Relationship Between Serum Isoflavone Levels and Disability-Free Survival Among Community-Dwelling Elderly Individuals: Nested Case–Control Study of the Tsurugaya Project

Abstract

Background.

The longer healthy life expectancy observed in Japan may be partly attributed to the Japanese diet. The researchers sought to examine whether serum isoflavone levels are associated with disability and death.

Methods.

The researchers used a nested case–control study to compare serum isoflavones (daidzein, genistein, glycitein, and equol) levels between 165 participants that died or were certificated as disabled (cases) and 177 controls. Disability was defined by certification of long-term care insurance. Conditional logistic regression models were used to calculate the risk of isoflavones for the composite outcome.

Results.

The proportion of cases was lower in the group with the highest levels of equol (34/91, 37%) compared with equol nonproducers (84/161, 52%). The risk of disability or death among equol producers remained reduced after adjusting for age and sex (odds ratio: 0.55, 95% confidence interval: 0.33–0.93). In a multivariate model, this risk was also unchanged (odds ratio: 0.51, 95% confidence interval: 0.27–0.96). There were no significant associations between daidzein, genistein, and glycitein with the composite endpoint.

Conclusions.

Higher serum equol levels, but not any other isoflavones, were inversely associated with the composite endpoint of disability and death. Although it cannot be concluded that equol per se has preventive effects on disability or death, higher equol levels appear associated with better health.

According to the Health Report published in 2004 by the World Health Organization, both healthy life expectancy at age 0 and 60 years were the longest in Japan compared with all other countries in the world (1). Therefore, it might be important to explore the determinants of the prolonged healthy life expectancy among Japanese. For instance, this longer healthy life expectancy may be, in part, attributed to the Japanese diet, which is high in foods such as fish, green tea, and soybean. In this study, the researchers focused on the relationship between soy isoflavones and disability-free survival.

A questionnaire survey is frequently used to estimate food consumption. However, in Japan, soybean is frequently used as a raw material in seasonings, such as miso paste and soy sauce. Therefore, it might be difficult to estimate the amount of soybean consumed from food frequency questionnaires. Indeed, the researchers previously found that there was a poor correlation between the assessment of soybean consumption by dietary record and food frequency questionnaire (2). Furthermore, studies on equol, a metabolite of daidzein that is produced by intestinal bacteria in some, but not all, adults have shown that those individuals who possessed equol-producing intestinal bacteria were more likely to benefit from soyfood consumption than those who did not (3,4). Importantly, equol production can only be assessed from blood or urine samples. Therefore, in the present study, the researchers decided to assess serum isoflavone as markers of soy intake.

Isoflavones, including genistein, daidzein, and glycitein, are found in soy and soy products (3,4). Isoflavones are known to have estrogenic effects, and consequently, may possess an ability to lower cholesterol and inhibit bone loss (3,4). Furthermore, emerging evidence suggests that isoflavones may be associated with lower risk of various cancers, including lung (5–7), prostate (8,9), and breast (10). Therefore, isoflavone levels may be associated with a lower risk of incident disability and mortality.

In the present study, the researchers used a nested case–control study design to investigate the relationship between serum isoflavone levels and risk of composite outcome of disability and death; a good indicator of healthy life expectancy.

Methods

Study Participants

As implemented in 2002 and 2003, the Tsurugaya Project was a comprehensive geriatric assessment of medical status, as well as physical and cognitive functions (11–17). The present study is based on data collected in 2002, as blood samples from that time period were available (16,17).

Of the 2,730 inhabitants aged 70 and older living in the Tsurugaya area of Sendai, Japan, 1,177 provided written informed consent to participate in the study. Because the researchers did not obtain agreement to review information regarding long-term care insurance (LTCI) in 2002, they requested agreement from the participants who underwent a comprehensive geriatric assessment in 2003. Of the 1,177 participants who underwent a comprehensive geriatric assessment in 2002, 671 underwent another comprehensive geriatric assessment in 2003, of which 657 agreed to a review of their LTCI information. The researchers excluded data from participants who were identified as having a disability on their LTCI certificate in 2003 (n = 55), participants who did not agree to their blood samples being analyzed or stored (n = 6), and participants who moved prior to being certified as disabled (n = 6). Of the 590 remaining participants, 208 developed a disability or died by June 30, 2009. The eligible participants were divided into eight strata according to sex and age (every 5 years; Table 1). Specifically, a select 178 cases (ie, participants that developed a disability or died) and 178 controls (ie, participants who lived without disability until June 30, 2009) were stratified. Because 14 serum samples (1 control and 13 cases) did not have sufficient serum to measure isoflavone levels (<1mL of serum), the researchers assessed a total of 342 participants in the present study (Table 1). The Ethics Committee of the Tohoku University Graduate School of Medicine approved the study protocol.

Serum Isoflavone Measurements

Blood samples collected under non-fasting conditions were immediately cooled at 4°C, centrifuged within 4 hours at 3,000g at 4°C for 10 minutes, and stored at −80°C. Concentrations of serum isoflavones, namely genistein, daidzein, glycitein, and equol, were measured using triple quadrupole tandem liquid chromatography–mass spectrometry (18). These measurements were determined at a clinical testing laboratory (SRL, Tokyo, Japan). Serum albumin, total cholesterol (TC), and casual glucose levels were also measured.

Other Measurements

Information regarding smoking status, drinking status, food intake, physical activity (PA), and history of disease was surveyed via a questionnaire, and drug information was confirmed by an experienced pharmacist. The participants were instructed to fill out a brief self-administered diet history questionnaire that included 75 food items with specified serving sizes described by natural portions or standard weight and volume measures of the servings commonly consumed in the study population. The mean daily intake of nutrients was calculated by using an ad hoc computer program developed to analyze the questionnaire (14). Participants indicated the mean frequency of consumption of green tea over the previous 1 month in terms of the specified serving size by selecting one of the eight frequency categories: almost never, less than 1 cup/wk, 1 cup/wk, 2–3 cups/wk, 4–6 cups/wk, 1 cup/d, 2–3 cups/d, and greater than or equal to 4 cups/d. Subsequently, the researchers summarized this information into three groups as follows: greater than or equal to 4 cups/d, 2–3 cups/d, and less than 2 cups/d (14). In terms of meat consumption, participants indicated the mean frequency of consuming a specified serving size of (1) chicken, (2) pork or beef, (3) ham, sausage, or bacon, and (4) liver over the previous 1 month by selecting one of the frequency categories: 2 times/d, 1 time/d, 4–6 times/wk, 2–3 times/wk, 1 time/wk, less than 1 time/wk, and none. These four frequencies were summed and a total meat consumption frequency was calculated. According to the distribution, the researchers classified participants into four groups of meat consumers: greater than or equal to 6.5 times/wk, 4.5–6.4 times/wk, 2.5–4.4 times/wk, and less than 2.5 times/wk. PA was first assessed by a self-reported single question on whether the participant had any PA in the past year. If “yes,” further questions were asked about the frequency and duration of walking, brisk walking, and sports. Each PA was classified into three categories on the basis of the frequency and duration of participation: (i) “high” PA (≥3–4 times/wk for ≥30 minutes each time), (ii) “low” PA (some PA in the past year, but not enough), and (iii) “none” (no PA). In this study, the researchers used three categories according to the distribution; participants who did any level of sports or high frequency of brisk walking, participants who did low frequency of brisk walking or any level of walking, and participants who did not have PA. Symptoms of depression were assessed via the Japanese version of the 30-item geriatric depression scale (12,14). The anthropometric variables (height and body weight) were recorded according to standard protocol. Body mass index was calculated as weight in kilograms divided by height in meters squared. Functional reach, which measures how far an individual can reach forward beyond their arm’s length while standing without losing balance, was measured and used as an indicator of physical function (13). The researchers used average stiffness of the right and left calcaneus as an indicator of bone mineral density. To assess stiffness, the researchers determined quantitative ultrasound parameters, such as the speed of sound (m/s), broadband ultrasound attenuation (dB/MHz), and the stiffness index (Stiffness), which was derived from speed of sound and broadband ultrasound attenuation. These parameters were measured in the right and left calcaneus using an Achilles Ultrasound Bone Densitometer (A-1000, GE-Lunar Corporation, Madison, WI) (19). Participants self-measured blood pressure at home using an automated device (HEM747IC: Omron Life Science Co. Ltd., Tokyo, Japan) (12). Participants were classified into groups based on the following categories: home hypertension, home borderline hypertension, and home normotension, according to the guidelines for home blood pressure (20). Participants prescribed antihypertensive medication were classified into the home hypertension group. The presence of diabetes was defined as a non-fasting blood glucose greater than or equal to 200mg/dL (11.1 mmol/L) or use of antidiabetic drugs. Impaired blood glucose was classified as non-fasting blood glucose between 140–199mg/dL (7.7–11.0 mmol/L). Participants were categorized into four TC groups: TC greater than or equal to 240mg/dL or use of cholesterol lowering drug, TC between 200–239mg/dL, TC between 160–199mg/dL, and TC less than 160mg/dL.

LTCI Certification

The researchers defined incident disability based on the LTCI certification system, which was launched as a national insurance scheme in April 2000 (21–23), and followed up those with certified incident disability until June 30, 2009.

Individuals aged 40–64 years and living in Japan, who were diagnosed with aging-related diseases (eg, Alzheimer’s disease and stroke), and those aged 65 and older, who were certified as requiring care, are eligible for benefits under the LTCI certification (24). To receive LTCI services, elderly individuals or their caregivers (family or professional) must contact the municipal government to have their care requirements officially certified (22). A trained local government official visits their home to evaluate nursing care needs via a questionnaire that assesses their current physical and mental status, and use of medical services (21). Standardized scores for physical and mental functioning, as well as the estimated amount of time required for care under nine categories (ie, grooming and/or bathing, eating, using the toilet, transferring, eating, assistance with instrumental activities of daily living, behavioral problems, rehabilitation, and medical services), are then calculated using software. Based on the national average values, it is decided whether applicants should be certified to receive LTCI services, and then, the system assigns a care needs level, which is determined by a certification board comprising physicians, nurses, and other experts in health and social services, who were appointed by the local mayor. The minimum standard for LTCI was care support level 1, which requires 25 minutes of total care/d (21,25).

Care needs are assessed according to seven levels, which closely correlate with the Barthel Index (Spearman’s coefficient: −0.86) and the Mini-Mental State Examination (Spearman’s coefficient: −0.42) (24,26). The outcome represents a comprehensive measure of disability among elderly individuals (26).

The Sendai City Municipal Authority provided annual information regarding LTCI certification, including the care level, date of certification, relocation, and death, between June 30, 2003 and June 30, 2009. The researchers defined incident disability as the certification of an individual by the LTCI to any level of care, and the date of disability as the first date of certification. Six participants were removed from the study due to relocation during the follow-up. The researchers used a composite outcome of disability and death, which can also be considered as an indicator of disability-free survival.

Statistical Analysis

The researchers classified participants into four groups based on quartiles of isoflavone levels. However, with respect to equol, almost half of participants were nonproducers of equol (ie, equol level < 1.0ng/mL), and consequently, the researchers used three categories, specifically, the nonproducers, lower half of equol producers (ie, equol level ≥ 1.0ng/mL), and upper half of equol producers.

The characteristics of cases and controls were compared using the χ² test or t test, as appropriate. Characteristics with respect to isoflavone levels were compared using the χ² test for categorical variables or ANOVA for continuous variables, as appropriate. A multiple logistic regression analysis was used to determine factors that predict equol production. This model included the following factors: smoking, alcohol drinking, blood pressure (ie, home hypertension, home borderline hypertension, and home normotension), blood glucose (diabetes, impaired blood glucose, and normal range), TC group, albumin, sex-specific quartile of functional reach, depression (geriatric depression scale ≥ 11), body mass index, sex-specific quartile of stiffness of calcaneus, history of cardiovascular disease, history of cancer, sex-specific quartile of total energy intake, green tea consumption, meat consumption, PA group, and serum daidzein concentration. The researchers also determined the factors that predict higher equol values in equol producers via a linear regression model with log-transformed equol and the above-mentioned factors.

A conditional logistic regression model on the age and sex strata was used to calculate the odds ratios (ORs) and 95% confidence intervals of isoflavones and risk of disability or death. The researchers used both crude and multiple adjusted models. In the multivariate model, the researchers adjusted for the potential confounders associated with isoflavone levels or incident disability or death mentioned above excluding serum daidzein level.

The researchers also calculated the risk of disability only (case = 142) or death only (case = 40). Furthermore, the researchers calculated the relationship of daidzein, genistein, and glycitein with composite outcome of disability and death among equol producers. The level of statistical significance was set at p < .05. All statistical analyses were performed using SAS software, version 9.1 (SAS Institute, Cary, NC).

Results

The baseline characteristics of control (ie, those that live without disability) and case (ie, those that developed a disability or died by the end of the follow-up period) groups are presented in Table 2. Due to age and sex matching, there were no differences observed in age and proportion of sex. TC levels, functional reach, stiffness, and PA were statistically lower among cases than controls (p < .05).

When the researchers compared the baseline characteristics of the participants according to each isoflavone type and their respective levels, there were no apparent age differences across all isoflavone groups. However, the proportions of women were generally lower in the higher ranges of all isoflavone levels. TC levels were also generally lower in the higher ranges of all serum isoflavone levels. The proportion of bone mineral density was significantly different in the glycitein group, but not in the daidzein, genistein, or equol groups. Although the proportion of women was lower at higher serum equol levels, the proportion of current smokers was also lower at the higher serum equol levels. When a multiple logistic regression analysis was performed to determine the predictors of equol producers, male gender was revealed as a significant predictor of higher equol production. Of all equol producers, men, nonsmokers, and participants with diabetes had higher log-transformed equol values. A higher concentration of daidzein also predicted higher log-transformed equol values (β = 0.003, p = .002).

The relationship between levels of different isoflavones and the composite endpoint of disability or death are presented in Table 3. There were no significant associations between daidzein, genistein, and glycitein with the composite outcome of disability or death after adjusting for age and sex. However, in the equol group, the risk of composite endpoint was lower with higher levels of equol, after control for age and sex (OR = 0.55; 95% confidence interval = 0.33–0.93). These associations remained unchanged when additional potential confounders were added to the model (OR = 0.51; 95% confidence interval = 0.27–0.96). Similarly, the relationship was unchanged when the researchers excluded participants who died without disability. Although participants with the highest isoflavone quartiles consistently showed lower risk of death (OR ≤ 0.46), this observation did not reach statistical significant due to the small number of deaths. The relationships between daidzein, genistein, and glycitein levels with the composite endpoint were also assessed among equol producers only (Table 4). The highest quartiles of the daidzein, genistein, and glycitein groups showed a nonsignificant trend for a lower risk of the composite endpoint (OR ≤ 0.86).

Discussion

The present nested case–control study is the first to show that higher levels of equol are associated with lower risk of disability or mortality. This inverse relation was also observed when the researchers compared participants with disability with controls. However, whether equol per se plays a causal role in increasing healthy life expectancy remains to be determined in future research.

There are several advantages of the present study. First, this study assessed comprehensive geriatric parameters, including physical function and depressive symptoms, which have been previously associated with incident disability or mortality. Second, the researchers used a nested case–control design, in which the measurement precedes the onset of the outcome, thus establishing the temporal relationship between the putative cause and the hypothesized effect. Third, the researchers used LTCI certification to assess disability, which is based on strictly established and uniform rules throughout Japan. This methodology enabled us to achieve higher follow-up rates, and eliminated potential selection bias in both the case and control groups. Nevertheless, this system is not perfect, as elderly individuals or their caregivers must initiate contact with the municipal government to receive LTCI services, and thus, some elderly individuals with disability may not be certified. However, this confounder would attenuate the relationship between equol levels and the composite outcome of disability and death. Therefore, the researchers’ conclusion that higher serum equol was associated with lower risk of composite endpoint of incident disability and death, should remain true. Another limitation of this study was that blood samples were collected in non-fasting conditions, potentially affecting serum isoflavone levels. Similarly, because equol production has been found to vary over time within individuals (27), the misclassification of equol status is a possibility. However, these limitations would only attenuate the relationship between equol levels and the composite outcome of disability and death.

In the present study, half of the participants were classified as equol producers, which corroborate the findings of previous reports from Asia (4,17). In the researchers’ attempt to determine the predictors of equol production, they were able to only identify sex. Interestingly, among all equol producers, men, nonsmokers, and participants with diabetes had higher equol values. However, previous studies in both Japan (28) and Europe (29) failed to find a relationship between smoking and equol levels. Additionally, a higher serum daidzein level predicted higher equol levels in equol producers. Therefore, a greater consumption of soy might increase equol level in equol producers. Thus, additional studies assessing the factors that affect equol levels are warranted.

Higher serum equol levels were found to be associated with a lower risk of disability and death. Initially, it was expected that this inverse association could be explained by superior bone mineral density (30). However, the correlation between equol levels and the bone mineral density of the calcaneus was not significant. Furthermore, adjusting for bone mineral density did not alter the risk of the composite endpoint of death and disability. Therefore, other mechanisms may play a role and should be considered. Unfortunately, the researchers’ study did not have any information with respect to the causes of disability or mortality, and consequently, they were not able to clarify the factors associated with reducing risk of the composite endpoint in the higher equol group than the other groups. There are, may be, several explanations for why equol was associated with disability and death. First, as mentioned in previous studies, the benefits of soybean consumption are greater in equol producers (4). Equol is known to have a stronger affinity for the estrogen receptor than any other isoflavone. Although statistical significance was not reached, the other isoflavones were also inversely associated with disability and death. Thus, this observation supports the above hypothesis. Also, it remains possible that rather than contributing to a better health outcome per se, equol may enhance bacterial activity or improve intestinal conditions, which in turn, contribute to better health. However, the researchers were not able to confirm this scenario in their study. Thus, to clarify whether equol per se decreases the risk of disability or mortality, future intervention studies on soybean intake among equol producers are warranted.

In conclusion, it was found that higher serum equol concentrations, and no other isoflavone, are independently associated with a lower risk of the composite endpoint of disability and death. However, whether equol per se has a direct causal effect on disability or mortality remains to be elucidated. Further studies, including randomized controlled trials, which clarify the role of equol in overall health, are warranted.

Funding

This study was supported by a Grant-in-Aid for Young Scientists (A) (21689018) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, a Health Sciences Research Grants for Health Service (H21-Choju-Ippan-001) from the Ministry of Health, Labour and Welfare, Japan, and the Japan Arteriosclerosis Prevention Fund.

Author Contributions

Study concept and design (A.H., K.O.-M., S.K., IT).

Acquisition of subjects and/or data (A.H., Y.S., Y.T., M.K,. T.T., K.O.-M., N.N., S.K., I.T.).

Analysis and interpretation of data (A.H., Y.S., Y.T., M.K., T.T., K.O.-M., N.N., S.K., I.T.).

Preparation of manuscript (A.H., A.F.).

Conflicts of Interest

There are no potential conflicts of interest that relate to the manuscript.

Acknowledgments

We would like to thank Misako Watanabe, Sumiko Aita, Yoshiko Nakata, Mika Wagatsuma, and Tomoko Muroi for their excellent technical assistance.

References

Author notes