Even in healthy individuals without apparent illness, we found statistically significant relationships between serum carotenoids and various CVD markers, and some of these relationships varied according to sex and age categories. To the best of our knowledge, this is the first study investigating the relationships between serum carotenoid levels and multiple CVD markers in healthy individuals by identifying differences in relation to sex and age.
4.1. Characteristics of the Study Subjects
As shown in Table 1
, the levels of many biomarkers appeared to increase with age in our subjects. For example, the average baPWV in the old group seemed to be relatively high compared to the cutoff value (1750 cm/s) in a previous study [23
]. The average baPWV in the Takashima Study (mean age: 58.9 years), based on an annual community health examination for the residents of Takashima City, Shiga Prefecture, Japan, was 1548 cm/s [24
], similar to the average of our participants aged 40 or above (i.e., 1520 cm/s; mean age: 58.0 years). Thus, there is no reason to think that the population in this study is not a general, relatively healthy population. According to the guidelines for hypertension issued by the Japanese Society of Hypertension, the mean SBP in advanced-age males falls within the range of high normal blood pressure (130–139 mmHg) [25
]; however, it does not reach the level of high blood pressure (140 mmHg or above) in Japanese subjects. In addition, BMI was higher in the middle-aged group for males compared to the old group, and it increased with age for females. Thus, the trend observed in this population was similar to that observed in the National Health and Nutrition Survey (Japan, 2017) [26
]. In total, 17.1% of the males and 18.0% of the females were on antihypertensive drugs, values that are lower than the results of the National Health and Nutrition Survey (31.8% for males and 25.9% for females). This may be due to the fact that we focused on a healthier population in this study. The serum concentrations of carotenoids were consistent with previous reports [27
]. These results suggest that the study population is a relatively general and healthy Japanese population.
Serum total carotenoid levels were significantly positively associated with vegetable intake in both males and females. In a recent systematic review, it was suggested that the concentrations of some carotenoids (lutein, β-cryptoxanthin, α-carotene, and β-carotene) are higher in the high-intake group of fruits and vegetables than in the low-intake group [28
]. A relationship between blood carotenoid levels and vegetable intake was also reported by Campbell et al. using a frequency survey Food Frequency Questionnaire [16
]. These previous studies are consistent with ours, suggesting that serum carotenoid concentrations could be an indicator of vegetable intake and that this could be expected to be helpful for nutritional guidance according to which non-invasive measurement of skin carotenoid levels has begun to be implemented [29
4.2. Relationship between Serum Total Carotenoid Levels and Markers of CVDs
Without age stratification, in males, seven CVD markers (baPWV, SBP, DBP, HOMA-IR, insulin, TGs, and HDL cholesterol) out of the nine were significantly associated with total carotenoids (Table 2
), whereas in females, eight markers (BMI, baPWV, SBP, HOMA-IR, blood insulin, FBG, TGs, and HDL cholesterol) were associated with total carotenoids (Table 3
). Many studies have shown that some CVD markers are healthier in those with higher vegetable intake [4
]. In addition, it has been suggested that the relationship between vegetable intake and some CVD markers is due to antioxidants such as carotenoids in the blood [32
]. These results are consistent with ours.
The best studies of physiological function of carotenoids have focused on quenching the activity of singlet oxygen, one of the reactive oxygen species (ROS) [34
]. ROS have been considered to elicit chronic inflammation and accelerate the onset and/or progression of CVDs [35
]. In the present study, we did not evaluate any circulating biomarkers of early endothelial dysfunction, which is the starting point of the atherosclerotic process and accompanies the progression of CVDs [36
]. We suggested that tomatoes can help prevent atherosclerosis by inhibiting vascular endothelial dysfunction in a primitive animal experiment [37
]. Tsitsimpikou et al. suggested that continuous intake of tomato juice significantly improves inflammation status and endothelial dysfunction [38
]. Regarding these mechanisms, reduction of the total expression of VCAM-1 and ICAM-1 by carotenoids (β-carotene and lycopene) has been reported to play an important role. [39
].In the future, we aim to elucidate the mechanism in detail by examining the relationship between internal carotenoid levels and markers such as E-selectin, ICAM-1, VCAM-1, and von Willebrand Factor as well as inflammatory markers. We found significant associations between hydrocarbon carotenoids (β-carotene, lycopene) and the level of baPWV. Arteriosclerosis progresses via the deposition of lipid in blood vessel walls, which is induced by foamy macrophages that infiltrate vascular endothelial cells and absorb oxidized LDLs without restriction [40
]. It has been reported that carotenes, which are mainly distributed in LDL cholesterol particles in the blood, improve the oxidative resistance of LDLs and reduce the levels of oxidized LDLs [41
]. They are also expected to have an inhibitory effect on the progression of arteriosclerosis and may contribute to the negative relationships observed between carotenoids and baPWV. Oxidative stress has also been reported to cause abnormal insulin secretion in pancreatic β-cells [43
]. Serum levels of lutein and β-carotene were found to be significantly negatively associated with diabetes-related markers (blood insulin, HOMA-IR) in both sexes, and they were also found to be negatively associated with FBG in females. Lutein and β-carotene might have a protective effect on pancreatic β-cells via an antioxidative effect.
There might be a possibility that serum total carotenoids are an alternative indicator of vegetable intake and that the associations between serum total carotenoid level and CVD markers observed in this study might be derived from some of the ingredients common in vegetables other than carotenoids. For example, dietary fibers in vegetables inhibit the increase of postprandial blood glucose levels and, thus, inhibit the progression of insulin resistance [44
]. It is also thought that relatively low energy intake might suppress weight gain and the deterioration of CVD markers. Potassium, which is abundant in vegetables in general, and γ-aminobutyric acid, which is particularly abundant in tomatoes, can also decrease hypertension [45
]. This suggests that these components are indirectly related to the significant relationship observed between carotenoids and blood pressure.
In addition to dietary habits, lifestyle habits, such as smoking, alcohol intake, and exercise, have a major impact on the development and progression of CVDs [47
]. Therefore, these three items were added to the adjustment factors in the analysis. However, no significant change was observed in the results, suggesting that the relationships observed in this study are independent of the lifestyles mentioned above. Obesity also underlies many CVDs [48
]. Therefore, BMI was added to the adjustment factors, but no significant change in the relationship between total carotenoids and CVD markers was observed in males or females, suggesting that these relationships are also independent of BMI. Thus, the associations that we found were probably due to the role of carotenoids and other ingredients of vegetables mentioned above.
However, these relationships were altered by age stratification (Table 2
; Table 3
). In young females, most of the relationships between carotenoids and CVD markers disappeared after adjusting for BMI. Thus, we speculate that obesity may have a negative impact on CVD markers, especially in young females, who are less likely to exhibit worsened premenopausal lipid-related markers.
In analyses with additional adjusted factors (lifestyle, BMI), a prominent number of significant relationships were observed between the serum total carotenoid levels and CVD markers in young males and middle-aged females. The levels of six CVD markers (BMI, baPWV, SBP, DBP, FBG, and TGs) were found to be higher in middle-aged males than in young. This suggests that the risk of CVDs starts increasing in young males. On the other hand, in females, menopause is known to alter lipid metabolism [49
]. Hence, middle age, at which menopause occurs, might be the time when the deterioration of CVD markers begins through changes in the lipid metabolism. High dietary intake of vegetables (i.e., leading to a high serum concentration of total carotenoids) during this period may delay or prevent the exacerbation of CVD markers.
Our findings, obtained by stratifying sex and age, should act as useful references for a new intervention or prospective study to confirm the causal correlation that indicates when we should consume more carotenoid-rich vegetables.
4.3. Relationship between Individual Carotenoid Levels and Markers of CVDs
The physiological effects of carotenoids, other than their ability to scavenge singlet oxygen, have been reported to vary according to the type of carotenoid. Therefore, the association between individual carotenoids and CVD markers was investigated ( Table 4
; Table 5
Both β-carotene and lutein are found in many green and yellow vegetables; however, the relevant CVD markers with β-carotene and lutein differed. Such differences may also be influenced by the differences in the types of vegetables containing these carotenoids and in the content of vegetables [50
]. However, it is presumed that these differences are also attributable to the distributions and physiological functions of these carotenoids in the body. For example, carotenes composed exclusively of carbon and hydrogen, such as β-carotene and lycopene, are found mainly in low-density lipoprotein (LDL) cholesterol particles in the blood, whereas xanthophylls containing hydroxyl groups in their structure, such as lutein, zeaxanthin, and β-cryptoxanthin, are distributed in relatively high proportions in HDL cholesterol particles [51
]. Thus, the role of HDL cholesterol particles as a carrier of xanthophylls might be one of the reasons for the significant positive relationship between them rather than any physiological function of xanthophylls.
Lycopene, which is highly hydrophobic and weakly abundant in HDL cholesterol, is positively associated with HDL cholesterol in both males and females. It has been suggested that lycopene increases HDL cholesterol subtypes HDL-2 and HDL-3 and increases the HDL cholesterol in general by decreasing the activity of cholesteryl ester transfer protein in the blood and increasing the activity of lecithin–cholesterol acyltransferase (LCAT) in obese middle-aged individuals [52
]. A mechanism of action has also been proposed whereby lycopene decreases LDL cholesterol by inhibiting 3-hydroxy-3-methylglutaryl-coenzyme A reductase [53
]. Friedewald’s equation indicates that (LDL cholesterol) = (total cholesterol)–(HDL cholesterol)–(TGs)/5 when the blood TG level is 400 mg/dL or lower, suggesting an association between the reduction of LDL cholesterol and the elevation of HDL cholesterol. Another study showed that diets rich in lycopene actually increase the HDL cholesterol [54
]. The positive relationship between lycopene and HDL cholesterol observed in this study may also be due to these physiological effects.
Serum levels of β-cryptoxanthin are significantly associated with multiple CVD markers in both males and females. This carotenoid, abundantly found in satsuma mandarin, was found to be associated with HOMA-IR in both sexes in the Mikkabi study [55
] and was also associated with pulse velocity in both males and females [56
]. In this study, it was hypothesized that β-cryptoxanthin has a protective effect against oxidative stress in the pathogenesis of insulin resistance. Additionally, β-cryptoxanthin has been reported to activate the nuclear factor erythroid 2-related factor 2 (Nrf2) pathway [57
]. Sulforaphane is a typical activator of Nrf2 that has been found to improve insulin resistance in a mouse model [58
]. In the present study, the serum concentrations of β-cryptoxanthin were significantly associated with diabetes-related markers in both sexes, but with baPWV only in females, although the subjects exhibited lower serum concentrations of β-cryptoxanthin compared to the population studied in the Mikkabi study.
The significant relationships between the serum concentration of β-cryptoxanthin and blood pressure or diabetes-related markers in females disappeared when we adopted BMI as the adjustment factor. This might suggest that obesity is a confounder for these associations. This is supported by previous reports, which showed that β-cryptoxanthin acts as an antagonist of peroxisome proliferation-activated receptor gamma (PPARγ) and may improve lipid metabolism and inhibit obesity via the inhibition of adipocyte hypertrophy [59
Future examinations aimed at clarifying the causal relationships between each individual carotenoid and CVD markers on the basis of this study might reveal the carotenoid-rich vegetables that we should eat to prevent CVDs.
4.4. Study Limitations
As this was a cross-sectional study, it is unclear whether the intake of carotenoids directly improves CVD markers. Therefore, an analysis using an interventional study or a prospective cohort study is required to approach causality.
As this study was performed via a resident-based health examination, the participants were all Japanese living in a narrow area. Therefore, reproducibility should be confirmed in a different country and/or race.
Some associations between carotenoids and markers of glucose metabolism in males disappeared in all age categories, not just in a particular age group, suggesting that this disappearance may be mainly due to a reduction in statistical power due to a decrease in the number of subjects, especially among males. As mentioned in the Introduction, we conducted a stratified analysis by age despite the limitation of decreasing power of detection to obtain information on when we should eat carotenoid-rich vegetables to prevent CVDs. Increasing the number of subjects is expected to increase detection power in the future.
The BDHQ used in this study, whose validity has been verified several times [19
], is often adopted in dietary surveys in Japan [61
]. However, since this is a frequency survey, there may be a slight discrepancy between the actual vegetable intake and the return error caused by recall bias.
In this study, only those who voluntarily underwent medical examinations were enrolled because we used the medical examination data of the residents, suggesting a selection bias.
We analyzed many CVD markers in combination with carotenoids. It cannot be denied that the significant relationships observed in this study were errors due to multiple comparisons. However, to avoid picking up such errors, we focused on what changed in the series, rather than a single result.