Cilostazol Treatment for Intracerebral Hemorrhage: Targeting Dural Lymphatics

Abstract

BACKGROUND:

The meningeal lymphatic system contributes to hematoma resolution and neurological recovery after intracerebral hemorrhage (ICH). We previously demonstrated that intraperitoneal cilostazol administration promotes dural lymphatic growth and enhances clearance of intracerebrally injected red blood cells; however, the antiplatelet effects of cilostazol raise safety concerns in ICH, potentially limiting clinical translation. Here, we evaluated whether oral cilostazol at clinical-equivalent doses in mice enhances dural lymphatic function and improves outcomes after ICH.

METHODS:

ICH was induced by collagenase injection to model microvascular rupture. Cilostazol was orally administered to ICH mice at clinically equivalent doses, given at different time points and for varying durations. Prolymphangiogenic effects of cilostazol were examined in healthy in vivo and ex vivo meninges. Flow cytometry and immunofluorescence were performed to assess lymphatic endothelial cell proliferation and dural lymphatic remodeling. Lymphatic function was evaluated by in vivo live imaging and histological analysis of fluorescent tracer drainage. Short- and long-term histopathologic and behavioral outcomes were also evaluated.

RESULTS:

Oral cilostazol treatment initiated either 3 hours or 3 days after ICH enhanced dural lymphatic coverage and function, accompanied by reduced hematoma volume, neuronal injury, synapse loss, and long-term neurological deficits. Cilostazol increased CD31⁺PDPN⁺ lymphatic endothelial cells in ex vivo meninges, induced dural lymphatic hyperplasia, and enhanced lymphatic drainage in healthy mice. Both pretreatment and delayed treatment of cilostazol did not increase hematoma volume but rather improved ICH outcomes.

CONCLUSIONS:

By promoting dural lymphatic remodeling and drainage, cilostazol facilitates hematoma resolution and neurological recovery after ICH. Cilostazol treatment does not exacerbate initial bleeding or provoke hematoma expansion in the experimental ICH. These findings support clinical translation and provide mechanistic and dosing rationale for an ongoing phase II randomized trial (URL: https://www.clinicaltrials.gov; Unique identifier: NCT06504576).

Intracerebral hemorrhage (ICH) is a severe stroke subtype with high mortality and substantial long-term disability.¹ Beyond the initial vascular rupture, secondary injury driven by mass effect from space-occupying hematomas and by toxic plasma components and red blood cell (RBC) metabolites contributes to progressive brain damage. As hematoma size is a significant predictor of prognosis in patients with ICH,² timely and effective hematoma evacuation and augmentation of endogenous clearance mechanisms have long been pursued as therapeutic strategies.^3–5 Recent data from the ENRICH trial (Early Minimally Invasive Removal of Intracerebral Hemorrhage) report favorable outcomes with hematoma evacuation for lobar ICH⁶; however, effective treatments for patients with deep ICH and for those who are not candidates for surgical intervention remain limited.⁷ In addition, few clinical trials aiming to enhance the endogenous hematoma absorption have successfully translated from preclinical studies.⁸ Therefore, it is essential to identify additional targets that can accelerate hematoma resolution and ultimately improve ICH recovery.

Lymphatic vessels within the dural meninges (hereafter referred to as dural lymphatics) mediate drainage of the cerebrospinal fluid (CSF) from the central nervous system (CNS) to help control brain waste clearance, immune surveillance, and injury responses.⁹ Accumulating evidence indicates that drainage from dural lymphatics to cervical lymph nodes assists in resolving hematomas, blood toxicity, and neuroinflammation in human and experimental models of hemorrhagic brain injuries, including ICH,¹⁰ subarachnoid hemorrhage,¹¹ and intraventricular hemorrhage.¹² We previously demonstrated that dural lymphatic ablation impedes hematoma clearance, whereas intracranial infusion of VEGF-C (vascular endothelial growth factor C), a well-known prolymphangiogenic growth factor, reduces hematoma volume in ICH animals.¹⁰ In the same study, pharmacological enhancement of dural lymphatic drainage by a Food and Drug Administration-approved PDE3 (phosphodiesterase type 3) inhibitor, cilostazol, improved histological and neurobehavioral outcomes after intraparenchymal injection of whole blood-isolated RBCs, providing proof of concept that targeting dural lymphatics may represent a viable therapeutic strategy in ICH.

Our previous work established that disruption of meningeal lymphatics impairs clearance, whereas their enhancement improves outcomes in experimental ICH. However, that study did not evaluate the safety, dosing, or translational feasibility of cilostazol, an antiplatelet agent clinically used to treat intermittent claudication and for secondary prevention of noncardioembolic stroke.^13–15 In contrast, the current study uses oral cilostazol at human equivalent doses, assessed across multiple treatment windows, to determine if augmenting dural lymphatics with cilostazol is safe and efficacious in reducing hematoma burden and improving neurological recovery to provide direct preclinical support for ongoing clinical testing. It also evaluated whether cilostazol exacerbates primary bleeding or increases the risk of hematoma expansion in acute ICH given its antiplatelet properties. Although the prolymphangiogenic effects in injured meninges were inferred in the prior study, this work provides cellular and functional evidence of increased lymphatic endothelial cell (LEC) proliferation and dural lymphatic remodeling following oral cilostazol treatment in physiological conditions.

Here, we systematically assessed the safety and therapeutic efficacy of oral cilostazol in a collagenase-induced ICH model, using clinical equivalent doses, multiple treatment windows, and different treatment durations. In parallel, we examined the direct pro-lymphangiogenic effects of cilostazol on dural lymphatics using ex vivo meningeal cultures, healthy mice, and ICH mice. We employed the collagenase model because it recapitulates ongoing microvascular disruption and hematoma enlargement over time, closely reflecting key pathophysiological features of acute ICH in patients.

Methods

Data Availability

Raw data supporting the findings of this study are available from the corresponding author on reasonable request. All detailed materials and methods are provided in the Supplemental Material.

Experimental Animals

All animal experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health and the ARRIVE guidelines 2.0 (Animal Research: Reporting of in vivo Experiments)¹⁶ and the protocols approved by the Institutional Animal Care and Use Committee at the National Taiwan University College of Medicine (approval number: 20220086). A total of 389 10- to 14-week-old male C57BL/6J (wild-type) mice from the Laboratory Animal Center of our institute were used in this study. Every effort was made to minimize animal suffering and the number of animals used. The animals were housed in a temperature- and humidity-controlled, specific pathogen-free facility under a 12-hour light/dark cycle. Twenty-nine mice died after collagenase injection surgery due to the natural mortality of this preclinical ICH model. Five warfarin-treated ICH mice died due to severe hemorrhage. Seventeen mice were excluded from statistical analysis due to failed surgery, defined as a neurological severity score below 3 on day 1 after ICH.¹⁷ Animals were randomly allocated to experimental groups using a coin toss randomization approach, and the number of animals used per experimental group is reported in Figure 1A. Behavioral tests (T.-Y.L. and W.-R.C.) and all other in vivo analyses (C.-K.L., Y.-C.H., Z.-R.J., and D.J.) were performed by investigators blinded to the experimental groups and treatment assignment. Group, effect sizes, and power were determined based on published data (see Statistical Analysis and Major Resources Table in the Supplemental Material, which also reports inclusion and exclusion criteria).

Figure 1.

Experimental design and animal assignment. A, Number of animals used in each experimental assignment. B, Schematic overview of the experimental design and subsequent analytical methods. b.i.d. indicates twice a day; CSZ, cilostazol; HC, healthy control; ICH, intracerebral hemorrhage; p.o., orally; Veh, vehicle; and W, warfarin. Created in BioRender. Jiang, D. (2026) https://BioRender.com/2f4dtyw.

Experimental Design

This study comprised 6 experiments, as outlined below and illustrated in Figure 1B. Studies have shown that the elimination half-time (t_1/2) of cilostazol is similar between rodents (t_1/2=10.7±3.4 hours)¹⁸ and humans (t_1/2≈11 hours).^19,20 The dose regimen was therefore determined based on (1) the human equivalent dose calculations indicating that around 100 mg in a 60 kg adult corresponds to ≈20 mg/kg in mice,²¹ and (2) standard clinical dosing practices, in which cilostazol is administered orally at 50 to 100 mg twice daily.¹³ Accordingly, mice were randomly assigned to receive freshly prepared 10, 20, or 40 mg/kg cilostazol (Sigma-Aldrich) or vehicle (0.5% carboxymethyl cellulose; Sigma-Aldrich) by oral gavage.

Experiment Evaluating Therapeutic Efficacy of Cilostazol Following ICH

To evaluate the therapeutic efficacy of cilostazol on dural lymphatic function, hematoma resolution, neurological deficits, and brain recovery after ICH, ICH mice received 10, 20, or 40 mg/kg cilostazol or vehicle via oral gavage starting 3 hours after ICH induction, then once daily for either 2 or 6 consecutive days. On days 3 and 7 after ICH, neurological performance and hematoma volume were assessed. Dural meninges and brain tissues were collected for immunofluorescence to analyze lymphatic vessel morphology as well as neuronal and synaptic loss. To determine lymphatic drainage, mice received an intracisterna magna injection of 0.5-μm FluoSpheres Carboxylate-Modified Microsphere (Thermo Fisher), and dural meninges and deep cervical lymph nodes (dCLNs) were harvested for immunofluorescence. LYVE (lymphatic vessel endothelial hyaluronan receptor)-1, a marker widely expressed by LECs, was used to visualize dural lymphatic vessels in whole-mount preparations and to quantify lymphatic vessel coverage and branching in the meninges.

Experiment Evaluating Therapeutic Efficacy of Cilostazol in Mitigating Long-Term Outcomes Following ICH

To evaluate the impact of cilostazol treatment on long-term outcomes, mice subjected to ICH received oral cilostazol at 20 mg/kg or vehicle once daily from 3 hours through day 7 after ICH. Neurological deficits were assessed on days 1, 3, 7, 14, and 21 after ICH, whereas cognitive functions were assessed on days 25 to 28. Hematoma volume and injury volume were measured on day 30 after ICH.

Experiment Evaluating the Prolymphangiogenic Effect of Cilostazol Ex Vivo

To examine the direct effects of cilostazol on meningeal LECs, dural meningeal cultures were freshly prepared from healthy mice and incubated with complete medium containing 10, 30, or 50 μmol/L cilostazol or 0.1% DMSO.²² After 24 hours of incubation, the ex vivo dural meninges were collected for flow cytometric analysis of cell viability and LEC populations.

Experiment Evaluating the Impact of Cilostazol on Dural Lymphatics Under Physiological Conditions

To understand the effect of cilostazol on dural lymphatics under physiological conditions, healthy mice orally received 20 mg/kg cilostazol or vehicle daily for 7 consecutive days. Dural meninges were collected for flow cytometry and whole-mount immunofluorescence. Flow cytometry was used to quantify dural LECs, defined as CD45⁻CD31⁺PDPN⁺ cells, with CD45 exclusion used to remove hematopoietic cells, CD31 used to identify endothelial cells, and PDPN used to further enrich for the lymphatic endothelial population. LYVE-1 whole-mount immunofluorescence was used to visualize dural lymphatic vessels and quantify lymphatic vessel complexity and coverage. Lymphatic drainage function was assessed as described above.

Experiment Evaluating the Effect of Cilostazol Pretreatment on Primary Hemorrhage

To explore whether the antiplatelet property of cilostazol increases bleeding risk after ICH, mice orally received 40 mg/kg cilostazol daily for 3 days before collagenase injection-induced ICH. Neurological performance, hematoma volume, and dural lymphatic vessel structure were assessed on day 3 after ICH. As a positive bleeding-risk control, mice received oral warfarin at 2 mg/kg²³ using the same route and dosing schedule as cilostazol, and hematoma volume was measured on day 3 after ICH.

Experiment Evaluating the Effect of Cilostazol Post-Treatment on Hematoma Expansion

To explore whether cilostazol post-treatment increases the risk of delayed hematoma expansion after ICH, mice were administered 40 mg/kg cilostazol beginning on ICH day 3 and continuing for 3 consecutive days. On day 7 post-ICH, neurological performance, hematoma volume, lymphangiogenesis, and lymphatic drainage were evaluated. As a positive control for this post-treatment paradigm, mice received oral warfarin at 2 mg/kg using the same route and dosing schedule as cilostazol, and hematoma volume was measured on day 7 after ICH.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism (version 10.2.3). Data are presented as mean±SEM or box plots showing the median (horizontal line), 25th and 75th percentiles (boxed range), and minimum and maximum values, with individual mice or biological replicates shown as circles. Sample sizes were determined based on pilot tests and prior work from our laboratory^10,24 at a power of 0.8, and were comparable to those in previous publications investigating hematoma resolution and iron toxicity in ICH.^4,25 Normality was assessed using the Shapiro-Wilk test. For comparisons between 2 groups, an unpaired Student t test was used for normally distributed data with equal variance, and Welch correction was applied when variances were unequal. For non-normally distributed 2-group comparisons, the Mann-Whitney U test was used. For comparisons among 3 or more groups, 1-way ANOVA followed by Dunnett posthoc test was used for normally distributed data with equal variances, whereas Brown-Forsythe ANOVA followed by a Dunnett T3 posthoc test was used when variances were unequal. For non-normally distributed data involving 3 or more groups, Kruskal-Wallis with Dunn posthoc test was employed. Repeated-measures 2-way ANOVA followed by Dunnett correction was used for longitudinal neurological assessments. A P value of <0.05 was considered statistically significant.

Results

Cilostazol Therapeutic Efficacy for Dural Lymphatic Function, Hematoma Resolution, and ICH Recovery

Our earlier proof-of-concept work with ICH showed that intraperitoneal administration of 10 mg/kg cilostazol increases RBC clearance and dural lymphatic vessel number at day 3 after intrastriatal injection of PKH-26-labeled RBCs.¹⁰ To evaluate the therapeutic efficacy and translational relevance of cilostazol in a clinically meaningful context, we selected dosing regimens and an oral delivery route corresponding to routine clinical use. Given that a daily dose of 100 to 200 mg cilostazol is commonly used for intermittent claudication and secondary prevention of stroke, and 20 mg/kg in mice is equivalent to ≈100 mg in a 60 kg human,^18–21 mice were treated with 10, 20, or 40 mg/kg cilostazol by oral gavage starting 3 h after ICH induction.

At day 3 after ICH, CNS-to-dCLN lymphatic drainage was significantly reduced (Figure S1). Acute treatment with 20 or 40 mg/kg cilostazol attenuated this early lymphatic dysfunction and was associated with smaller hematoma volumes (Figures S2A and S3), as well as increased dural lymphatic branch number and coverage area compared with vehicle-treated mice (Figures S2B and S4). In addition, meningeal lymphatic function has been reported to regulate cerebral edema, and impaired dural lymphatic drainage has been associated with reduced cerebrospinal fluid/interstitial fluid exchange and macromolecule clearance, thereby worsening brain injury.⁹ Therefore, we examined brain water content at 72 hours after ICH. Cilostazol treatment at 20 or 40 mg/kg reduced brain water content in the ipsilateral striatum of ICH mice (Figure S2C). These findings suggest that early pharmacological enhancement of dural lymphatic function mitigates the initial decline in lymphatic drainage and facilitates more efficient hematoma resolution.

Next, we assessed the effects of cilostazol at the late, day 7, time point post-ICH. At the recovery stage of the collagenase ICH model,^4,5,24,26 we found that daily oral administration of 20 and 40 mg/kg cilostazol significantly induced dural lymphatic hyperplasia at the left and right transverse sinuses (hotspots), characterized by increased LYVE-1-positive lymphatic vessel complexity (sprouts and loops) and coverage area (Figure 2A and 2B; Figure S5). These hotspots are critical regions for macromolecule uptake and typically show high branching and complexity along the transverse sinuses.²⁷ Consistent with these structural changes, fluorescent tracer drainage to the dCLNs was significantly increased in mice treated with 20 or 40 mg/kg cilostazol compared with vehicle-treated controls (Figure 2C and 2D).

Figure 2.

Cilostazol treatment promoted dural lymphatic hyperplasia and drainage after collagenase-induced intracerebral hemorrhage (ICH). A, Representative whole-mount meningeal images of the left transverse sinus (LTS) and right transverse sinus (RTS) showing histological changes in LYVE (lymphatic vessel endothelial hyaluronan receptor)-1-positive lymphatic vasculature in vehicle (Veh)- and 10, 20, and 40 mg/kg cilostazol (CSZ10, CSZ20, and CSZ40)-treated ICH mice at day 7. Yellow arrowheads indicate enlarged dural lymphatic vessels. B, Quantification of lymphatic vessel branch number and LYVE-1 coverage at the LTS and RTS of Veh- and CSZ10, CSZ20, and CSZ40-treated ICH mice. Veh, n=22; CSZ10, n=7; CSZ20, n=17; CSZ40, n=18. C, Representative images of whole-mount meninges and deep cervical lymph nodes (dCLNs) showing drainage of fluorescent beads in Veh-, CSZ10-, CSZ20-, and CSZ40-treated ICH mice at day 7. D, Quantification of mean fluorescence intensity (MFI) of beads and percentage of bead coverage in dCLNs of Veh-, CSZ10-, CSZ20-, and CSZ40-treated ICH mice. Veh n=9, CSZ10 n=5, CSZ20 n=10, CSZ40 n=9. Data are presented as box plots showing minima, maxima, interquartile range (box bounds), and median (black line). Each circle represents 1 mouse. Kruskal-Wallis test with Dunn test (B and D).

Cilostazol-induced dural lymphatic expansion and enhanced drainage were accompanied by reduced hematoma volume in ICH mice treated with 20 or 40 mg/kg cilostazol compared with vehicle-treated controls (Figure 3A; Figure S6). Functionally, these mice demonstrated improved neurological performance on the corner turn test, forelimb placing test, hindlimb adduction test, and composite neurological severity score (Figure 3B). Histologically, cilostazol-treated mice exhibited increased neuronal survival in the perihematomal region (Figure 3C and 3D), along with preservation of presynaptic and postsynaptic puncta, as assessed by VGLUT1 and PSD95 immunoreactivity (Figure 3E and 3F). Synaptic integrity, quantified by VGLUT1-PSD95 apposition, was also significantly increased in mice treated with 20 or 40 mg/kg cilostazol at day 7 post-ICH (Figure 3F). Interestingly, we observed that 10 mg/kg cilostazol produced only a subtle, albeit statistically significant, expansion of dural lymphatic coverage at day 7. However, this rudimentary lymphatic structure remodeling did not translate into enhanced dural lymphatic function or improved pathological sequelae in ICH mice (Figures 2 and 3). Together, these findings show that 20 and 40 mg/kg cilostazol represent effective doses for ICH treatment, promoting hematoma resolution through augmentation of dural lymphatic function.

Figure 3.

Cilostazol treatment enhanced hematoma resolution, neurobehavioral recovery, and neuron survival after collagenase-induced intracerebral hemorrhage (ICH). A, Representative coronal sections showing hematoma and quantification of hematoma volume in vehicle (Veh)- and 10, 20, and 40 mg/kg cilostazol (CSZ10, CSZ20, and CSZ40)-treated ICH mice at day 7. Veh n=18, CSZ10 n=7, CSZ20 n=15, CSZ40 n=11. B, Corner turn test, forelimb placement test, hindlimb adduction test, and neurological severity score in healthy control (HC) and Veh- and CSZ-treated ICH mice at day 7. HC n=10, Veh n=23, CSZ10 n=10, CSZ20 n=21, CSZ40 n=14. C, Representative immunofluorescence of neuronal marker NeuN (red) with nuclear marker DAPI (blue) in the Veh- and CSZ-treated mice at day 7 after ICH. The boxed areas indicate regions of interest in high-magnification images. D, Quantification of NeuN⁺ neurons in the ipsilateral hemispheres of Veh- and CSZ-treated ICH mice at day 7. Veh n=6, CSZ10 n=4, CSZ20 n=6, CSZ40 n=7. E, Representative immunofluorescence of vGLUT1 (green) and PSD95 (red), with corresponding 3D spot reconstructions. Veh n=6, CSZ10 n=4, CSZ20 n=6, CSZ40 n=7. F, Quantification of the number of VGLUT1 (top) and PSD95 (middle) puncta and of VGLUT1/PSD95 colocalized puncta to their total number (bottom) in the Veh- and CSZ-treated ICH brain at day 7. Veh n=6, CSZ10 n=4, CSZ20 n=6, CSZ40 n=7. Data are presented as box plots that show minima, maxima, interquartile range (box bounds), and median (black line). Each circle represents 1 mouse. Kruskal-Wallis test with Dunn test (A and B); 1-way ANOVA with Dunnett test (forelimb placement test of B, D, and F). Created in BioRender. Jiang, D. (2026) https://BioRender.com/2f4dtyw.

Cilostazol Induces Dural Lymphatic Hyperplasia

Although cilostazol has been reported to stimulate growth in cultured human LECs,²² it is unclear whether cilostazol directly promotes dural LEC proliferation in vivo. To address this question, we first examined the effects of cilostazol in ex vivo meningeal cultures (Figure 4A). Treatment with 30 or 50 μmol/L cilostazol for 24 hours did not affect overall cell viability but significantly increased the proportion of CD31⁺PDPN⁺ LECs compared with vehicle-treated controls (Figure 4B), providing direct evidence of a pro-lymphangiogenic effect on meningeal lymphatic endothelium. Since the therapeutic efficacy of 20 and 40 mg/kg cilostazol in ICH animals was comparable, we chose the lower dose to examine dural LEC responses and dural lymphatic morphology and function in healthy animals. After 7 days of oral administration of 20 mg/kg cilostazol, analysis of dural tissue-resident cells revealed an increased percentage of LECs in cilostazol-treated animals compared with the vehicle group (Figure 4C). Healthy mice receiving cilostazol also displayed increased sprouting and greater dural lymphatic coverage at hotspots (Figure 4D; Figure S7). Functionally, cilostazol-treated mice showed elevated dural lymphatic drainage, as evidenced by greater fluorescent bead drainage to dCLNs through cerebrospinal fluid circulation (Figure 4E). Collectively, these findings demonstrate that cilostazol directly promotes dural LEC proliferation and lymphatic network remodeling, resulting in enhanced lymphatic drainage capacity.

Figure 4.

Cilostazol induced dural lymphatic endothelial cell proliferation, sprouting, expansion, and enhanced lymphatic drainage. A, Schematic of the ex vivo meningeal culture procedure. B, Representative contour plots and quantifications of total viable cells and lymphatic endothelial cells (LECs) in the DMSO- and 10, 30, and 50 μmol/L cilostazol (CSZ)-treated healthy dural meninges after 24 h. Veh n=12, 10 μmol/L n=9, 30 μmol/L n=12, 50 μmol/L n=9. C, Representative contour plots and quantification of LECs in the meninges of vehicle (Veh)- and cilostazol (CSZ)-treated healthy mice at day 7. Veh n=10, CSZ20 n=10. D, Representative whole-mount meningeal images of the left transverse sinus (LTS) and right transverse sinus (RTS) depicting LYVE-1 (lymphatic vessel endothelial hyaluronan receptor)-1-positive lymphatic vasculature in Veh- and 20 mg/kg cilostazol (CSZ20)-treated healthy mice at day 7 (top). Yellow arrowheads indicate enlarged dural lymphatic vessels. Quantifications of lymphatic vessel branch number and LYVE-1 coverage at the LTS and RTS of Veh- and CSZ20-treated healthy mice (bottom). Veh n=4, CSZ20 n=4. E, Representative images of whole-mount meninges and deep cervical lymph nodes (dCLNs) showing drainage of fluorescent beads in Veh- and CSZ20-treated healthy mice at day 7 (left). Quantifications of mean fluorescence intensity (MFI) of beads and percentage of bead coverage in the dCLNs of Veh- and CSZ20-treated healthy mice (right). Veh n=4, CSZ20 n=4. Data are presented as box plots that show minima, maxima, interquartile range (box bounds), and median (black line). Each circle represents 1 mouse. Student t test (C, D, beads MFI of E); Mann-Whitney U test (beads coverage of E); Brown-Forsythe ANOVA with Dunnett T3 test (live cells of B); 1-way ANOVA with Dunnett (PDPN⁺ cells of B). Created in BioRender. Jiang, D. (2026) https://BioRender.com/2f4dtyw.

Cilostazol Does Not Induce Hematoma Expansion After ICH

Use of oral antithrombotic agents increases the risk of recurrent ICH, and patients with antithrombotic-associated ICH often present with larger baseline hematomas and a higher chance of hematoma expansion.^28,29 To determine whether cilostazol treatment might worsen primary hemorrhage, we pretreated the mice with 40 mg/kg cilostazol for 3 days before ICH induction. Because discontinuation of antithrombotic agents has been recommended to improve clinical outcomes in ICH,⁷ cilostazol treatment was discontinued after ICH induction, and function and histological outcomes were evaluated at the acute phase on day 3. In a complementary paradigm, designed to evaluate whether cilostazol increases the risk of hematoma expansion, mice were post-treated with 40 mg/kg cilostazol starting on day 3 after ICH, and outcomes were evaluated on day 7. Warfarin treatment was included in both paradigms as a positive control for hematoma expansion.²³ Neither pretreatment nor delayed post-treatment with cilostazol increased hematoma volume. Instead, cilostazol-treated mice in both paradigms exhibited significantly smaller hematomas compared with vehicle-treated controls (Figure 5A; Figures S8 through S10), whereas warfarin-treated mice exhibited enlarged hematomas (Figures S8 and S9).

Figure 5.

Cilostazol pretreatment or post-treatment did not enlarge hematoma volume after collagenase-induced intracerebral hemorrhage (ICH). A, Representative coronal sections and quantifications of hematoma volume in ICH mice receiving either cilostazol pretreatment (vehicle [Veh] or 40 mg/kg cilostazol [CSZ40] for 3 days before collagenase injection), assessed at day 3, or post-treatment (Veh or CSZ40 from day 3 to 6 after collagenase injection), assessed at day 7 after ICH. Pretreatment, Veh n=9, CSZ40 n=9; post-treatment, Veh n=10, CSZ40 n=12. B, Corner turn test, forelimb placement test, hindlimb adduction test, and neurological severity score of ICH mice receiving either cilostazol pretreatment, assessed at day 3, or post-treatment, assessed at day 7 after ICH. Pretreatment, Veh n=9, CSZ40 n=9; post-treatment, Veh n=10, CSZ40 n=12. C, Representative whole-mount meningeal images of the left transverse sinus (LTS) and right transverse sinus (RTS) depicting LYVE (lymphatic vessel endothelial hyaluronan receptor)-1-positive lymphatic vasculature in mice receiving either cilostazol pretreatment, assessed at day 3, or post-treatment, assessed at day 7 after ICH (top). Yellow arrowheads indicate enlarged dural lymphatic vessels. Quantifications of lymphatic vessel branch number and LYVE-1 coverage at the LTS and RTS of Veh- and CSZ40-pretreated ICH mice (bottom). Pretreatment, Veh n=5, CSZ40 n=8; post-treatment, Veh n=10, CSZ40 n=12. Each circle represents 1 mouse. D, Representative images of whole-mount meninges and deep cervical lymph node (dCLNs) showing fluorescent bead drainage at day 7 in ICH mice post-treated with Veh or CSZ40. Quantifications of bead mean fluorescence intensity (MFI) and percentage bead coverage in the dCLNs of Veh- and CSZ40-treated ICH mice. Veh n=10, CSZ40 n=12. Data are presented as box plots that show minima, maxima, interquartile range (box bounds), and median (black line). Each circle represents 1 mouse. Student t test (A, pretreatment of C, forelimb placement of B in post-treatment); Student t test with Welch correction (lymphatic vessel branch number of C in post-treatment, and D); Mann-Whitney U test (pretreatment of B, corner turn, hindlimb adduction, and neurological severity score of B in post-treatment, LYVE-1 coverage of C in post-treatment).

The observed reductions in hematoma burden of cilostazol-treated ICH mice were accompanied by improved neurological performance on the hindlimb adduction test, forelimb placement test, and neurological severity score (Figure 5B). Consistent with these functional benefits, cilostazol treatment, administered either before or after ICH, induced dural lymphatic hyperplasia in the injured meninges, as evidenced by increased lymphatic branch number and coverage area (Figure 5C; Figure S7). Delayed cilostazol treatment from days 3 to 6 post-ICH also significantly enhanced fluorescent tracer outflow to the dCLNs (Figure 5D), indicating improved CNS lymphatic drainage. Together, these results demonstrate that cilostazol does not exacerbate primary hemorrhage or hematoma expansion, and instead improves hematoma clearance and neurological recovery through promotion of dural lymphatic remodeling and function.

Cilostazol Alleviates Long-Term Residual RBC Burden, Brain Damage, and Behavioral Deficits After ICH

Having established that cilostazol did not provoke hematoma enlargement or delayed hematoma expansion, we next asked whether 20 mg/kg cilostazol provides long-term benefit beyond the acute phase of ICH (Figure 6A). At day 30 after ICH, Ly76-positive residual RBC burden (hematoma volume) was markedly reduced in cilostazol-treated mice compared with vehicle-treated controls (Figure 6B). Cilostazol treatment also reduced brain injury volume, as assessed by Luxol fast blue staining (Figure 6B). Functionally, cilostazol-treated mice showed sustained improvement in neurological severity scores from day 3 through day 21 after ICH compared with vehicle-treated mice (Figure 6C). When assessed for long-term cognitive functions, vehicle-treated ICH mice exhibited impaired performance in the novel object recognition test, Y-maze forced alternation test, and social novelty test compared with healthy controls. Although sociability was comparable between groups, cilostazol treatment improved performance in these assays, indicating partial recovery of object recognition memory, spatial working memory, and social recognition memory (Figure 6D through 6F). Together, these findings indicate that cilostazol treatment reduces long-term residual RBC burden and brain tissue injury and is associated with sustained improvements in neurological and cognitive outcomes after ICH.

Figure 6.

Cilostazol treatment enhanced long-term intracerebral hemorrhage (ICH) recovery. A, Schematic overview of the experimental design and subsequent analytical methods. B, Representative Ly76 immunofluorescence images and quantification of Ly76-positive residual RBC volume in Veh- and CSZ20-treated ICH mice at day 30 (left). Representative images and quantification of Luxol fast blue staining in Veh- and CSZ20-treated ICH mice at day 30 (right). Dotted lines delineate regions of injury. C, Neurological severity score of healthy controls (HC) and Veh- and CSZ20-treated ICH mice at days 1, 3, 7, 14, and 21 post-ICH. HC n=15, Veh, n=12, CSZ20 n=12. D, Novel object recognition test of HC-, Veh-, and CSZ20-treated ICH mice at day 30 post-ICH. HC n=15, Veh, n=12, CSZ20 n=12. E, Y-maze (forced alternation) test of HC, Veh-, and CSZ20-treated ICH mice at day 30 post-ICH. HC n=15, Veh, n=12, CSZ20 n=12. F, Three-chamber social interaction test assessing sociability and social novelty in HC and Veh- and CSZ20-treated ICH mice at day 30 post-ICH. HC n=15, Veh, n=12, CSZ20 n=12. Data are presented as box plots that show minima, maxima, interquartile range (box bounds), and median (black line). Each circle represents 1 mouse. Student t test with Welch correction (B); repeated-measures 2-way ANOVA with Dunnett correction (C); Kruskal-Wallis test with Dunn test (D and F); 1-way ANOVA with Dunnett test (E). Created in BioRender. Jiang, D. (2026) https://BioRender.com/2f4dtyw.

Discussion

ICH remains the most serious and least treatable form of stroke,²⁸ and current pharmacological and surgical interventions largely focus on limiting hematoma-originated damage in the brain.^6,30,31 In this study, we report 4 principal findings. First, oral cilostazol administered at doses of 20 and 40 mg/kg per day, corresponding to clinically used doses of 100 and 200 mg/d in humans, enhanced dural lymphatic function, facilitated intraparenchymal hematoma clearance, preserved neuronal and synaptic integrity, and improved neurobehavioral recovery after ICH. Notably, cilostazol at 20 mg/kg per day also reduced long-term RBC burden and brain injury, accompanied by improved cognitive outcomes in the chronic phase after ICH. Second, cilostazol exerted a pro-lymphangiogenic effect, promoting dural LEC proliferation and lymphatic vasculature hyperplasia. Third, cilostazol pretreatment did not increase hematoma volume. Finally, delayed cilostazol post-treatment did not induce hematoma expansion. Together, these findings extend prior observations supporting the therapeutic potential of cilostazol in hemorrhagic cerebrovascular conditions,^32–34 adding to the broader evidence that improving the speed and efficiency of hematoma resolution may favorably influence ICH outcomes.^3–5

Mass effect from hematomas disrupts neural architecture and causes brain cell loss in the perihematomal region. The subsequent iron toxicity and inflammatory cascades triggered by RBC metabolites further contribute to neurological complications and poor clinical outcomes.^28,35,36 Strategies to enhance brain hematoma clearance either through surgical evacuation or boosting microglia/macrophage phagocytic and sequestration functions are, therefore, a major focus in ICH research. Although targeting professional phagocyte-mediated hematoma resolution remains at an early translational stage, exploratory analyses from the MISTIE III trial and the recently completed ENRICH trial demonstrated that lower final (postsurgery) hematoma volume correlates with better outcomes.^6,30 These pioneer studies support the broader hypothesis that optimizing strategies aimed at blood product removal may ultimately improve functional recovery and independence in ICH patients.³⁷

Effective treatments for ICH remain inadequate, underscoring critical knowledge gaps in the mechanisms governing the dynamic phases of ICH injury and repair. The CNS lymphatic system and its drainage function have emerged as a key modulator in various CNS diseases.⁹ In line with these notions, we earlier observed that the CNS-to-cervical lymph node lymphatic drainage is impaired at day 3 after ICH. Early enhancement of dural lymphatic drainage facilitates the removal of intracranial debris, metabolites, and hematomas, dampens neuroinflammation, and benefits functional outcomes in the late stage of ICH.^10,38,39 Although restoring dural lymphatic function for ICH treatment remains largely a hypothetical therapeutic concept, the present work, together with other emerging studies, supports the idea that the dural lymphatic system represents a promising avenue for developing pharmacological and noninvasive strategies to improve brain repair and long-term ICH outcomes.^9,10,38,39

Cilostazol, a selective PDE3 inhibitor, is widely used in Asia-Pacific countries for stroke prevention, although its only federally approved indication remains intermittent claudication.¹³ PDE3 inhibition elevates intracellular cAMP that activates PKA (protein kinase A), a signaling axis previously shown to upregulate VEGF-C expression in mouse LECs and promote proliferation and network formation in human lymphatic microvascular endothelial cells.^22,40 This signaling opens the possibility that cilostazol, through activation of cAMP/PKA, may modulate lymphatic biology within the dural meninges. Notably, the extent to which cilostazol influences dural lymphatic vessel development and drainage under physiological conditions has not been previously defined. Although mechanistic work remains necessary to delineate the precise downstream targets, our ex vivo data provide direct evidence that cilostazol exerts a pro-lymphangiogenic effect, as reflected by increased dural LEC proliferation. In addition, the in vivo studies conducted in both healthy and ICH animals further demonstrate that cilostazol treatment augments dural lymphatic network complexity and enhances lymphatic drainage capacity, independent of baseline physiological or pathological status. These observations agree with a previous study showing that cilostazol promotes drainage of macromolecules and reduces Aβ deposition in aged Alzheimer disease mice.⁴¹

Dose-response analyses further revealed that daily treatment with 20 or 40 mg/kg cilostazol was required to achieve functionally meaningful enhancement of dural lymphatic drainage and hematoma resolution. In contrast, 10 mg/kg produced only modest structural lymphatic remodeling without corresponding gains in neurological recovery, indicating only rudimentary structural remodeling. However, these low-dose findings reinforce the notion that cilostazol possesses intrinsic prolymphangiogenic activities at the dural meninges. Together, these observations suggest that cilostazol acts not only as a vasculoprotective and antiplatelet agent but also as a modulator of dural lymphatic remodeling and function, thereby aiding hematoma resolution and ICH recovery. These results establish a mechanistic basis for cilostazol’s capacity to enhance dural lymphatic growth and drainage, providing a foundation for its therapeutic utility in ICH and other CNS diseases characterized by pathogenic substance deposition.

Approximately half of the first-time ICH occur in patients receiving ongoing oral antithrombotic drugs, and both larger baseline hematoma volume and greater hematoma expansion are well-recognized prognostic factors of poor clinical outcomes.^2,36 Given these considerations, the antiplatelet effects of cilostazol raise understandable concerns regarding its clinical translatability for ICH treatment. Clinically, hematoma expansion typically occurs during the first several hours of onset.^7,28 In the collagenase ICH model, hematoma growth continues for at least 24 hours before stabilizing.⁵ Leveraging this temporal profile, we evaluate the safety of cilostazol with respect to the risk of hematoma growth using the collagenase ICH, under both pretreatment and delayed-treatment paradigms.

Neither pretreatment nor post-treatment increased hematoma volume; instead, cilostazol significantly reduced hematoma size. These findings are consistent with converging clinical and preclinical evidence demonstrating the potential therapeutic efficacy of cilostazol in hemorrhagic cerebrovascular conditions. Cohort studies in subarachnoid hemorrhage indicated that cilostazol does not increase hemorrhagic complications and may even confer benefit.^42,43 Additionally, cilostazol was associated with a reduced risk of secondary hemorrhagic stroke in the CSPS 2 trial (Cilostazol for Prevention of Secondary Stroke)⁴⁴ and has been shown to protect against hemorrhagic transformation and poststroke ICH.⁴⁵ Consistent with a prior study showing that pretreatment with a human-equivalent dose of 100 mg/d reduces acute hematoma volume in an ICH mouse model,³² our results corroborate and extend those findings by demonstrating that 40 mg/kg cilostazol (equivalent to 200 mg/d in humans) poses no detrimental effect on hematoma expansion and is effective for hematoma resolution. The pleiotropic effects of cilostazol, including preservation of blood-brain barrier integrity, protection of endothelial function, and antithrombotic function, may collectively contribute to the observed reduction of hematoma burden.⁴⁶ Although additional studies using next-generation sequencing approaches, such as single-cell or single-nucleus RNA-seq, will help delineate cell-type-specific actions of cilostazol, the present work together with previous literature supports the conclusion that cilostazol does not increase hematoma size and is not likely to exert adverse impact on hematoma expansion.⁴⁷

Notably, delayed cilostazol treatment initiated 3 days after ICH induction remained effective in reducing hematoma volume, although improvement in certain behavioral domains was limited. This suggests that some functional outcomes may require earlier interventions or longer treatment duration. Importantly, the efficacy of delayed treatment argues against cilostazol interfering with collagenase-mediated hemorrhage induction and supports the existence of a clinically relevant therapeutic window. These features enhance cilostazol’s translational appeal and suggest compatibility with current clinical management strategies, including surgery. This observation also highlights the need for future investigations to define the optimal therapeutic duration and timing of cilostazol treatment for ICH recovery.

Several limitations warrant consideration. First, our histological analyses focused on dorsal dural lymphatic vessels. Therefore, whether basal or nasal lymphatic networks respond similarly remains unknown. Potential systemic effects of pharmacologically modulating dural lymphatics on peripheral lymphatic vasculature have yet to be evaluated. In addition, although the bead drainage experiments demonstrated enhanced tracer delivery to the dCLNs and increased uptake along dural lymphatic vessels, alternative routes, including cribriform plate and perineural pathways, may also contribute to the clearance of CSF- and blood-derived materials. Future studies incorporating tissue-clearing imaging, cribriform plate analysis, and route-specific lymphatic blockade will help determine the relative contributions of individual drainage pathways to hematoma clearance after ICH. Second, although ex vivo experiments demonstrated direct pro-lymphangiogenic effects, our investigation primarily relied on histological assessments. Future studies are needed to map the molecular pathways through which cilostazol promotes LEC proliferation in the context of ICH. This information would be useful for anticipating potential adverse effects associated with long-term cilostazol treatment in patients. Third, studies were conducted exclusively in male mice; future work should examine sex- and age-dependent responses and incorporate common ICH comorbidities, including hypertension, cerebral amyloid angiopathy, and diabetes, to enhance translational relevance. Lastly, recent pharmacovigilance analyses have reported signals of hemorrhagic transformation or ICH when cilostazol is combined with alteplase.⁴⁸ Therefore, our findings should not be interpreted as evidence that cilostazol is universally safe across all hemorrhagic or thrombolytic clinical contexts. Given that the collagenase-induced ICH model reflects progressive microvascular bleeding and may not fully recapitulate larger-vessel bleeding in human ICH, careful safety monitoring remains warranted in future clinical studies.

Conclusions

Oral cilostazol did not exacerbate bleeding and induce hematoma expansion under the dosing regimens tested and may present a mechanistically informed therapeutic strategy to enhance dural lymphatic drainage, accelerate hematoma resolution, and improve neurological recovery after ICH. By targeting CNS-to-cervical lymph node lymphatic function, this approach addresses a critical unmet need in ICH treatment. The present findings provide essential mechanistic and dosing rationale supporting ongoing and future randomized clinical trials, including an ongoing phase II study that is evaluating cilostazol for hematoma clearance after ICH.

ARTICLE INFORMATION

The graphic abstract was created in BioRender. Jiang, D. (2026) https://BioRender.com/2f4dtyw.

The podcast and transcript are available at https://www.ahajournals.org/str/podcast.

Acknowledgments

The authors thank the Imaging Core and Flow Cytometric Analysis and Sorting Core of the First Core Lab at the National Taiwan University College of Medicine for technical support in image acquisition, flow cytometry, and data analysis.

Disclosures

None.

Supplemental Material

ARRIVE Checklist

Supplemental Materials and Methods

Figures S1–S10

References 10, 17, 24, 26, 49, 50

Funding Statement

This work was supported by the National Science and Technology Council of Taiwan (NSTC 112-2628-B-002-027, NSTC 113-2628-B-002-007, NSTC 114-2628-B-002-006, and NSTC 114-2320-B-002-061 to Dr Chang; NSTC 112-2923-B-002-001-MY3 and NSTC 113-2628-B-002-013-MY3 to Dr Tsai), National Taiwan University (112L7846, 113L7829, and 114L7808 to Dr Chang), and National Taiwan University Hospital Bei-Hu Branch (112-115T001 to Dr Tsai). Dr McCullough was supported by the American Heart Association (24IAUST1196589) and the National Institutes of Health (R35NS132265).