Skip to content

Treatment

TL;DR — No drug reverses established vascular brain injury. A Cochrane network meta-analysis of eight trials/4,373 participants found donepezil 5 mg changed ADAS-Cog by −0.92 points, donepezil 10 mg by −2.21, galantamine by −2.01, and rivastigmine by 0.03; the review judged the significant effects unlikely to be clinically important (Battle 2021, PMID 33704781). Donepezil 10 mg and galantamine increased adverse events (OR 1.95 and 1.57) (Battle 2021, PMID 33704781). As of that 2021 review, no pharmacological treatment was recommended to improve cognition or function in VCI; the 2011 AHA/ASA statement had already noted that no specific VCI treatment was FDA-approved (Battle 2021, PMID 33704781; Gorelick 2011, PMID 21778438; O'Brien 2015, PMID 26595643). Treatment is therefore a bundle: prevent further injury, treat contributing conditions, consider a time-limited symptomatic trial, rehabilitate deficits, and support function and caregivers.

Cholinesterase-inhibitor evidence

Drug/dose Cognitive effect vs placebo Adverse events Certainty/interpretation
donepezil 5 mg ADAS-Cog MD −0.92 (−1.44 to −0.40) OR 1.22 (0.94–1.58) high/moderate; slight
donepezil 10 mg MD −2.21 (−3.07 to −1.35) OR 1.95 (1.20–3.15) moderate/high; benefit–harm tradeoff
galantamine 16–24 mg MD −2.01 (−3.18 to −0.85) OR 1.57 (1.02–2.43) moderate; slight
rivastigmine 3–12 mg MD 0.03 (−3.04 to 3.10) OR 3.21 (0.36–28.88) low/very low; uncertain

All values are from Battle et al. (2021, PMID 33704781). Eight trials enrolled possible/probable VaD or other VCI (donepezil n=2,193; rivastigmine n=800; galantamine n=1,380) without modern biomarker separation of mixed Alzheimer disease. Mean ages were 72.2–73.9 years. Search date was 19 August 2020.

Individual trials

Trial N/duration Finding
Wilkinson donepezil 616/24 weeks ADAS-Cog differences −1.65 (5 mg) and −2.09 (10 mg); CIBIC-plus improved in 39% (5 mg) and 32% (10 mg) vs 25% placebo; AE withdrawal 8.8% placebo, 10.1% 5 mg, 16.3% 10 mg (Wilkinson 2003, PMID 12939421)
Román donepezil 974/24 weeks V-ADAS-Cog LS difference −1.156 (−1.98 to −0.33); global coprimary not significant; 11 deaths on donepezil 5 mg vs 0 on placebo (Román 2010, PMID 20395618)
Auchus galantamine 788/26 weeks ADAS-Cog −1.8 vs −0.3; ADL no difference (0.7 vs 1.3); AE discontinuation 13% vs 6% (Auchus 2007, PMID 17664404)
Mok rivastigmine 40/26 weeks no significant efficacy outcome; withdrawal 30% vs 15% (Mok 2007, PMID 19300631)
Black donepezil (307 study) 603/24 weeks ADAS-Cog effect size −1.90 (5 mg, P=0.001) and −2.33 (10 mg, P<0.001); CIBIC-plus significant for 5 mg only (P=0.014), CDR-SB for 10 mg only (P=0.007); ADFACS −1.31 for both doses (P=0.02); AE withdrawal 11.1% placebo, 11.1% 5 mg, 21.8% 10 mg (P=0.005) (Black 2003, PMID 12970516)
Wilkinson open-label extension 885 enrolled of 1,219 eligible / 30 weeks after the two 24-week trials ADAS-Cog reduction 0.6–1.15 points from double-blind baseline at week 54 in continuous-donepezil patients; late starters did not catch up; 14.4% discontinued for adverse events (Wilkinson 2010, PMID 19623601)

Black 2003 and Wilkinson 2003 are the two sibling 24-week donepezil trials (studies 307 and 308) on which the drug's vascular-dementia case originally rested, and Black is the one that also showed a functional benefit — ADFACS −1.31 at both doses — which Wilkinson did not (Black 2003, PMID 12970516; Wilkinson 2003, PMID 12939421). The pattern across both is that global and functional coprimaries separated inconsistently and at different doses, which is precisely the inconsistency that stopped a regulatory indication. The 30-week open-label extension adds a point that is easy to over-read in either direction: patients on donepezil for 54 continuous weeks stayed 0.6–1.15 ADAS-Cog points below their double-blind baseline, and patients who started donepezil only in the extension did not catch up (Wilkinson 2010, PMID 19623601). Non-catch-up in an open-label extension is compatible with a disease-modifying interpretation and equally with irreversible decline during the placebo period plus selection of completers; the design cannot distinguish them, and 14.4% withdrew for adverse events over 30 weeks.

Román 2010 used only donepezil 5 mg (2:1 randomization) and is therefore not a 10 mg confirmation of Wilkinson (Román 2010, PMID 20395618). These results explain why statistical cognitive differences did not translate into a robust vascular-dementia indication.

Memantine

The vascular-dementia memantine evidence rests on two 28-week sibling trials run in parallel in France and the UK, both using memantine 10 mg twice daily against placebo in probable VaD.

Trial N (randomized / ITT) Entry MMSE ADAS-Cog result Global result
MMM 300 (France) 321 / 288 12–20 memantine +0.4 points gained vs placebo −1.6 points lost; difference 2.0 (95% CI 0.49–3.60) CIBIC-plus improved-or-stable 60% vs 52%, P=0.227 (Orgogozo 2002, PMID 12105362)
MMM500 (UK, 54 centres) 579 / 548 10–22 change from baseline differed by −1.75 points (95% CI −3.02 to −0.49) CGI-C no significant difference (Wilcock 2002, PMID 12409683)

Both trials therefore separated on the cognitive coprimary and failed on the global coprimary — the same pattern that later sank the donepezil program (Román 2010, PMID 20395618). Tolerability was close to placebo: 77% of memantine and 75% of placebo participants had any adverse event in MMM500, dizziness 11% vs 8% (Wilcock 2002, PMID 12409683). Pooling the two trials, a 2026 VaD-specific meta-analysis put memantine at ADAS-Cog/11 MD −2.20 (95% CI −3.24 to −1.15; 2 trials, n=752; moderate certainty) with no benefit on the Gottfries-Brane-Steen global scale (MD −1.93, 95% CI −4.69 to 0.84; n=595) (Cheng 2026, PMID 42635820). The dementia-wide Cochrane memantine review includes these vascular data but does not establish consistent functional benefit for pure VaD (McShane 2019, PMID 30891742). Two older summaries of the same program report the subgroup claims by severity and by small- versus large-vessel imaging (Möbius 2003, PMID 16191242).

Memantine's distinguishing feature in VCI is acceptability rather than efficacy: in a dose-resolved network meta-analysis of 11 double-blind RCTs it ranked best on tolerability while matching galantamine on cognition (Shi 2022, PMID 35048806).

How large are the drug effects, really — two incompatible readings

Three independent quantitative syntheses of overlapping trial sets disagree by roughly a factor of five on the size of the same effects, and the disagreement is methodological, not empirical.

Synthesis Method Donepezil 5 mg vs placebo, cognition Galantamine vs placebo, cognition
Cochrane (Battle 2021, PMID 33704781) pairwise, raw ADAS-Cog points MD −0.92 (−1.44 to −0.40) MD −2.01 (−3.18 to −0.85)
Frequentist NMA (Shi 2022, PMID 35048806) standardized mean differences SMD −1.11 (−1.88 to −0.34) SMD −1.99 (−3.03 to −0.95) for 24 mg
VaD-specific meta-analysis (Cheng 2026, PMID 42635820) scale-specific, RoB 2 + GRADE (not pooled) MD −2.01 (−3.18 to −0.85) on ADAS-Cog/11; −2.51 (−3.87 to −1.15) on ADAS-Cog/13

An SMD near −1.1 corresponds to a large effect by conventional benchmarks, whereas the same comparison expressed in raw units is under one ADAS-Cog point. Because the two analyses draw on largely the same trials, the divergence must come from the standardization denominator rather than from new data, and readers should treat the SMD ranking as a within-network ordering, not as an effect magnitude. Shi 2022 also credits donepezil 10 mg with executive and global benefit (CDR-SB −0.25, 95% CI −0.44 to −0.06; EXIT25 −1.47, 95% CI −2.79 to −0.15) that the Cochrane review does not endorse (Shi 2022, PMID 35048806; Battle 2021, PMID 33704781). This is the single most consequential unresolved measurement dispute in VaD pharmacotherapy: it decides whether the drugs are marginal or moderately useful.

The most recent VaD-restricted synthesis found only 16 placebo-controlled parallel-group trials totalling 5,668 participants, and all but one were published in 2012 or earlier — the field has essentially stopped running placebo-controlled VaD drug trials (Cheng 2026, PMID 42635820).

The Western-null versus Asia-positive split

Reviews restricted to English-language, low-risk-of-bias trials conclude that nothing is worth recommending (Battle 2021, PMID 33704781; Cheng 2026, PMID 42635820). A Bayesian network meta-analysis that added CNKI, Wanfang, and grey literature assembled 194 RCTs across 21 anti-VaD drugs and ranked agents that appear nowhere in the Western guideline literature at the top: on MMSE, butylphthalide, huperzine A, edaravone, rivastigmine, and memantine; on activities of daily living, huperzine A, butylphthalide, Tianzhi granule, nicergoline, and idebenone (Dang 2024, PMID 39239652). A second network meta-analysis of 16 studies (5,599 participants) ranked the Chinese herbal compound sailuotong first on ADAS-Cog (MD −3.00, 95% CI −4.50 to −1.50; SUCRA 88.5%) and memantine first on MMSE (MD 1.23, 95% CI 0.23–2.23; SUCRA 80.8%), while flagging the network as sparse and the underlying trials as small (Li 2025, PMID 41404461).

The two literatures are not reconcilable by simply averaging them. The Asian networks contain hundreds of small, often single-centre, frequently unblinded trials whose pooled effects would be practice-changing if real; the Western reviews exclude most of them on risk-of-bias grounds and are left with almost nothing. Neither position has been tested by an adequately powered independent replication outside its own region, and this is a live gap rather than a settled question.

Drugs with a positive pivotal trial and no confirmation

Agent Pivotal result Status
Cerebrolysin (porcine brain peptide, IV) 242 VaD participants; ADAS-Cog+ improved 10.6 vs 4.4 points (LS mean difference −6.17, P<0.0001); CIBIC+ 2.84 vs 3.68 (difference 0.84, P<0.0001); combined responders 67.5% vs 27.0% (Guekht 2011, PMID 20656516) Cochrane identified 6 trials/597 participants but the cognitive pooling is 3 trials/420 people (SMD 0.36, 0.13–0.58) and the global-response pooling is 2 trials/379 (RR 2.69, 1.82–3.98), both very low quality; no new eligible trial appeared between the 2013 and 2019 reviews; where funding was reported it was industry (Cui 2019, PMID 31710397)
Actovegin (deproteinized calf-blood haemodialysate) ARTEMIDA: 503 patients randomized within 1 week of supratentorial ischaemic stroke with MoCA ≤25; ADAS-Cog+ change at 6 months −6.8 vs −4.6, treatment difference −2.3 (95% CI −3.9 to −0.7; P=0.005) (Guekht 2017, PMID 28432265) Recurrent ischaemic stroke was the most frequent serious adverse event and was non-significantly more common on active drug; the authors themselves called for confirmatory trials, and none has reported (Guekht 2017, PMID 28432265)
DL-3-n-butylphthalide (NBP) 281 participants with subcortical VCI-no-dementia at 15 Chinese centres, 200 mg tid × 24 weeks; ADAS-Cog change −2.46 vs −1.39 (P=0.03); CIBIC-plus improved in 57.1% vs 42.1% (P=0.01) (Jia 2016, PMID 26086183) A Cochrane review of antithrombotic therapy included Jia 2016 because of NBP's putative antiplatelet effect and rated that result very low certainty (adjusted MD −1.07, 95% CI −2.02 to −0.12 on ADAS-Cog-12), judging the difference possibly not clinically relevant; MMSE, CDR, and function showed no difference (Kwan 2022, PMID 35833913)
Ginkgo biloba EGb 761 In dementia overall, 6-month Cochrane estimates favour ginkgo on the Syndrom-Kurztest (MD −1.86, 95% CI −3.48 to −0.24; 9 studies, 2,801 participants) and on the ADL International Scale (MD −0.19, 95% CI −0.35 to −0.03; 8 studies, 2,571) (Wieland 2026, PMID 41641880) Heterogeneity is extreme (I² 96% and 91%) and certainty low; in MCI, ginkgo probably has no effect on CDR (MD −0.03, 95% CI −0.06 to 0.01) or ADAS-Cog (MD −0.07, 95% CI −0.67 to 0.51) at 6 months (moderate certainty). The Cochrane dementia global-clinical-status estimate (MD −0.06, 95% CI −1.00 to −0.20) is internally inconsistent — the point estimate lies outside the stated interval — and is not used here. The VaD-restricted pooling gives SKT MD −2.65 (95% CI −5.17 to −0.12) with I² 92.8% and very low certainty (Cheng 2026, PMID 42635820)

Serious-adverse-event risk for ginkgo is the best-characterised safety datum in this table: RR 0.95 (95% CI 0.82–1.09) in MCI at up to 12 months, high-certainty evidence (Wieland 2026, PMID 41641880). The pattern across the table is consistent — a single large positive trial, an unreplicated effect size several times larger than anything cholinesterase inhibitors achieve, and a systematic review that downgrades it to very low certainty rather than refuting it. Absence of confirmation is not evidence of absence of effect; it is evidence that nobody has funded the confirmation.

Calcium-channel and cholinergic-precursor combinations

A Cochrane review of nimodipine for primary degenerative, mixed, and vascular dementia found mixed published evidence and did not establish a routine VCI indication (López-Arrieta 2000, PMID 10796495). The one modern test, the CONIVaD pilot, randomized 62 patients with SVD and mild-to-moderate cognitive impairment to nimodipine 30 mg tid plus either choline alphoscerate 600 mg bid or placebo for 12 months; the primary endpoint (≥2-point MoCA loss) and every secondary endpoint were null (Salvadori 2021, PMID 33855653). Its most informative result is a feasibility finding: 22% dropped out, and while 96% took more than three-quarters of the assigned choline alphoscerate, only 15% did so for nimodipine — a thrice-daily regimen in this population is close to untestable (Salvadori 2021, PMID 33855653). A systematic map of 118 VCI studies (19,223 participants, 63 intervention types, 125 outcome measures) concluded that trial design remains too heterogeneous to rank treatments (Masserini 2023, PMID 37539725).

Exercise: the only non-drug intervention with randomized VCI-specific trials

Every trial above tests a drug. The two randomized trials that recruited subcortical vascular cognitive impairment specifically and tested exercise reach small positive results with the same caveats as the drug literature — and, notably, effect sizes in the same range.

Trial Design Primary result
Aerobic training (Liu-Ambrose 2016, PMID 27760869, NCT01027858) 70 adults with mild subcortical ischaemic VCI (mean age 74, 51% female), 6-month thrice-weekly progressive aerobic training vs usual care plus education, single-blind, with 6-month post-cessation follow-up ADAS-Cog −1.71 points (95% CI −3.15 to −0.26, P=0.02) at 6 months, not sustained at 12 months (−0.63, −2.34 to 1.07, P=0.46); no effect on EXIT-25 or ADCS-ADL; 6-minute walk +30.35 m (5.82–54.86) and diastolic BP −6.89 mm Hg (−12.52 to −1.26)
Progressive resistance training (Liu-Ambrose 2026, PMID 41795685, NCT02669394) 91 adults with SVD and MCI randomized to 12 months of progressive resistance training vs balance-and-tone control; 76 completed ADAS-Cog-Plus −0.18 (95% CI −0.35 to −0.01, P=0.04); effect present in females (−0.27, −0.49 to −0.05, P=0.02) but not males; C-reactive protein −2.93 (−5.36 to −0.49, P=0.02); no other secondary outcome significant

Three observations. The aerobic ADAS-Cog effect of −1.71 points is numerically within the range of pooled cholinesterase-inhibitor estimates, but populations, controls and trial methods differ, and no head-to-head trial has compared exercise with a cholinesterase inhibitor. Second, the between-group aerobic effect was no longer statistically detectable six months after the intervention stopped; this does not by itself establish a maintenance requirement. Third, the resistance-training trial used an active comparator (balance and tone) rather than usual care, which may contribute to its smaller estimate, although ADAS-Cog-Plus is not directly interchangeable with ADAS-Cog. The sex-specific effect is unexplained and, on a subgroup of 91 participants, provisional.

Citicoline is the other non-cholinergic agent with a dedicated post-stroke cognitive trial. In an open-label randomized parallel study, 347 first-ever ischaemic stroke patients (mean age 67.2) were assigned 6 weeks post-stroke to citicoline 1 g/day for 12 months or no citicoline, with neuropsychological evaluation at 1, 6 and 12 months. Citicoline-treated patients had better outcome in attention–executive function (OR 1.721, 95% CI 1.065–2.781 at 6 months; OR 2.379, 1.269–4.462 at 12 months) and temporal orientation (OR 1.780, 1.020–3.104; OR 2.155, 1.017–4.566), with better but non-significant functional outcome (mRS ≤2: 57.3% vs 48.7%, P=0.186) and 10.7% discontinuation (Alvarez-Sabín 2013, PMID 23406981). The trial was open-label with a no-treatment control and only 199 of 347 completed the 1-year neuropsychological assessment, so the odds ratios above carry both performance and attrition bias; it belongs in the same category as the Cerebrolysin and Actovegin results in the table above — a positive unreplicated signal in a field that has stopped funding confirmation.

Practical symptomatic-trial framework

Step Action
1 define target: cognition, behavior, or function
2 review bradycardia, syncope, weight, GI, interactions
3 discuss uncertain clinical magnitude and off-label status
4 obtain baseline patient/informant measure
5 titrate and monitor
6 stop if harms or no meaningful benefit

This is research synthesis, not individualized prescribing advice.

Antithrombotics and vascular treatment

Antiplatelets and anticoagulants prevent indicated vascular events; they are not cognitive enhancers. In SPS3, among 2,916 recent lacunar-stroke participants followed a median 3.0 years, neither dual antiplatelet therapy nor a lower systolic target (<130 vs 130–149 mm Hg) changed CASI Z-score trajectories (Pearce 2014, PMID 25453457). ASPREE found no dementia, MCI, or cognitive-decline benefit of 100 mg aspirin versus placebo in 19,114 older adults over a median 4.7 years (HR for dementia triggers 1.03, 95% CI 0.91–1.17) (Ryan 2020, PMID 32213642). WMH alone is not an indication for aspirin (Wardlaw 2021, PMID 34414301).

Anticoagulation in AF requires net-benefit assessment, particularly with lobar microbleeds or CAA (Charidimou 2017, PMID 29117953).

Mechanism-targeted SVD drugs

LACI-2 randomized 363 people with lacunar stroke to isosorbide mononitrate (ISMN), cilostazol, both, or neither for 1 year (NCT03451591). Recruitment and 12-month retention (358/363, 98.6%) met feasibility; 257/272 (94.5%) took ≥50% of allocated drug (Wardlaw 2023, PMID 37222252).

Contrast Effect (95% CI)
ISMN alone, composite aHR 0.80 (0.59–1.09)
cilostazol alone, composite aHR 0.77 (0.57–1.05)
ISMN, recurrent stroke aOR 0.23 (0.07–0.74)
ISMN, cognitive impairment aOR 0.55 (0.36–0.86)
cilostazol, dependence aHR 0.31 (0.14–0.72)
combination, composite aHR 0.58 (0.36–0.92)
combination, any cognitive impairment aOR 0.44 (0.23–0.85)

These secondary signals require confirmatory trials before practice change (Wardlaw 2023, PMID 37222252). A prespecified 6-month analysis showed the separation appears early rather than only at 1 year: ISMN versus control reduced the composite (adjusted OR 0.74, 95% CI 0.55–0.99) and improved stroke impact (Mann-Whitney difference −0.15, 95% CI −0.25 to −0.05); cilostazol improved cognition on the DSM-5 seven-level scale (adjusted common OR 0.64, 95% CI 0.41–0.99); and the combination improved cognition (adjusted common OR 0.40, 95% CI 0.21–0.78), mood (Zung mean difference −6.94, 95% CI −12.25 to −1.64), and global stroke impact (Mann-Whitney difference −0.23, 95% CI −0.37 to −0.09) (Bath 2026, PMID 42535538). A ClinicalTrials.gov query on 2026-08-31 found no registered phase-3 successor to LACI-2 under that registry; LACI-2 itself is registered as ISRCTN14911850 (Bath 2026, PMID 42535538).

The cilostazol case illustrates how far a drug can travel on regional evidence. Twenty randomized trials totalling 10,505 participants — almost all conducted in Asia-Pacific countries — show cilostazol reduces recurrent ischaemic stroke (OR 0.68, 95% CI 0.57–0.81), haemorrhagic stroke (OR 0.43, 95% CI 0.29–0.64), death (OR 0.64, 95% CI 0.49–0.83), and systemic bleeding (OR 0.73, 95% CI 0.54–0.99), at the cost of headache and palpitations; the benefit was larger in trials with >40% lacunar stroke and with treatment longer than 6 months, but the data were explicitly insufficient to judge effects on cognition or imaging (McHutchison 2020, PMID 32646330). Against that, the COMCID trial randomized 159 people with MCI (MMSE 22–28, CDR 0.5) to cilostazol 50 mg twice daily or placebo for up to 96 weeks and found no cognitive benefit: least-squares mean MMSE change at 96 weeks −1.8 with cilostazol versus −1.3 with placebo, with one subdural haematoma on active drug (Saito 2023, PMID 38048134). A drug with a robust stroke-prevention signal and a null dedicated cognition trial is exactly the profile that makes "vascular prevention equals cognitive prevention" an assumption rather than a finding.

Nitric-oxide-pathway drugs are being probed for cerebrovascular physiology rather than cognition. PASTIS found no significant single-dose tadalafil 20 mg effect on subcortical CBF in 55 people with symptomatic SVD (greatest WMH increment +9.8%, P=0.096) (Pauls 2022, PMID 35135037). The OxHARP three-way crossover trial (75 enrolled, 65 with valid primary data) tested sildenafil 50 mg thrice daily against placebo and against cilostazol: cerebral pulsatility, the primary outcome, was unchanged (0.02, 95% CI −0.01 to 0.05; P=0.18), but sildenafil improved cerebrovascular reactivity on transcranial ultrasound (0.83 cm/s per mm Hg, 95% CI 0.23–1.42; P=0.007) and within white-matter hyperintensities on BOLD MRI (0.07, 95% CI 0–0.14; P=0.043), and increased perfusion inside WMH (1.82 mL/100 g/min, 95% CI 0.5–3.15; P=0.008) (Webb 2024, PMID 38832504). The trial is a clean demonstration that the mechanistic target chosen for SVD drug development — pulsatility versus reactivity versus perfusion — determines whether a drug looks active or inert.

Rehabilitation and nonpharmacologic management

Intervention Target Evidence boundary
cognitive rehabilitation task-specific deficits/compensation heterogeneous small trials
exercise fitness, vascular health, cognition ACSM-adherent protocols: cognition SMD 0.53 (0.13–0.94); low/indeterminate adherence SMD 0.09 (−0.13 to 0.32) (Ye 2024, PMID 39666076)
occupational therapy routines, safety, environment function-centered
speech-language therapy aphasia/communication stroke phenotype-specific
caregiver education cueing, routines, burden essential but under-measured
sleep/mood treatment reversible amplifiers treat comorbidity, not lesion

Fourteen exercise trials (1,333 participants) were too protocol-heterogeneous for a single ranking (Ye 2024, PMID 39666076).

Behavioral and psychological symptoms

Before medication, assess pain, infection, sleep, delirium, sensory loss, environment, and caregiver interaction. Dementia-wide randomized data show atypical antipsychotics increase all-cause death (118/3,353 [3.5%] vs 40/1,757 [2.3%]; OR 1.54, 95% CI 1.06–2.23) over generally 10–12 weeks (Schneider 2005, PMID 16234500). Pooled risperidone/olanzapine dementia trials also raised cerebrovascular adverse-event reporting; observational comparisons versus typical agents are less consistent (Herrmann 2005, PMID 15697324). A 2026-08-31 PubMed search for vascular-dementia-specific behavioral RCTs identified one small trial: 30 older adults with VaD and clinically significant BPSD received cannabidiol 300 mg/day or placebo for 4 weeks, with reductions in NPI (interaction p=0.05) and BPRS (p<0.05), no cognitive or functional change, and mild adverse events (Pessoa 2026, PMID 41277041). That signal is not a basis for practice. Antipsychotic death and stroke estimates still come from mixed-dementia samples and should be treated as relevant safety evidence rather than VaD-specific efficacy.

Regulatory and recommendation status

A live 2026-08-31 PubMed search did not identify a subsequent vascular-dementia-specific pivotal trial that would overturn the Cochrane conclusion of no recommended cognitive drug. Regulator labels were not independently retrieved. Guideline differences on cholinesterase inhibitors are value judgments under small effects, not contradictory large-effect readings (Battle 2021, PMID 33704781).

Open questions

  • Why has no trial compared aerobic exercise (ADAS-Cog −1.71) directly against a cholinesterase inhibitor (−0.92 to −2.01) in subcortical vascular cognitive impairment? (Liu-Ambrose 2016, PMID 27760869; Battle 2021, PMID 33704781)
  • Is the loss of the aerobic-exercise effect within six months of stopping a reason to frame exercise as maintenance therapy rather than a treatment course? (Liu-Ambrose 2016, PMID 27760869)
  • Is the female-specific resistance-training effect real, and if so is it mediated by the inflammatory change (CRP −2.93)? (Liu-Ambrose 2026, PMID 41795685)
  • Did donepezil non-catch-up in the 54-week extension indicate disease modification or irreversible decline during placebo? (Wilkinson 2010, PMID 19623601)
  • Does citicoline's attention–executive benefit survive a blinded, placebo-controlled replication? (Alvarez-Sabín 2013, PMID 23406981)
  • What patient-important threshold should replace statistical ADAS-Cog significance? (Battle 2021, PMID 33704781)
  • Do Alzheimer-biomarker-positive mixed cases account for cholinesterase response? (Román 2010, PMID 20395618)
  • Can LACI-2 secondary signals for ISMN/cilostazol be confirmed for cognition and dependence? (Wardlaw 2023, PMID 37222252)
  • Which cognitive rehabilitation components transfer to independent function? (Masserini 2023, PMID 37539725)
  • Why do standardized-mean-difference and raw-scale syntheses of the same cholinesterase-inhibitor trials differ roughly fivefold, and which reading should guide practice? (Shi 2022, PMID 35048806; Battle 2021, PMID 33704781)
  • Are the Asia-Pacific network-meta-analysis leaders (butylphthalide, huperzine A, sailuotong) real effects or risk-of-bias artifacts? (Dang 2024, PMID 39239652; Li 2025, PMID 41404461)
  • Would a confirmatory trial of Cerebrolysin or Actovegin replicate effect sizes several times larger than any cholinesterase inhibitor has achieved? (Guekht 2011, PMID 20656516; Guekht 2017, PMID 28432265; Cui 2019, PMID 31710397)
  • Does cilostazol's stroke-prevention benefit translate into cognitive benefit, given a null dedicated MCI trial? (McHutchison 2020, PMID 32646330; Saito 2023, PMID 38048134)

References

  1. Battle CE, et al. Cholinesterase inhibitors for vascular dementia and other vascular cognitive impairments. Cochrane Database Syst Rev. 2021;2:CD013306. PMID 33704781
  2. Wilkinson D, et al. Donepezil in vascular dementia: a randomized, placebo-controlled study. Neurology. 2003;61:479-86. PMID 12939421
  3. Román GC, et al. Randomized, placebo-controlled, clinical trial of donepezil in vascular dementia. Stroke. 2010;41:1213-21. PMID 20395618
  4. Auchus AP, et al. Galantamine treatment of vascular dementia: a randomized trial. Neurology. 2007;69:448-58. PMID 17664404
  5. Mok V, et al. Rivastigmine in Chinese patients with subcortical vascular dementia. Neuropsychiatr Dis Treat. 2007;3:943-8. PMID 19300631
  6. Möbius HJ, Stöffler A. Memantine in vascular dementia. Int Psychogeriatr. 2003;15 Suppl 1:207-13. PMID 16191242
  7. McShane R, et al. Memantine for dementia. Cochrane Database Syst Rev. 2019;3:CD003154. PMID 30891742
  8. López-Arrieta JM, Birks J. Nimodipine for primary degenerative, mixed and vascular dementia. Cochrane Database Syst Rev. 2000;2002:CD000147. PMID 10796495
  9. Charidimou A, et al. Brain microbleeds, anticoagulation, and hemorrhage risk. Neurology. 2017;89:2317-2326. PMID 29117953
  10. Ye Y, et al. Effects of exercise on patients with vascular cognitive impairment based on ACSM recommendations. J Neurol. 2024;272:31. PMID 39666076
  11. Masserini F, et al. Therapeutic strategies in vascular cognitive impairment. Alzheimers Dement. 2023;19:5795-5804. PMID 37539725
  12. Gorelick PB, et al. Vascular contributions to cognitive impairment and dementia. Stroke. 2011;42:2672-713. PMID 21778438
  13. O'Brien JT, Thomas A. Vascular dementia. Lancet. 2015;386:1698-706. PMID 26595643
  14. Wardlaw JM, et al. Isosorbide mononitrate and cilostazol treatment in patients with symptomatic cerebral small vessel disease: LACI-2. JAMA Neurol. 2023;80:682-692. PMID 37222252
  15. Pauls MMH, et al. The PASTIS trial: testing tadalafil for possible use in vascular cognitive impairment. Alzheimers Dement. 2022;18:2393-2402. PMID 35135037
  16. Pearce LA, et al. Effects of long-term blood pressure lowering and dual antiplatelet treatment on cognitive function in patients with cerebral small vessel disease. Lancet Neurol. 2014;13:1177-85. PMID 25453457
  17. Ryan J, et al. Randomized placebo-controlled trial of the effects of aspirin on dementia and cognitive decline. Neurology. 2020;95:e320-e331. PMID 32213642
  18. Wardlaw JM, et al. ESO Guideline on covert cerebral small vessel disease. Eur Stroke J. 2021;6:CXI-CLXII. PMID 34414301
  19. Schneider LS, et al. Risk of death with atypical antipsychotic drug treatment for dementia. JAMA. 2005;294:1934-43. PMID 16234500
  20. Herrmann N, Lanctôt KL. Do atypical antipsychotics cause stroke? CNS Drugs. 2005;19:91-103. PMID 15697324
  21. Orgogozo JM, et al. Efficacy and safety of memantine in patients with mild to moderate vascular dementia: a randomized, placebo-controlled trial (MMM 300). Stroke. 2002;33:1834-9. PMID 12105362
  22. Wilcock G, et al. A double-blind, placebo-controlled multicentre study of memantine in mild to moderate vascular dementia (MMM500). Int Clin Psychopharmacol. 2002;17:297-305. PMID 12409683
  23. Cheng T, et al. Pharmacological interventions targeting cognition and related clinical outcomes in vascular dementia: a systematic review and meta-analysis of randomized controlled trials. Eur J Clin Pharmacol. 2026;82:240. PMID 42635820
  24. Shi X, et al. Comparative efficacy and acceptability of cholinesterase inhibitors and memantine based on dosage in patients with vascular cognitive impairment: a network meta-analysis. Curr Alzheimer Res. 2022;19:133-145. PMID 35048806
  25. Dang C, et al. Pharmacological treatments for vascular dementia: a systematic review and Bayesian network meta-analysis. Front Pharmacol. 2024;15:1451032. PMID 39239652
  26. Li W, et al. Network meta-analysis of the efficacy of pharmacological treatments for post-stroke cognitive impairment and vascular cognitive impairment. Front Neurol. 2025;16:1683496. PMID 41404461
  27. Cui S, et al. Cerebrolysin for vascular dementia. Cochrane Database Syst Rev. 2019;2019:CD008900. PMID 31710397
  28. Guekht AB, et al. Cerebrolysin in vascular dementia: improvement of clinical outcome in a randomized, double-blind, placebo-controlled multicenter trial. J Stroke Cerebrovasc Dis. 2011;20:310-8. PMID 20656516
  29. Guekht A, et al. ARTEMIDA trial: a randomized controlled trial to assess the efficacy of Actovegin in post-stroke cognitive impairment. Stroke. 2017;48:1262-1270. PMID 28432265
  30. Jia J, et al. The effects of DL-3-n-butylphthalide in patients with vascular cognitive impairment without dementia caused by subcortical ischemic small vessel disease. Alzheimers Dement. 2016;12:89-99. PMID 26086183
  31. Kwan J, et al. Antithrombotic therapy to prevent cognitive decline in people with small vessel disease on neuroimaging but without dementia. Cochrane Database Syst Rev. 2022;7:CD012269. PMID 35833913
  32. Wieland LS, et al. Ginkgo biloba for cognitive impairment and dementia. Cochrane Database Syst Rev. 2026;2:CD013661. PMID 41641880
  33. Salvadori E, et al. Efficacy and safety of the association of nimodipine and choline alphoscerate in the treatment of cognitive impairment in patients with cerebral small vessel disease: the CONIVaD trial. Drugs Aging. 2021;38:481-491. PMID 33855653
  34. Saito S, et al. Efficacy and safety of cilostazol in mild cognitive impairment: a randomized clinical trial. JAMA Netw Open. 2023;6:e2344938. PMID 38048134
  35. Bath PM, et al. Isosorbide mononitrate/cilostazol for lacunar cerebral small vessel disease—outcomes at 6 months in the LACI-2 randomized controlled trial. J Am Heart Assoc. 2026;15:e046235. PMID 42535538
  36. McHutchison C, et al. Cilostazol for secondary prevention of stroke and cognitive decline: systematic review and meta-analysis. Stroke. 2020;51:2374-2385. PMID 32646330
  37. Webb AJS, et al. Cerebrovascular effects of sildenafil in small vessel disease: the OxHARP trial. Circ Res. 2024;135:320-331. PMID 38832504
  38. Pessoa RMP, et al. Effects of cannabidiol on behavioral and psychological symptoms of vascular dementia: a randomized, double-blind, placebo-controlled clinical trial. J Psychopharmacol. 2026;40:380-387. PMID 41277041
  39. Black S, et al. Efficacy and tolerability of donepezil in vascular dementia: positive results of a 24-week, multicenter, international, randomized, placebo-controlled clinical trial. Stroke. 2003;34:2323-30. PMID 12970516
  40. Wilkinson D, et al. The long-term efficacy and tolerability of donepezil in patients with vascular dementia. Int J Geriatr Psychiatry. 2010;25:305-13. PMID 19623601
  41. Liu-Ambrose T, et al. Aerobic exercise and vascular cognitive impairment: a randomized controlled trial. Neurology. 2016;87:2082-2090. PMID 27760869
  42. Liu-Ambrose T, et al. Resistance training and subcortical vascular cognitive impairment: a 12-month randomized trial. Alzheimers Dement. 2026;22:e71245. PMID 41795685
  43. Alvarez-Sabín J, et al. Long-term treatment with citicoline may improve poststroke vascular cognitive impairment. Cerebrovasc Dis. 2013;35:146-54. PMID 23406981