possible explanation is that HE occurring prior to the initial CT could not be assessed in this study. It should be acknowledged that the occurrence of HE during the early phase of ICH may result in a large hematoma by the time of hospital arrival.

Although smaller hematomas are generally associated with better outcomes, the present results indicate that hematoma volume was not a significant predictor of expansion in this population. Therefore, even small hematomas in patients treated with factor Xa inhibitors warrant consideration of reversal therapy to prevent expansion and improve outcomes.

Delcourt et al. (4) reported that a hematoma volume increase of ≥6 mL or ≥33% significantly worsened outcomes (mRS 3–6), with adjusted odds ratios of 1.72 and 1.67, respectively. Consistent with this finding, the present study demonstrated that HE was associated with poor functional outcomes in patients taking factor Xa inhibitors.

Andexanet alfa achieves rapid and effective reversal of factor Xa inhibitors; however, the optimal criteria for its administration have yet to be defined (16).

The predictors of HE identified in this study were an irregular or heterogeneous hematoma morphology and ICH onset within 4 h after factor Xa inhibitor intake. The former likely reflects incomplete hemostasis of the hematoma, while the latter indicates high factor Xa inhibitor activity. In the presence of these factors, reversal of factor Xa inhibitors should be considered. However, it must be recognized that HE may occur even in the absence of these predictors, and the indication for reversal should therefore be carefully considered. By contrast, basing reversal decisions on hematoma volume or the interval from ICH onset—both of which have been reported as predictors of HE in general ICH—is not appropriate.

Thromboembolic complications were observed in 4.4% of patients during anticoagulation therapy interruption due to ICH in this study. Andexanet alfa may reduce the duration of treatment interruption. Whether a shortened discontinuation period of factor Xa inhibitors can effectively prevent thromboembolic events requires further investigation.

The limitations of this study include its retrospective design and the fact that it was conducted at a single center. Future research involving larger, multicenter clinical studies is warranted.