Systematic Review on the Efficacy and Safety of Herbal Medicines for Vascular Dementia

We present a systematic review of existing research that aims to assess the efficacy and safety of herbal medications (HM), as either monotherapy or adjunct to orthodox medications (OM), mainly comprised of cholinesterase inhibitors, for vascular dementia (VaD). We included 47 studies conducted in mainland China, each testing different HM. Of 43 HM monotherapy studies, 37 reported HM to be significantly better than OM or placebo; six reported similar efficacy between HM and OM. All four HM adjuvant studies reported significant efficacy. No major adverse events for HM were reported. Heterogeneity in diagnostic criteria, interventions and outcome measures hindered comprehensive data analysis. Studies suggested that HM can be a safe and effective treatment for VaD, either alone or in conjunction with OM. However, methodological flaws in the design of the studies limited the extent to which the results could be interpreted. Thirty most commonly used herbal constituents, including Rhizoma Chuanxiong (Chuanxiong in Chinese), Radix Polygoni Multiflori (Heshouwu in Chinese) and Radix Astragali (Huangqi in Chinese). were ranked. Further multi-center trials with large sample sizes, high methodological quality and standardized HM ingredients are necessary for clinical recommendations to be made.


Introduction
Vascular dementia (VaD) is one of the most common forms of dementia after Alzheimer's disease (AD) [1], and the most frequent cause of dementia in the elderly [2]. First described as arteriosclerotic dementia [3], VaD is defined as loss of cognitive function resulting from ischemic, hemorrhagic brain lesions (such as border zone infarcts and ischemic periventricular leukoencephalopathy) or hypoperfusion, due to cerebrovascular disease or cardiovascular pathology [4]. Incomplete microangiopathic infarcts due to fibrohyalinosis are regarded as the major pathophysiological manifestation [5] of VaD. While AD is characterized by memory impairment, VaD is characterized by executive dysfunction [6] and behavioral psychological symptoms such as apathy, abulia, opposition, agnosia [7], anxiety [8], depression [9], and suicidal thoughts [9]. Cognitive impairment is relatively mild as compared to AD.
VaD accounts for approximately 30% of dementia in the world today [10]. In Europe, out of 3.7 million of people with clinical dementia, 800,000 have a diagnosis of VaD [11]. The prevalence rate of VaD is around 1-4% in Western developed nations [12]. Recently, in mainland China, a nationwide investigation found the prevalence of VaD to be around 0.8% [13]. The total annual cost (direct, illness related and cost arise from informal care) of dementia in developing countries is estimated to be at least USD 73 billion [10]. A study in Denmark revealed the annual cost per demented person to be DKK 77,000 (approximately USD 14,114) [14]. Thus, the total healthcare cost for VaD patients is highest among all other forms of dementia [15], and the frequency of VaD is increasing exponentially for people over the age of 65 years old [16]. If current trends continue, VaD will become an increasingly significant public health problem in the 21st century. Drugs currently used in the treatment of VaD include cholinesterase inhibitors (donepezil, rivastigmine, and galantamine) [17] and non-cholinergics (memantine, nimodipine, hydergine, nicergoline, CDP-choline, folic acid [18], posatirelin, propentofylline, and pentoxifylline [19]). These orthodox medications (OM) have some efficacy [20]. Preventive therapeutic strategies aiming at reducing cerebrovascular risk factors [17] are also considered by patients likely to develop VaD. As yet, there is no compelling evidence that any of these strategies are effective, and no single intervention can be recommended for the prevention of VaD [21]. This creates a difficult and frustrating situation for sufferers of the disease, their caregivers, and healthcare providers [18], as well as for healthy people hoping to avoid developing VaD.
Owing to the limitations of OM and therapeutic prevention, some patients resort to herbal medications (HM). Traditionally, a number of herbs have been used for cognitive disorders. For example, Artemisia absinthium (Wormwood) was used in traditional European medicine to restore cognitive functions [22]. Melissa officinalis (Lemon balm), also widely used in Europe, has been claimed to restore memory [23]. Since the 16th century, Europe, Salvia lavandulaefolia (Spanish sage) and Salvia officinalis (common sage) have been reported as being effective for improving memory [22]. Bacopa monniera (water hyssop) has been used in Ayurvedic medicine to improve memory and intellectual functions. Centella asiatica (Asiatic pennywort), another Ayurvedic remedy, when combined with milk, is also given to improve memory [24]. Withania somnifera root is classed among the rejuvenative tonics in Ayurvedic medicine and is known to sharpen memory [25]. Codonopsis pilosula root (Dangshen in Chinese), Biota orientalis leaves (Cebaiye in Chinese), and Polygala tenuifolia root (Yuanzhi in Chinese) have been used in traditional Chinese medicine (TCM) for amnesia [26,27]. Some compounds with cognitionimproving properties have been isolated from various plants. EGb 761, an extract from the leaves of the tree Ginkgo biloba, originally used in Western medicine for circulatory disorders [28], shows reversal of decline in cognitive function and of cerebral insufficiency in numerous studies [29], and is now mainly used in VaD as well [30]. In another study, hyperforin, isolated from Hypericum perforatum, a herb used in Portuguese folk medicine, appears to enhance cognitive function [31].
In China, a nation with its own system of medicine that has been continuously documented over two thousand years, the incorporation of Chinese herbal medicine (CHM) with Western medicine in the treatment of dementia has become a standard in recent decades. Salvia miltiorrhiza Bge. (Danshen in Chinese) and Pueraria thomsonii Benth. (Gegen in Chinese), commonly used herbs in the Chinese materia medica for the treatment of cardiocerebrovascular symptoms, are well tolerated and effective in improving vascular function and structure. Thus, either one might be able to effectively intervene in the pathophysiological cascade of VaD [32]. An animal study revealed that glossy privet fruit (Ligustrum lucidum Ait.), a kidney-tonifying Chinese herbal medicine, inhibits neural cell apoptosis following the onset of vascular dementia by reducing apoptotic signals induced by cerebral ischemia/hypoxia [33]. A number of proprietary herbal medicines may also be effective for VaD. According to a study, Chunghyul-dan, which possesses therapeutic effects for microangiopathy, could be useful to inhibit the development of VaD [34]. Huperzine A (HupA), a cholinesterase inhibitor naturally derived from the Chinese herb Lycopodium serratum or Huperzia serrata, has even better penetration through the blood-brain barrier, higher oral bioavailability, and longer duration of AChE inhibitory action than tacrine, donepezil, and rivastigmine [35]. Its anticholinesterase activity is stronger than galantamine (a commonly used drug to treat Alzheimer's disease and various memory impairments) [26]. Its ability to improve memory deficits in elderly people with VaD and AD has been demonstrated, with minimal peripheral cholinergic side effects and no unexpected toxicity [35]. These reports suggest that comprehensive investigation of the efficacy and safety of HM is worthwhile; the results could lead to better treatment of VaD as well as effective prevention.
In a previously conducted systematic review [36], we looked for clinical trials for Alzheimer's disease (AD), and found an even larger number of clinical trials conducted on VaD patients, mainly from mainland China. Because a large proportion of patients have both VaD and AD pathologies [37], and because the current OM treatments for both types of dementia are similar, we have systematically reviewed the clinical trials of HM conducted on VaD patients in this study.

Objective
This systematic review was conducted to assess the safety and efficacy of HM, as either monotherapy or adjunct to OM in the treatment of VaD.

Inclusion Criteria.
All published studies reporting randomized, controlled clinical trials comparing HM as monotherapy or adjuvant therapy, with placebo or OM as controls, were included. No restriction on the language of publication was imposed. As there is a lack of a single, specific criterion for the diagnosis of VaD [38], we accepted the use of the following instruments: Diagnostic and Statistical Manual of Mental Disorders (DSM-III, DSM-III-R [39], and DSM-IV [40]), the International Classification of Diseases, 10th Revision (ICD-10) [41]; the State of California Alzheimer's Disease Diagnostic and Treatment Centers (ADDTC) scale [42]; the Hachinski Ischemic Scale (HIS) [43]; the National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherche et l'Enseignement en Neurosciences scale (NINDS-AIREN) [44]. Trials with participants possessing other forms of dementia (Alzheimer's disease, Lewy body dementia, and frontotemporal dementia) were excluded. There was no restrictions on the ethnicity, gender, age, or disease duration of the participants in the trials. The HM interventions could be either (1) a single herb, (2) a preparation containing multiple herbs, (3) extracts from an herb, or (4) proprietary herbal products. They had  [48], regardless of language and publication status. Hand-search of a list of Chinese and English journals was carried out to find the latest studies. We also referred to the reference lists of relevant papers to identify potential studies. Two independent reviewers (K. W. Chan and S. C. Man) assessed the trials for their eligibility. The inclusion of trials was confirmed upon consensus of reviewers. Risk of bias assessment of the trials was performed according to the revised Consolidated Standards of Reporting Trials (CONSORT) statement [49]. Any disagreement was settled by discussion.

Description of Included Studies.
Using the search strategy as described, 116 studies were identified. Upon full-text examination we excluded 69, on the basis that (1) 13 were not randomized controlled trials, (2) four were repeat publications, (3) three did not state their inclusion criteria, (4) one included other forms of dementia, (5) 44 did not disclose adequate baseline information, and (6) four involved the use of non-HM intervention, such as acupuncture. These disqualifications left 47 studies for this systematic review.
There were a total of 3725 participants (2423 male, 1302 female) in the 47 included trials. Among them, three had cross-over design while the remaining were parallel design studies. The age of participants ranged from 45-89 years old, and their disease duration ranged from two months to 12 4 Evidence-Based Complementary and Alternative Medicine years. Thirty-two studies were performed in a single center; one was performed in multiple centers; 14 trials did not give this information. The duration of trials lasted from one to seven months. All of the trials were conducted in mainland China, and all the subjects were of Chinese ethnicity. (Table 2). We adopted the checklist of items suggested by CONSORT [50] in the evaluation of methodological quality (risk of bias). The checklist was divided into five sections, namely, title and abstract, introduction, methods, results, and discussion.
With regard to method, none of the 47 studies reported details for sample size calculations. Eight [51-54, 57, 59-61] did not clearly state their objective of study. All but one study [62] poorly reported their randomization and blinding. Statistical method was not reported in 13 studies [51-53, 58, 59, 63-70].
For the results section, only two studies [62,71] reported with a flowchart. Patient recruitment, outcomes and estimation, and ancillary analyses were mentioned by the majority of studies. More than half of the studies did not report adverse event count.
In the discussion section, both the interpretation and overall evidence were adequately reported by the studies; generalizability, however, was not sufficiently illustrated by 13 studies. Table 2). All of the included studies claimed to have allocated participants randomly to study groups. Six [72][73][74][75][76][77] reported the use of computer-generated sequences; the other 41 studies did not provide any description of how randomization was achieved. (Table 2). Except for one study [71] which clearly stated that it did not use any blinding methods, 32 studies did not report whether they applied blinding or not. For those studies which reported the use of allocation concealment, 6 were single blind [52,57,66,[78][79][80][81], and 8 were double blind [55,56,62,76,[82][83][84][85]. (Table 3). Only two studies [82,86] used a single diagnostic criterion to select participants. Other studies used two or more diagnostic criteria. Among them DSM III and IV were the most commonly adopted (42 studies used it) [39,40]. The other criteria commonly used, listed in descending order of frequency, were HIS [43] [88], Functional Activities Questionnaire (FAQ) [89], and self-derived criteria (SELF). Furthermore, some studies also carried out diagnostic imaging such as CT and magnetic resonance imaging (MRI) for participant selection. (Table 3). A number of batteries were employed to evaluate the baseline characteristics and outcome measures. The most commonly employed set of evaluative questionnaires included MMSE [45] (38 studies); HDS [47] (23 studies); ADL [46] (20 studies); Berg Balance Scale (BBS 8 studies); SELF (5 studies); memory quotient (MQ 1 study); geriatric dementia scale (GDS 1 study). Moreover, diagnostic imaging such as electroencephalography (EEG 6 studies) and CT (2 studies) were carried out. A number of studies also took into account the changes in hemodynamics (21 studies) and transcranial doppler (TCD 3 studies) as part of the outcome measures. Lastly, biochemical analysis, such as changes in the level of superoxidase dismutase (SOD), malondialdehyde (MDA), homocysteine (HCY), testerone (T), and 17 beta-estradiol (E2) were also adopted in assessing the efficacy and safety of interventions in the studies.

Baseline Characteristics and Outcome Measures
The different batteries used in the studies resulted in variation in outcome measures. As the data were not suitable for meta-analysis, only qualitative appraisal could be carried out. (Table 3). There were altogether 43 trials testing herbal medicine as a monotherapy for VaD. Among them 15 studies compared different HM preparations with Piracetam alone (a nootropic agent). One study compared HM with another stronger nootropic compound Aniracetam. One study compared HM with Piracetam + hydergine and one study compare HM with Piracetam + Vitamin E + respiratory stimulant Duxil (Almitrine). Hydergine, (also known as ergoloid mesylates, another nootropic agent), was tested alone against HM in 11 studies. Five studies reported HM having similar efficacy to these nootropics; the remaining 23 claimed HM to be significantly better.

Herbal Medicine as Monotherapy
Seven studies compared HM with Duxil (Almitrine) alone, a respiratory stimulant originally used to treat patients with chronic obstructive pulmonary disease. In one study, HM is compared with Duxil + Nimodipine (a dihydropyridine calcium channel blocker for the treatment of high blood pressure). One study reported HM to have similar efficacy with Duxil; the other seven claimed that HM is better than Duxil.
Furthermore, HM was compared with Nimodipine in two studies, Huperzine A in one study, cerebroprotein hydrolysate in one study, and placebo in three studies. All of these studies concluded that HM is better than the control intervention. (Table 3). Four trials evaluated HM as an adjunct therapy for VaD. Two of them evaluated the adjunct effect of the CHM decoction BuYangHuanWuTang. Wang compared it with the coadministration of Piracetam and Nimodipine; Yan compared it with the co-administration of Aniracetam, Nimodipine, together with the injection of cerebroprotein hydrolysate. Shi and Wang studied the CHM decoction which, according to TCM theory, could "tonify the kidney, activating blood," and tested its adjunct effect with nimodipine + hydergine (In Table 2              TCM theory, the brain is considered an outgrowth of "kidney" energy. Neurodegenerative disorders such as dementia are caused by stagnation of "blood," accumulation of "phlegm," and deficiency of the "kidney." In order to resist or halt the condition, TCM treatment targets the nourishment of the kidney by means of "kidney invigorating," "blood activating" and "phlegm dissipating" herbal decoctions [110]. Guo et al. [53] studied another CHM BuNaoTongQiao decoction, which possesses nootropic properties, according to TCM theory, and compared it with Duxil. All these studies reported that when HM is used together with Western medications, both the efficacy and safety of OM could be enhanced. (Table 3). Among those 43 studies which tested HM as monotherapy, 25 studies did not report any cases of withdrawal. Ten studies claimed they did not observe any adverse events in groups treated with HM. Eight studies reported a number of mild adverse events, such as mouth dryness, sore throat, constipation, nausea, loss of appetite, and dyspepsia. These adverse events could be resolved without treatment. Serious adverse events were not observed. Occurrence of adverse events remained unclear in the four studies which tested HM as adjunct therapy. The dropouts or withdrawals were unclear in 44 out of 47 studies. Wu et al. [83] reported two dropouts in the course of intervention. Liu et al. [71] reported 46 dropouts during his trial. Cui et al. [77] reported five dropouts.

A Wide Variety of Herbal Remedies.
Thirty-one out of 47 studies tested herbal mixtures prepared in the form of granules or capsules. Fifteen studies tested their herbal mixtures in the form of decoctions. One study tested the extract from a single herb. As some of the studies tested the same herbal mixture, altogether 42 different herbal mixtures were tested among these 47 studies (Table 4). These herbal mixtures or extracts, according to the TCM theory, have the ability to "tonify the "kidney," activate blood." Despite the absence of pharmacological studies to verify their safety, these studies reported encouraging effects and high safety profiles. Upon further analysis of the constitutional ingredients in these herbal formulas, we ranked the 30 most commonly used herbal constituents together according to the dosages ( Table 5).
Rhizoma Chuanxiong, originates from the plant Ligusticum chuanxiong Hort., which is used in TCM to "remove blood stasis." Chemical analysis shows that it possesses an alkaloid named ligustrazine, which has antioxidant, antiinflammatory, antifibrosis, and immune-modulative properties [111]. A clinical study is being carried out to evaluate its effect on patients' recovery from cerebral vascular accidents [112]. The root of Polygonum multiflorum Thunb. (Radix Polygoni Multiflori in English, the Chinese name is Heshouwu), commonly known as fleece flower root, is another popular HM used to treat premature aging and dementia. Past studies have shown it to have activity that may contribute to cardiovascular protection [113]. Longterm pretreatment with it may protect the brain against focal cerebral ischemia [114]. One animal study also suggests that it has anti-oxidant properties [115], with the capacity to prevent cognitive deficits [116], possibly even to promote learning and enhance memory [117]. A medical team in Taiwan is proposing a phase II clinical trial to assess the efficacy and safety of a new drug derived from it [118]. Radix Astragali (Huangqi), from Astragalus membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao, a commonly used herb to "vitalize spleen Qi" and "treat circulatory disorders" in TCM, possesses various components (astragalus saponins, astragalus polysaccharide) demonstrated to have anti-oxidation properties. These, together with its anti-cholinergic property, have been suggested to be the source of its significant antidementia effect [119]. A clinical trial is being carried out to determine the effect of an HM capsule with Radix Astragali (Huangqi in Chinese) as the main constituent on ischemic stroke [112]. Radix Salviae Miltiorrhizae (Danshen in Chinese), the root of Salvia miltiorrhiza Bge., is used to "activate blood and resolve stasis" according to TCM. A laboratory study of its triterpenoids-enriched extract revealed that its antiatherogenic property was mediated by an anti-inflammatory mechanism [120]. Another animal study of Radix Salviae Miltiorrhizae (Danshen) reports it to reduce the area of cerebral infarct in ischemia-reperfusion injured rats, suggesting it has potential in the treatment of cerebral infarct in humans [121]. Radix Ginseng (Rensheng), the root of Panax ginseng C.A. Mey., is a popular notifying herb in TCM, and its ginsenosides have been found to have protective effects on memory via antiapoptosis in a rat model with vascular dementia [122], and to stimulate angiogenesis and tissue regeneration [123], suggesting that it has potential to help VaD patients. In a Korean clinical study, Radix Ginseng was reported to be clinically effective in improving the  cognitive performance of AD patients [124]. Rhizoma Acori Tatarinowii (Shichangpu), also named grassleaf or sweet-flag rhizome, the rhizome of Acorus tatarinowii Schott., is used in TCM for resuscitation after coma. Pharmacological studies suggest this effect may be due to the increase in permeability of the blood-brain barrier [125]. In another pharmacological study, the fruit of Cornus officinalis Sieb. et Zucc. (Shanzhuyu in Chinese), which is used in TCM to "tonify the kidney," was found to possess an extract that has protective effects against oxidative stress-induced neurotoxic processes [126].
Other experimental reports have indicated that the triterpenoid saponins from the roots of Polygala tenuifolia Willd.
(Yuanzhi) possess neuroprotective effects [127,128]. Study on extracts of Alpiniae Oxyphyllae Miq. (Fructus Alpinae Oxyphyllae in English, the Chinese name is Yizhi) have found evidence that it protects neurons against ischemia-induced cell death [129] and that it prevents glutamate-induced apoptosis in cortical neurons [130]. In another study, it was reported that Rhizoma Polygonati (known as Huangjing in Chinese, used in TCM as a notifying agent) could improve learning and memory in a scopolamine-induced mouse model of dementia by reducing the damaging effects of cerebral ischemia and anti-oxidation, having similar effects to those provided by vitamin E [131]. In a study to examine the anti-oxidative and neuroprotective effects of Paeonia lactiflora Pall. (Baishao in Chinese), it was found to suppress the hydrogen peroxide-induced apoptosis in PC12 cells, suggesting that it could be a new antioxidant useful in the prevention of neuronal diseases [132]. Rhizoma Gastrodiae from Gastrodia elata Bl. (Tianma in Chinese), a classic HM used to "extinguish wind and arrest convulsions" in TCM theory, possesses vasodilating [133], anti-inflammatory, and antiangiogenic activities [134], suggesting a potential VaD treatment. The total alkaloids found in Radix Codonopsis (Dangshen in Chinese used in TCM to "tonify Qi") have been reported to potentiate neurite outgrowth induced by nerve growth factor in PC12 cells [135]. Glossy privet fruit, from Ligustrum lucidum Ait. (Nüzhenzi in Chinese), a kidney-tonifying HM, can inhibit cell apoptosis by reducing apoptotic signals induced by cerebral ischemia/hypoxia [33].

Study Weaknesses.
Though all the studies reported promising results of HM in the treatment of VaD, they demonstrated a number of weaknesses as well. The evidence drawn from the studies was insufficient for us to confirm the safety and efficacy of HM, because of the following issues (1) The sample sizes of the studies ranged from 18 to 300, and none of them reported sample size calculations, as suggested by the CONSORT statement. Treatment effects can be exaggerated when sample size is inappropriate, and thus the results of these studies may not be conclusive.
(2) Different diagnostic criteria were used in the studies. Some of these criteria were even self-derived and their validities remained unknown. This produced much discrepancy.
(3) Differences in the baseline characteristics of the subjects limit the extent to which results can be compared with each other.
(4) Though all of the studies claimed to have participants allocated randomly, only a few reported the method of randomization. For those studies without detailed descriptions of randomization, we could not rule out the possibility of bias. Furthermore, unclear descriptions of allocation concealment, dropouts, and intention-to-treat analysis further hamper the ability to assess the validity of the evidence reported by these studies.
(5) Outcome measures varied and were incomplete in the studies. Some investigators employed self-developed scales, which could not be, or had not been, independently evaluated for their sensitivity and specificity. The validity is further questionable due to insufficient or inappropriate statistical treatment. Though metaanalysis techniques such as vote-counting may have been used for the analysis of the data, we avoid to do so because (1) the statistical significance or size of the results of the individual studies are ignored; and (2) vote-counting takes no account of the differential weights given to each study. [136] 18 Evidence-Based Complementary and Alternative Medicine (6) Different HM were tested in the 47 studies included here, with great variation in terms of composition, dosage, and duration of interventions. This renders comparison of the studies impossible, and thus quantitative analysis could not be carried out.
(7) A number of studies (30 out of 47) did not mention safety issues. The investigators of these studies may have underestimated possible adverse events, and the safety of HM in these studies could not be guaranteed.

Implications for Further Studies.
Regarding the published studies, methodology quality is the leading should concern. It is recommended that future clinical studies follow the guidelines as suggested by CONSORT to minimize bias as well as to ensure high validity, statistically reliable results and to permit comparison with other studies. Researchers should explicitly report methods for calculation of sample size. Widely recognized diagnostic criteria and outcome measures should be used. It is highly recommended to incorporate medical imaging techniques (such as perfusion computed tomography) to confirm the diagnosis of VaD. Appropriate statistical analyses should be carried out for baseline data and outcome results; long-term followup is also recommended and highly desirable. Our review has identified the individual herbs that appear most frequently in formulas for VaD. The top five are Rhizoma Chuanxiong (Chuanxiong), Radix Polygoni Multiflori (Heshouwu), Radix Astragali (Huangqi), Radix Ginseng (Renshen), and Rhizoma Acori Talarinowii (Shichangpu). The clinical efficacy and safety of these herbs, over centuries of use and during recent controlled studies, are a powerful combination of attributes. We believe that further highquality clinical studies on these individual constituents, as well as the herbal mixtures resulted, could lead to the discovery of new drugs for effective treatment and prevention of VaD.

Conclusion
Currently available RCTs suggested that HM might be more effective and safer than OM for treatment of VaD. However, these studies have a number of weaknesses, mainly due to their methodological insufficiencies. With regard to the reports that did meet our selection criteria, the results indicated that HM, in a predominance of instances, can be superior to OM and useful in the treatment of VaD. Further multicenter trials with large sample sizes, high methodological quality, and standardized HM ingredients are needed to confirm the value of HM in treating VaD, in order to establish specific clinical recommendations.