Alzheimer disease is the most common cause of dementia worldwide [1]. It is estimated that, in persons aged 65 years and above, mild cognitive impairment is present in up to 20% and dementia or major cognitive impairment is present in approximately 1% to 25% [2]. The number of people with dementia is likely to reach 44 million worldwide by 2030 [3]. The current major treatment for Alzheimer disease is drug therapy with an acetylcholinesterase inhibitor, such as donepezil [1]. These types of drugs have been found to be effective in treating dementia of all severities; however, they have no proven disease-modifying effect and, therefore, do not improve long-term outcomes [3]. In addition, their effect on patients with mild cognitive impairment due to Alzheimer disease (MCI-AD) is unproven. Furthermore, their use has been associated with adverse events such as nausea, vomiting, diarrhea, muscle spasms, insomnia, abnormal dreams, headaches, peripheral edema, weight loss, syncope, fatigue, asthenia, and tremors [4].

The search for other effective and safe treatments has led to an interest in herbs and extracts that have anecdotally been used to prevent and treat memory loss. Among many such herbs, Bacopa monnieri (also known as Brahmi, bacopa, or water hyssop) has attracted much attention [5-7]. This herb has been mentioned and extensively used for many centuries in Ayurveda (traditional Indian system of complementary medicine) for improving cognitive ability and preventing memory loss [8,9].

Several in vitro and animal studies have shown a neuroprotective effect of this plant and its extracts [6,10-12]. However, the use of this herb in humans has yielded conflicting results, and its role in treating patients with established Alzheimer disease is not clear. In addition, most studies [13-15] have been performed on healthy adults. Many studies (eg, [16]) from the Ayurveda stream have been anecdotal, and many others (eg, [17]) have methodologically weak designs. Systematic reviews [13,18-20] on B. monnieri have included studies on healthy individuals and have not included studies conducted in the last 5 years.

The major active substances in B. monnieri with potential effects on cognition and memory are bacoside-A, bacoside-B, alkaloids, and flavonoids [5]. The therapeutic effects of bacosides demonstrated in preclinical and in vitro studies include enhancement of neurotransmission, potentiation of synaptic activity, and repair of damaged neurons by upregulating neuronal synthesis and kinase activity [6]. Holcomb et al [10] found a reduction in amyloid levels using B. monnieri extract in mice models. Cognitive enhancement and neuroprotective effects have been found in Alzheimer disease animal models [21-24]. Studies have found that amnesia induced by diazepam in mice could be reversed using B. monnieri [25] and that L-arginine N(omega)-nitro-L-arginine–induced anterograde and retrograde amnesia could be reversed using B. monnieri [26]. Recently, changes in metabolites in plasma, urine, and feces in healthy individuals after consuming B. monnieri essence for 12 weeks were found using liquid chromatography mass spectrometry and that aminoacyl-transfer RNA, aromatic amino acids, and branched-chain amino acid biosynthetic pathways were mainly related to the metabolites identified in all 3 types of samples [27]. Aromatic amino acids, particularly phenylalanine, are metabolites found to decrease in level in the plasma of healthy community-dwelling participants aged 70 years and older who later progressed to amnestic mild cognitive impairment or Alzheimer disease compared to normal controls [28].

Considering the potential of B. monnieri as a neuroprotective agent and the gap in the existing literature connecting B. monnieri and dementia, we aimed to perform a systematic review to determine whether it has beneficial effects on cognitive impairment due to Alzheimer disease and identify gaps in the literature.

We aimed to address the following question: What are the effects (benefits and harms) of B. monnieri on individuals with dementia due to Alzheimer disease?

Our objectives were to (1) determine the clinical efficacy and safety of B. monnieri in persons with mild, moderate, or severe dementia due to Alzheimer disease or with MCI-AD; and (2) compare the efficacy and safety of different doses of B. monnieri.

A comprehensive search strategy (Multimedia Appendix 1) was employed to identify all relevant studies. We did not place any restrictions on language or publication status (published and in press) during the search. PubMed, Embase, Cochrane Library, clinical trial registries (World Health Organization, Australia-New Zealand, United States, and South Africa), the metaRegister of Controlled Trials, and CINAHL were searched from inception to January 2021. We also searched for studies in the reference lists of all studies included in the pool of retrieved papers. Where possible, we contacted authors to obtain the full text of the papers. The study has been reported according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, and the protocol is available on the International Prospective Register of Systematic Reviews (CRD42020169421).

We included randomized and quasi-randomized controlled trials that compared B. monnieri, its extract or active ingredients (at any dosage), with a placebo or a cholinesterase inhibitor among adults with dementia due to Alzheimer disease and MCI-AD. Cohort studies, case-control studies, case reports, systematic reviews, policy papers, letters to the editor, correspondence, and nonhuman studies were excluded. We also excluded trials that were confounded by treatment or a control group receiving another active treatment that has not been factored into the randomization. Furthermore, studies must report one or more of the following outcomes to be eligible for inclusion: cognitive function (determined by the change from baseline in the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), Mini-Mental State Examination, Postgraduate Institute Memory Scale, or any culturally adapted or validated tools to assess cognition), activities of daily living using scores such as the Alzheimer Disease Cooperative Study, clinician-rated global impression tests, behavioral symptoms, and safety as measured by incidence of adverse effects, dependency, or death.

Following the search, 2 authors (VVY and AB) screened titles, abstracts, and full-text papers independently and extracted data from each paper using standardized data extraction forms. Discrepancies were resolved by consensus between the 2 authors or by a third author (AA). Data on study characteristics, participant characteristics (age, gender, healthy adults, patients with Alzheimer disease), study setting, use of B. monnieri (formulation, dose), placebo, any active comparator, and outcomes of interest were extracted. Two authors (AB and BM) used the Cochrane risk-of-bias tool for randomized controlled trials [29] to determine the risk of bias for all eligible papers, and a third author (AA) independently assessed the papers that were found to be eligible to be included in this systematic review in terms of the level of concern (ie, low, some, or high). We did not group studies comparing B. monnieri with the placebo and those comparing B. monnieri with cholinesterase inhibitors since no meaningful comparisons were possible due to the heterogeneity of outcomes. Since none of the outcomes of interest was reported for all studies, we did not pool them during analysis.

We used ReviewManager (version 5.4; The Cochrane Collaboration) for data analysis. Wherever possible, an intention-to-treat analysis was planned. Due to explicit clinical and methodological heterogeneity between available studies, we performed no statistical tests for heterogeneity. The data available from eligible studies precluded a meta-analysis, and we performed an interpretative synthesis of data from individual studies. The GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework was used to summarize the certainty of evidence [30,31]. We did not attempt to ascertain publication bias since there were only a few eligible studies.

Overall, 164 studies were identified by the search. The abstracts of all studies were screened, and 5 [32-36] were found to be eligible (Figure 1 and Multimedia Appendices 2 and 3).

The primary objective of the study conducted by Prabhakar et al [32] was to determine if B. monnieri improved the memory of patients with Alzheimer disease and MCI-AD compared with donepezil [32]. The diagnosis of Alzheimer disease and MCI-AD was aided by magnetic resonance imaging of the brain, fluorodeoxyglucose-positron emission tomography of the brain, and cerebrospinal fluid biomarkers (beta amyloid and total tau). B. monnieri was administered once daily at a dose of 300 mg, while 10 mg of donepezil was given once daily for 12 months. Although a sample size of 48 patients (24 in each arm) was planned, only 34 patients (17 in each arm) could be recruited after 45 months, due to which the study was terminated. The primary outcome was the difference in change of the ADAS-Cog score and Postgraduate Institute Memory Scale score from baseline after 12 months of treatment between the 2 treatment groups. However, patients were followed up for changes in scores at 3, 6, and 9 months of treatment. Change from baseline of neuropsychological tests such as the verbal fluency-controlled oral word test and animal names test, quality of life-Alzheimer disease, activities of daily living inventory, adherence to treatment, and adverse events were secondary outcomes. The authors reported attempts to follow up on all patients with whom contact was lost during the study period, and an intention-to-treat analysis was used. Missing data were handled by multiple imputations due to loss of follow-up. A total of 4 and 9 patients were lost to follow-up in the donepezil and B. monnieri arms, respectively, at the end of 12 months. There were differences in baseline characteristics (more patients with Alzheimer disease in the donepezil arm and more patients with MCI-AD in the B. monnieri arm), which were adjusted for during analyses.

Cicero et al [33] compared the effect of a combination of agents (B. monnieri, L-theanine, Crocus sativus, copper, folate, vitamin B complex, and vitamin D) over 2 months and those of a placebo in improving cognitive functions in older adult patients. They included 30 participants with Mini-Mental State Examination scores between 20 and 27 or self-perceived cognitive impairment (whether the impairment was dementia or mild cognitive impairment was not reported). The primary outcome was change in Mini-Mental State Examination score from baseline at 2 months. The Perceived Stress Questionnaire and Self-Rating Depression Scale scores were other outcomes.

Sadhu et al [34] investigated the efficacy of a polyherbal test formulation composed of extracts of B. monnieri, Hippophae rhamnoides, and Dioscorea bulbifera (total dose of 500 mg) on cognitive functions [34]. The test formulation was compared with a placebo in healthy adults without dementia and the test formulation was compared with donepezil (10 mg twice a day) in older adult patients (aged 60-75 years) with Alzheimer disease (n=123; deterioration of memory in at least 3 of the following: poor orientation, poor judgment and problem-solving, difficulty in community affairs, inability to function independently at home, or difficulties in personal care) for 12 months. Subsequently, the participants underwent a clinical screening using the Dementia Rating Scale-2 before being randomized to test formulation or donepezil. The primary outcome was cognitive function assessed by a composite of mental status (Mini-Mental State Examination), verbal memory, complex psychomotor tests, and attention or executive functions at 12 months.

Raghav et al [35] studied the efficacy of B. monnieri extracts in patients with age-associated memory impairment and no evidence of dementia or psychiatric illness [37]. Participants were adults older than 55 years of age with memory loss in daily activities and a logical subset score <6 on the Wechsler Memory Scale. Patients with a Mini-Mental State Examination score >24 were excluded. Eligible patients (N=40; B. monnieri group: n=20; placebo group: n=20) were randomized to receive 125 mg of B. monnieri extract or a placebo twice a day for 12 weeks followed by the placebo for both groups for another 4 weeks. The outcomes were the Mini-Mental State Examination and the Wechsler Memory Scale (subsets Logical Memory, Visual Reproduction, and Paired Associated Learning).

Barbhaiya et al [36] studied the effects of Bacomind(an enriched phytochemical combination containing 450 mg of standardized B. monnieri extract) on 65 individuals aged 50 to 75 years with self-reported memory impairment (Mini-Mental State Examination score >24) for at least 1 year. The study duration was 24 weeks (drug administered for 12 weeks, then no drugs for 12 weeks). Neuropsychological assessment was performed at baseline, week 12, and week 24. Outcomes were analyzed for attention, memory, and speed of information processing. Three patients were lost to follow-up for the second visit, and 15 patients were removed as outliers. The final analysis was completed for 44 patients (Bacomind: n=23; placebo: n=21). Further, 3 patients were lost to follow-up during the study.

All 5 studies were deemed to have a high risk of bias.

Different tools were used in the studies to determine functional outcomes. Prabhakar et al [32] found no significant difference in the change in activities of daily living scores such as Alzheimer Disease Cooperative Study activities of daily living at any time during follow-up between the donepezil and B. monnieri arms. Similarly, no changes were noted in quality of life measured using Quality of Life Patient and Informant questionnaires. Sadhu et al [34] used the Functional Activity Questionnaire and reported a significant difference in change at 12 months between the donepezil arm (mean 9.801, SD 1.458) and the B. monnieri formulation arm (mean 11.873, SD 2.751; P<.001). Raghav et al [35] did not report any functional or quality of life–related outcomes. Cicero et al [33] reported significant improvement in Self-Rating Depression Scale scores in the B. monnieri formulation arm. Barbhaiya et al [36] did not report any functional outcomes.

Prabhakar et al [32] reported no significant differences in the number of patients who experienced one or more adverse events. No major adverse events were reported either. There were 3 deaths (2 in donepezil arm and one in B. monnieri arm) reported due to myocardial infarction. Raghav et al [35] reported diarrhea in 1 patient and headache in 2 patients in the placebo arm. One participant of the B. monnieri arm was reported to have experienced rashes. Sadhu et al [34] reported nausea, constipation, and drowsiness; however, no data on the number of participants with these events were reported. In the study conducted by Cicero et al [33], 1 participant was reported to have an aftertaste following ingestion of the B. monnieri formulation. Barbhaiya et al [36] did not report any adverse events.

We used the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach to assess the certainty of the evidence in the included studies, using the criteria outlined in the Cochrane Handbook [37]. The certainty of the evidence for the reported outcomes was very low. We assessed the quality of evidence for 4 major clinically relevant outcomes (ADAS-Cog, Postgraduate Institute Memory Scale, Mini-Mental State Examination, and Wechsler Memory Scale). We judged that all 5 included studies had a high risk of bias, downgrading the evidence by 2 levels. We downgraded one level for impreciseness due to the small sample size and wide CI. Hence, the overall certainty of evidence was very low.

There was no high-quality evidence for the benefits of B. monnieri compared with a placebo or donepezil for cognitive function, functional outcomes, or adverse events. All 5 studies were heterogeneous with respect to doses of B. monnieri, B. monnieri as part of a polyherbal combination, use of a placebo or donepezil as the control group, duration of treatment (2 months to 12 months), cognitive tests to assess primary outcomes, and the diagnostic criteria for Alzheimer disease and mild cognitive impairment.

Prabhakar et al [32] was able to demonstrate a statistically better outcome in the donepezil group compared to the B. monnieri group, even though there was no difference at 3, 6, and 9 months. Any statistical significance in this study was limited by a small sample size (34 patients) due to poor recruitment (the trial was stopped prematurely because of this issue).

Raghav et al [35], Sadhu et al [34], and Cicero et al [33] were able to demonstrate improvements in one or more facets of cognition using B. monnieri compared to the placebo or donepezil. All 3 studies had small sample sizes. Furthermore, duration of treatment and follow-up were also short. Since Sadhu et al [34] and Cicero et al [33] used a polyherbal preparation and combined nutraceuticals (with B. monnieri as one component), improvements noted cannot be attributed to B. monnieri alone. These 2 studies [33,34] did not use brain imaging; hence, the accuracy of Alzheimer disease diagnosis cannot be ascertained.

Although Barbhaiya et al [36] reported significant improvements in several tests of cognition, removal of 15 participants after randomization as outliers without giving any reasonable explanation and the absence of 33% of participants from the final analysis are major issues.

A previous meta-analysis [13] that evaluated the efficacy of B. monnieri for cognitive performance included studies with both healthy participants and individuals with memory impairment (518 participants from 9 studies); however, there were only 2 trials with cognitively impaired patients [35,36]. Similar to our observation, Barbhaiya et al [36] described an overall dropout of around 33%. However, there is no mention of the exclusion of 15 participants as outliers. A meta-analysis performed on data from 437 participants showed a shortened duration taken to complete the Trail B test (–17.9 ms, 95% CI –24.6 to –11.2; P<.001) and decreased choice reaction time (10.6 ms, 95% CI –12.1 to –9.2; P<.001). However, the Trail B test results were based on a single study with 46 healthy volunteers while the decreased choice reaction time was based on a subgroup analysis of 2 studies on healthy volunteers (46 and 62 participants) using 300 mg of B. monnieri. Hence, it cannot be truly considered as pooled estimates of efficacy in those with dementia. We did not perform a meta-analysis since the heterogeneity of data from the studies on people with dementia precluded meaningful pooling.

Another systematic review [18] on the effectiveness of B. monnieri as a nootropic, neuroprotective, or antidepressant supplement included studies involving healthy volunteers and those with dementia and depression. Three studies evaluated B. monnieri in Alzheimer disease or mild cognitive impairment [33,34,38], of which Goswami et al [38] was a nonrandomized study. A meta-analysis was not performed in this review.

In a systematic review by Cicero et al [19], the meta-analysis by Kongkeaw et al [13] was cited, and no new studies were included apart from those included by Kongkeaw and colleagues. Brioschi Guevara et al [20] had included 2 studies (Raghav et al [35] and Barbhaiya et al [36]) in their systematic review and described these studies as follows: “two old small studies on B. monnieri in individuals with memory complaints suggest a potential effect on some aspect of memory function or on attention tests that still need to be confirmed.” No data or study characteristics were mentioned in the review.

Though previous systematic reviews [13,19] had included most of the studies that this review also found eligible, they had not specifically addressed the question of the efficacy of B. monnieri in persons with Alzheimer disease. Inclusion of healthy volunteer studies and nonrandomized studies in these reviews makes it difficult to draw conclusions about similarities, and although significant improvements in specific tests had been noted in pooled analyses, they were mostly based on data from subgroups or single studies.

We have thoroughly searched for and analyzed all the available evidence critically. Only 5 eligible trials were identified due to the use of stringent inclusion criteria. Due to severe heterogeneity in the included studies in terms of criteria for diagnosis, cognitive tests used, B. monnieri formulations (including polyherbal), duration of treatment, and lack of confidence in the diagnosis of dementia in the included patients, we did not conduct a meta-analysis.

First, the use of very stringent inclusion criteria led to few eligible trials. However, this also means that the review question has been addressed specifically without any dilution of intent. Second, 1 author (VVY) was involved in a trial included in this review (ie, Prabhakar et al [32]). This potential source of bias was addressed because author VVY did not participate in the risk-of-bias assessment of trials (performed independently by authors AB and BM). Third, as with any systematic review, it is possible that some studies might have been missed. We ensured the inclusion of all potential studies by searching multiple databases. Moreover, independent screening of search output by 2 authors ensured that bias was minimized in assessing eligibility. We followed the guidance provided in the Cochrane Handbook to minimize potential biases in the review process [34]. Fourth, because only 5 trials were included in this review, we could not use funnel plots to assess the risk of publication bias.

As discussed above, this review found that there was no high-quality evidence for the benefits of B. monnieri compared with a placebo or donepezil for cognitive function, functional outcomes, or adverse events. All 5 studies had heterogeneity with respect to the B. monnieri dosage used in the trials, B. monnieri used as part of a polyherbal combination, use of a placebo or donepezil as the control group, duration of treatment (2 months to 12 months), cognitive tests used to assess primary outcomes, and diagnostic criteria used for Alzheimer disease and mild cognitive impairment. These hindered the generation of high-quality evidence for the use of B. monnieri in Alzheimer disease and MCI-AD. Based on these results, we opine the following design changes for future trials of B. monnieri in patients with Alzheimer disease.

The evidence obtained from the present systematic review is of very low certainty. The evidence from 5 trials suggests that there is no difference between B. monnieri and a placebo or donepezil in the treatment of Alzheimer disease or mild cognitive impairment. No major safety issues were reported in the trials included in this review.