2023 Volume 30 Issue 6 Pages 589-600
Aim: Circulating omega-3 and omega-6 polyunsaturated fatty acids may to contribute to cardiovascular health at the population level. Over a decade, we investigated changes in the serum eicosapentaenoic acid (EPA) to arachidonic acid (AA) ratio, and in serum concentrations of the individual fatty acids, in a Japanese community.
Methods: Community surveys took place in 2002–2003 and 2012–2013 in a rural area of Japan. The community surveys included 3,194 and 3,220 community dwellers aged ≥ 40 years who did not take EPA medication in 2002–2003 and 2012–2013, respectively. Fatty acid fractionations in serum were measured using a gas chromatography method. Changes in the serum EPA/AA ratio over time were examined using linear mixed models.
Results: Overall, the average serum EPA/AA ratio decreased over the 10 years. A decreasing trend in the serum EPA/AA ratio occurred in all age groups except participants aged ≥ 80 years, with larger decreases in the younger age groups. A similar decline in serum EPA/AA ratio occurred in participants with and those without lipid-lowering therapy. Serum EPA concentrations were slightly increased in the whole population but remained stable or even decreased in participants aged 40–69. In contrast, the average serum AA concentrations increased in all age groups.
Conclusion: In a Japanese community, the serum EPA/AA ratio decreased over 10 years at the population level, especially in middle-aged participants.
See editorial vol. 30: 587-588
Omega-3 and omega-6 polyunsaturated fatty acids are associated with cardiovascular health1, 2). Eicosapentaenoic acid (EPA) is an omega-3 polyunsaturated fatty acid found mainly in fish oil and exerts triglyceride-lowering and anti-inflammatory effects via enhanced lipoprotein metabolism3, 4). Arachidonic acid (AA) is an omega-6 polyunsaturated fatty acid found in various foods, including meat, eggs, and fish, and is involved in neurogenesis and, therefore, essential for growth and development5). AA has complex biological effects on cardiovascular health, as it is the precursor to various metabolites that promote or suppress inflammation2). However, excessive circulating AA can cause atherosclerotic cardiovascular disease and thromboembolism6). EPA and AA have mutual antagonistic effects; EPA inhibits the formation of prostaglandin E2 from AA and produces the less-inflammatory prostaglandin E3 7). Additionally, the synthetic pathways for EPA and AA share the same set of enzymes8). Thus, the ratio of the circulating level of EPA to AA (EPA/AA ratio) in blood could reflect systemic inflammation, platelet aggregation, endothelial function, and oxidative stress levels7, 9).
Studies on the association between EPA/AA ratio and cardiovascular risks have been conducted primarily in Japan7). In clinical studies, the circulating EPA/AA ratio has been considered a marker for predicting cardiovascular risks. Lower circulating EPA/AA ratios have been treated as a treatment target in patients with a history or high risk of cardiovascular diseases10-15). Population-based epidemiological studies have also shown that lower circulating EPA/AA ratios are associated with increased cardiovascular risks and death16-18). There is a debate about the usefulness of fatty acid ratios as cardiovascular risk markers in past studies conducted in other countries19). Previous studies in Japan10-18) show that the circulating EPA/AA ratio may be an indicator of cardiovascular risk at a population level.
Circulating fatty acid concentrations in a population shouldchange along with changes in food intake. The traditional Japanese diet is characterized by high fish and seafood consumption, and the circulating EPA levels of Japanese populations are among the highest in the world20, 21). Japanese populations also have a high level of AA intake22). However, dietary patterns among Japanese adults have continued to change to the present day. Evidence shows marked and continuous Westernization since the 2000s, characterized by increasing meat consumption and decreasing fish consumption23). Therefore, a decline in the omega-3 to omega-6 fatty acid ratio over time can be anticipated, but data on changes in the circulating EPA/AA ratio at a population level across time are limited24, 25).
In this study, we sought to examine changes in the circulating EPA/AA ratio and the individual fatty acid levels at a population level over a decade in a Japanese community. We anticipated that the balance of circulating fatty acid levels might be useful for assessing residual cardiovascular disease risk at a population level.
The present study was conducted using data from the Hisayama Study. The Hisayama Study is a population-based study investigating cardiovascular disease and its risk factors. Details of the design of the Hisayama Study have been described elsewhere26). We used data from surveys conducted in 2002–2003 (hereafter the 2002 survey) and in 2012–2013 (hereafter the 2012 survey). The health examination included 3,328 residents aged ≥ 40 between 2002 and 2003 (participation rate: 77.6%). We excluded 30 individuals who did not consent to enter the study, 86 individuals who did not fast overnight, 2 with no measurement of serum fatty acids, and 16 prescribed EPA medication (i.e., ethyl icosapentate). The remaining 3,194 participants were included in the 2002 survey. In 2012–2013, 3,396 residents aged ≥ 40 years participated in the health examination (participation rate: 72.5%). After excluding 6 participants who did not consent, 133 individuals who did not fast overnight, 14 with no serum fatty acid measurement, and 23 who were taking EPA medication, 3,220 participants were included in the 2012 survey. In total, 4,251 participated in either the 2002 or the 2012 survey, and 2,163 participants took part in both surveys. None of the participants was prescribed omega-3 fatty acid (a combination of EPA and docosahexaenoic acid [DHA]) ethyl agents, as these agents first became commercially available in Japan in early 2013. The Kyushu University Institutional Review Boards for Clinical Research approved this study, and we obtained written informed consent from all participants.
Measurement of Fatty AcidsBlood samples were drawn from study participants after at least 12 h of fasting. The serum specimens were separated, frozen, and stored at −80℃ until used. Serum fatty acid levels of the samples collected from the 2002 survey were measured in 2010, and those of the 2012 survey were measured in 2017 using the gas chromatography method (SRL, Tokyo). The samples were processed according to the method described elsewhere16). For the 2002 survey, 24 fatty acids fractionations, including EPA, DHA, and AA concentrations, were measured27). For the 2012 survey, four fractionations (EPA, AA, DHA, and dihomo-γ-linolenic acid) were measured. The measurements of fatty acid fractions were commercially available. Both measurements of fatty acid fractions in the 2002 and 2012 surveys were performed by a clinical laboratory using an established and standardized method. The reproducibility (coefficient of variation) for serum EPA, DHA, and AA measurements in this method were 4.4%, 2.3%, and 3.8%, respectively28).
CovariatesInformation on medical history, medications for hypertension, diabetes, dyslipidemia, smoking habits, alcohol intake, and regular exercise was obtained by a self-administered questionnaire. Trained interviewers checked the questionnaire at the examination. Smoking and drinking habits were categorized by whether the participant currently smoked or drank. We defined any sports or other exercise (≥ 3 times/week) in leisure time as regular exercise. Height and weight were measured in light clothes, and body mass index was calculated as weight (kg) divided by the height (m) square. Blood pressure was taken with an automated sphygmomanometer (BP-203RV III; Omron Colin, Tokyo) in a seated position, and in the analysis, we used the average of three consecutive measurements. Hypertension was defined as blood pressure ≥ 140/90 mmHg and/or current use of antihypertensive medications. Plasma glucose levels were measured by using the hexokinase method. Hemoglobin A1c levels were measured by latex aggregation immunoassay in 2002 and by high-performance liquid chromatography in 2012. Because the standardization for hemoglobin A1c measurement in the 2002 survey was based on the reference materials of the Japan Diabetes Society, the values were converted to National Glycohemoglobin Standardization Program equivalent values to allow direct comparison with the 2012 Survey by the following the formula: HbA1c (%)=1.02×HbA1c in 2002 survey (%) ＋0.25%. Diabetes was defined as fasting plasma glucose levels ≥ 126 mg/dL, 2-h postload plasma glucose levels ≥ 200 mg/dL, hemoglobin A1c ≥ 6.5%, or current use of oral glucose-lowering medications or insulin. Serum total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides were measured using the enzymatic method, and serum non-HDL cholesterol levels were calculated by subtracting HDL cholesterol from total cholesterol values. Dyslipidemia was defined according to the diagnostic criteria proposed in the Japan Atherosclerosis Society Guideline 2017: serum LDL cholesterol ≥ 140 mg/dL, serum HDL cholesterol ＜40 mg/dL, or serum triglycerides ≥ 150 mg/dL, or non-HDL cholesterol ≥ 170 mg/dL or use of lipid-lowering agents29). Serum high-sensitivity C-reactive protein levels were analyzed using a modified version of the Behring Latex-Enhanced CRP assay on a Behring Nephelometer BN-100 (Behring Diagnostics, Westwood, MA, USA).
Statistical AnalysisThe difference in the characteristics between the 2002 survey and the 2012 survey was tested by mixed effect models to consider the partially overlapping sampling between the 2002 and 2012 surveys. We specified a normal distribution and identity link function for continuous variables (i.e., linear mixed models) and a binomial distribution and logit link function for categorical variables (i.e., logistic mixed models). Serum triglycerides and high-sensitivity C-reactive protein levels were natural-log-transformed before fitting the model because of the skewed distributions.
The change in the serum EPA/AA ratio in the population over time was modeled by using a linear mixed model. The serum EPA/AA ratio was log-transformed before the analysis due to its skewed distribution. An unstructured residual variance-covariance matrix parameterized through its Cholesky root was specified to model the repeated measurement in an individual, and the Kenward–Roger method was used to determine denominator degrees of freedom. The first set of covariates included age groups (40–49, 50–59, 60–69, 70–79, and ≥ 80 years) and sex. Additionally, the interactions of time with age group and with sex were examined, and the interaction of time and age group was significant: thus, a time×age-group interaction term was also added to the age- and sex-adjusted model. The same analysis was made for participants who took lipid-lowering agents in each cohort (n=315 in the 2002 survey and n=736 in the 2012 survey) and for those who did not take lipid-lowering agents.
To further examine whether changes in the serum EPA/AA ratio reflect changes in prevalent diseases, treatments, and lifestyle behaviors, the following variables were additionally included as time-varying covariates: hypertension, diabetes, dyslipidemia, serum high-sensitivity C-reactive protein, body mass index, current smoking, current drinking, and regular exercise. All analyses were repeated for serum EPA and AA levels (log-transformed) to investigate the contribution of individual fatty acids to the change in serum EPA/AA ratio. Moreover, the analyses were repeated for the serum DHA/AA ratio and serum DHA to examine whether marine omega-3 fatty acids show similar trends. All statistical analyses were performed using SAS software (v.9.4; SAS Institute, Cary, NC). A two-tailed p-value ＜0.05 was considered statistically significant in all analyses.
The demographic characteristics of the study population are shown in Table 1. Participants in the 2012 survey were older and were likely to have lower mean values of diastolic blood pressure, fasting and postload plasma glucose levels, serum LDL cholesterol and non-HDL cholesterol concentrations, high serum sensitivity C-reactive protein concentrations and body mass index, and higher mean values of hemoglobin A1c levels and serum HDL cholesterol concentrations. Moreover, participants in the 2012 survey had higher frequencies of hypertension, diabetes, dyslipidemia, antihypertensive medication, glucose-lowering medication, lipid-modifying medication use, and regular exercise, and lower frequencies of current smoking.
| Variables | 2002 survey (n= 3,194) | 2012 survey (n= 3,220) | p value |
|---|---|---|---|
| Age, years | 61.5 (12.4) | 64.3 (13.0) | ＜0.001 |
| 40–49 | 19.2 | 15.7 | |
| 50–59 | 27.3 | 19.4 | |
| 60–69 | 25.6 | 29.7 | |
| 70–79 | 19.4 | 21.8 | |
| ≥ 80 | 8.5 | 13.4 | |
| Male sex, % | 43.0 | 42.8 | ＜0.001 |
| Systolic blood pressure, mmHg | 131.9 (21.1) | 130.4 (19.5) | 0.90 |
| Diastolic blood pressure, mmHg | 78.5 (12.0) | 76.6 (11.3) | ＜0.001 |
| Antihypertensive medication use, % | 23.9 | 37.1 | ＜0.001 |
| Hypertension, % | 43.6 | 53.0 | ＜0.001 |
| Fasting plasma glucose, mg/dL | 109.2 (23.0) | 105.8 (20.5) | ＜0.001 |
| Post-load plasma glucose, mg/dL b) | 146.1 (68.3) | 144.6 (62.0) | ＜0.001 |
| Hemoglobin A1c, % | 5.4 (0.8) | 5.7 (0.7) | ＜0.001 |
| Glucose-lowering medication use, % | 5.3 | 9.4 | ＜0.001 |
| Diabetes, % | 17.7 | 18.2 | ＜0.001 |
| Serum HDL cholesterol, mg/dL | 62.6 (16.2) | 64.5 (17.4) | ＜0.001 |
| Serum LDL cholesterol, mg/dL | 123.0 (31.5) | 120.2 (32.2) | ＜0.001 |
| Serum triglycerides, mg/dL a) | 97 (69–143) | 99 (71–140) | 0.17 |
| Serum non-HDL cholesterol, mg/dL | 141.3 (35.7) | 137.0 (35.9) | ＜0.001 |
| Lipid-modifying medication use, % | 9.9 | 22.9 | ＜0.001 |
| Dyslipidemia, % | 50.7 | 57.6 | ＜0.001 |
| Serum high sensitivity C-reactive protein, mg/L a) | 0.47 (0.23–1.03) | 0.40 (0.19–0.92) | ＜0.001 |
| Body mass index, kg/m2 | 23.1 (3.4) | 23.0 (3.6) | ＜0.001 |
| Current smoking, % c) | 21.8 | 15.7 | ＜0.001 |
| Current drinking, % c) | 43.6 | 47.8 | 0.74 |
| Regular exercise habit, % c) | 10.8 | 14.2 | ＜0.001 |
Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein. Values are shown as means (standard deviations) or frequencies except where noted.
P for difference was calculated by a linear or logistic mixed model without covariates.
a) Values are shown as the median (interquartile range), and the p value was calculated by testing the difference of natural-log transformed mean values between the 2002 and 2012 surveys.
b) In the 2002 and 2012 surveys, 263 and 559 participants were excluded, respectively, due to missing values.
c) One participant in the 2012 survey was excluded due to missing values.
Table 2 shows the serum EPA/AA ratio and serum EPA and AA concentrations in the 2002 and 2012 surveys. In this population, the serum EPA/AA ratio decreased over time, with the geometric means (95% confidence intervals) of the serum EPA/AA ratio being 0.40 (0.39–0.41) in the 2002 survey and 0.32 (0.32–0.33) in the 2012 survey (unadjusted p＜0.001). The serum EPA concentrations (geometric mean with 95%CI) were slightly increased from 59.0 (57.8–60.2) µg/mL in the 2002 survey to 65.1 (63.9–66.3) µg/mL in the 2012 survey (unadjusted p＜0.001). Additionally, the serum AA concentrations were substantially increased, from 147.6 (146.4–148.9) µg/mL in the 2002 survey to 199.9 (198.2–201.5) µg/mL in the 2012 survey. The changes in EPA/AA ratios by age groups are shown in Fig.1. The serum EPA/AA ratios were decreased in the 2012 survey compared to the 2002 survey in all age groups, except in the oldest (≥ 80 years) group, with larger decreases in the younger age groups (p for time×age-group interaction ＜0.001). Additionally, similar trends were observed among participants with and those without prescribed lipid-lowering agents (Supplementary Fig.1).
| 2002 survey (n= 3194) | 2012 survey (n= 3,220) | p value | |
|---|---|---|---|
| Serum EPA/AA ratio | 0.40 (0.39–0.41) | 0.32 (0.32–0.33) | ＜0.001 |
| Serum EPA concentration, μg/mL | 59.0 (57.8–60.2) | 65.1 (63.9–66.3) | ＜0.001 |
| Serum AA concentration, μg/mL | 147.6 (146.4–148.9) | 199.9 (198.2–201.5) | ＜0.001 |
Abbreviations: EPA, eicosapentaenoic acid; AA, arachidonic acid. Values are shown as the unadjusted means (95% confidence intervals).
Estimates were obtained from a linear mixed model, where values were log-transformed prior to analysis and back-transformed to obtain geometric means and their 95% confidence intervals for presentation.
Abbreviations: EPA, eicosapentaenoic acid; AA, arachidonic acid.
Estimates were obtained from a linear mixed model, where values were log-transformed before analysis and back-transformed to obtain geometric mean values for presentation. The model included time, age groups, the interaction term of time × age groups, and sex as a fixed effect.
＊p＜0.05 vs. the corresponding age group in the 2002 survey.
Abbreviations: EPA, eicosapentaenoic acid; AA, arachidonic acid.
Estimates were obtained from a linear mixed model, where values were log-transformed before analysis and back-transformed to obtain geometric mean values for presentation. The model included time, age groups, the interaction term of time × age groups, and sex as a fixed effect.
＊p＜0.05 vs. the corresponding age group in the 2002 survey.
Fig.2 shows the serum EPA and AA concentrations changes for each age category. The change in serum EPA concentrations varied across age groups, with the concentrations decreasing in those aged 50–59, and increasing in the groups older than 70 years. The serum AA concentrations increased in all age groups, with a greater increase observed in the younger age groups (p for time×age-group interaction =0.004). These trends remained unchanged after multivariable adjustment for concurrent cardiometabolic profile and lifestyle behaviors (Supplementary Table 1).
Abbreviations: EPA, eicosapentaenoic acid; AA, arachidonic acid.
Estimates were obtained from a linear mixed model, where values were log-transformed before analysis and back-transformed to obtain geometric mean values for presentation. The model included time, age groups, the interaction term of time × age groups, and sex as a fixed effect.
＊p＜0.05 vs. the corresponding age group in the 2002 survey.
| Exp (β) estimates (95% CI) a) | p value | |
|---|---|---|
| Serum EPA/AA ratio | ||
| Total population | 0.77 (0.75–0.79) | ＜0.001 |
| By age group | ||
| 40–49 | 0.68 (0.65–0.73) | ＜0.001 |
| 50–59 | 0.66 (0.63–0.69) | ＜0.001 |
| 60–69 | 0.74 (0.71–0.78) | ＜0.001 |
| 70–79 | 0.85 (0.80–0.90) | ＜0.001 |
| ≥ 80 | 0.94 (0.87–1.01) | 0.07 |
| Serum EPA concentrations | ||
| Total population | 1.04 (1.01–1.06) | 0.003 |
| By age group | ||
| 40–49 | 0.96 (0.91–1.02) | 0.23 |
| 50–59 | 0.90 (0.86–0.95) | ＜0.001 |
| 60–69 | 1.01 (0.96–1.06) | 0.69 |
| 70–79 | 1.13 (1.07–1.19) | ＜0.001 |
| ≥ 80 | 1.21 (1.12–1.30) | ＜0.001 |
| Serum AA concentrations | ||
| Total population | 1.35 (1.34–1.36) | ＜0.001 |
| By age group | ||
| 40–49 | 1.41 (1.37–1.45) | ＜0.001 |
| 50–59 | 1.37 (1.34–1.40) | ＜0.001 |
| 60–69 | 1.36 (1.33–1.39) | ＜0.001 |
| 70–79 | 1.33 (1.30–1.36) | ＜0.001 |
| ≥ 80 | 1.29 (1.25–1.33) | ＜0.001 |
Abbreviations: EPA, eicosapentaenoic acid; AA, arachidonic acid; CI, confidence interval.
The estimates represent the effect of time in the whole group and in each age group derived from the linear mixed models, where the serum EPA AA ratio, serum EPA concentrations, and serum AA concentrations were log-transformed prior to the analysis due to skewed distributions. The model included time, age group, interaction of time × age-group, sex, and time-varying variables including hypertension, diabetes, dyslipidemia, high-sensitivity C reactive protein, body mass index, current smoking, current drinking, and regular exercise.
a) The coefficients were exponentiated for presentation so that the values can be interpreted as relative changes (i.e., levels in the 2012 survey/levels in the 2002 survey). The exponentiated coefficients of ＞1.00 indicate an increase in values over time, and those of ＜1.00 indicate a decrease.
Additional analyses showed decreased serum DHA/AA ratios in all age groups, including the ≥ 80 years group (Supplementary Fig.2). As with the serum EPA concentrations, there was no secular change in serum DHA concentrations among the youngest group (40–49 years). There was a greater increase in serum DHA levels over time among the more advanced age groups (p for time×age-group interaction ＜0.001; Supplementary Fig.3).
Abbreviations: DHA, docosahexaenoic acid; AA, arachidonic acid.
Estimates were obtained from a linear mixed model, where values were log-transformed before analysis and back-transformed to obtain the geometric mean values for presentation. The model included time, age groups, the interaction term of time × age groups, and sex as a fixed effect.
＊p＜0.05 vs. the corresponding age group in the 2002 survey.
Abbreviation: DHA, docosahexaenoic acid.
Estimates were obtained from a linear mixed model, where values were log-transformed before analysis and back-transformed to obtain geometric means and their 95% CIs for presentation. The model included time, age groups, the interaction term of time × age groups, and sex as a fixed effect.
＊p＜0.05 vs. the corresponding age group in the 2002 survey.
This study described the changes in serum EPA/AA ratios at a population level in Japan, based on data from over 4,000 community residents. An overall decline in serum EPA/AA ratios was observed in this population. The decline in serum EPA/AA ratio was more pronounced in the younger age groups due to a marked increase in AA and a nonsignificant change in EPA in the younger age groups. Our findings suggested that cardiovascular risk might be elevated, reflected as circulating EPA/AA ratios, especially in middle age.
We demonstrated a decrease in the serum EPA/AA ratio over time in the analyzed population as a whole. In agreement with this observation, a previous study in 889 Japanese patients hospitalized with suspected acute coronary syndrome also demonstrated a decrease in the serum EPA/AA ratio over 10 years25). Another study in Japan demonstrated that serum AA concentrations increased from 2006 to 2012 among 878 community dwellers, while serum EPA concentrations were unchanged, and the serum EPA/AA ratio was not reported24). In the present study, using data from a relatively large sample size of approximately 4000 community residents, we found a decreasing trend in the EPA/AA ratio. We further confirmed similar results both in participants who took lipid-lowering agents and those who did not.
Increasing trends in serum AA concentrations were consistently observed in the present study and the previous studies of Japanese populations24, 25). A study in Japanese community dwellers demonstrated an increase in serum AA concentrations from approximately 150 µg/mL to ＞200 µg/mL24), which were comparable to the levels in the present study. In contrast, findings on secular change in serum EPA concentrations seemed to vary between the present study and the previous studies. We observed that serum EPA concentrations increased slightly in the whole population but not in those aged between 40 and 69 years. A study on acute coronary syndrome patients also reported a decrease in plasma EPA concentrations regardless of age25). In contrast, in the study mentioned above of community dwellers, the serum EPA concentration did increase in those aged ＜60 years, while the proportion of EPA to total fatty acid did not change over time24). The reason for the discrepancy between studies was unclear, but it was likely attributable to the different enrollment procedures and participation rates. Additionally, our findings were also inconsistent with those from a population-based study of 722 UK adults aged 40–79, in which increasing trends in plasma phospholipid EPA levels and decreasing trends in AA levels occurred30). The discrepancy may be due to the difference in the recent trends in intakes of dietary sources of fatty acids, such as fish and seafood, between populations.
A possible reason for the decrease in the serum EPA/AA ratio and the changes in individual fatty acids may be the recent change in the intake of food sources of fatty acids in Japanese populations. According to the national surveys, seafood intake in the general Japanese population decreased from 88.2 g/day in 2002 to 70.0 g/day in 2012 31, 32), and the decrease in population-level seafood intake was more pronounced in younger age groups31, 32). For example, seafood intake in the 40–49 age group decreased to 60.4 g/day in 2012 versus 91.8 g/day in 2002. Intake in the ≥ 70 age group decreased slightly but remained virtually unchanged, at 96.1 g/day in 2002 and 86.5 g/day in 2012. Our findings on the differences in secular change in serum EPA concentrations across age groups were consistent with the results of these national surveys. The similar trends in DHA we observed further implied the influence of fish and seafood intake on omega-3 fatty acids at a population level. Additionally, the abovementioned national surveys also demonstrated that meat intake increased over time regardless of age group31, 32). Dietary meat intake increased from 85.0 g/day to 105.6 g/day among those aged 40–49 years and from 44.6 g/day to 58.3 g/day among those aged ≥ 70, representing increases of 24% and 31%, respectively. Our observation of the general increase in serum AA concentrations may reflect these recent increasing trends in meat intakes. Meanwhile, Otsuka et al. reported that the proportion of serum AA (% total fatty acid) increased over a 13-year study period24). This increase remained even after adjusting for the dietary intake of AA, suggesting that fatty acid concentrations cannot be explained by dietary changes alone, but may be influenced by in vivo changes as well. Further longitudinal studies are warranted to examine the factors that determine changes in circulating fatty acid concentrations.
We found that secular changes in the serum EPA/AA ratio were similar between participants who took prescribed lipid-lowering agents and those who did not. This emphasizes the practical importance of monitoring the circulating EPA/AA ratio when evaluating residual cardiovascular risk at the population level. Despite the great variety in blood concentrations of omega-3 and omega-6 polyunsaturated fatty acids among populations in different counties20), consistent associations of low circulating EPA/AA ratio with inflammation and chronic diseases were reported in both Westerners and Japanese16, 33). Thus, the circulating EPA/AA ratio could be a universal marker of atherosclerotic diseases in community-dwelling adult populations. However, most of the previous studies on circulating EPA/AA ratios and cardiovascular health have been conducted in populations at high cardiovascular risk10-13, 34), and the evidence in low-risk populations is limited and still controversial16, 18). Further studies are warranted to investigate the potential predictive value of the circulating EPA/AA ratio and its changes for the risk of future cardiovascular events in nonpatients.
The present study has several strengths. The survey was conducted in the town of Hisayama, which has characteristics similar to those in the national statistics26, 35), on a large sample size with a high health-checkup participation rate, which may strengthen the generalizability of the findings to the Japanese population. The analysis using a mixed model allowed us to describe the trends at a population level without excluding data from participants with missing values. Limitations should also be noted. First, only four fractions of fatty acids were measured in the 2012 survey, so relative concentrations to total fatty acids could not be calculated. Second, we could not examine the influence of the dietary intakes of fatty acids, including the food sources of fatty acids and the use of fatty acid supplements, on the observed changes in serum fatty acids, as the two surveys used different dietary survey methods and thus were not comparable. Third, the sample storage period was different for the 2002 survey (approximately 8 years) and the 2012 survey (approximately 5 years), which might have affected the comparability of serum fatty acid concentrations. However, it has been reported that long-term storage at −80℃ for 10 years or less does not have much influence on fatty acid profiles36). Therefore, this is unlikely to change our conclusions.
The serum EPA/AA ratio decreased over time in community-dwelling adults. A decline in serum EPA/AA ratios, especially in middle age, might indicate an increase in the potential risk of cardiovascular disease at a population level. Thus, monitoring serum EPA/AA ratios would be of great importance among individuals who take lipid-lowering agents and those who do not evaluate residual cardiovascular risk at the population level.
We thank the staff of the Division of Health of Hisayama for their cooperation in this study. The statistical analyses were performed using the computers offered under the category of General Projects by the Research Institute for Information Technology, Kyushu University. We also thank KN International for proofreading the manuscript.
This study was supported in part by the Ministry of Education, Culture, Sports, Science and Technology of Japan (JSPS KAKENHI grant nos. JP21H03200, JP19K07890, JP20K10503, JP20K11020, JP21K07522, JP21K11725, JP21K10448, JP22K07421, and JP22K17396); by a Health and Labour Sciences Research Grant of the Ministry of Health, Labour and Welfare of Japan (no. JPMH20FA1002); and by a grant from the Japan Agency for Medical Research and Development (no. JP22dk0207053). This study was also sponsored by Mochida Pharmaceutical (Tokyo).
Dr. Ninomiya received research funding from Mochida Pharmaceutical. The other authors declare no conflict of interest.
