Associations between Traditional Chinese Medicine Body Constitution and Cardiovascular Disease Risk in a White population

Abstract

Background

Traditional Chinese medicine (TCM) has guided generations of practice on disease treatment and health maintenance. The TCM principles include the framework of body constitution. However, no study has assessed the body constitution in US population.

Methods

This is an ancillary study of the Personalized Prevention of Colorectal Cancer Trial which conducted in US in 2012-2016. 191 white participants were evaluated for body constitution type using a self-administered Traditional Chinese Medicine Questionnaire (English version). The body constitution subtypes and cardiovascular disease (CVD) risk were assessed.

Results

Fifty-seven (29.8%) were identified as balanced constitution (BC), while Blood-stasis (17.3%), Qi-deficient (13.6%), and inherited-special constitutions (10.5%) were the pre-eminent pathologic subtypes. Additional analyses investigated the relationship between CVD risk and body constitution subtypes. No major types of TCM body constitution were associated with the GCRS and other CVD biomarkers.

Conclusions

It is important to understand the underlying mechanisms contributing to these differences, which may not only help to understand the underlying mechanism for TCM, but also help to identify novel factors or mechanisms for CVD risk, prevention and treatment.

1. Introduction

In Traditional Chinese Medicine (TCM), a primary framework of health maintenance and surveillance hinges upon the classification of individuals into nine categories of “body constitutions” : Balanced constitution (BC), Qi-deficiency constitution, Yang-deficiency constitution, Yin-deficiency constitution, Phlegm-dampness constitution, Damp-heat constitution, Blood-stasis constitution, Qi-stagnation constitution, and Inherited-special constitution[1, 2]. These body constitutions help to draw a picture of an individual’s health status and have been used on a provider-specific basis to identify patients who may be at risk for future disease, such as cardiovascular disease (CVD). However, TCM has a very limited role in Western medical practice and part of the reason for this is a simple lack of data concerning its application in Western populations. What is more, there have been few studies regarding the relationship between TCM principles and contemporary models of health and disease.

CVD remains the leading cause of death worldwide, in particular due to its subtypes of coronary artery disease (CAD), as well as its manifestations of acute coronary syndrome (ACS) and myocardial infarction (MI) [3, 4]. The landmark Framingham Heart study provided data to establish what are now well-known risk factors of CVD, including cigarette smoking, cholesterol, blood pressure, and others [5]. Understanding and identifying these risk factors have led to major advances in health maintenance, including the advent of routine blood pressure monitoring, regular exercise, and lipid control. Nonetheless, heart disease remains the leading cause of death in the US. Thus, there is a need to identify other early markers of CVD or populations at greatest risk to further preventive cardiology. The relationship of TCM body constitutions and markers of chronic disease, including dyslipidemia, hypertension, diabetes, etc., have further been studied and helped foster increased interest in the TCM body constitution system, as they might serve as an early indicator of needed intervention for specific chronic illnesses [6–8].

All of these previous studies using TCM body constitution have been conducted in Chinese populations except for one study of 400 White college students attending three Beijing universities in China [9]. In addition, none of the previous studies have examined the associations between TCM body constitutions and CVD risk. To that end, the aim of this study is to investigate the ability for TCM body constitutions to identify participants at higher risk for CVD per general cardiovascular risk score (GCRS), an updated version of the original Framingham risk score (FRS), in a US population of white individuals; whether TCM body constitution is a better correlate of biomarkers of CVD (C-reactive protein (CRP) and uric acid) [10, 11] than the GCRS; and whether TCM body constitution is associated with weight, body mass index and waist-to-hip ratio. In the same way that the Framingham study gave physicians insight on which health parameters may predict later heart disease, we seek to determine the potential value of utilizing TCM body constitutions in the clinical setting to predict CVD development [12, 13].

2. Methods

We conducted a cross-sectional study using baseline data from the Personalized Prevention of Colorectal Cancer Trial (PPCCT; NCT01105169 at ClinicalTrials.gov). The primary aim of the PPCCT was to examine the effect of Mg supplementation on expression of carcinogenesis biomarkers in colorectal mucosa. Between 2010 and 2017, the PPCCT enrolled 250 participants, aged 40 to 85 years, with high risk of colorectal cancer from Vanderbilt University Medical Center. Inclusion criteria: we included aged 40 to 85 participants who had history of adenomas or hyperplastic polyps diagnosed from 1998 to 2014 or with a high risk of colorectal cancer, known TRPM7 rs8042919 genotype, and daily calcium intake between 700–2000 mg/day and a ratio of daily intake of calcium and magnesium greater than 2.6 based on two 24-hour dietary recalls. Exclusion criteria: those with a history of colectomy, inflammatory bowel disease, any organ transplantation, cancer other than non-melanoma skin cancer, gastric bypass, chronic renal diseases and hepatic cirrhosis, chronic ischemic heart disease, diarrhea, type I diabetes mellitus, pituitary dwarfism; current use of lithium carbonate therapy, blood anticoagulant drugs, digoxin and licorice; no contact information and informed consent or who were breastfeeding or pregnant were excluded from the study [14–16]. The study was approved by the Vanderbilt Institutional Review Board.

2.1. Constitution in Traditional Chinese Medicine Questionnaire

A self-administered Traditional Chinese Medicine Questionnaire (TCMQ) was implemented for a subset of participants enrolled from May 31, 2012 to Jan 30, 2016 (n=191). The original version of the standardized Constitution in TCMQ was developed and validated in China with sensitivity and specificity of predicting body constitution types ranging from 42.7% to 82.7% [17, 18]. The TCMQ was translated to English by the research team and tested in cognitive debriefing interviews before it was applied in the study. Eligible participants were asked to complete the self-administered Traditional Chinese Medicine Questionnaire (TCMQ) at home before clinical visits. Our trained research staff answered question during clinical visits and checked the completeness and quality of the questionnaire. The TCMQ consists of 60 questions, which fall into nine subscales that correspond to one of nine TCM body constitution types: BC (8 Items), Qi-deficiency (8 Items), Yang-deficiency (7 Items), Yin-deficiency (8 Items), Phlegm-dampness (8 Items), Damp-heat (6 Items), Blood-stasis (7 Items), Qi-stagnation (7 Items), and Inherited-special (7 Items). Six of the questions exhibit overlap between different subscales.

The 60-item questionnaire is graded on a 5-point Likert scale, ranging from 1 (not at all) to 5 (very much). Each of the nine subscales within the TCMQ assess one type of the TCM body constitution individually. A total score of each subscale is obtained by summing relevant item scores [19]. Then, the transformed score was generated for each type by using the following equation.

$$Transformscore=100\times \frac{\mathrm{O}\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}-\mathrm{N}\mathrm{u}\mathrm{m}.\mathrm{o}\mathrm{f}\mathrm{i}\mathrm{t}\mathrm{e}\mathrm{m}}{\mathrm{N}\mathrm{u}\mathrm{m}.\mathrm{o}\mathrm{f}\mathrm{i}\mathrm{t}\mathrm{e}\mathrm{m}\times 4}$$

Following the criteria, a higher score in a specific TCMQ body constitution subscale indicates a higher likelihood of the corresponding body constitution type, with a score of 30 being a “threshold” for case definition. To ultimately classify a participant within a specific TCM body constitution, the following algorithm is applied: when 1) the score for the BC subscale is greater than or equal to 60 and other type scores are less than 30, the study participant is diagnosed as “balanced constitution” which is considered the BC; when 2) the score for an imbalanced body constitution type (all other body constitutions) is greater than or equal to 40, then the participant is regarded as one or more of eight imbalanced types; when 3) a type score is between 30-40, the diagnosis will be made by a well-trained Chinese Medicine Practitioner. The experienced Chinese Medicine Practitioner made the diagnosis for those with the compound constitution based on their TCMQ as well as health assessment to ensure each person had one constitution.

For the reliability and validity of the self-administered Traditional Chinese Medicine Questionnaire (English Version), the construct validity was confirmed with scaling success rates that ranged from 75.0 to 100% (Supplemental Table 1), and confirmatory factor analysis indicate an acceptable model fit. The results of reliability for test-retest reliability (intra-class correlation coefficients) were 0.7-0.8 (Supplemental Table 2), the Cronbach’s alphas ranged from 0.44 to 0.72 during three-month period.

2.2. General cardiovascular risk score

Participants completed a telephone interviewer-administered survey at baseline to solicit information on medical history, tobacco and alcohol use, and other risk factors for colorectal cancer. Persons who smoked cigarettes regularly, measured by packs per day, during the previous 12 months were classified as smokers. During each clinic visit, the participant’s use of medications and biosamples were collected. Weight, height, waist and hip circumference as well as systolic and diastolic blood pressure, were measured. Body mass index (BMI) (kg/m²) and waist hip ratio (WHR) were calculated. Serum samples were assayed for a lipid profile (low-density lipoproteins cholesterol (LDL-C), high-density lipoproteins cholesterol (HDL-C), total cholesterol (TC), and triglycerides, c-reactive protein (CRP), and uric acid at the Vanderbilt Lipid Laboratory which is standardized by the Centers for Disease Control and Prevention for lipid analysis (Supplemental materials). Those used antihypertensive medication, blood pressure categorization was made accordingly. Participant characteristics were then assigned specific point values, with total points being then converted to a general cardiovascular risk score, representing a person’s 10yr risk of CVD. This score is a derivative of the original Framingham risk score (FRS) produced by the same group. However, it was reported to be better in predicting CVD risk, which may be explained by the modeling of risk factors as continuous variables (as opposed to categories in the model developed by Wilson et al) [20]. A risk assessment calculator based on the Cox regression model of proportional hazards was utilized to obtain participant GCRS [21].

Females:

$$\text{Risk factors}=\left(\text{ln}\left(\text{Age}\right)*2.32888\right)+\left(\text{ln}\left(\text{Total cholesterol}\right)*1.20904\right)-\left(\text{ln}\left(\text{HDL cholesterol}\right)*0.70833\right)+\left(\text{ln}\left(\text{Systolic BP}\right)*\text{Hypertension medication factor}\right)+\text{Cig}+\text{DM}-26.1931$$

$$\text{Risk}=100*\left(1-0.95012\text{exp}\left(\text{Risk factors}\right)\right)$$

Hypertension medication factor: 2.76157 (not on medication) vs 2.82263 (on medication)

Cig = 0.52873

DM = 0.69154

Males:

$$\text{Risk factors}=\left(\text{ln}\left(\text{Age}\right)*3.06117\right)+\left(\text{ln}\left(\text{Total cholesterol}\right)*1.12370\right)-\left(\text{ln}\left(\text{HDL cholesterol}\right)*0.93263\right)+\left(\text{ln}\left(\text{Systolic BP}\right)*\text{Hypertension medication factor}\right)+\text{Cig}+\text{DM}-23.9802$$

$$\text{Risk}=100*\left(1-0.88936\text{exp}\left(\text{Risk factors}\right)\right)$$

Hypertension medication factor: 1.93303 (not on medication) vs 1.99881 (on medication)

Cig = 0.65451

DM = 0.57367

HDL: high-density lipoproteins; BP: blood pressure; Cig: cigarette smoking status; DM: Diabetes.

2.3. Statistical analyses

Mean ± standard deviation for continuous demographic variables and percentage for categorical demographic variables were presented in Table 1 and Table 2. We performed generalized linear model for continuous variables or Pearson chi-squared tests for categorial variables to evaluate the differences among the nine types of TCM body constitution. We further evaluated the impact of significant differences among BC group, Low CVD risk group and the remaining group adjusting for age and sex. All P values are two sided and statistical significance was determined using an alpha level of 0.05. The data analyses used software SAS Enterprise Guide 7.1.

Table 1:

Baseline Characteristics of Participants Overall and Grouped by TCM BC Type

Characteristic	Overall (n=191)	Balanced (n=57)	Qi-deficiency (n=26)	Yang-deficiency (n=9)	Yin-deficiency (n=12)	Phlegm-dampness (n=16)	Damp-heat (n=6)	Blood-stasis (n=33)	Qi-stagnation (n=12)	Inherited-special (n=20)	P value
Age (y), Mean ± SD	60.0 ± 7.8	61.0 ± 6.9	62.6 ± 7.8	55.0 ± 4.2	58.7 ± 5.7	60.4 ± 9.7	61.5 ± 9.3	58.2 ± 8.1	59.3 ± 7.9	59.6 ± 9.8	0.28
BMI (kg/m2), Mean ± SD	30.3 ± 6.7	28.9 ± 6.7	32.4 ± 7.2	31.4 ± 7.1	32.9 ± 7.2	34.2 ± 6.3	30.5 ± 6.3	28.1 ± 4.0	31.9 ± 7.4	29.4 ± 7.1	0.03
Gender, n (%)											<0.0001
Male	100 (52.4)	42 (73.7)	16 (61.5)	1 (11.1)	2 (16.7)	6 (37.5)	5 (83.3)	8 (24.2)	8 (66.7)	12 (60)
Systolic Blood Pressure (mmHg), Mean ± SD	127.5 ± 15.3	129.1 ± 13.2	128.5 ± 18.3	117.7 ± 14.7	120.0 ± 16.3	130.4 ± 10.9	134.2 ± 9.8	128.9 ± 17.8	120.7 ± 16.6	128.0 ± 13.7	0.16
Total cholesterol (mg/dL), Mean ± SD	204.7 ± 57.7	198.7 ± 54.7	198.1 ± 52.3	191.3 ± 37.3	216.8 ± 59.7	226.8 ± 70.9	162.2 ± 43.0	215.4 ± 69.9	201.1 ± 38.5	209.2 ± 55.8	0.36
HDL (mg/dL), Mean ± SD	61.2 ± 21.3	59.3 ± 20.8	58.8 ± 16.2	66.2 ± 9.3	61.0 ± 16.2	65.4 ± 36.2	44.7 ± 15.9	66.9 ± 23.3	58.3 ± 16.9	62.2 ± 18.5	0.42
LDL (mg/dL), Mean ± SD	118.5 ± 41.9	115.6 ± 43.2	112.9 ± 39.0	107.7 ± 28.9	126.3 ± 43.8	130.4 ± 40.5	96.5 ± 29.0	123.8 ± 49.3	120.8 ± 30.2	121.5 ± 44.8	0.72
C-reactive Protein	3.1 ± 3.9 (n=101)	2.1 ± 2.1 (n=35)	3.5 ± 3.6 (n=16)	1.1 ± 1.1 (n=3)	7.3 ± 4.8 (n=6)	5.1 ± 5.8 (n=5)	11.5 ± 15.5 (n=2)	2.0 ± 1.2 (n=16)	2.6 ± 3.7 (n=6)	3.3 ± 3.4 (n=12)	0.003
Uric acid	6.2 ± 1.7 (n=101)	6.6 ± 1.8 (n=35)	6.5 ± 2.0 (n=16)	5.2 ± 1.5 (n=3)	6.0 ± 1.4 (n=6)	7.2 ± 1.3 (n=5)	5.8 ± 0.8 (n=2)	6.5 ± 0.8 (n=16)	5.4 ± 1.3 (n=6)	5.6 ± 1.3 (n=12)	0.19
General cardiovascular risk score	18.2 ± 14.1	20.6 ± 14.2	21.9 ± 14.9	6.4 ± 3.8	13.2 ± 13.3	17.7 ± 12.5	27.4 ± 20.4	12.9 ± 10.4	20.0 ± 17.0	20.1 ± 14.7	0.01
Smoking status, n (%)											0.09
Never	102 (53.4)	28 (49.1)	16 (61.5)	6 (66.7)	6 (50)	10 (62.5)	4 (66.7)	22 (66.7)	4 (33.3)	6 (30)
Ever/current	89 (46.6)	29 (50.9)	10 (38.5)	3 (33.3)	6 (50)	6 (37.5)	2 (33.3)	11 (33.3)	8 (66.7)	14 (70)
Drinking status, n (%)											0.22
Never	70 (36.7)	17 (29.8)	13 (50)	4 (44.4)	6 (50)	8 (50)	2 (33.3)	12 (36.4)	5 (41.7)	3 (15)
Ever	42 (21.9)	13 (22.8)	4 (15.4)	0 (0)	4 (33.3)	3 (18.8)	3 (50)	5 (15.2)	4 (33.3)	6 (30)
Current	79 (41.4)	27 (47.4)	9 (34.6)	5 (55.6)	2 (16.7)	5 (31.3)	1 (16.7)	16 (48.5)	3 (25)	11 (55)
Antihypertensive medication use, n (%)	78 (40.84)	21 (36.9)	17 (65.4)	2 (22.2)	6 (50)	6 (37.5)	4 (66.7)	12 (36.4)	4 (33.3)	6 (30)	0.16
Lipid control medication use, n (%)	68 (35.6)	19 (33.3)	13 (50)	2 (22.2)	4 (33.3)	3 (18.8)	5 (83.3)	12 (36.4)	2 (16.7)	8 (40)	0.10
Glycemic control medication use, n (%)	22 (11.52)	6 (10.6)	5 (19.23)	1 (11.1)	2 (16.7)	2 (12.8)	2 (33.3)	0 (0)	1 (8.3)	3 (15)	0.31
Physically active ≥ 2 days/week, n (%)	162 (84.8)	52 (91.2)	22 (84.6)	6 (66.7)	9 (75)	12 (75)	4 (66.7)	28 (84.9)	11 (91.7)	18 (90)	0.38
Education college or beyond	173 (90.6)	53 (92.9)	24 (92.3)	9 (100)	11 (91.7)	12 (75)	5 (83.3)	32 (96.9)	9 (75)	18 (20)	0.18

Continuous variables were presented as mean ±SD, P values were calculated using GLM model; categorical variables presented as n (%), P values were calculated using chi-square test.

LDL: low-density lipoproteins; HDL high-density lipoproteins; TCM: Traditional Chinese medicine

Table 2:

Stratified analysis of Balanced group, low CVD risk group, and remaining groups

Characteristics	Balanced (n=57)	Low CVD risk group (n=54) *	Remaining groups (n=80) †	P 1	P 2
Age (y), Mean ± SD	61.0 ± 6.9	57.8 ± 7.1	60.8 ± 8.7	0.047	-
BMI (kg/m2), Mean ± SD	28.9 ± 6.7	29.7 ± 5.7	31.7 ± 7.0	0.04	0.03
Gender (n), %				<0.0001	-
Male	42 (73.7)	11 (20.4)	47 (58.8)
Female	15 (26.3)	43 (79.6)	33 (42.3)
Systolic Blood Pressure (mmHg), Mean ± SD	129.1 ± 13.2	125.1 ± 17.4	128 ± 15.2	0.37	0.91
Total cholesterol (mg/dL), Mean ± SD	198.7 ± 54.7	211.7 ± 63.2	204.4 ± 56.1	0.50	0.70
HDL (mg/dL), Mean ± SD	59.3 ± 20.8	65.4 ± 20.0	59.8 ± 22.3	0.23	0.23
LDL (mg/dL), Mean ± SD	115.6 ± 43.2	121.7 ± 45.1	118.5 ± 39.2	0.75	0.72
CRP (mg/L), Mean ± SD	2.1 ± 2.1	3.5 ± 4.3	3.8 ± 4.6	0.15	0.19
CRP level (elevated), n (%)	0 (0)	3 (12)	4 (9.8)	0.13
Uric acid (mg/dL), Mean ± SD	6.6 ± 1.8	5.5 ± 1.3	6.3 ± 1.6	0.04	0.65
Uric acid level (elevated), n (%)	23 (65.7)	7 (28)	21 (51.2)	0.02
General cardiovascular risk score	20.6 ± 14.2	11.9 ± 10.5	20.7 ± 15.0	0.0005	0.40
Smoking (n), %				0.34	-
Never	28 (49.1)	34 (62.9)	40 (50.0)
Ever	25 (43.9)	17 (31.5)	30 (37.5)
Current	4 (7.02)	3 (5.6)	10 (12.5)
Drinking (n), %				0.54	-
Never	17 (29.8)	22 (40.7)	31 (38.8)
Ever	13 (22.8)	9 (16.7)	20 (25)
Current	27 (47.4)	23 (42.6)	29 (36.3)
Antihypertensive medications, %	21 (36.8)	20 (37.0)	37 (46.3)	0.43	-
Lipid control medications, %	19 (33.3)	18 (33.3)	31 (38.8)	0.74	-
Glycemic control medications, %	6 (10.6)	3 (5.6)	13 (16.3)	0.16	-

Continuous variables were presented as mean ±SD, P values were calculated using GLM model, P1 not justified. P2 adjusted for age and sex.

Categorical variables presented as n (%), P₁ values were calculated using chi-square test.

^*Consists of pooled participants from Yang-deficient, Yin-deficient, and Blood-Stasis constitution subtypes;

^†Consists of pooled participants from Qi-deficient, Phlegm-dampness, Damp-heat, Qi-stagnation, and Inherited-special constitution subtypes.

CRP elevated:>3.0mg/L, uric acid elevated:>7.0 mg/dL

CVD: cardiovascular disease; CRP: C-reactive Protein.

3. Results

In our study, BC (29.8%) was the predominant body constitution type, while the three most common pathologic body constitution types were Blood-stasis (17.3%), Qi-deficient (13.6%), and inherited-special (10.5%) (see Supplemental Table 3).

The characteristics of the study participants are shown overall and by body constitution type in Table 1. Among the 191 participants, the median age was 60, a minority were active smokers, and a majority were using anti-hypertensive and anti-lipid medications, both of which reflect a higher risk of CVD. There were statistically significant differences in the distribution of several factors across the various body constitution types. These included BMI, gender, CRP, and GCRS. Specifically, the BC had a generally lower average BMI (28.9), CRP (2.2), and GCRS (20.6), as well as lower female predominance (26.3%) compared to other groups.

3.1. Comparison of cardiovascular disease risk among TCM subtypes

Among participants using hypertensive medication, the three most common body constitution types were BC (26.9%), Qi-deficient (21.8%), and Blood-stasis (15.4%) (results not shown in table). On the other hand, the predominant types among participants using lipid control medications were BC (27.9%), Qi-deficient (19.1%), Blood stasis (17.7%), and inherited-special (11.8%) (results not shown in table). The GCRS averages, which represent a 10-year risk of cardiovascular disease, were statistically significantly different across body constitution types (p=0.01). Specifically, individuals in Yang-deficiency (6.4), Blood stasis (12.9), and Yin-deficiency (13.2), had a substantially lower risk of cardiovascular disease than other groups (Table 1). These body constitution types were also more common among women than among men.

To further investigate the characteristics of the three body constitution types with lower CVD risk, we combined individuals of these body constitution types into one low-risk group, and compared this new group against both the normal “balanced constitution” type and the combination of the other 5 TCM groups (Table 2). As previously noted, the low-risk group had a significantly higher proportion of women. Additionally, individuals in these groups were younger (57.8 y, p=0.047) than their counterparts in the rest of the study (61.0 y and 60.8 y, respectively). They also tended to have a higher average serum HDL-C (p=0.228), which is known to be protective against CVD, compared to counterparts. This low-risk group also had the highest percentage of never smokers and never drinkers compared to the other groups, although this difference was not statistically significant. We also performed multivariate analysis adjusting for age and sex. However, we did not find the major types of TCM body constitution were associated with the GCRS.

3.2. Stratified analyses by CRP and Uric acid levels

Based on previous studies which linked elevated CRP and uric acid to increased CVD risk, we classified our study population into “normal” and “elevated” groups. Specifically, individuals with CRP levels >10 mg/dL and uric acid >5.5 mg/dL were sorted into elevated CRP and uric acid groups, respectively. Average GCRS were 14.7 and 19.9 in the elevated and normal CRP groups, respectively; on the other hand, average GCRS was 21.7 and 16.8 in the elevated and normal uric acid groups, respectively. Analyses were then conducted to identify if specific TCM subtypes were associated with elevated biomarker levels. While there were no noticeable differences in the distribution of individuals with elevated CRP levels, the low-risk CVD group had a significantly lower level of uric acid (Table 2). The statistical significance disappeared after adjusting for age and sex.

4. Discussion

All previous studies of body constitution type have been conducted in Chinese populations except for a study of White college students in China [9]. The present study, to the best of our knowledge, is the first to be done in an American population [22]. The major types of TCM body constitution in our study of white individuals in the US were different from the types among Chinese in China[23]. Furthermore, in this cross-sectional study, we newly examined the relationships between the nine body constitution types with predictive models or markers of CVD, in particular the GCRS.

Our study of white individuals in the US had a large proportion of Blood-Stasis (17.3%), Qi-deficient (13.6%), and inherited-special types (10.5%). In a previous large study with 8,448 participants conducted within Chinese population, the most common pathologic body constitution subtypes are Qi-deficient (13.4%), Damp-heat (9.1%) and Yang-deficient (9.0%) [23]. The difference between our study of white individuals in the US and this large study conducted in Chinese population may be due to multiple factors. First, our study participants may not represent the general US population. Thus, our finding needs to be confirmed by future larger studies conducted in US population. Secondly, this difference could be due to different genetic background. Thirdly, the difference could be caused by regional environmental factors as our study was conducted in one state in the US. There was one previous study of Caucasian college students attending three Beijing Universities in China [9]. We found the distribution of major types of TCM body constitution in this earlier study is also different from that in our study. However, the White college students recruited in this earlier study were mostly younger than 50 years old and lived in Beijing, a Northern city in China while most of the participants in our study were older than 50 years and lived in a single US state. Thus, the differences could be caused by age and regional environmental factors.

Another potential explanation for these discrepancies may lie in identifying how biomarkers of common diseases contribute to an individual’s TCM subtype. In our study, significant differences were found among the nine different body constitution types by various different biomarkers and vitals for cardiovascular disease. Some of these findings were expected, such as lower BMIs in the “healthy” Gentleness controls and the classically thinner Blood-stasis compositions [24]. However, the elevated BMIs in most other body constitution types, including the Phlegm-dampness and Yin-deficient types stand in contrast to findings in studies conducted in Chinese populations [25]. Again, the reason is unclear. For example, White, and in particular White Americans, peoples tend to have a higher overall BMI than Asian populations [26]. If BMI contributes to a person’s TCM body constitution type, then it may help to explain the differences in proportion of TCM body constitution subtypes between an American and Chinese populations. Taken a step further this could reflect differences in disease prevalence, such as cardiovascular disease, from one population to another, as previous studies found the incidence rates of cardiovascular disease and cancers (i.e. breast cancer, colorectal cancer and prostate cancer) are much higher in US populations than in Chinese populations [27, 28]. Zhu et al. also identified which TCM subtypes were more represented among major chronic diseases, including hypertension, hyperlipidemia, and hyperglycemia. Thus, further understanding the underlying mechanism for the difference in TCM body constitution type between US and Chinese population may help to identify the novel risk factors contributing to common disease etiology.

Another primary research focus of this study was the GCRS, which represents an individual’s percent risk of cardiovascular disease (CVD) in 10 years, with a higher score necessarily translating to a higher percent risk. Again, our analyses revealed significant differences in the overall GCRS across body constitution types. Constitutions with the lowest average GCRS included Yang-deficient (6.4), Blood-stasis (12.9), and Yin-deficient (13.2), while Damp-heat (27.4) and Qi-deficient (21.9) recorded the highest GCRS.

Further subgroup analysis of the three low-CVD risk groups (Yin/Yang-deficient and BS) revealed a significantly lower average age and BMI, as well as a higher predominance of women, compared to both the BC group and the remaining 5 groups. This biologically makes sense in the context of the Framingham model, which typically associated men, particularly older and more obese men, with increased CVD risk. Another finding was a significantly lower uric acid level among the low-CVD risk group. While uric acid was not a component of the initial Framingham study, more recent literature has reported on an association between uric acid and CVD risk, specifically an association between elevated uric acid (>5.5) and increased CVD risk [29]. However, the significant difference did not remain after adjusting for age. Thus, it is very likely that the association was confounded by age. Mediation analysis was also conducted, which revealed no significant effect mediation by sex. Thus, the loss of statistical significance when stratified by sex may be due to reduced sample size as we also found the patterns still remained in the stratified analysis.

The major limitation of the study was the lack of a population-based sample which may have affected the distribution and sample size of body constitution group resulting in some limited statistical power in stratified analysis. Secondly, some parts of the survey were not easily translatable to English from the original Chinese version due to the cultural differences, with some questions, such as those involving “sighing without reason” or “thick saliva” perhaps being less readily answered by our exclusively American population. However, this only appeared in limited survey questions which may not materially affect the directions of results. Also, this non-differential misclassification caused by the translation in cultural difference usually biases the results to the null. In other words, the true associations may be stronger than the ones we observed. Thirdly, the limitation is that we did not collect family history of cardiovascular disease and cannot adjust for this covariate in the current study. Further studies are warranted to examine whether family history of CVD contributes to the TCM body constitution types related to CVD risk. Finally, although the intra-class correlation coefficients ranging from 0.7 to 0.8 and the Cronbach’s alphas ranging from 0.44 to 0.72 are reasonable for reliability, these findings indicate that TCM body constitution types could vary during the three-month period in a study of White individuals in the US. Further studies are warranted to identify the factors contributing to the changes.

The majority of previous studies conducted in Mainland China which have investigated a similar topic have noted significant associations between CAD and particular TCM body constitution types, in particular Yin/Yang deficiency as well as Phlegm-dampness [30, 7]. These findings have helped practitioners in China to better identify patients at risk for CVD and therefore initiate a preventive health regimen. Additional studies have also postulated associations between body constitution types and chemokine receptor/ligan expression, which would hold implications for a whole host of pathologies, including immunology and oncology. Translating these promising findings to an American population could be beneficial, given the prevalence of CVD in the United States. However, our findings found although there are similarities, there are substantial differences between these two populations. Future studies should aim to understand the underlying mechanisms contributing to these differences, which may not only help to understand the underlying mechanism for TCM, but also help to identify novel factors or mechanisms for CVD risk, prevention and treatment.

In conclusion, we found that the major types of TCM body constitution were not associated with the GCRS and other CVD biomarkers. Future larger studies in US populations, particularly population-based studies including multiple racial/ethnic groups are warranted to confirm our findings.

Abbreviations

ACS

acute coronary syndrome

BC

balanced constitution

BMI

Body mass index

CAD

coronary artery disease

CVD

cardiovascular disease

CRP

c-reactive protein

eGFR

estimated glomerular filtration rate

GCRS

General cardiovascular risk score

HDL-C

high-density lipoproteins cholesterol

LDL-C

low-density lipoproteins cholesterol

MDRD

Modification of Diet in Renal Disease

MI

myocardial infarction

TC

total cholesterol

TCM

Traditional Chinese Medicine

WHR

waist hip ratio

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.