The clinical efficacy of salvia officinalis

The clinical efficacy of salvia officinalis An evaluation of the clinical efficacy of Salvia officinalis, Salvia lavandulaefolia and Melissa officinalis for the prophylaxis, management and amelioration of cognitive dysfunction: with particular reference to Alzheimers disease and non-Alzheimer-type senile dementias. 1. Introduction Dementia is a collection of symptoms caused by a chronic, global deterioration of cognitive function. It can occur at any age but is most prevalent in the elderly and increases with age (Beers et al. 2006: 1811). Around 5% of people over 65, 25% over 80 and 45% over 85 have some form of dementia (Knapp et al. 2007: 10; Collins 1997: 185). The population is aging and whereas today there is an estimated 700,000 people in the UK suffering from dementia, this number is set to increase to more than a million by 2025. The huge impact dementia has on society, devastating families and costing around  £17-18 million annually cannot be overstated (Knapp et al. 11). Early identification and safe, effective, intervention is therefore important. Dementia may be classified as Alzheimers or non-Alzheimer-types (Beers et al. 2006: 1811). The most common dementia is Alzheimers disease (AD) (Grossman et al. 2006: 985), affecting around 20 million people worldwide (Akhondzadeh et al. 2003: 53) and accounting for around 62% of dementias (Knapp et al. 2007: 11). Non-Alzheimer-type dementias typically affecting those over 60 include vascular dementia (27%), Lewy body dementia and fronto-temporal dementia (Knapp et al. 29). Cognitive disorders are treated allopathically with drugs that have yet to show real benefits and have a number of side-effects and contraindications. The need for safer, more effective treatments has led to increasing interest in the use of herbs for their management (Akhondzadeh and Abbasi 2006: 117). A variety of herbs, for example Salvia officinalis, Rosmarinus officinalis, Mellissa officinalis, Ginkgo biloba (Heinrich et al 2004: 234), Withania somnifera (Howes et al. 2003: 12), Centella asiatica (Chevallier 1996: 78) and Panax ginseng (Mantle et al. 2000: 207) have long-standing traditional use as memory-enhancing herbs. Consequently a number of clinical studies have been conducted to assess the efficacy of some of these herbs, most notably Ginkgo biloba, Salvia spp. and Mellissa officinalis, in the treatment of cognitive disorders. Of these, only clinical trials of Gingko biloba have been extensively reviewed (Birks and Grimley Evans 2002; Ernst et al. 1999; Oken et al. 1998). This present review aims to fill this gap by providing up-to-date information on whether clinical studies of Salvia spp. and Mellissa officinalis support their traditional use as cognition enhancers. To inform herbal practice it will evaluate clinical studies to assess whether the results have determined safe, effective herbal strategies and prescription for prophylaxis, management and amelioration of cognitive decline. 2. The literature review 2.1. Background: clinical presentation and pathology Although much scientific progress has been made since 1907 when Alois Alzheimer first described a case of dementia with â€Å"peculiar patches† disseminated throughout the cerebral cortex (Collins 1997: 185), there is still much to learn about the aetiology and pathogenesis of Alzheimers disease and other dementias (Knapp et al. 2007: 11). The onset of dementia is insidious, often beginning as mild cognitive impairment (MCI) and progressing to severe dementia over time (Loveman et al. 2006: 4). In the early stages, episodes of mild forgetfulness or misplacing possessions are often attributed to normal aging. Patients commonly suffer from anomic aplasia and agnosia but retain language comprehension (Collins 1997: 186). Dementia becomes more apparent when sufferers are unable to learn new information, to register the content of a conversation, or to recall recent events or the names of family members. Unlike those with benign forgetfulness, dementia patients are unaware of their amnesia. Frequently, there are mood changes, depression and other psychologic disturbances. Language comprehension fails (aphasia) and eventually patients may simply repeat what they hear or be unable to speak at all. Visuospacial deficits usually occur at a late stage (Collins 1997: 186). Those affected have difficulty in copy drawing simple obj ects. Differential diagnosis between MCI subtypes of various and complex aetiologies is challenging (Kidd 1999: 145). As some MCI subtypes are reversible (Levey et al. 2006: 992) prophylaxis for dementia could potentially encompass a range of varied or unknown aetiologies and risk factors. Knowledge of these and an awareness of differing clinical presentations are therefore important (Levey et al: 991). Additionally, an understanding of current orthodox treatment strategies and key neurochemical impairments in dementia can inform herbal practice of the most likely therapeutic actions of herbs. 2.1.1. Alzheimers disease As clinical studies have indicated that mild to moderate Alzheimers disease (AD) responds better to allopathic drugs than severe AD (Levey et al: 2006: 993), to prevent transition of MCI to AD early diagnosis is important. Evidence suggests that MCI associated with memory loss most commonly leads to AD (Levey et al. 991) and results of a clinicopathologic study of 80 subjects with MCI through to autopsy suggest that depression is one of the first features of AD (Galvin et al. 2005: 763). Formation of diffuse neuritic senile plaques in the brain is characteristic of AD but as these can only be determined from biopsy (Collins 1997: 186) probable diagnosis is made by clinical neuropsychological testing (Grossman et al. 2006: 986) such as the Mini Mental state Examination (MMSE) (Alzheimers Society 2002: 436), while magnetic resonance imaging can corroborate diagnosis by identifying areas of temporal neuronal loss (Vandenberghe and Tournay 2004: 347). Progression of AD is unremitting for around 5-10 years until death ensues. In the final stages sufferers may develop apraxia, with difficulty in performing familiar tasks. A common cause of death is pneumonia when patients eventual difficulty with eating results in aspiration pneumonia (Collins 1997: 186). The loss of faculties has been ascribed to both structural and neurochemical abnormalities (Perry et al. 1996: 1063). Senile plaques in the brains of AD patients contain amyloid and tau protein (microtubule associated protein) (Collins 1997: 188). Since isolation of b-amyloid peptide from cerebral vessels in AD patients (Wong et al. 1984: 8729), the accepted hypothesis for the pathogenesis of AD has been the ‘amyloid hypothesis, which proposes that AD is due to excessive formation of extracellular b-amyloid (Ab?) from amyloid precursor protein (APP), a membrane protein in neurons (Grossman et al. 2006: 986). It is thought that Ab molecules initiate a toxic cascade long before plaque forms by causing an inflammatory reaction, disrupting synaptic function and causing neurons to degenerate (Grossman et al. 986) with a loss of cholinergic fibres in the basal forebrain. In vitro results suggest that Ab enters mitochondria and induces free radical damage (Reddy 2006: 9). Intracellular neurofibrillary tangles are believed to be formed by abnormal phosphorylation of tau proteins (Tanzi and Bertram 200 5: 545), particularly in the hippocampus and neocortex, areas of the brain involved in memory (Mantle et al. 2000: 202). To date, thirteen genes have been implicated in AD (Bertram et al. 2007: 17). Of sporadic late onset Alzheimers up to 40% of cases may be due to a faulty gene on chromosome 21, ApoE4, an isoform of the ApoE gene that encodes for apolipoprotein, an astrocytic protein that may play a role in the reparative process in the brain. ApoE4s pathogenetic mechanism may be to enhance amyloid deposits within tissue by accelerating cleavage of b-peptide (Collins 1997: 189). Possession of a gene implicated in AD does not necessarily result in its development, the likelihood of which is further complicated by the potential role of environmental factors such as viruses and toxins in combination with genetic factors (Bird 2005: 864). 2.1.2. Vascular dementia Vascular dementia (VaD) is any type of dementia caused by cerebral blood vessel disease (Micieli 2006: S37). Onset of VaD is usually abrupt. Imaging may reveal areas of multiple infarcts (Collins 1997: 191) but their presence does not necessarily imply dementia (Grossman 2006: 987). According to Looi and Sachdev (1999) it is not possible to differentiate between AD and VaD with neuropsychological testing. Speech and language difficulties associated with vascular dementia may be mild or there may be a more pronounced aphasia as in multi-infarct VaD (Collins 1997: 191). 2.1.3. Frontal lobe dementia Frontal lobe dementia or Picks disease is uncommon and is characterised by neuronal loss and gliosis. Rarely, there are fibrillary inclusion bodies in the neurons. Presentation of frontal lobe dementia differs from AD in that the first symptoms are a change in personality rather than memory loss (Collins 1997: 193). 2.1.4. Lewy body dementias Dementia with Lewy bodies may differ to AD in its presentation in that patients suffer from marked visual hallucinations. Additionally, cognition tends to fluctuate between normality and confusion. Parkinsonian features such as shuffling gait, tremor, bradykinesia and rigidity are prevalent. Sleep behaviour disorder, such as acting out attacking themes, may appear years before other signs of the disease (Grossman et al. 2006: 989). 2.2. Risk factors Factors believed to pose a risk for developing dementia include cardiovascular disease, being female, a family history of dementia, Downs syndrome, older age, head trauma, diabetes and lower educational standards (Collins 1997: 186, 188; Lebson et al. 1997: 301). 2.2.1. Cardiovascular disease: Patients may have more than one type of dementia concurrently (Beers et al. 2006: 1811). This is compounded by results of a number of epidemiological studies suggesting that cardiovascular disease increases the risk of developing AD (Stampfer 2006: 12). Using transcranial Doppler ultrasonography Sun et al. (2007: 152) demonstrated diminished cerebral blood flow velocities in MCI patients who also carried the ApoeE4 allele. Risk factors for VaD are believed to include artherogenic factors such as hypertension, hyperlipidaemia, diabetes, and cigarette smoking (Micieli 2006: S38). Conversely, there are indications from clinical trials that nicotine has a protective effect for AD (Breteler et al. 1992: 71). Results of a randomised, double-blind, multicentred trial in which subjects with hypertension were treated with antihypertensives or placebo suggest that hypertension is a risk factor for developing both AD and VaD. Antihypertensives reduced risk by 55%. The results were significant as subjects had similar characteristics, the sample size was large (3228) and equally divided into placebo and treatment groups. Median follow-up was 3.9 years (Forette et al. 2002: 2047). 2.2.2. Head trauma: A meta-analysis by Fleminger et al. (2003: 858) replicated earlier findings by Mortimer et al. (1991) that head injuries pose a risk for AD but only in males, thought to be due to an early protective effect of oestrogens in females (Fleminger et al. 860). Bias may have been introduced into both studies as informants recalled the injuries. 2.2.3. Diabetes mellitus: Given that diabetes mellitus (DM) is a known risk factor for vascular disease it is not surprising that most studies on the development of vascular dementia in DM patients have shown a positive association (Biessels 2004: 10). Studies on DM as a risk factor for AD, however, have yielded conflicting results, possibly due to study limitations such as small sample sizes and selection bias (Leibson et al. 1997: 301). Large longitudinal studies may be more reliable. A population-based historical cohort study of 1,455 cases followed over 9,981 person years found a statistically significant positive association (Leibson et al. 304). According to results from the Framlingham Study, diabetes may not be an independent risk factor for developing AD but risk is strongly associated with possession of the ApoE4 genotype (Akomlafe et al. 2006: 1551). 2.2.4. Hormones: Women are twice more likely than men to suffer from AD. Although this may be partly due to women having a longer life expectancy (Beers et al. 2006: 1814) there is evidence to suggest that a decline in endogenous oestrogen in later life plays a role in its pathogenesis. Oestrogen is believed to stimulate cholinergic activity, reduce oxidative stress related cell damage, reduce vascular risks, reduce Ab formation and promote synaptic activity (Zandi et al. 2002: 2123; Hoskin et al. 2004: 141). Evidence from studies to determine whether oestrogen-containing hormone replacement therapy (HRT) in women has a protective effect on the brain, however, is conflicting (Colucci et al. 2006: 1376) but this may be due to differences in methodology and confounding factors (Resnick and Henderson 2002: 2171). For example, in one large prospective study that found a positive correlation between HRT use and a significant reduction in AD development, patients with dementia were asked q uestions regarding previous use of HRT (Zandi et al. 2124) yet accurate recall in a dementia sufferer cannot be guaranteed. Results of a retrospective case-control study suggesting the likelihood of women developing AD increases with number of pregnancies (Colucci et al. 2006: 1375) could be of little value. Cases with previous head injuries, low educational standards, both considered risk factors for AD (Collins 1997: 186; Fleminger et al. 2003: 858), and those who had used HRT, were not excluded from the study. There is evidence to suggest testosterone may delay AD onset in men. Men over 32 years of age who were free from AD at baseline (n = 574) were followed for a mean of nineteen years (Moffat et al. 2004: 188). Long-term free testosterone levels were significantly lower in men who developed AD. Due to conflicting results and confounding factors in the research the clinical evidence for risk factors for dementia is inconclusive. However, although more research is needed the results can assist in informing herbal practice. 2.3. Orthodox treatment strategies As cholinergic neurotransmitters are believed to have a role in memory function (Grossman et al. 2006: 985) symptomatic treatment for subtypes of dementia is similar and focuses on acetylcholinesterase (AChE) inhibition with drugs such as donepezil, rivastigmine and galantamine (Loveman et al. 2006: 8). According to Delagarza (2003: 1366) loss of cholinergic neurons causes a decrease in acetylcholine and subsequent drop in AChE with a compensatory rise in butylcholinesterase (BChE). Nicotinic receptors also decrease. Rivastigmine also inhibits BChE; galantamine also acts on nicotinic receptors. Depression in dementia is treated with non-anticholinergic antidepressants as anticholinergic drugs exacerbate symptoms (Beers et al. 2006: 1814). Another drug, memantine, a N-methyl-D-aspartic acid (NMDA) receptor antagonist (Grossman et al. 987), licensed to treat moderate to severe AD, acts by modulating the action of the neurotransmitter glutamate, which is believed to be associated with c holinergic damage and neurodegeneration when secreted in excess (Loveman et al. 2006: 8). Dizziness, diarrhoea, headaches, nausea and vomiting were found by a meta-analysis of dementia drugs to be common adverse events with anti-cholinesterases and memantine (Loveman et al. 2006: 49). Furthermore, their long-term benefits are inconclusive (Loveman et al. 145). Similarly, their use for vascular or Lewy body dementia is controversial as a review of clinical trials data deems there is insufficient evidence for their efficacy. Trials were of generally poor quality and with inconsistent findings (Maggini et al. 2006: 457). Other potential drugs for AD include 70 new compounds formulated to interfere with the toxic amyloid cascade or to target inflammation, oxidation or apoptosis (Grossman et al 2006: 987). As g-aminobutyric acid (GABA) agonists can impair memory GABA antagonists are also being developed (Association of the British Pharmaceutical Industry). 2.4. Potential herbal treatment strategies In view of the hypothesised pathological sequelae, risk factors and current orthodox treatment of dementias, efficacious herbs for these conditions could potentially have one or more of AChE-inhibiting or cholinergic, antidepressant, hypotensive, hypoglycaemic, antioxidant, anti-inflammatory, GABA modulator, nicotinic agonist, testosterogenic and oestrogenic actions. According to Kennedy and Scholey (2006: 4614) orthodox AChE inhibitors are not well tolerated by patients as they are toxic alkaloids and European plants traditionally used for cognitive enhancement may therefore provide non-alkaloid safer alternatives. To this end Salvia officinalis, Salvia lavandulaefolia and Melissa officinalis, members of the Labiatae family (Lamiaceae), have been extensively investigated in vitro. 2.4.1. Salvia spp. Salvia is the largest genus in the Labiatae family with over 700 species. The most common European species are Salvia officinalis L (garden or common sage) (Figure 1) and Salvia lavandulaefolia Vahl (Spanish sage), both of which originate on the shores of the Mediterranean (Kennedy and Scholey 2006: 4614). S. officinalis is an aromatic, evergreen shrub up to 75 cm in height with greyish-green oblong to lanceolate opposite leaves covered in a fine down. It has bluish-violet, two-lipped flowers arranged in whorls (Wildwood 1998: 202). S. lavandulaefolia has narrower leaves and a lower spreading habit (Sergei Savelevs Database). Sage was used in medieval Europe as a tisane for prolonging life and is a traditional spring tonic for strengthening weak constitutions (Lipp 1996: 63). According to Culpepper (1826: 147) ‘Sage is of excellent use to help the memory, warming and quickening the senses and an old country remedy, which indicates its efficacy for inflammation: ‘A sunburnt face is eased by washing with sage tea (Page 1978: 41). Other traditional uses are for headaches and migraine (Page: 34). The major active constituents of the leaves of both species are believed to be the volatile oils (1.0-2.8%), containing monoterpenes such as a-pinene, b-pinene, 1-8-cineole, camphor, geraniol and thujone (Kennedy and Scholey 2006: 4615). S. officinalis contains around 50% a- and b- thujone whereas only traces have been found in S. lavandulaefolia. As thujone, a terpenoid ketone, is potentially neurotoxic, S. lavandulaefolia may provide a safer alternative than S. officinalis to orthodox dementia drugs (Perry et al. 1999: 530). However, S. officinalis is toxic only at doses of over 15 g (Grainger-Bissett and Wichtl 2001: 441) but the oil should not be ingested. Both species contain polyphenolic compounds including rosmarinic acid, methyl carnosate, luteolin, luteolin-7-0-glucoside and caffeic acid (Kennedy and Scholey 4615), triterpenes eg oleanic acid and the flavonoids 5-Methoxysalvagenin (Barnes et al. 2002: 408) and hispidulin (Johnston and Beart 2004: 809). 2.4.2. Melissa officinalis M. officinalis L (balm, lemon balm) (Figure 2) originates from the eastern Mediterranean region and western Asia and is now widely cultivated in the west (Grainger Bissett and Wichtl 2001: 329). It is a bushy perennial, about 60 cm high with bright green, lemon-scented leaves in opposite pairs. Small labiate flowers grow in whorls and change colour from pale yellow to white or pale blue. Fresh leaves should be collected when young (Wildwood 1998: 175). It has been in medicinal use as a nervous system restorative for over 2000 years (Kennedy and Scholey 2006: 4617). The London Dispensary (1696 cited in Grieve 1931) states: ‘An essence of Balm, given in Canary wine every morning will renew youth, strengthen the brain John Evelyn wrote: ‘Balm is sovereign for the brain, strengthening the memory and powerfully chasing away melancholy (cited in Grieve 1931). There are no known contraindications or adverse effects (Barnes et al. 2002: 339). M. officinalis contains 0.2-0.3% essential oil (EO) consisting of over 70 components including around 60% monoterpenoid aldehydes and over 35% sesquiterpenes. The principle monoterpenes include citronellol, neral, geranial, methyl citronellate, ocimene; major sesquiterpenes include b-caryophylene and germacrene D. The herb also contains flavonoids, caffeic and chlorogenic glycosides, polyphenolic acids such as rosmarinic acid, and triterpenes (Granger Bissett and Wichtl 2001: 330). 2.5. Possible mode of action of phytochemical constituents in dementia 2.5.1. Antioxidant properties Numerous studies have been conducted on Salvia officinalis in a search for natural antioxidants to use in the food industry. Consequently, results of chemical tests on purified extracts of the herb have suggested that phenolic compounds rosmarinic acid, carnosic acid, carnosol, carnosoic acid, rosmadiol, rosmanol, epirosmanol, isorosmanol, galdosol methyl carnosate, 9-erythrosmanol and luteolin-7-0-glucopyranoside have significant antioxidant activity (Bertelsen et al 1995: 1272; Cuvelier et al. 1994: 665; Pizzale et al. 2002: 1651; Miura et al. 2002: 1848; Wang et al. 1998: 4869). S. lavandulaefolia dried leaf extracts in ethanol, chloroform and water, and various EO monoterpenes were assayed for antioxidant properties in phospholipid microsomes. The extracts and monoterpenes a-pinene, b-pinene, 1-8-cineole, camphor and geraniol and thujone all showed significant antioxidant activity (Perry et al. 2001: 1351). The extracts showed greater antioxidant activity than any individual monoterpenes, which suggested a synergistic effect (Perry et al. 1352). Ferreira et al. (2006: 35) measured the antioxidant properties of EOs, decoctions and ethanolic extracts of M. officinalis and S. officinalis relative to b-carotene. The EO and decoctions of both herbs showed significant antioxidant activity. Lima et al (2007) found methanolic and aqueous extracts of S. officinalis prevented lipid peroxidation in hepatoma cells. As there were more phenolics in the methanol extract it was thought there were other antioxidant compounds in the aqueous extract. Ethanolic EO extract from dried M. officinalis investigated for its ability to inhibit lipid peroxidation in vitro showed a dose-dependent (10-20 mg) 80-90% protection of linoleic acid from peroxyl radical attack. As no rosmarinic acid was detected in the EO the antioxidant action was attributed to squalene (Marongiu et al. 2004: 790). Considering there are potentially 70 constituents in the EO it is unlikely that this would have been the only active phytochemical but composition of the oil varies according to harvesting, origin and climate (Grainger-Bissett and Wichtl 2001: 329). Interestingly, M. officinalis prepared as a tea demonstrated significant antioxidant capacity, which corresponded to high phenolic content, when assayed with the ABTS (2,2/-azinobis 3-ethylbenzothiazoline-6-sulfonic acid) radical decolourisation assay (Ivanova et al. 2005: 147). 2.5.2. Anti-inflammatory activity Chloroform, aqueous and ethanol extracts and monoterpenes of S. lavandulaefolia, were tested for their ability to inhibit formation of pro-inflammatory eicosanoids thromboxane B2 (TXB2) and leukotriene B4 (LTB4) in leucocytes (Perry et al. 2001: 1348). The chloroform and ethanol extracts showed significant inhibition of LTB4. Alpha-pinene and geraniol showed weak selectivity for LTB4 and TXB2 respectively (Perry et al. 1351). The results support the traditional use of S. lavandulaefolia as an anti-inflammatory herb but indicate that it is the sum of the whole plant phytochemicals acting in synergy that are likely to contribute to this action. A standardised ethanolic extract containing 9.9% rosmarinic acid (RA) from the leaves of S. officinalis reduced Ab-induced neuronal cell death, Ab-induced lipid peroxidation, reactive oxygen species formation, DNA fragmentation and tau protein hyperphosphorylation in vitro (Iuvone et al. 2006: 1143). Kimura et al (1987) found rosmarinic acid (RA) had the ability to inhibit pro-inflammatory cytokines in human polymorphonuclear leucocytes (PMNs) in vitro. As both species contain RA these results suggest antioxidant, anti-inflammatory and neuroprotective properties of M. officinalis and the Salvia spp. against Ab-induced neurotoxicity. 2.5.3. Oestrogenic activity A range of concentrations of EO, ethanolic, chloroform and aqueous extracts and isolated monoterpenes of S. lavandulaefolia were assayed in yeast culture for oestrogen-binding properties. The EO showed weak oestrogenic activity at low concentrations. The aqueous and ethanolic fractions and geraniol showed significant oestrogenic activity (Perry et al. 2001: 1352). The results of this experiment support S. lavandulaefolias use as an oestrogenic herb. The effects of S. officinalis in combination with Medicago sativa were assessed on menopausal symptoms related to oestrogen deprivation. Hot flushes and night sweats were completely eliminated in 20 out of 30 women (De Leo et al. 1998: 207). These effects were attributed to dopaminergic actions but it is not clear for which herb. S. officinalis does, however, contain geraniol found to be oestrogenic in vitro (Perry et al. 2001: 1352). 2.5.4. Acetylcholinesterase inhibitory activity M. officinalis EO demonstrated strong AChE inhibition in homogenised human brain tissue but ethanolic extract of the dried leaf had no effect. Ethanolic fresh leaf extract showed a weak effect (Perry et al. 1996: 1064). Conversely, when EOs and ethanolic extracts of M. officinalis were assayed in solution with AChE negligible results were obtained for its inhibition by EO and significant results for its ethanolic extract (Ferreira et al. 2006: 34). Dried, reconstituted ethanolic, ethyl acetate or aqueous extracts of M. officinalis, yielding 10mg/ml, demonstrated weak AChE inhibitory activity when assayed in a chemical system using thin layer chromatography (Salah and Jà ¤ger 2005: 146). The herbs were purchased from local suppliers in the Lebanon so their quality is unknown. S. officinalis EO and ethanolic extract assayed in solution with AChE showed moderate AChE inhibitory activity (Ferreira et al. 2006: 34). Moderate (dose-dependent) AChE and weak BChE inhibition was demonstrated by ethanolic extracts of fresh and dried S. officinalis and S. lavandulaefolia in human brain homogenates. The EOs had significant effects but not the individual constituents (camphor, thujone, cineole, caffeic acid and borneol) (Perry et al. 1996: 1066). The findings suggest a major synergistic effect of the constituents (Perry et al. 2000: 895), which was later confirmed by Savelev et al. (2003: 667). The results for camphor conflict with another experiment in which S. lavandulaefolia EO and isolated monoterpenes a-pinene, 1-8-cineole and camphor demonstrated AChE inhibitory activity in human erythrocytes. Ethanolic extracts of dried S. officinalis, S. lavandulaefolia and M. officinalis were assayed for acetylcholine (ACh) receptor activity in human brain homogenate. All plants demonstrated ACh receptor activity and M. officinalis had the highest nicotinic displacement value (Wake et al. 2000: 108). 2.5.5. GABA modulation Methanol extract from S officinalis leaves revealed the flavonoids apigenin, hispidulin and cirsimaritin functioning as benzodiazepine receptor-active components (Kavvadias et al. 2003: 113), suggesting a potential calming effect for the herb, which may be relevant to AD. 2.6. Evaluation of in vitro studies According to the results all three herbs may have AChE inhibitory, anti-inflammatory and antioxidant properties, and S. lavandulaefolia and S. officinalis may have and oestrogenic properties (Appendix I, Table 1, page 36) and a sedative effect for S. officinalis. Although these results are interesting in vitro systems cannot be extrapolated to humans and clinical evidence is necessary to support findings. For example, they cannot determine effective human dosage or mode of administration. They largely do not account for potential synergistic effects of the herbs nor do they provide an indication of in vivo physiological, pathological and genetic, or environmental, influences. Furthermore, the extent to which phytochemicals in herbs are effective in dementia may depend upon their bioavailability in the brain (Anekonda and Reddy 2005: 371). It is worth noting, however, that as terpenoids tend to be lipophilic they are able to cross the blood brain barrier (Houghton and Howes 2005: 12). Some results are conflicting but they may depend on methodological quality and design. The experiments cited above vary widely in their approach with regard to extraction methods and assay methods. Savelev (2003: 667) has demonstrated how two different methods used for exploring interactions between the same agents may give different results when applied to the same set of data. Consistency of results may also be affected by differences in harvesting times and quality of herbs. Results for M. officinalis are particularly inconsistent but, according to Perry et al. (1996: 1068) most commercial sources of the EO are adulterated. Additionally, variation in media composition is known to affect the outcome of in vitro tests (Maurer and Kuschinsky 2006: 73). Consequently, in vitro experiments can only provide an indication of the clinical efficacy of therapeutic interventions. However, despite the inherent difficulties of in vitro research with herbs, there is considerable consistency with their potential value in dementia prophylaxis and management (Appendix I, Table I, page 36). Promising results in vitro of constituents of plants traditionally used to enhance memory, and subsequent interest in their potential actions in the brains of human patients, has generated clinical trials of M. officinalis and Salvia spp. for dementia. These will be reviewed. 3. Method A computerised literature search was conducted on the Allied and Complementary Medicine Database (AMED) including CINAHL Database, EMBASE, Pascal Biomed, Biological Abstracts, RCN Journals Database and IPA (International Pharmaceutical Abstracts); PubMed, the Cochrane Collaboration, Bandolier, the NHS Centre for Reviews, The National Research Register, ADEAR (Alzheimers Disease Education and Referral Centre database), PLoS (Public Library of Science), Herbalgram and Alt HealthWatch as well as hand-searching in books and journals. Literature searches dated back to 1985 and the final search was in April 2007. Key words in medical subject headings (MeSH) for an initial search in various Boolean combinations were: memory, cognitive dysfunction, dementia, Alzheimers, herbal, botanicals, phytotherapy, complementary and alternative. Also, in a second search these MeSh terms were entered with key herbs: Salvia, sage, Melissa and lemon balm. Inclusion criteria Controlled clinical trials, observational studies and case reports. Herbs for which there are at least two clinical studies in relation to cognitive enhancement. Exclusion criteria Due to the limitations and ethical considerations of animal experiments the review is restricted to human trials. Trials with combined preparations are excluded. Due to time constraints and a restriction to papers in the English language a complete systematic review is not viable at this time. To eliminate