Gender Differences in the Associations of Plasma Pyridoxal 5′-Phosphate with Plasma Polyunsaturated Fatty Acids among US Young and Middle-Aged Adults: NHANES 2003–2004

Vitamin B6-restricted diets and low plasma pyridoxal 5′-phosphate (PLP) status altered plasma polyunsaturated fatty acids (PUFA) compositions. Evidence suggests the role of gender in the metabolism of vitamin B6 and PUFA. However, no epidemiologic study examined the impact of gender on the relationship between vitamin B6 and PUFA status in adults. Thus, we investigated whether there were gender differences in the association of vitamin B6 intake and plasma PLP concentration with plasma PUFA concentrations and ratios (eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), arachidonic acid (AA), EPA + DHA, EPA/AA, (EPA + DHA)/AA) in US young/middle-aged adults. In total, 864 participants (20–59 years; 484 men, 380 women) from the National Health and Nutrition Examination Survey (NHANES) 2003–2004 were used for this cross-sectional study. Nutrient intakes were estimated from two 24 h recalls and supplement questionnaires; plasma PLP and PUFA were measured. Multivariate linear regression was utilized to obtain unstandardized (b) and standardized (β) coefficients. Covariates included demographic, socioeconomic, dietary variables, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, and BMI. There were significant interactions between gender and PLP on EPA (P-interaction = 0.004), DHA (P-interaction = 0.020), EPA + DHA (P-interaction = 0.010), EPA/AA (P-interaction = 0.002), (EPA + DHA)/AA (P-interaction = 0.004), whereas no interaction between gender and B6 intake existed. In gender-stratified analyses, in men, PLP was positively associated with EPA (β = 0.138, b = 0.104, p = 0.0004), DHA (β = 0.101, b = 0.058, p = 0.036), EPA + DHA (β = 0.125, b = 0.073, p = 0.005), EPA/AA (β = 0.144, b = 0.099, p = 0.0002), (EPA + DHA)/AA (β = 0.123, b = 0.068, p = 0.005). However, no associations between PLP and PUFA existed in women. In conclusion, gender differences were found in the relationships between plasma PLP and plasma EPA, DHA, EPA + DHA, EPA/AA, and (EPA + DHA)/AA, with significant direct associations in men only among US young/middle-aged adults.

In male rats fed B6-restricted diets, linoleic acid (LA; 18:2n-6) increased, but arachidonic acid (AA; 20:4n-6) decreased in plasma and liver phospholipids compared to controls [4]. Similarly, another study reported an increase in LA and a decrease in AA in liver microsomal and plasma total lipids in B6-deficient male rats [7]. Tsuge et al. [8] further showed the lower contents of eicosapentaenoic acid (EPA; 20:5n-3), docosahexaenoic acid (DHA; other health-related data were collected [32]. In the MEC laboratory, blood specimens were processed and shipped to corresponding laboratories for analysis [34]. From the total 10,122 participants in the NHANES 2003-2004, participants aged ≥20 years with fasting (≥8 h) blood samples were used in this study (n = 4034). Of these, 1829 participants had measurements of plasma concentrations of ALA, LA, EPA, DHA, or AA. Of these, participants aged 20-59 years numbered 1141. The data from adults aged 20-59 years were used since the aging process may alter lipid metabolism by reducing fat oxidation and free fatty acid mobilization [35,36] and changing blood PUFA levels [37][38][39].
Consequently, the final resulting analytic sample size was 864 participants (484 men, 380 women). Depending on nonpositive weights, missing data (the responses of "refused" or "don't know" were treated as missing), and/or model covariates, the samples for descriptive and regression analyses were available from 674 to 864 participants. Of the final analytic sample, the proportion of men was higher than women (55.4% for men vs. 44.6% for women), which came from excluding women with the current use of HRT, oral/injectable contraceptives, or with pregnancy or lactation. A flow chart describing the sample selection is presented in Figure 1. via a computer-assisted personal interview (CAPI) system [32]. The participants were invited to the mobile examination center (MEC), where anthropometric, reproductive history, and other health-related data were collected [32]. In the MEC laboratory, blood specimens were processed and shipped to corresponding laboratories for analysis [34]. From the total 10,122 participants in the NHANES 2003−2004, participants aged ≥20 years with fasting (≥8 h) blood samples were used in this study (n = 4034). Of these, 1829 participants had measurements of plasma concentrations of ALA, LA, EPA, DHA, or AA. Of these, participants aged 20−59 years numbered 1141. The data from adults aged 20−59 years were used since the aging process may alter lipid metabolism by reducing fat oxidation and free fatty acid mobilization [35,36] and changing blood PUFA levels [37][38][39].
Consequently, the final resulting analytic sample size was 864 participants (484 men, 380 women). Depending on nonpositive weights, missing data (the responses of "refused" or "don't know" were treated as missing), and/or model covariates, the samples for descriptive and regression analyses were available from 674 to 864 participants. Of the final analytic sample, the proportion of men was higher than women (55.4% for men vs. 44.6% for women), which came from excluding women with the current use of HRT, oral/injectable contraceptives, or with pregnancy or lactation. A flow chart describing the sample selection is presented in Figure 1.

Assessment of Vitamin B6 and PUFA Intakes from Food and Supplements
According to the NHANES documentation [46], collection of two days of dietary intake data was performed by trained staff; a first 24 h recall interview at MEC was used for the first day's data and a second 24 h recall interview by telephone 3-10 d after the MEC  [46].
Personal interview data on the use of vitamins, minerals, and other dietary supplements (supplement name, ingredients, amounts, and serving size) were recorded by the interviewer [47]. Participants were asked duration, frequency, and daily amount in the past 30 days of dietary supplement use [48]. NCHS obtained product label information for dietary supplements from the manufacturer/retailer, the Internet, company catalogs, and the Dietary Supplement Label Database (DSLD) containing the full label contents from supplement products in the US [48]. The ingredient amounts in supplement products were determined by matching the name and the manufacturer of the product to those in a database developed by NCHS and the National Institutes of Health (NIH)'s Office of Dietary Supplements [48]. Using various dietary supplement files, we identified participants who reported the use of any dietary supplements containing the ingredients of vitamin B6, EPA, DHA, and ALA. The average daily intake of these nutrients from dietary supplements was estimated for participants, using the number of days that the use of supplements was reported, the amount taken per day, and the serving size unit from the supplement product label.
The mean dietary daily intakes of vitamin B6, ALA, LA, EPA, DHA, AA, total fat, and total energy were calculated from the averages of both the first and the second day dietary recall interviews. Octadecatrienoic acid (linolenic acid; 18:3), which includes mainly α-linolenic acid and lesser quantities of γ-linolenic acid (18:3n-6) [49], was examined. Octadecadienoic acid (LA), eicosatetraenoic acid (AA), EPA, and DHA were also assessed. Total daily intakes of vitamin B6, EPA, DHA, and ALA from food and dietary supplements were estimated by combining the mean dietary daily intake from dietary recall data and the average daily intake from dietary supplement data.

Definition of Gender
The terms of gender and sex are often used interchangeably in the scientific literature [50,51]. Sex is related to the biological distinctions that differ between men and women based on reproductive functions. In contrast, gender refers to socially and culturally constructed differences between men and women on the basis of the sex of the individual or the personal identification of an individual's own gender [50,51]. Based on the usage of gender in the codebook of NHNAES [52], the term of gender was used in this study, although this study's possible suggested mechanisms relevant to the function of vitamin B6 and PUFA are inherent to the biological differences between men and women. Homogeneous, nonradioactive, enzymatic assay (A/C Diagnostics, San Diego, CA, USA) was used for measuring plasma PLP concentration (nmol/L) [53]. The mean intraassay coefficient of variation (CV) from 7.8% to 8.3% and the mean inter-assay CV from 12.0% to 13.1% were reported [53]. Assay values below the lower limit of detection were replaced with 7.1 nmol/L (the detection limit of 10.09 nmol/L divided by the square root of 2) [53,54]. Plasma PLP was dichotomized by 20 nmol/L as a categorical variable because plasma PLP ≥ 20 nmol/L is the definition of vitamin B6 adequacy used to set the current estimated average requirements (EAR) and recommended dietary allowances (RDA) of vitamin B6 [55].

Plasma Polyunsaturated Fatty Acids (PUFA)
Blood samples for fatty acid concentration measurements (µmol/L) were collected from participants ≥20 years after fasting for ≥8 h. Gas chromatography-mass spectrometry was used to measure plasma fatty acids. Modified Lagerstedt et al. method was performed to measure plasma total fatty acid concentrations [56]. The mean intra-assay CV (standard deviation (SD)) for analytes was reported as 9% (10%) and the mean inter-assay CV (SD) for analytes as 8% (10%) [56]. In this study, we excluded subjects missing data on any individual fatty acids since this would influence the calculation of the sum of EPA and DHA and the ratios of EPA/AA and (EPA + DHA)/AA [57].

Study Covariates
Since plasma PUFA levels may be influenced by the following demographic, socioeconomic variables, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, BMI [58][59][60], and menopausal status [61], they were selected as covariates in this study. In addition, since the consumption of fatty acids and total fat can influence PUFA profiles [57,62,63], dietary factors including total fat intake and PUFA intake were included as model covariates.

Socioeconomic Factors
Family poverty income ratio (PIR), an index for the ratio of annual family income to the poverty threshold, was dichotomized by 1.3. Participants with PIR below 1.3 are eligible for the Supplemental Nutrition Assistance Program [64]. Educational attainment was categorized into high school graduation or less and some college or higher.

Dietary Factors
Dietary variables, such as total intakes of vitamin B6, EPA, DHA, ALA from food and supplements, dietary intakes of LA and AA from food, and total fat intake from food, were used as covariates.

Other Factors
Cigarette smoking status was categorized as a never smoker who smoked <100 cigarettes in life; a former smoker who did not smoke at the time of interview among those who had smoked ≥100 cigarettes in life; and a current smoker who reported ongoing smoking. Alcohol consumption was categorized as an abstainer who had <12 drinks of any type of alcoholic beverage in life; a former drinker who had ≥12 drinks in life or any one year, but none in the past 12 months; a current drinker who had ≥12 drinks in life, and drank ≥1 time in the past 12 months.
Physical activity level expressed on the metabolic equivalent of task (MET) score was categorized as <500, 500-1000, and ≥1000 MET min/week, which was calculated from the frequency and the duration of household/yard work, transportation, and leisure time. Prescription medication use was defined as a positive response to the question of taking any prescription medication in the past month.
Body mass index (BMI) was categorized as <18.5, 18.5-24.9, 25-29.9, and ≥30 kg/m 2 . Menopause was defined to be when a woman did not have a menstrual period in the past 12 months.

Statistical Methods
Appropriate sample weights were applied to account for complex survey design and the unequal probability of selection, noncoverage, and nonresponse bias. Variance estimates were computed using Taylor series linearization accounting for the complex sample design. All tests were two-sided, and the significance level was set at p < 0.05. Statistical analyses were performed with SAS 9.4 (SAS Institute Inc., Cary, NC, USA).
The dietary and the total intakes of vitamin B6, ALA, LA, EPA, DHA, and AA were energy-adjusted using the residual method to employ the regression model with total Nutrients 2021, 13, 477 6 of 18 caloric intake as the independent variable and absolute nutrient intake as the dependent variable [65].
Plasma PLP and PUFA variables were natural log-transformed to improve normality due to skewed distributions as they were used as dependent variables. We tested the normality of the distributions of plasma PLP and PUFA (EPA, DHA, AA, EPA + DHA, EPA/AA, (EPA + DHA)/AA) variables using quantile-quantile (Q-Q) plots and skewness values in the univariate analysis. Those variables were highly skewed except plasma AA, which were slightly skewed. To improve normality, plasma PLP and PUFA variables were natural log-transformed. After the transformation, we tested the normality using the same method described above, and those variables improved normality, approaching near-normal. The use of log-transformed AA did not make significant differences in the findings compared with the use of the original metric AA.
For descriptive statistics of the aforementioned covariates for all study participants, men and women each, we estimated frequencies and sample-weighted percentages with standard errors (SE) for categorical variables and arithmetic means (nutrient intake variables) and geometric means (plasma variables) with SE for continuous variables. The characteristics between men and women were compared using Rao-Scott F-adjusted chisquare tests for categorical variables and t-tests for continuous variables.
If vitamin B6 intake and plasma PLP are included together as independent variables in the model, this may lead to the unreliable estimation of regression coefficients in the fitted model due to possible multicollinearity [66]. In addition, the inclusion of B6 intake and PLP together as independent variables did not make material differences in the findings. Therefore, reported findings were from regressions using two separate models for vitamin B6 intake and plasma PLP, respectively.
To test whether gender modifies the relationship between vitamin B6 intake and plasma PLP, respectively, and plasma PUFA, we added the interaction terms of gender*B6 intake and gender*PLP each in the fully adjusted model before performing the genderstratified analysis. While the gender*B6 intake interaction was not significant, the gender*PLP interaction was significant. Thus, we evaluated the relationship between PLP and plasma PUFA using gender-stratified analysis.
Gender-stratified bivariate and multivariate linear regression analyses were performed to examine whether the association of plasma PLP concentration with plasma PUFA concentrations (EPA, DHA, AA, EPA + DHA) and ratios (EPA/AA, (EPA + DHA)/AA) differed by gender. Natural log-transformed plasma PLP and PUFA variables of interest were used in regression models after inspecting residual plots of gender-stratified models.
In the gender-specific models, the covariates were sequentially introduced in the following models. Model 0 was unadjusted. Model 1 was adjusted for demographic variables (age, race/ethnicity), BMI, dietary variables (total intakes of vitamin B6, EPA, DHA, ALA from food and supplements, dietary intake of LA and AA from food, total fat intake from food), and menopausal status (only for women). Model 2 was adjusted for all variables in model 2, plus socioeconomic variables (PIR, educational attainment), physical activity level, cigarette smoking status, alcohol consumption, and prescription medication use.
Unstandardized regression coefficients (b) with 95% confidence intervals were estimated. Standardized regression coefficient (β) was employed to express the change in log-transformed plasma PUFA concentrations and ratios in SD for 1 SD of change in log-transformed plasma PLP concentration. For each model, the coefficient of determination, R 2 , was quantified to measure the percentage of the total variability in plasma PUFA concentrations and ratios that the model explains. In the adjusted models 1 and 2, R 2 was adjusted for the model's number of predictors.
No multicollinearities were detected among independent variables in the fully adjusted gender-stratified regression models (for men: variance inflation factors (VIF) = 6.9 for total EPA intake, VIF = 7.3 for total DHA intake, VIF < 2.3 for the other variables; for women: VIF = 6.2 for total EPA intake, VIF = 7.4 for total DHA intake, VIF < 3.5 for the other variables). Table 1 presents the participants' characteristics by gender. There were more men than women in the study population (p = 0.003). Over one-third of women were in the range of 40-49 years of age, with more than a quarter of men in the range of 40-49 years of age (p = 0.046). Gender was distributed similarly for non-Hispanic White, and there was a tendency of a higher proportion of men for Hispanic and Other groups and a lower proportion of men for non-Hispanic Black compared to women (p = 0.09). The prevalence of overweight was higher for men than for women, while women were more likely to be obese than men (p = 0.0004). Compared with men, women were more likely to have a low family income (p = 0.0496) and were less likely to consume alcohol (p = 0.002). Men tended to be more physically active than women (p = 0.06). More women used vitamin B6 supplements (p = 0.018) and prescription medications (p = 0.011) than men. Approximately one-fifth of the women were postmenopausal (p < 0.0001). On the other hand, there were no differences between men and women for educational attainment, cigarette smoking status, and n-3 PUFA supplement use.  1 Rao-Scott F-adjusted chi-square tests for examining whether there are differences in proportions between men and women across categories of each characteristic. 2 Rao-Scott F-adjusted chi-square tests for examining whether there are differences in proportions between premenopausal and postmenopausal status. The supplement use of vitamin B6 and n-3 PUFA, respectively, was defined as a positive response to the question of taking any dietary supplements containing vitamin B6 and n-3 PUFA (eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), α-linolenic acid (ALA)) each in the past month. Table 2 displays distributions of the energy-adjusted dietary (from food) and total intakes (from food and supplements) of vitamin B6 and PUFA, concentrations and ratios of plasma PUFA, and concentrations of PLP by gender. Distributions of the original metric nutrient intakes were presented in Table S2. Using the residual method [65], nutrient intakes were evaluated in relation to the mean total energy intake of all study participants; the resulting measures of energy-adjusted nutrient intakes were independent of total energy intake. Due to the application of this method, overall, the energy-adjustment resulted in changes in the mean dietary intake estimates of the nutrients, with the decreased energyadjusted mean values from the original metric mean values in men and the increased energy-adjusted means from the original metric means in women. The original metric dietary vitamin B6 intake was lower in women than in men, but the energy-adjusted dietary vitamin B6 intake was similar between them, although the difference was not significant. A similar pattern was observed in the case of total vitamin B6 intake.

Distributions of the Intakes of Vitamin B6 and PUFA and the Concentrations of Plasma PUFA and PLP by Gender
There was a large increase from 2 mg of dietary vitamin B6 to 5 mg of total vitamin B6 from food and supplements, indicating some participants were taking high dosages of vitamin B6 supplements. We observed that, among vitamin B6 supplement users (n = 222 without 21 nonpositive dietary sample weight), the participants greater than 95 percentiles consumed more than 58.3 mg/d of total vitamin B6 up to 187.0 mg/d (median: 2.5 mg/d, interquartile range (IQR): 1.5-7.6 mg/d; data not shown), which could affect the large increase. Meanwhile, men and women had similar total vitamin B6 intake values from food and supplements, although a higher proportion of women took vitamin B6 supplements than men. This can be explained by the greater proportion of high dosage of vitamin B6 supplement intake in men than in women. For example, among vitamin B6 supplement users, six men and two women consumed more than 100 mg/d of vitamin B6 supplements, with five extreme observations of four men and one woman (149.4, 149.5, 150.7, 152.6, 187.0 mg/d; data not shown).
A mean total energy intake was lower in women than in men (p < 0.0001) after adjusting for demographic factors (age, race/ethnicity), socioeconomic factors (PIR, educational attainment), physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, BMI, and menopausal status. A mean dietary ALA intake was higher in women than in men (p = 0.046), whereas mean dietary EPA, DHA, and AA intakes were lower in women than men (p = 0.022, p = 0.027, p = 0.019, respectively). Similarly, a mean total ALA intake from food and supplements was higher in women (p = 0.036), but a mean total DHA intake tended to be greater in men than in women (p = 0.06). Differences between men and women did not exist in dietary intakes of vitamin B6 and LA from food and total intakes of vitamin B6 and EPA from food and supplements. Higher geometric means of plasma DHA concentration and (EPA + DHA)/AA ratio were observed in women than in men (p = 0.043, p = 0.034 each) after adjusting for demographic and socioeconomic factors, total fat intake, total intakes of EPA, DHA, ALA, dietary intakes of LA and AA, BMI, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, and menopausal status. A geometric mean plasma EPA concentration tended to be higher in men than in women (p = 0.09). There were no differences between men and women in plasma concentrations of ALA, LA, AA, EPA + DHA, and EPA/AA ratio.
A geometric mean plasma PLP concentration was lower in women than in men after adjusting for demographic and socioeconomic factors, total energy intake, total vitamin B6 intake, BMI, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, and menopausal status (p = 0.0002). Vitamin B6 deficiency, defined by plasma PLP concentration <20 nmol/L, was more common among women than men, with more than a quarter of women with vitamin B6 deficiency (p < 0.0001).

No Interaction Effects between Gender and Vitamin B6 Intake on Plasma PUFA Concentrations and Ratios
To assess whether gender modifies the association between vitamin B6 intake and plasma PUFA, we included an interaction term of gender*B6 intake in multivariate linear regression models fully adjusted for demographic and socioeconomic factors, total fat intake, total intakes of EPA, DHA, and ALA, dietary intakes of LA and AA, BMI, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, and menopausal status. There was no significant interaction between gender and vitamin B6 intake on EPA (Pinteraction = 0.37), DHA (Pinteraction = 0.11), AA (Pinteraction = 0.86), EPA + DHA (Pinteraction = 0.14), EPA/AA (Pinteraction = 0.34), and (EPA + DHA)/AA (Pinteraction = 0.11), respectively (data not shown). These results indicate that the relationship between vitamin B6 intake and plasma EPA, DHA, AA, EPA + DHA, EPA/AA, (EPA + DHA)/AA did not differ between men and women.
The relationships between plasma PLP concentration and plasma PUFA concentrations and ratios by gender are presented in Table 3. In gender-stratified bivariate and multivariate regression models, plasma PLP was positively associated with plasma EPA, DHA, EPA + DHA, EPA/AA, (EPA + DHA)/AA, respectively, in men only. In the fully adjusted model  Based on the results of the coefficient of determination, adjusted R 2 (aR 2 ), for men in the partial model 1, 24%, 27%, 29%, 29%, and 32% of the variance in EPA, DHA, EPA + DHA, EPA/AA, and (EPA + DHA)/AA, respectively, is explained by demographic, dietary factors and BMI. After further adjustment in the full model 2, in men, the additional 3%, 4%, 3%, 3%, 4% of the variance in EPA, DHA, EPA + DHA, EPA/AA, (EPA + DHA)/AA each is further explained by socioeconomic factors, physical activity level, cigarette smoking status, alcohol consumption, and prescription medication use, in addition to the model 1 covariates (aR 2 = 0.27, aR 2 = 0.31, aR 2 = 0.32. aR 2 = 0.32, aR 2 = 0.36, respectively).
Lastly, the relationship between PLP and AA was not affected by gender. No significant association existed between PLP and AA in all models for both men and women.

Discussion
This present study observed gender differences in the relationships between plasma PLP and plasma EPA, DHA, EPA + DHA, EPA/AA, and (EPA + DHA)/AA, with the significant positive associations in men only, but not in women, among US young and middle-aged adults after adjusting for demographic, socioeconomic and dietary factors, physical activity level, cigarette smoking status, alcohol consumption, prescription medication use, and BMI. In addition, the association between vitamin B6 intake and the selected PUFA was not affected by gender. To the best of our knowledge, this study is the first population-level observational study to report the gender differences in the relationship between vitamin B6 status, measured by plasma PLP, and plasma PUFA levels in adults aged 20-59 years.
Compared with men, a mean plasma DHA concentration was higher in women, which is supported by the studies describing that women had higher contributions of DHA to plasma lipids than men [14,20,24]. In addition, vitamin B6 intake was similar in both men and women in this study, but the mean plasma PLP level was lower in women than in men (33.5 nmol/L for women, 51.1 nmol/L for men). The differences in plasma PLP levels between them may be partially attributable to the prevalence of vitamin B6 deficiency (defined by plasma PLP < 20 nmol/L) for women, which was approximately three times higher than for men (29.0% for women, 10.4% for men).
Vitamin B6 serves as a coenzyme in the form of PLP for various metabolic reactions, including the synthesis of hemoglobin and neurotransmitters, interconversion of amino acids, gluconeogenesis, and metabolism of tryptophan, one-carbon units, and nucleic acids [67][68][69]. PUFA, such as EPA, DHA, and AA, provide important structural components of cell membranes [14,15] and exert immunomodulatory functions in humans [15,17]. EPA (n-3 PUFA) and AA (n-6 PUFA) can be synthesized from the precursors, ALA and LA, respectively, in the series of desaturation and elongation reactions catalyzed by elongases and desaturases such as ∆6-desaturase (D6D) and ∆5 desaturase (D5D) [16,24]. DHA is further converted from docosapentaenoic acid (DPA; 22:5n-3) via chain elongation, desaturation (by D6D), and peroxisomal β-oxidation [24]. D6D is the rate-limiting step for the biosynthesis of longer chain PUFA, including EPA and DHA; it is influenced by PLP status [12,70,71]. Although the underlying mechanisms of how vitamin B6 regulates D6D has not been fully understood yet [71], it may be speculated that inadequate vitamin B6 status could result in lowering the activity of D6D, thereby leading to the altered compositions of n-3 and n-6 PUFA.
An animal study showed that the D6D activity in male rats fed B6-deficient diets was lower than in the pair-fed control, suggesting low vitamin B6 status disturbed metabolic conversions from LA to AA and from ALA to EPA and DHA, with more reduction in DHA [8]. An in vitro study demonstrated that the relative mRNA expressions of FADS2 (D6D) and FADS1 (D5D) genes were lower in vitamin B6-restricted human hepatoma cells than in control [71]. Further, a case-control study with healthy men and women (21-60 years) reported that D6D activity plus D5D activity (EPA/ALA; enzyme activity measured by a product/precursor ratio) and the second D6D activity (DHA/DPA) were lower in a group with low serum vitamin B6 status (pyridoxine + PLP <3 µg/L; n = 21) than in a control group with normal B6 status (n = 22), although plasma PUFA compositions in the group with low B6 status were not different from the control group [9]. These previous findings imply that vitamin B6 status may influence fatty acid desaturation, possibly via PLP-dependent D6D catalyzing PUFA synthesis [5,8,9,71]. Therefore, in part, these may explain the positive associations between plasma PLP and PUFA for men in this study (Table 3).
This study revealed the significant positive associations between plasma PLP and plasma EPA, DHA, EPA/AA, and (EPA + DHA)/AA in men alone, not in women. D6D activity may be regulated by nutritional and non-nutritional factors, including glucose, alcohol, age, estrogen, insulin, etc. [24,72,73]. Besides, the status of iron, which is located at every terminal protein of the D6D enzyme complex [74], was shown to influence the activity Nutrients 2021, 13, 477 13 of 18 of D6D [75][76][77][78]. Evidence from animal and human studies has indicated that low iron levels, such as iron depletion or iron deficiency, may adversely affect PUFA synthesis [75][76][77][78][79]. It is noteworthy that the aforementioned study by Krajcovicova-Kudlackova et al. [9] demonstrated that the D6D activity (EPA/ALA, DHA/DPA) was positively correlated with serum iron levels. Moreover, the loss of D6D activity and the inhibition effect on PUFA synthesis were more pronounced in the group with low serum iron levels (<12 µmol/L for men, <10 µmol/L for women; n = 16) than in the group with low vitamin B6 level. Although data on the relationship between iron and PUFA metabolism are limited, the evidence thus far suggests that low iron status as well as low vitamin B6 status might negatively affect PUFA metabolism [9,[75][76][77][78][79][80]. Furthermore, the present study showed that, compared with men, women had not only the lower mean plasma PLP concentration but also the low serum iron concentration, with the greater prevalence of low vitamin B6 and iron status for women ( Table 2; Table S3), which is in agreement with the studies showing the greater prevalence of low iron status in women than in men [81,82]. Thus, the non-significant association of vitamin B6 status and PUFA in women, unlike men, in this study might be explained possibly due to the interaction between iron and PUFA metabolism [75][76][77][78][79]. The combination of vitamin B6 deficiency and low iron status might impact much more adversely on PUFA synthesis in women than in men. To confirm this, future research is necessary to explore whether iron could be another contributing factor to the gender differences in the relationship between vitamin B6 and PUFA metabolism.
Since plasma PLP can be reduced by inflammation [40], the differences in plasma PLP concentration between men and women may be, in part, mediated by inflammation. The proportion of elevated C-reactive protein (CRP) concentration (≥3 mg/L CRP), indicative of inflammation [83], was higher in women than in men (data not shown).
In particular, EPA and AA may play a critical role in regulating inflammatory responses by serving as precursors of EPA-derived and AA-derived eicosanoids, respectively [15]. AA promotes platelet aggregation and inflammatory reactions, whereas EPA exerts anti-inflammatory effects and acts as an antagonistic regulator against AA [15]. EPA and DHA also generate resolvins, which have anti-inflammatory and inflammationresolving properties [15]. Further, circulating EPA level was inversely associated with inflammatory markers and cardiovascular disease (CVD) incidence [27]; plasma EPA/AA is considered predictive of CVD risk [84,85] and chronic inflammation, with a higher EPA/AA ratio corresponding to lower levels of inflammation [85]. Besides, vitamin B6 status was inversely associated with inflammation [25,26]. This study revealed significant positive associations between plasma PLP and EPA, DHA, and EPA/AA in men (Table 3), which may imply a possible interrelationship between vitamin B6 status and blood PUFA levels and inflammatory diseases in men. However, the interconnection among vitamin B6, PUFA, and inflammation is unclear, and it remains to be answered whether inflammation in relation to vitamin B6 and PUFA would differ by gender.
This study has several limitations. First, there are remaining suspected confounders that were not measured to examine the association between plasma PLP and plasma PUFA. This study did not include lifestyle habits, such as vegetarianism, veganism, folate, vitamin B12, homocysteine, and methylmalonic acid in the assessment. Vegetarianism and veganism may influence plasma PUFA concentrations, and the adherence to such diets may increase the risk of developing vitamin B12 deficiency [86] and lower n-3 PUFA [87]. Vitamin B6 participates in one-carbon metabolism, and the associated metabolites, such as folate, vitamin B12, and homocysteine, are relevant to vitamin B6 deficiency [40]. For this reason, the observed association could still be influenced by residual confounding effects.
Second, in NHANES, fatty acids in plasma and serum were measured in two survey cycles, 2003-2004 and 2011-2012, respectively, while plasma or serum PLP was measured in four different cycles, 2003-2004, 2005-2006, 2007-2008, and 2009-2010. Thus, we utilized the data of fatty acids and PLP from the 2003-2004 cycle, which collected both of them together. However, since, over the past two decades, the use of fish oil supplements increased in US adults [88], it is possible that this could impact the population's level of PUFA status. Besides, in the 2003-2004 cycle, plasma fatty acids were measured as part of a surplus specimen project. Only about 45% of the participants with fasting blood samples had the measurements of plasma ALA, LA, EPA, DHA, or AA. This may result in the potential sample selection bias in this study.
Third, we did not exclude any outliers in the analysis. Five participants were taking unusually high dosages of vitamin B6 supplements. The relative standard error (RSE) was 10.3% with the outliers included and 7.4% with the outliers excluded. The estimate with the outliers may still be considered reliable since the RSE did not exceed 20%. For these reasons, those unusually high total vitamin B6 intake values could be valid and therefore were retained in this study; however, these influential observations may introduce bias.
Fourth, the smaller sample size of the group of women (n = 380) than the group of men (n = 484) in the gender-stratified analysis might lead to larger sampling variation in women than in men, possibly influencing the results of the nonsignificant association between plasma PLP and plasma PUFA in women and the significant association in men.
Fifth a cross-sectional survey design only demonstrates statistical associations but does not provide causality, making it difficult to establish mechanisms underlying the observations in this study. Sixth, we only utilized plasma fatty acids since other sources of fatty acids, such as erythrocytes, were not available in the NHANES database; unavailability of other direct and functional vitamin B6 biomarkers in the 2003-2004 survey cycle makes analyses limited to plasma PLP.
Nevertheless, this study has strengths. First, the rich, multiethnic NHANES dataset allowed us to use multiple covariates, which may reduce potential sources of bias. Second, although the method of 24 h dietary recall to assess dietary information is subject to have measurement error associated with the dietary recall method (e.g., underreporting or overreporting of food intakes) [89], the 2 day non-consecutive 24 h recall used in this study is considered to perform better than a single 24 h recall since it allows for reducing intra-individual variability in nutrient intakes [90]. Third, since the 2003-2004 cycle is unique to provide data for vitamin B6 intake and plasma PLP and PUFA together, this study took an opportunity to evaluate the relationship between vitamin B6 intake and status and plasma PUFA in the US representative sample.
In conclusion, the significant positive associations between plasma PLP and plasma EPA, DHA, EPA + DHA, EPA/AA, and (EPA + DHA)/AA were found in men only, not in women, among US young and middle-aged adults. Future large-scale prospective studies are necessary to confirm the relationship between gender, vitamin B6, and PUFA status.   Institutional Review Board Statement: Ethical review and approval were waived for this study because the NHANES datasets are de-identified and publicly available, and their use does not require IRB approval.
Informed Consent Statement: Patient consent was waived because, although the NHANES datasets were collected with informed consent, the publicly available data are de-identified, and these are secondary analyses of the publicly available datasets.