Pharmacology, Biochemistry and Behavior 75 (2003) 651 – 659

www.elsevier.com/locate/pharmbiochembeh

Review article
Salvia for dementia therapy: review of pharmacological activity
and pilot tolerability clinical trial
Nicolette S.L. Perrya, Chloe Bollenb, Elaine K. Perryb,*, Clive Ballardc
a
Department of Pharmacology and Toxicology, University of Otago, Dunedin, New Zealand
b
Medicinal Plant Research Centre, MRC Building, Centre for Development in Clinical Brain Ageing, Newcastle General Hospital,
Westgate Road, Newcastle Upon Tyne NE4 6BE, UK
c
Wolfson Research Centre, Newcastle General Hospital, Westgate Road, Newcastle Upon Tyne, UK
Received 19 November 2002; received in revised form 7 April 2003; accepted 30 April 2003

Abstract

S. lavandulaefolia Vahl. (Spanish sage) extracts and constituents have demonstrated anticholinesterase, antioxidant, anti-inflammatory,
oestrogenic and CNS depressant (sedative) effects all of which are currently relevant to the treatment of Alzheimer’s disease (AD). The
essential oil inhibits the enzyme acetylcholinesterase (AChE) from human brain tissue and bovine erythrocyte and individual monoterpenoid
constituents inhibit AChE with varying degrees of potency. In vivo AChE inhibition of select brain (striatal and hippocampal over cortical)
AChE was obtained following oral administration of the essential oil to rats. In a study in healthy volunteers essential oil administration
produced significant effects on cognition. In a pilot open-label study involving oral administration of the essential oil to patients with AD, a
significant increase in diastolic and systolic blood pressure was observed in two patients, however this may have been due primarily to
preexisting hypertension and there were no abnormalities in other vital signs or blood samples during the trial period. Although an open label
trial is not free from practice effects or rater – caregiver expectations, statistically significant differences between baseline and 6 weeks
treatment were a reduction in neuropsychiatric symptoms and an improvement in attention.
D 2003 Published by Elsevier Inc.

Keywords: S. lavandulaefolia; Essential oil; Alzheimer’s disease; Tolerability; Monoterpenoids; Anticholinesterase; Antioxidant; Anti-inflammatory;
Oestrogenic

1. Introduction anti-inflammatory, antioxidative and oestrogenic mecha-

nisms, nicotinic receptors, nerve growth factors and the
The majority of FDA-approved drugs for Alzheimer’s formation of neurofibrillary tangles and Ah plaques—the
disease (AD) (e.g., tacrine, donepezil, rivastigmine, galant- hallmarks of the diseased brain (Cutler and Sramek, 2001;
amine) act by countering the cholinergic deficit associated Fluck et al., 2000; Wolfe, 2001; Zandi and Breitner, 2001). It
with the cognitive dysfunction and are based on inhibition of is increasingly evident that preventative and symptomatic
the enzyme acetylcholinesterase (AChE) (Giacobini, 2000, treatment of AD will become a multitarget drug strategy.
2001; Talesa, 2001). Peripheral cholinergic (gastrointestinal) The premise of traditional herbal medicine is treatment of
adverse effects for currently used cholinesterase inhibitors the whole disorder as opposed to single isolated symptoms
are common as well as other side effects such as hepatotox- of the disease. Numerous herbal extracts, containing several
icity (tacrine). More recently, the uncompetitive NMDA (N- active constituents and often more than one plant species,
methyl-D-aspartate) antagonist memantine that improves have been used to treat CNS-related disorders. Amongst
functioning and behavioural symptoms in patients with AD these, Salvia species (S. officinalis L. and S. lavandulaefolia
has been approved (Helmuth, 2002). Other targets include Vahl., S. miltiorrhiza Bung.) are prominent for their reputed
beneficial effects on memory disorders, depression and
cerebral ischaemia (Howes et al., 2003; Perry et al.,
* Corresponding author. Tel.: +44-191-273-5251; fax: +44-191-273-
2000a,b). Several studies have now been undertaken to
5291. scientifically investigate the traditional (CNS-related) use
E-mail address: e.k.perry@ncl.ac.uk (E.K. Perry). of S. lavandulaefolia (a sage species that contains only trace

0091-3057/03/$ – see front matter D 2003 Published by Elsevier Inc.

doi:10.1016/S0091-3057(03)00108-4
652 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659

amounts of the convulsant thujone) relevant to the treatment 1998; Dorman et al., 1995; Lu and Foo, 2001; Malencic et
of Alzheimer’s disease; these have centred on the activity of al., 2000; Perry et al., 2001; Zupko et al., 2001) while
the essential (volatile) oil (Perry et al., 1996, 2000, 2001, camphor (20 –30% of essential oil) has demonstrated pro-
2002; Savelev et al., 2003). oxidant effects in a liposome peroxidation preparation
(Perry et al., 2001) (Table 1). The herb constituent rosmar-
inic acid displays more potent radical scavenging activity
2. Pharmacological activities of Salvia lavandulaefolia than trolox (a derivative of a-tocopherol) (Lu and Foo,
relevant to Alzheimer’s disease 2000, 2002) and the inhibition of liposome peroxidation by
the monoterpenoids is considered weak, though significant
2.1. Antioxidant compared to the standard antioxidant propyl gallate (Perry et
al., 2001). Given that the monoterpenoids with antioxidant
Many Salvia species and their isolated constituents activity in this study were present in the essential oil at a
possess significant antioxidant activity in enzyme-depen- slightly higher relative percentage (collectively over 30%)
dent and enzyme-independent systems (Dorman et al., 1995; than camphor (27% of essential oil), the pro-oxidant activity
Hohmann et al., 1999; Lu and Foo, 2001; Malencic et al., of camphor may not have its effect in the whole essential oil.
2000; Zupko et al., 2001). A number of these have exhibited Further to this any potential synergistic or antagonistic
effects relevant to potential treatment of CNS related dis- interactions could change the antioxidant profile of a whole
orders (Howes et al., 2003), the flavonoid apigenin for extract and further studies, using other systems, should be
example has been shown to protect neurons against Ah undertaken as potential antioxidant activity can depend on
induced toxicity (Wang et al., 2001). S. lavandulaefolia the systems used and may also depend on concentration
ethanolic extracts (both the water-soluble and chloroform- (Motohashi et al., 2002).
soluble fractions), individual constituents of the essential oil
[the monoterpenoids 1,8-cineole (1), linalool (5), a(2)- and 2.2. Anti-inflammatory activity
h-pinene] and herb [the phenolic monoterpenoid carvacrol
(7), the flavone luteolin (8) and the phenolic rosmarinic acid In addition to antioxidant activity many Salvia species
(10)] have been reported to be antioxidant (Adam et al., and their isolated constituents demonstrate anti-inflamma-

Table 1
In vitro activities of extracts and constituents of S. lavandulaefolia relevant to the treatment of Alzheimer’s disease
Constituent (% essential oil) Antioxidanta Anti-inflammatoryb Oestrogenicc AntiChEd (IC50) Sedative/CNS depressante
EtOH extract B5 mg/ml B50 Ag/ml B1.25 mg/ml – –
CHCl3 fraction B5 mg/ml B50 Ag/ml  – –
H2O fraction B5 mg/ml B50 Ag/ml B5 mg/ml – –
Essential oil – – B0.13 Ag – 1.3 mg/ml B0.03 mg/ml –
1,8-Cineole (1) (15 – 30%) B0.1 M  0.2 M  0.001 – 2.25 mM B0.4 – 7 mM –
a-Pinene (2) (4 – 7%) B0.1 M B0.2 M  0.001 – 2.25 mM B0.67 mM –
h-Pinene (5 – 12%) B0.1 M –  0.001 – 2.25 mM B1.5 mM –
Borneol (3 – 15%) – – – B>2 mMe –
Camphor (3) (20 – 30%)  pro-oxidant  0.2 M e B>5 mM –
Geraniol (4) ( < 1%)  B0.2 M B0.1 – 2 mM B>5 mM –
a-Terpineol (6) ( < 1%) – – – Bf 5 mM Be (in vivo)
g-Terpinene ( < 1%) – – – B>5 mM –
Caryophyllene epoxide (10) – – – B>2 mMe –
Linalool (5) ( < 1%) Be – – B>5 mMe Be
Carvacrol (7) Be Be – – –
Rosmarinic acid (10) Be Be – – –
Luteolin (8) Be Be – – –
Genkwarin – Be – – –
Cirsimartin – Be – – –
Salvigenin – Be – – –
Cirsiliol (11) – – – – Bd
– = not assessed.
Unless otherwise stated, data are taken from Perry et al. (1996, 2000b, 2001).
a
Inhibition of lipid peroxidation in bovine phospholipid liposomes.
b
Inhibition of formation of TXB2 and LTB4 (calcium-ionophore-stimulated rat leucocytes).
c
Oestrogenic activity using a recombinant oestrogen-inducible screen in yeast.
d
Inhibition of bovine erythrocyte AChE using colorimetric assay of Ellman et al. (1961).
e
Adam et al., 1998; Atanasovo-Shopova et al., 1973; Buchbauer et al., 1993; Dorman et al., 1995; Lu and Foo, 2001, 2002; Malencic et al., 2000; Paladini
et al., 1999; Silva Brum et al., 2001; Tinwell et al., 2002; Zupko et al., 2001).
 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659 653

tory properties (Bingol and Sener, 1995; Howes et al., 2.4. Cholinesterase inhibition
2003; Maklad et al., 1999) (Table 1). Carvacrol (7) and a-
pinene have been shown to be anti-inflammatory constitu- A further activity, and one that originally stimulated
ents of S. lavandulaefolia (Bingol and Sener, 1995; Wagner research into the putative CNS beneficial effects of S.
et al., 1986), in addition, the latter monoterpenoid has been lavandulaefolia, is its ability to inhibit AChE. S. lavandu-
shown to inhibit the enzyme cycloxygenase, an activity that laefolia essential oil inhibits human brain (autopsy tissue)
may be of particular relevance to anti-inflammatory treat- AChE in vitro (IC50 0.07 mg/ml) (Perry et al., 1996) and
ment of AD (Zandi and Breitner, 2001). Other constituents bovine erythrocyte AChE in vitro (IC50 0.03 –5 mg/ml)
that have shown anti-inflammatory activity in various in (Perry et al., 2000b; Savelev et al., 2003). This inhibition has
vitro assays are the flavones genkwarin and luteolin (8) and been confirmed in vivo; oral administration of S. lavandu-
the 6-hydroxy flavones cirsimaritin and salvigenin and laefolia essential oil once daily for 5 days to rats resulted in
several polyphenolics including rosmarinic acid (10) (Bin- decreased striatal AChE activity at a lower dose and de-
gol and Sener, 1995; Wagner et al., 1986). S. lavandulae- creased striatal and hippocampal AChE activity at a higher
folia ethanolic extracts (the chloroform-soluble fraction dose; at both doses there was no change in the AChE activity
over the water-soluble fraction) and monoterpenoids pres- in the cortex (Perry et al., 2002). Thus, it is apparent that
ent in the essential oil, a-pinene (2) and geraniol (4), have following oral administration to rats one or more constituents
demonstrated weak (significant) inhibition of eicosanoid of S. lavandulaefolia essential oil or their metabolites, reach
synthesis, though there may be more potent constituents the brain (crossing the gastrointestinal and blood – brain
present in minute quantities in the essential oil (Perry et al., barriers) to inhibit AChE in select brain areas, consistent
2001). The effect of the total oil would be expected to be with evidence of inhibition of the brain enzyme in vitro.
weak, but in favour of inhibition of 5-lipoxygenase inhibi- The major (commercially obtained) monoterpenoid con-
tion, since a-pinene (5% of essential oil tested) showed stituents of S. lavandulaefolia essential oil inhibit bovine
weak selectivity for the pro-inflammatory eicosanoid leu- erythrocyte AChE with varying degrees of potency (Table
kotriene B4 (LTB4), whereas geraniol ( < 1% of essential 1) (Miyazawa et al., 1997; Perry et al., 2000b; Ryan and
oil) showed weak selectivity for thromboxane B2 (Perry et Byrne, 1988; Savelev et al., 2003). 1,8-Cineole (15 – 30% of
al., 2001). LTB4 is produced via the enzyme 5-lipoxyge- essential oil) and a-pinene (4– 7%) show the most inhibition
nase, the gene for which is up-regulated during neuro- (IC50 0.4 – 0.7 mM) that is less active by factor of at least
degeneration and although the role of this inflammatory 103 compared to the anticholinesterase (antiChE) alkaloid
mediator in AD is not entirely apparent, selective inhibition physostigmine. It was assumed that the antiChE activity of
may be therapeutically relevant (Sugaya et al., 2000). the major monoterpenoid constituents (camphor, 1,8-cine-
Screening in other anti-inflammatory assays that use medi- ole, borneol a- and h-pinene) were responsible for the
ators more specific to AD would be valuable, such as the antiChE activity of the whole essential oil. However, con-
classic complement cascade surrounding the site of Alz- stituents combined in a naturally occurring ratio were
heimer amyloid plaques and inhibition of interleukin-1h significantly less potent than that of the whole oil (Perry
and the inflammatory cytokine S-100h, levels of which are et al., 2000b; Savelev et al., 2003). It was proposed that the
also increased in the Alzheimer brain (Zandi and Breitner, monoterpenoids may act synergistically to inhibit AChE and
2001). a recent study (Savelev et al., 2003) examined putative
interactions (synergism, antagonism) or zero-interaction
2.3. Oestrogenic activity between the antiChE constituents of S. lavandulaefolia
essential oil. It was concluded that the inhibitory activity
In addition to antioxidant and anti-inflammatory activity, of S. lavandulaefolia essential oil results from complex
dose-dependent oestrogenic activity was demonstrated in an interactions, which produce both synergistic and antagonis-
ethanolic extract (which appeared to be concentrated in the tic responses between the constituent terpenoids. For exam-
water-soluble fraction) of S. lavandulaefolia (Perry et al., ple, minor synergy was found between 1,8-cineole (1) and
2001). The essential oil and constituent monoterpenoid a-pinene (2) and 1,8-cineole (1) and caryophyllene epoxide
geraniol ( < 1% of essential oil) also showed oestrogenic (10) and antagonism was found to occur between 1,8-
activity, the latter of which was dose dependent and weak in cineole (1) and camphor (3).
comparison to the standard oestrogenic compound, 17h- This antiChE activity S. lavandulaefolia essential oil
oestradiol (Table 1). The monoterpenoid camphor (3) was terpenoid constituents is of interest since previously identi-
shown to be inactive in yeast oestrogen receptor trans- fied antiChEs are amines, the naturally occurring type being
activation assay (Tinwell et al., 2002). The potential oestro- alkaloids and these compounds were not known to exist in
genic activity of the S. lavandulaefolia extracts, essential oil the Salvia species tested. Other nonmonoterpenoid com-
and constituents requires further investigation in an in vivo pounds present in S. lavandulaefolia herb may inhibit AChE
environment where metabolic processes and the effective- since the disesquiterpene gossypol (Ryan and Byrne, 1988),
ness of a compound-receptor complex in producing a gene the coumarin auraptene and the sesquiterpenoids nookatone
response may alter oestrogenic activity (Green et al., 1998). (Howes et al., 2003) and caryophyllene epoxide (10) (1% S.
654 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659

lavandulaefolia essential oil) (Savelev et al., 2003) are also cardiovascular (myocardial infarction in previous 3 months)
weak inhibitors of AChE. diseases; haematological or oncological disorder within
previous 2 years, history of alcohol or drug abuse, vitamin
2.5. Tolerability trial of S. lavandulaefolia essential oil in B12 or folate deficiency, alanine transaminase >180 U/l,
Alzheimer’s disease haemoglobin < 10, white blood cell < 4, creatinine >180; a
high bp (>100 mm Hg diastolic). Also because of the
Since the above effects of S. lavandulaefolia (essential possible hypoglycaemic effect of sage (Zarzuelo et al.,
oil) are currently relevant to treatment of AD and a recent 1990) patients with any form of diabetes were excluded.
placebo-controlled, double-blind, balanced, cross-over de-
sign study of S. lavandulaefolia in healthy volunteers 3.2. Intervention
(Kennedy et al., 2002) demonstrated that the essential oil
produced a number of significant effects on cognition 3.2.1. Extract
(including improvements in both immediate and delayed Capsules contained 50 Al essential oil of S. lavandulaefo-
word recall scores that were coupled with decrements in lia plus 50 Al of sunflower oil (Helianthus annuus) (as a
both accuracy and speed of attention and were associated carrier) and were maintained at 4 jC. The essential oil was
with reductions in self-rated ‘alertness’ and ‘calmness’) commercially extracted by Advanced Phytonics (Olway
clinical trials in AD are indicated. It was considered initially Works, Healey Road, Ossett, WF5 8LT, U.K.) from the
important to assess tolerability in patients with AD. A flowering tops of the plant (grown in Granada, Spain and
(phase-II) pilot open-label trial was carried out which, harvested in early Spring of 1999) and encapsulated by Power
although there was no placebo control, also endeavoured Health (Poklington, U.K.). The gas chromatography profile
to determine potential efficacy and to identify appropriate of the capsulated material (analysed 5 months after capsu-
drug dose ranges in order to maximize the potential for lation, by the Scottish Agricultural Council, Auchincruive,
demonstrating efficacy in subsequent extended clinical Ayr, Scotland) showing the main constituents of the essential
trials. This involved a limited number of patients (n = 11) oil is provided in Fig. 1 [BP20 column (25 m  0.25 mm; film
for a short period of time (6 weeks). thickness, 0.25 Am; with polar, bonded stationary phase of
polythene glycol); 50– 200 jC at 5 jC min 1 + 10 min;
injection and FID temperature, 250 jC; injection volume, 200
3. Methods Al]. GC analysis demonstrated 94 peaks, the major peaks
were of borneol, camphene, camphor (3), 1,8-cineole (1) and
3.1. Patient cohort a-terpineol (6), with only a trace of thujone consistent with
that in literature (Fournier et al., 1993).
The trial (conducted in 1999 in Newcastle Upon Tyne,
U.K.) included 11 patients (1 male and 10 female), aged 3.2.2. Dosage
76 –95 years in whom a diagnosis of mild to moderate Week 1: one capsule at 8 a.m.; Week 2: one capsule at
probable Alzheimer’s disease (NINCDS/ADRDA criteria; 8 a.m. and one capsule at 7 p.m.; Weeks 3 – 6: as above with
McKhann et al., 1984) had been established. Patients had a one additional capsule at 12:30 p.m. Capsules were dis-
MMSE (Mini-Mental State Examination) score of between pensed in weekly amounts in dark glass bottles with an audit
10 and 26 and Neuropsychiatric Inventory (NPI ) scores for of capsules taken by pill counts, checking medication sheets
items 3 and 9 were 0. All patients had vision and hearing and questioning the caregiver at each visit.
sufficient for compliance with testing procedures. The study
received Newcastle Joint Ethical Committee approval, and 3.3. Study design
all patients and their carers were fully informed and pro-
vided written and oral consent. Use of other drugs continued Vital signs (weight, physical and neurological examina-
(exceptions below) with no change in prescription during tion and heart rate), adverse events, capsule counts and drug
the trial period. Nine patients were in residential homes, two dispensation were recorded at the end of every week.
living at home—one was capable of taking medication Complete assessments of primary outcome measures were
herself and the other was living with her daughter. conducted at baseline and after 6 weeks of treatment. At
screening and at termination physical and extensive neuro-
3.1.1. Exclusion logical examinations were carried out, as well as complete
Patients with the following were excluded: baseline heart laboratory tests. Blood samples were screened for B12, red
rate below 50; use of tricyclic antidepressants, thioridazine cell folate, free T4, TSH, sodium, potassium, chloride,
or related neuroleptic; investigative drugs taken in the last 4 calcium, phosphate, urea, creatinine, albumin, alkaline
weeks; cholinergic therapy within the previous 6 months phosphatase, bilirubin, plasma glucose, cholesterol, ALT,
(and known hypersensitivity to cholinesterase inhibitors); full blood count, number and state of red blood cells (RBC)
clinically significant gastrointestinal, renal, hepatic, endo- (haemoglobin, mean corpuscular volume, mean corpuscular
crine, pulmonary, neurological (e.g., epilepsy, stroke) or haemoglobin, mean corpuscular haemoglobin concentration,
 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659 655

1
Fig. 1. Gas chromatographic profile of the major constituents of S. lavandulaefolia essential oil used (BP20 column; 50 – 200 jC at 5 jC min + 10 min;
injection and FID temperature 250 jC; injection volume 200 Al).

platelets, mean plasma volume, human calcitonin) and drug-induced cognitive performance change (Simpson et
AChE. Blood was collected in heparinised tubes, spun, al., 1988) and assesses performance on a number of tests
plasma withdrawn and stored at 20 jC until analysed. including word presentation, immediate word recognition
RBC AChE and plasma butyrylcholinesterase (BuChE) and simple reaction time. NPI, developed to assess psycho-
activity was assessed using a modified version of the pathology in dementia patients evaluates 12 neuropsychiat-
colorimetric method of Ellman et al. (1961; see Perry et ric disturbances common in dementia including delusions,
al., 2000b) and the catalytic concentration of enzyme (b) hallucinations, agitation, depression, anxiety and also appe-
expressed in units per litre (U/l) using the formula: b (U/ tite and eating changes in patients and is sensitive to
l) = F./e  DA/.Dt; where F = dilution factor for the RBC in treatment effects. Paired samples t tests were used to assess
sodium phosphate buffer; e = extinction coefficient in posttreatment differences from baseline for all the assess-
mM 1cm 1; DA = absorbance per minute. ments; P values of .05 or less were considered significant.

3.4. Outcome measures

4. Results
MMSE, performed at baseline and 6 weeks, is a simpli-
fied form of the cognitive mental status examination suitable 4.1. Tolerability
for AD patients and assesses response to drug treatment and
includes 11 questions (test time 5 – 10 min) (McShane et al., There were no patient withdrawals during the trial and no
1997; Folstein et al., 1975). The Cognitive Drug Research patient experienced any adverse physical or neurological
(CDR) computerised cognitive assessment system is a effects throughout the study. Blood samples, taken at the
validated, well-tolerated assessment that is sensitive to end of the trial, showed no statistically significant changes in
656 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659

any measure in any patient. Although not statistical signifi- between baseline and 6 weeks (Table 2). The CDR results
cant, there was an increase in bilirubin after 6 weeks treat- indicated no apparent adverse effects on cognition and there
ment, the mean increase between values before and after were several interesting trends indicating an improvement in
treatment was 1.30 Amol/l ( P=.057). Blood pressure was memory and attention that in one instance reached statistical
measured weekly and there were no changes in 9 of the 11 significance (Table 2). Thus, using a one-tailed paired t test
patients. In two patients with a history of hypertension, there (with level of significance set at .01 in view of the multiple
were increases in blood pressure at 3 weeks (highest dose); tests), vigilance task performance (VIGACC; detection of
diastolic rose to 100 and 110 mm Hg and systolic rose, in only target digits) improved ( P=.01). However, there were also
one of these, to 200 mm Hg which when included in total significantly more false alarms (i.e., responses when the
analysis gave a mean increase (n = 11) in diastolic bp of 15.09 target stimuli were not present) made during the number
( P=.025) and systolic bp of 6.19 ( P=.049). vigilance test (VIGFA; P=.03). There was also a trend for
improvement on an immediate word recognition task
4.2. Blood cholinesterases ( P=.14). There was a statistically significant improvement
in the NPI measured after 6 weeks compared to baseline
Blood cell samples were screened for RBC AChE and ( P=.024; Table 2) and this was also evident in the patient with
plasma BuChE and although results were variable there was the highest blood pressure.
a mean decrease (14.4%) in AChE activity as a result of
treatment (Table 2). Mean BuChE activity showed no
overall change (Table 2), with seven patients showing a 5. Discussion
decrease (3635.6 F 634.8 to 3354 F 477.8 U/l) and four
showing an increase (2793 F 651.9 to 3232.3 F 907.5 U/ This is the first open-label trial to evaluate the tolera-
l). In vitro (bovine erythrocyte) AChE inhibition by the bility of oral administration of S. lavandulaefolia essential
same batch of S. lavandulaefolia essential oil gave an IC50 oil in patients with AD. In addition to S. lavandulaefolia
value 0.116 Al/ml. essential oil proving safe at this dosage (patients, with the
exception of the two with a history of hypertension, did not
4.3. Clinical outcome measure experience any side effects while taking the capsules) there
were significant improvements in NPI and CDR scale
Although the trial was not originally designed to assess assessment. This is remarkable given the small heteroge-
efficacy in terms of cognitive function or behavioural fea- nous sample size, diverse MMSE scores and the variability
tures, the following observations were made. The mean in performance associated with a dementia population.
MMSE for the whole group remained statistically the same Given that two patients experienced an increase in blood
pressure (and since the patient population was heterogenous
Table 2
and limited) investigation into the long-term effects of S.
Mean differences in demographics and outcome measures following oral lavandulaefolia essential oil on blood pressure would be
administration of S. lavandulaefolia essential oil appropriate. The fact that sage (species not specified)
Assessment Baseline After treatmenta should not be administered to people with high blood
mean F S.D. mean F S.D. pressure is recorded in one of the medical herbals (Bartram,
(x – y) (x – y) 1995) and in future efficacy studies such patients (with a
MMSE 19.36 F 4.20 (10 – 26) 19.82 F 3.52 (14 – 26) history of hypertension) should be excluded or at least
NPI 5.70 F 6.25 (0 – 20) 2.70 F 5.29 (0 – 16)b monitored and withdrawn from the trial in the event of
CDR increased blood pressure. Although this side effect may
VIGACCc 81.21 F 35.50 (7 – 100) 83.70 F 32.34 (27 – 100)d
VIGFAc 1.91 F 3.24 (0 – 11) 3.44 F 4.93 (0 – 14)
impact the ultimate effectiveness of S. lavandulaefolia
IRECSDc 523 F 412 (22 – 1495) 972 F 914 (108 – 3142) essential oil as a potential treatment for AD, the increase
IRECRTMc 1748 F 1322 (470 – 1700) 2054 F 1319 (518 – 5574) in blood pressure in two patients coincided with adminis-
AChE (U/L)e 16,765 F 4808 14,353 F 5048 tration of the highest dose of (50 Al 3  /day). An equiv-
(10,268 – 20,952) (9939 – 26,182) alent dose of herb (S. lavandulaefolia leaf containing
BuChE (U/L)e 3329 F 742 (1988 – 4592) 3310 F 623 (2102 – 4308)
approximately 2% of essential oil) would be 2.5 g 3  per
Data are shown as mean F S.D. (data range).
a per day and in future trials a dose less than 50 Al 3  /day is
S. lavandulaefolia essential oil (6 weeks).
b
P=.0243; t = 2.324. likely to be the most appropriate since a lower dose of
c
VIGACC = vigilance task: ability to respond to target digits; essential oil has been shown to have cognitive effects in
VIGFA = vigilance task: false alarms; IRECSD= immediate recognition young volunteers (Tildesley et al., 2003). In addition the
standard deviation; IRECRTM = immediate recognition median reaction whole herb may have different/ no blood pressure effects
time.
d and a better and larger selection of patients may clarify the
P=.014; t = 2.684.
e
Catalytic units/l = F./o  DA/.Dt; where F = dilution factor for the effects of S. lavandulaefolia on hypertension.
RBC in sodium phosphate buffer; o = extinction coefficient in These initial clinical data demonstrate that S. lavandu-
mM 1cm 1; DA = absorbance per minute. laefolia essential oil results in improvements in memory and
 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659 657

attention in patients with AD and improves memory in cannot be excluded since thujone (present only in trace
healthy volunteers (Kennedy et al., 2002) and this corrob- quantities in S. lavandulaefolia) has been shown to compet-
orates the use of sage (S. lavandulaefolia) for memory itively antagonise GABAA receptor agonists (Hold et al.,
related disorders recorded in English medical herbals (Perry 2000) and camphor (3) (20 – 30% of essential oil) has con-
et al., 2000a). vulsive effects at high doses (Emery and Corban, 1999).
While a placebo controlled trial is now required, these Salvia species are widely used for their anxiolytic and
findings together with those of Tildesley et al. (2003) dem- sedative properties and this use is supported by in vitro and
onstrate that following oral administration of S. lavandulae- in vivo data that show a structurally diverse number of
folia essential oil significant cognitive or behavioural effects constituents [e.g., caffeic acid ethyl esters, cirsiliol (11),
can be observed in human subjects. Although it is possible carnosic acid, miltirone, linalool (5)] act as CNS depressants
that peripheral side effects, practice and rater – caregiver at the GABA chloride channel and/or at glutamate binding
expectations could play a role, it can be suggested that S. sites (Chang et al., 1991; Dentali and Hoffman, 1990; Silva
lavandulaefolia constituents and/or their metabolites reach Brum et al., 2001; Maklad et al., 1999; Marder et al., 1996;
the brain (crossing the gastrointestinal and blood – brain Rutherford et al., 1992; Silva Brum et al., 2001; Paladini et
barriers) and exert an effect on cognition. The biochemical/ al., 1999). Cirsiliol (11), one of many plant flavonoids that
neuronal systems affected are likely to include the cholinergic have been shown to have affinity for the benzodiazepine
system since the essential oil constituents inhibit erythrocyte receptor site (Ki 200 AM) (Paladini et al., 1999), is present in
AChE in vitro (Perry et al., 2000b) and rat brain AChE in vivo S. lavandulaefolia herb and is unlikely to be present in the
(Perry et al., 2002) and although results require duplication, essential oil. It is thus worth considering a combination of
blood samples from the present study suggest a trend towards herb extract and essential oil in future studies since potential
inhibition of erythrocyte AChE following oral administration antioxidant [carvacrol (7)], anti-inflammatory [rosmarinic
in humans. In addition, the glutamatergic system may be acid (9), genkwarin, luteolin (8), cirsimartin and salvigenin]
involved in the CNS effects of S. lavandulaefolia since the and other CNS-active constituents [cirsiliol (11)] are likely to
monoterpenoids linalool (5) and terpineol (together < 11% of be present in whole herb extracts (Table 1, Fig. 2). It is
the essential oil), have been shown to have in vivo CNS important in future studies to determine the mechanism of
depressant action (Atanasovo-Shopova et al., 1973; Buchba- action of (monoterpenoid) constituents present in essential
uer et al., 1993), and linalool (5) has been shown to compet- oils in general, since constituents and/or metabolites enter the
itively antagonise [3H]glutamate binding to receptors (Silva blood, cross the blood –brain barrier and reach the brain to
Brum et al., 2001) (Table 1). An effect of other constituents of have an observable CNS effect following inhalation and/or
the essential oil on glutamatergic and GABAergic systems topical application and since they are used in aromatherapy

Fig. 2. Constituents present S. lavandulaefolia and their putative pharmacological activities.

658 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659

worldwide (e.g., Atanasovo-Shopova et al., 1973; Buchbauer Acknowledgements

et al., 1993; Buckle, 1993; Jirovetz et al., 1990; Yamada et al.,
1994). The authors wish to acknowledge the following:
Thus, although evidence so far indicates that S. lavandu- Advanced Phytonics (AP; Maz Nicola and Rachel Price)
laefolia may be appropriate to the treatment of AD (Fig. 2), for provision of essential oil, extracted by the AP method;
further in vitro and in vivo studies are necessary to determine Katerina Svoboda (Department of Plant Biology, Scottish
structure –activity relationships and pharmacological inter- Agricultural Council, Auchincruive, Scotland) for analysis
actions between (monoterpenoid) constituents responsible of essential oil; CDR (Keith Wesnes and Gareth Ayre) for
for the observed effect. For example, in addition to the provision of cognitive/attentional test battery and data
potential interactive (antagonistic) antiChE effect of the analysis; additional clinical support from Katherina Reichelt
monoterpenoid camphor (20 –30% of essential oil) (Savelev and Katherine Wallace; permission from the directors of St
et al., 2003), camphor has shown pro-oxidant effects in one Catherine’s and White House Residential, Elderly Care
study (Perry et al., 2001) and is associated with hepato- and homes in Newcastle; blood cholinesterase analyses by Dr.
neurotoxicity at high doses (Aliye et al., 2000; Jimenez et al., Ed Okello; encapsulation by Power Health, Poklington; and
1983). It could be recommended that this constituent be Murray Craig (Department of Pharmacology and Toxicol-
extracted from the essential oil in further studies. A further ogy, University of Otago, New Zealand) for graphics
example of the variability of in vitro (isolated) constituent assistance.
effects compared to in vivo (combined) effects is that a
thujone 7-hydroxy metabolite attains much higher brain
levels than a-thujone (the naturally occurring form) and is References
less toxic to mice and less potent in a GABAA receptor
binding study (Hold et al., 2000). In addition information is Adam JN, Sivropoulou A, Kokkini S, Lanaras T, Arsenakis M. Antifungal
needed on the pharmacokinetics of S. lavandulaefolia con- activities of Origarnum vulgare subsp. Hirtum, Mentah spicata, Lav-
andula angustifolia and Salvia fruticose essential oils against human
stituents—how they are metabolised, what the maximum pathogenic fungi. J Agric Food Chem 1998;46:1739 – 45.
duration of action and half-life is and further how they may Akhondzadeh S, Noroozian M, Mohammadi M, Ohadinia S, Jamshidi AH,
interact with other drugs since monoterpenoids have been Khani M. Salvia officinalis extract in the treatment of patients with mild
shown to inhibit enzymes of the cytochrome P4502B sub- to moderate Alzheimer’s disease: a double blind, randomized and pla-
family (De-Oliveira et al., 1997). Issues concerning dosage cebo-controlled trial. J Clin Pharm Ther 2003;28(1):53 – 9.
Aliye UC, Bishop WP, Sanders KD. Camphor hepatotoxicity. South Med J
and extract form (herb and/or essential oil) are complex. 2000;93(6):596 – 8.
Complete correlation of the relative percentage of active Atanasovo-Shopova S, Rusinov KS, Boicheva I. Central neurotropic effects
constituents in preparations with the knowledge of pharma- of lavender essential oil: II. Effects of linalool and terpineol. Chem
cological interactions and biological effects may well be Abstr 1974;81:58351.
impractical based on present knowledge. Future trials should Bartram T. Encyclopaedia of herbal medicine. 1st ed. London: Grace; 1995.
p. 379.
therefore endeavour to control for variations in the chemical Bingol LM, Sener B. A review of terrestrial plants and marine organ-
constituents and provide quantitative chemical and bioactive isms having anti-inflammatory activity. Int J Pharmacogn 1995;33(2):
profiles in all reported studies. 81 – 97.
Buchbauer G, Jirovetz L, Jager W, Plant C, Dietrich H. Fragrance com-
pounds and essential oils with sedative effects upon inhalation. J Pharm
Sci 1993;82(6):660 – 4.
6. Conclusion Buckle J. Aromaratherapy. Nurs Times 1993;89(20):32 – 5.
Chang HM, Chui KY, Tan FWL, Yang Y, Zhong ZP. Structure – activity
S. lavandulaefolia has pharmacological activities relevant relationship of miltirone, an activi benzodiazepine receptor ligand iso-
to dementia therapy. In a pilot clinical trial, after 6 weeks of lated from Salvia miltiorrhiza Bunge (Danshen). J Med Chem 1991;34:
1675 – 92.
treatment with S. lavandulaefolia essential oil in patients with
Cutler N, Sramek JJ. Review of the next generation of Alzheimer’s disease
AD, statistically significant differences included an improve- therapeutics: challenges for drug development. Prog Neuropsychophar-
ment in accuracy for the detection of target digits (VIGACC), macol Biol Psychiat 2001;25:24 – 57.
an increase in the number of false alarm responses during the Dentali SJ, Hoffman JJ. 16-Hydroxycarnosic acid, a diterpene from Salvia
number of vigilance test (VIGFA) and a lower NPI score on apiana. Phytochemistry 1990;29(3):993 – 4.
both main items. Patients did not experience any adverse De-Oliveira ACAX, Ribeiro-Pinto LF, Paumgartten FJR. In vitro inhibition
of CYP2B1 monooxygenase by h-myrcene and other monoterpenoid
effects while taking S. lavandulaefolia essential oil capsules compounds. Toxicol Lett 1997;92:39 – 46.
although there was an increase in systolic and diastolic blood Dorman HJP, Deans SG, Noble RC. Evaluation in vitro of plant essential
pressure in patients with predisposing hypertension1. oils as natural antioxidants. J Essent Oil Res 1995;7:645 – 51.
Ellman GL, Courtney KD, Andres V, Featherstone RM. A new and rapid
colorimetric determination of acetylcholinesterase activity. Biochem
1
During the preparation of this paper a 4-month placebo-controlled Pharmacol 1961;7:88 – 95.
clinical trial of a salvia officinalis extract demonstrating significant Emery DP, Corban JG. Camphor toxicity. J Paediatr Child Health 1999;35:
improvement in cognition in mild/moderate Alzheimer’s disease (19 on 105 – 8.
sage, 20 on placebo) has been published, (Akhondzadeh et al., 2003). Fluck E, File SE, Rymer J. Ten years of hormone replacement therapy and
 N.S.L. Perry et al. / Pharmacology, Biochemistry and Behavior 75 (2003) 651–659 659

performance in tasks associated with frontal lobe function. Br J Clin Perry N, Howes M-J, Houghton P, Perry E. Why sage may be a wise
Pharmacol 2000;50:384 – 5. remedy: effects of Salvia on the nervous system. In: Kintzios SE, editor.
Folstein MF, Folstein SE, McHugh PR. ‘‘Mini Mental State’’, a practi- Sage: the genus Salvia. Netherlands: Harwood; 2000a. p. 207 – 23.
cal method for grading the cognitive state of patients for the clinician. Perry NSL, Houghton PJ, Theobald AE, Jenner P, Perry EK. In-vitro in-
J Psychiat Res 1975;12:189 – 98. hibition of erythrocyte acetylcholinesterase by Salvia lavandulaefolia
Fournier G, Pages N, Casperec I. Contribution to the study of Salvia lav- essential oil and constituent terpenes. J Pharm Pharmacol 2000b;52:
andulaefolia essential oil: potential toxicity attributable to sabinyl ace- 895 – 902.
tate. Planta Med 1993;59:96 – 7. Perry NSL, Houghton PJ, Sampson J, Theobald AE, Hart S, Lis-Balchin
Giacobini E. Cholinesterase inhibitors: from Calabar bean to Alzheimer M, et al. In-vitro activity of S. lavandulaefolia (Spanish sage) relevant
therapy. In: Giacobini E, editor. Cholinesterases and cholinesterase in- to treatment of Alzheimer’s disease. J Pharm Pharmacol 2001;53:
hibitors. London: Martin Dunitz; 2000. p. 181 – 226. 1347 – 56.
Giacobini E. Do cholinesterase inhibitors have disease-modifying effects in Perry NS, Houghton PJ, Jenner P, Keith A, Perry EK. Salvia lavandulae-
Alzheimer’s disease? CNS Drugs 2001;15(2):85 – 91. folia essential oil inhibits cholinesterase in vivo. Phytomedicine
Green PS, Gridley KE, Simpkins JW. Nuclear estrogen receptor-independ- 2002;9(1):48 – 51.
ent neuroprotection by estratrienes: a novel interaction with glutathione. Rutherford DM, Nielsen MPC, Hansen SK, Witt MR, Bergendorff O,
Neuroscience 1998;84:7 – 10. Sterner O. Isolation and identification from Salvia officinalis of two
Helmuth L. New therapies. New Alzheimer’s treatment that may ease the Diterpenes which inhibit t-butylbicyclophosphorol [35S]thionate bind-
mind. Science 2002;297(5585):1260. ing to chloride channels of rat cerebrocortical membranes in vitro.
Hohmann J, Zupko I, Redei D, Csanyi M, Falkay G, Mathe I, et al. Protec- Neurosci Lett 1992;135:224 – 6.
tive effects of the aerial parts of Salvia officinalis, Melissa officinalis, Ryan MF, Byrne O. Plant – insect coevolution and inhibition of acetylcho-
and Lavandula angustifolia and their constituents against enzyme-de- linesterase. J Chem Ecol 1988;14(10):1965 – 75.
pendent and enzyme-indepenent lipid peroxidation. Planta Med 1999; Savelev S, Okello E, Perry NSL, Wilkins RM, Perry EK. Synergistic and
65:576 – 8. antagonistic interactions of anticholinesterase terpenoids in Salvia lav-
Hold KM, Sirisoma NS, Ikeda T, Narahashi T, Casida JE. a-Thujone (the andulaefolia essential oil. Biochem Pharmacol Behav 2003;75:661 – 8,
active component of absinthe): g-Aminobutyric acid type A receptor this issue.
modulation and metabolic detoxification. PNAS 2000;97(8):3826 – 31. Silva Brum LF, Elisabetsky E, Souza D. Effect of linalool on [2H] MK801
Howes M-JR, Perry NSL, Houghton PJ. Plants with traditional uses and and [3H] muscimol binding in mouse cortical membranes. Phytother
activities relevant to the management of Alzheimer’s disease and other Res 2001;15:422 – 5.
cognitive disorders. Phytother Res 2003;17(1):1 – 18. Simpson PM, Wesnes KA, Christmas L. A computerised system for the
Jimenez JF, Brown AL, Arnold WC, Byrne WJ. Chronic camphor ingestion assessment of drug-induced performance changes in young, elderly or
mimicking Reye’s syndrome. Gastroenterology 1983;84:394 – 8. demented populations. Proc BPS 1988;711P – 2P.
Jirovetz L, Buchbauer G, Jager W, Raverdino V, Nikiforov A. Determina- Sugaya K, Uz T, Kumar V, Manev H. New anti-inflammatory treatment
tion of lavender oil fragrance compounds in blood samples. Fresenius J strategy in Alzheimer’s disease. Jpn J Pharmacol 2000;82:85 – 94.
Anal Chem 1990;338:922 – 3. Talesa VN. Acetylcholinesterase in Alzheimer’s disease. Mech Ageing Dev
Lu Y, Foo LY. Antioxidant activities of polyphenols from sage (Salvia 2001;122:1961 – 9.
officinalis). Food Chem 2001;75:197 – 202. Tildesley NTJ, Kennedy DO, Perry EK, Ballard CG, Savelev S, Wesnes
Lu Y, Foo LY. Polyphenolics of Salvia—a review. Phytochemistry 2002;59: KA, Scholey AB. Salvia lavandulaefolia (Spanish Sage) enhances
140 – 77. memory in healthy young volunteers. Biochem Pharmacol Behav
Maklad YA, Aboutabi EA, El-Sherei MM, Meselhy KM. Bioactivity stud- 2003;75:669 – 74, this issue.
ies of Salvia transsylvannica (Schur ex Griseb) grown in Egypt. Phyt- Tinwell H, Lefevre PA, Moffat GJ, Burns A, Odum J, Spurway TD, et
other Res 1999;13:147 – 50. al. Confirmation of uterotrophic activity of 3-(4-methylbenzylidine)-
Malencic DJ, Gasic O, Popovic M, Boza P. Screening for antioxidant camphor in the immature rat. Environ Health Perspect 2002;110(5):
properties of Salvia reflexa Hornem. Phytother Res 2000;14:546 – 8. 533 – 6.
Marder M, Viola H, Wasowski C, Wolfman C, Waterman PG, Medina JH, Wagner H, Wierer M, Bauer R. In vitro inhibition of prostaglandin biosyn-
et al. Cirsiliol and caffeic acid ethyl ester, isolated from Salvia guar- thesis by essential oils and phenolic compounds. Planta Med 1986;
anitica are competitive ligands from the central benzodiazepine recep- 184 – 7.
tors. Phytomedicine 1996;3(1):29 – 31. Wang C-N, Chi C-W, Lin Y-L, Chen C-F, Shiao Y-J. The neuroprotective
McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadian EM. effects of phytoestrogens on amyloid h protein-induced toxicity are
Clinical diagnosis of Alzheimer’s disease: report of the NINCDS- mediated by abrogating the activation of caspase cascade in rat cortical
ADRDA work group under the auspices of Department of Health neurones. J Biol Chem 2001;276(7):5287 – 95.
and Human Services Task Force on Alzheimer’s disease. Neurology Wolfe MS. Secretase targets for Alzheimer’s disease: identification and
1984;34:939 – 44. therapeutic potential. J Med Chem 2001;44(13):2039 – 59.
McShane R, Keene J, Fairburn C, Jacoby R, Hope T. Issues in drug treat- Yamada K, Mimaki Y, Sashida Y. Anticonvulsive effects of inhaling lav-
ment for Alzheimer’s disease. Lancet 1997;350:886 – 7. ender oil vapour. Biol Pharm Bull 1994;17(2):359 – 60.
Miyazawa M, Watanabe H, Kameoka H. Inhibition of acetylcholinesterase Zandi PP, Breitner JCS. Do NSAIDs prevent Alzheimer’s disease? And, if
activity by monoterpenoids with a p-menthane skeleton. J Agric Food so, why? The epidemiological evidence. Neurobiol Aging 2001;22:
Chem 1997;45:677 – 9. 811 – 7.
Motohashi N, Shirataki Y, Kawase M, Tani S, Sakagami H, Satoh K, Zarzuelo A, Risco S, Gamez MJ, Jimenez J, Camara M, Martinez MA.
et al. Cancer prevention and therapy with kiwifruit in Chinese Hypoglycemic action of Salvia lavandulifolia Vahl. ssp. oxyodon: a
folklore medicine: a study of kiwifruit extracts. J Ethnopharmacol contribution to studies on the mechanism of action. Life Sci 1990;47:
2002;81:357 – 64. 909 – 15.
Paladini AC, Marder M, Viola H, Wolfman C, Wasowski C, Medina JH. Zupko I, Hohmann J, Redei D, Falkay G, Janicsak G, Mathe I. Antioxidant
Flavonoids and the central nervous system: from forgotten factors to activity of leaves of Salvia species in enzyme-dependent and enzyme-
potent anxiolytic compounds. J Pharm Pharmacol 1999;51:519 – 26. independent systems of lipid peroxidation and their phenolic constitu-
Perry N, Court G, Bidet N, Court J, Perry E. European herbs with chol- ents. Planta Med 2001;67:366 – 8.
inergic activities: potential in dementia therapy. Int J Geriatr Psychiatry
1996;11:1063 – 9.