OUP user menu

A randomized cross-over study of a traditional Japanese medicine (kampo), yokukansan, in the treatment of the behavioural and psychological symptoms of dementia

Katsuyoshi Mizukami, Takashi Asada, Toru Kinoshita, Katsuaki Tanaka, Kazuki Sonohara, Ryuhei Nakai, Kiyoshi Yamaguchi, Haruo Hanyu, Kiyoshi Kanaya, Tetsuya Takao, Masakatsu Okada, Sumio Kudo, Hayato Kotoku, Masahiko Iwakiri, Hirofumi Kurita, Toshihiro Miyamura, Yosuke Kawasaki, Koji Omori, Kazumasa Shiozaki, Toshinari Odawara, Tatsuya Suzuki, Shizuru Yamada, Youichi Nakamura, Kenji Toba
DOI: http://dx.doi.org/10.1017/S146114570800970X 191-199 First published online: 1 March 2009


The effectiveness and safety of yokukansan (TJ-54), a traditional Japanese medicine (kampo) for the treatment of the behavioural and psychological symptoms of dementia (BPSD), were evaluated in 106 patients diagnosed as having Alzheimer's disease (AD) (including mixed-type dementia) or dementia with Lewy bodies. Patients were randomly assigned to group A (TJ-54 treatment in period I and no treatment in period II; each period lasting 4 wk) or group B (no treatment in period I and TJ-54 treatment in period II). BPSD and cognitive functions were evaluated using the Neuropsychiatric Inventory (NPI) and the Mini-Mental State Examination (MMSE), respectively. Activities of daily living (ADL) were evaluated using Instrumental Activities of Daily Living (IADL) in outpatients and the Barthel Index in in-patients. For the safety evaluation, adverse events were investigated. Significant improvements in mean total NPI score associated with TJ-54 treatment were observed in both periods (Wilcoxon test, p=0.040 in period I and p=0.048 in period II). The mean NPI scores significantly improved during TJ-54 treatment in groups A and B (p=0.002 and p=0.007, respectively) but not during periods of no treatment. Among the NPI subscales, significant improvements were observed in delusions, hallucinations, agitation/aggression, depression, anxiety, and irritability/lability. The effects of TJ-54 persisted for 1 month without any psychological withdrawal symptoms in group A. TJ-54 did not show any effect on either cognitive function or ADL. No serious adverse reactions were observed. The present study suggests that TJ-54 is an effective and well-tolerated treatment for patients with BPSD.

Key words
  • Behavioural and psychological symptoms of dementia
  • traditional Japanese medicine (kampo)
  • yokukansan


It has been reported that the behavioural and psychological symptoms of dementia (BPSD), including aggression, agitation, screaming, wandering, hallucinations, and delusions occur in 20–80% of patients with dementia (Lawlor, 2004). These symptoms impair the activities of daily living (ADL) in patients with dementia, impose great burdens on caregivers, hasten hospital admission (Steele et al. 1990), and increase care costs (Beeri et al. 2002). Although some atypical antipsychotic agents have been used for the treatment of BPSD (Herrmann & Lanctôt, 2006), adverse reactions such as deterioration of cognitive function, somnolence, extrapyramidal symptoms, and gait disturbance are often present (Schneider et al. 2006). Recently, an increased mortality rate was reported in elderly patients with dementia who were using atypical antipsychotic agents (Schneider et al. 2005) as well as in such patients who were using conventional antipsychotic agents (Wang et al. 2005). There is therefore an urgent need to develop or find an effective BPSD treatment regimen that can be used more safely.

Kampo is the name given to Japanese traditional herbal medicine. Around the 5th century, Chinese herbal medicines were first introduced into Japan. Since then, through the accumulation of a vast body of clinical experience, kampo has developed and has been systematized. Kampo has attracted attention due to its purported ability to effectively treat disease while maintaining a favourable quality of life (QoL). More specifically, some papers have highlighted the efficacy of kampo in the treatment of dementia (Iwasaki et al. 2004; Terasawa et al. 1997; Yamaguchi et al. 2004).

Yokukansan (yi-gan san in Chinese) extract (TJ-54, Tsumura Co., Japan) is a kampo medicine containing JP Atractylodes lancea rhizome, JP poria sclerotium, JP Cnidium rhizome, JP Uncaria hook, JP Japanese angelica root, JP Bupleurum root, and JP Glycyrrhiza (JP: Japanese Pharmacopoeia) (Fig. 1). It has been prescribed for the treatment of emotional distress and agitation in infants. Recently, Iwasaki and colleagues reported that TJ-54 alleviates BPSD and problems with ADL in patients with dementia (Iwasaki et al. 2005a, b) and proposed its utilization as a new remedy to treat BPSD. We conducted this study in order to evaluate further the effectiveness and safety of TJ-54.

Fig. 1

Chemical profile of TJ-54 analysed by three-dimensional HPLC. Each peak of TJ-54 in the HPLC profile was identified by comparison of the retention times and UV spectra of chemically defined standard compounds.



We recruited patients aged 55–85 yr who met the diagnostic criteria for Alzheimer's disease (AD) using DSM-IV criteria (APA, 1994) and the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association criteria (NINCDS/ADRDA; McKhann et al. 1984), regardless of co-existence of vascular lesions (so-called mixed-type dementia). We also recruited patients who were diagnosed as having dementia with Lewy bodies (DLB) (‘probable DLB’ based on the diagnostic criteria of McKeith et al. 1996). They were further selected based on an additional inclusion criterion that they showed, in the Neuropsychiatric Inventory (NPI, 10 items; Cummings et al. 1994), a score of ⩾6 for at least one of ten items at baseline. Patients with serious physical conditions or patients with delirium caused by a disease other than dementia were excluded.

The concomitant drugs prohibited in this study were antipsychotic agents, mood stabilizers, anxiolytic agents (other than benzodiazepine anxiolytic agents) and kampo other than the study drug.

This study was carried out in compliance with the ethical principles embodied in the Helsinki Declaration (1975). The study protocol was approved by the internal review board at each study site. Written informed consent was obtained from each patient (if capable of giving consent) or his/her proxy consent provider prior to participation in this study.

Study design

This study was conducted using a cross-over method, with a 4-wk treatment period (TJ-54 7.5 g t.i.d.) and a 4-wk non-treatment period. The subjects were randomly assigned, using a central registration method, into group A (TJ-54 treatment in period I and non-treatment in period II) or goup B (non-treatment in period I and TJ-54 treatment in period II) (Fig. 2). A total of 106 subjects were randomized to either group A (n=54) or group B (n=52) according to a randomization schedule using outpatient/in-patient as a stratification factor and the institution as a block factor. This study was conducted at 20 Japanese medical institutions from October 2005 to April 2007. The physician in charge at each institution made a diagnosis and evaluated the clinical state, and the assessment was carried out by a rater independent of the physician.

This study was conducted as an open-label study since it was impossible to prepare a suitable placebo due to the unique flavor and odour of TJ-54.

Outcome measurement

The subjects were assessed three times, at baseline, at 4 wk (at the end of period I), and at 8 wk (at the end of period II) after starting the study. BPSD and cognitive functions were assessed using the NPI (10 items) and the Mini-Mental State Examination (MMSE; Folstein et al. 1975). Assessment of ADL was performed using the Instrumental Activities of Daily Living (IADL; Mahoney & Barthel, 1965) in outpatients, and using the Barthel Index (Lawton & Brody, 1969) in in-patients. Regarding NPI, the test–retest reliability was evaluated by intra-class correlation (ICC) calculated via a fitting linear mixed-effect model for the observations at each visit (Verbeke & Molenberghs, 2000).

Since pseudoaldosteronism has previously been reported as an adverse reaction to TJ-54, in addition to a routine survey of adverse events during the study, serum potassium levels and the degree of leg oedema were determined at baseline, at the end of period I, and at the end of period II.

Statistical analysis

The analysis set (full analysis set; FAS) included all the enrolled subjects except those who did not return to the study site after their first visit (subjects with no available data) and those whose drug-taking compliance appeared to be poor. Inter-group comparisons were performed according to the standard 2 × 2 cross-over method, using the change after cross-over as the response (Jones & Kenward, 1989). The changes throughout period I were also analysed and compared between the two groups. Where data were missing, the last observation carried forward (LOCF) method was used.

The Wilcoxon test (with continuity correction) was applied for efficacy evaluation using a two-sided type I error of 5%. The 95% confidence interval (CI) for each efficacy parameter was estimated by analysis of covariance (ANCOVA) using the baseline value and the category of outpatient/in-patient as covariates. In addition, for descriptive purposes, the intra-group changes in periods I and II were analysed by the signed-rank sum test using a two-sided type I error of 5%. No adjustment for multiplicity was performed.



The patients' backgrounds for all of the 106 subjects enrolled are shown in Tables 1 and 2. The average age was 78.7±5.4 yr for outpatients and 78.5±6.7 yr for in-patients. AD was the most common type of dementia in both outpatients and in-patients. The mean total NPI score at baseline was 27.1±12.7 and 24.2±14.8 for outpatients and in-patients, respectively. The mean MMSE score at baseline was 16.1±6.0 and 9.6±6.8 for outpatients and in-patients, respectively.

View this table:
Table 1

View this table:
Table 2

The FAS comprised 103 subjects and three subjects were excluded. One subject did not return to the study site after their first visit, and two subjects were excluded based on the physician's judgement regarding compliance.


Of the subjects who completed the study, 45 were in group A and 43 were in group B (Fig. 2). After cross-over (in period II), a significant difference was observed in the NPI score between TJ-54 treatment and non-treatment (p=0.048; point estimate of treatment effect: −1.4, 95% CI −3.0 to 0.2). Also in period I, the changes in NPI score were significantly different between TJ-54 treatment and non-treatment (p=0.040; point estimate of treatment effect: −5.2, 95% CI −10.1 to −0.3). The NPI score improved significantly in the TJ-54 treatment period in both groups (group A: from 24.0 to 19.7, p=0.002; group B: from 28.6 to 23.5, p=0.007), while no significant improvement was seen in the non-treatment period (Table 2). No rebound phenomenon of BPSD after the last dose of TJ-54 was seen in group A. The ICC of NPI was estimated as 0.731.

Among the subscales of NPI, agitation/aggression and irritability/lability were alleviated significantly by TJ-54 treatment in both groups, while delusions, hallucinations, depression, and anxiety were alleviated significantly in only one of the two groups (Table 3).

View this table:
Table 3

In subgroup analysis in terms of disease type, no significant difference was observed, but a significant alleviation of BPSD by TJ-54 treatment was seen in the AD subgroup (including patients with mixed-type dementia) (Table 2). In patients with DLB, a significant alleviation by TJ-54 treatment was only seen in group A (Table 2), but when the two groups were combined, the NPI score decreased in 11/13 patients with DLB (Fig. 3). No significant differences after TJ-54 treatment were seen in the MMSE, IADL, or Barthel Index (Table 2).

Fig. 3

Changes of Neuropsychiatric Inventory (NPI) score in patients with dementia with Lewy bodies. (a) Period I; (b) period II.


Throughout the study period, adverse events for which a causal relationship with TJ-54 could not be ruled out (adverse reactions) were noted in six subjects. Three subjects developed gastrointestinal symptoms (vomiting/diarrhoea, nausea, epigastric distress), and these patients stopped taking TJ-54. Thereafter, their symptoms disappeared immediately. Glycyrrhiza contains glycyrrhizin, which has a facilitating action on potassium excretion in the renal tubules but there is a risk that hypokalaemia may be induced. In fact, following TJ-54 treatment, the average serum potassium level in the subjects decreased by 0.20 mequiv/l, although the value remained within the normal range. Two subjects developed hypokalaemia (3.8→3.4, 4.1→2.2 mequiv/l), and one of the two also showed sedation. In these two patients, however, the potassium level quickly returned to the baseline level after they stopped taking TJ-54. In addition, one subject had leg oedema. No serious adverse reactions, such as extrapyramidal tract symptoms, anticholinergic symptoms, or delirium were observed.


The effect of TJ-54 on BPSD was recently reported by Iwasaki et al. (2005b). They conducted a 4-wk, randomized, observer-blinded comparative study in 52 elderly patients with BPSD, reporting that TJ-54 significantly alleviated BPSD compared with non-treatment (p<0.001). This finding is consistent with the present randomized cross-over multi-centre study using a larger sample size in which a significant improvement in total NPI score after 4-wk treatment with TJ-54 was demonstrated. It is noteworthy that in this cross-over study, the mean total NPI score in group A did not deteriorate after withdrawal of TJ-54; i.e. there was no rebound-related deterioration observed for at least 4 wk after the last dose of TJ-54. This finding suggests that the effect of TJ-54 may persist over a certain period. The reason for this apparent carry-over effect remains unclear, but one conceivable explanation is that the patient–caregiver relation improved due to alleviation of BPSD, and that this might have contributed, at least in part, to the observed carry-over effect. In addition, a recent study demonstrated that repeated treatment with yokukansan induced down-regulation of 5-HT2A receptor in the mouse prefrontal cortex (Egashira et al. 2008). Thus, there is a possibility that this mechanism may also be involved in the carry-over effect. The precise mechanism involved in the carry-over effect is an issue that requires future clarification. However, since the NPI score was significantly improved by TJ-54 treatment in period II, over and above the improvements present in group A, the present study demonstrated that TJ-54 effectively alleviates BPSD.

The present study, TJ-54 significantly alleviated BPSD in the 90 subjects with AD in both periods I and II. Moreover, in the 13 patients with DLB, statistically significant alleviation was seen in period I, and 11 of these patients showed a decrease in their NPI scores (Fig. 2). Iwasaki and colleagues also reported that hallucinations disappeared in 12/15 DLB patients after TJ-54 treatment (Iwasaki et al. 2005a). Both the present study and their study suggest that TJ-54 might be a promising agent in the treatment of BPSD in DLB patients.

In the NPI subscales, TJ-54 was found to alleviate various symptoms, such as delusions, hallucinations, agitation/aggression, depression, anxiety, and irritability/lability. In particular, agitation/aggression and irritability/lability were alleviated by TJ-54 treatment in both periods I and II, suggesting that these symptoms may be the most appropriate symptoms to target with TJ-54. Uncariae Uncis cum Ramaulau (JP Uncaria hook), which is one of the components of TJ-54, acts on 5-HT1A receptors (Kanatani et al. 1985). It has been reported that the density of 5-HT1A receptors in the brains of AD patients correlates negatively with aggressive behaviour (Lai et al. 2003) and that partial agonists of 5-HT1A have a beneficial effect on agitation and aggression (Cantillon et al. 1996; Sato et al. 2007). Therefore, it is possible that TJ-54 might have alleviated BPSD through its action on the 5-HT1A receptor. Uncariae Uncis cum Ramaulau also has an activating effect on the acetylcholine neurotransmitter system (Murakami et al. 2005). Furthermore, Angelicae radix (Japanese angelica root, another component of TJ-54, reportedly acts on the GABA transmission system (Liao et al. 1995). In addition, recent studies have demonstrated that TJ-54 inhibits 5-HT2A receptor activation (Egashira et al. 2008) and also modulates excitatory neurotransmitter systems (Takeda et al. 2008). It is, therefore, reasonable to suppose that these neurotransmission-regulatory effects of TJ-54 can alleviate various psychological and behavioural symptoms.

Seven adverse events were observed in six patients in this study, and three patients discontinued TJ-54 treatment because of gastrointestinal symptoms. In addition, two patients developed hypokalaemia. However, these symptoms disappeared immediately after discontinuation or completion of treatment, and no serious adverse effects were observed. Furthermore, no patients showed adverse reactions such as extrapyramidal symptoms, drowsiness, delirium, or hypotension, which are often observed in patients taking antipsychotic agents. The safety profile and the beneficial effects of TJ-54 observed in the present study show great promise for elderly patients with dementia, especially patients with DLB characterized by hypersensitivity to an antipsychotic agent (McKeith et al. 1996). Our results demonstrate that TJ-54 treatment would have no adverse effects on cognitive function and ADL. The study by Iwasaki et al. (2005b) reported improvement of ADL and absence of adverse effects on cognitive function. Thus, together these two studies suggest that TJ-54 would be effective and well tolerated in the treatment of BPSD in elderly patients.

In summary, the results of our preliminary study suggest that TJ-54 would be effective and well tolerated in patients with BPSD and that no rebound-related deterioration would occur after withdrawal of TJ-54. However, our study had several limitations. First, it was not a placebo-controlled study, and the sample size was relatively small. Second, the duration of both the non-treatment and the TJ-54 treatment periods was only 4 wk. In future, a double-blind, placebo-controlled study with a larger sample size will be necessary to confirm the effects and safety of TJ-54 on BPSD.



Statement of Interest:

Kyorin University, University of Tsukuba, and Nippon Medical School each receive funding from Tsumura & Co. not exceeding $10 000 per annum and not related to this study. K. Mizukami, T. Asada, T. Kinoshita, K. Tanaka, K. Sonohara, R. Nakai, K. Yamaguchi, H. Hanyu, K. Kanaya, M. Okada, M. Iwakiri, Y. Kawasaki, K. Shiozaki, T. Suzuki, S. Yamada, Y. Nakamura, and K. Toba have received personal compensation from Tsumura & Co. not exceeding $10 000 per annum.


View Abstract