ABSTRACT
Background
Insulin regulates fatty acids (FAs) in the blood; conversely, FAs may mediate insulin sensitivity and are potentially modifiable risk factors of the diabetogenic state.
Objective
The objective of our study was to examine the associations between plasma concentrations of FAs, fasting plasma glucose (FPG), and glycated hemoglobin (HbA1c) among individuals (n = 1433) in the NHANES (2003–2004).
Design
Plasma concentrations of 24 individual FAs were considered individually and in subgroups, per chemical structure. Study participants were categorized in diabetogenic groups: Group 1 (HbA1c ≥6.5% or FPG ≥126 mg/dL), Group 2 (HbA1c 5.7% to <6.5% or FPG 100 to <126 mg/dL), and Group 3 (HbA1c <5.7% and FPG <100 mg/dL). We assessed associations between diabetogenic groups and plasma FAs in multivariate multinomial regressions (with Group 3 as the reference).
Results
Overall, 7.0% of study participants were in Group 1; 33.3% were in Group 2. Plasma concentrations of several individual FAs, including even-chain saturated FAs (SFAs; myristic, palmitic, stearic acids) and monounsaturated FAs (MUFAs; cis-vaccenic, oleic acids), were respectively associated with greater odds of Groups 1 and 2 status, adjusting for covariates. Higher concentrations of SFA and MUFA subgroups (highest compared with lowest quartile) were associated with increased odds of Group 2 status [SFAs adjusted OR (aOR): 1.51 (95% CI: 1.05, 2.18); MUFAs aOR: 1.78 (95% CI: 1.11, 2.85)]. Higher eicosapentaenoic acid plasma concentration was associated with decreased odds of Group 1 status [quartile 4 aOR: 0.41 (95% CI: 0.17, 0.95)].
Conclusions
Higher plasma concentrations of SFAs and MUFAs, primary de novo lipogenesis products, were associated with elevated FPG and HbA1c in a nationally representative study population in the United States. Additional studies are necessary to elucidate potential causal relationships between FAs (from endogenous production and dietary consumption) and diabetogenic indicators, as well as clinical implications for managing diabetes and prediabetes.
Keywords: fatty acids, glycated hemoglobin, fasting plasma glucose, United States
INTRODUCTION
Globally, >420 million people have diabetes (1, 2); moreover, estimates indicate that 629 million adults will have diabetes in 2045 (2). Among adults worldwide, the age-standardized prevalence of diabetes was 8.5% in 2014, which is nearly double the prevalence in 1980 (1). Nutrition therapy is a key lifestyle modification that is recommended for the prevention and management of diabetes (1, 3, 4). However, there is no universally recommended diet or eating pattern, aside from general guidelines such as lower calories and replacing saturated fats with unsaturated fats (1, 3, 4). One major research gap is determining the respective influences of individual fatty acids (FAs) in the etiology of the diabetogenic state.
This research gap stems from the complex and bidirectional putative linkages between the diabetogenic state and FAs (5). In healthy individuals, insulin is a key hormone that regulates the synthesis and storage of lipids (such as FAs), carbohydrates, and proteins (5). In a dysregulated state (e.g., insulin resistant), altered macronutrient metabolism may result in elevated circulating FA concentrations (5). Conversely, FAs are hypothesized to affect diabetogenic and metabolic indicators through several mechanistic pathways, including inflammation (6), oxidative stress (7), and hepatic de novo lipogenesis (8, 9). Prior studies have found heterogeneous results of the effects of FA dietary interventions and diabetogenic indicators (Supplemental Tables 1–4) (10).
There is limited literature regarding the associations of circulating concentrations of FAs (individually and as subgroups) and diabetogenic indicators among externally valid study populations. The generalizability of findings is particularly important, in order to account for the numerous factors that are likely to affect circulating FAs and the diabetogenic state. Examples of these factors include: interindividual variability in lipid absorption and metabolism (e.g., de novo lipogenesis) (11, 12), diet, and heterogeneous interventions and study populations in prior studies.
Therefore, our study objective was to assess the associations between plasma concentrations of FAs (individually and in subgroups), fasting plasma glucose (FPG), and glycated hemoglobin (HbA1c) in a nationally representative sample in the United States.
METHODS
Study population
In this cross-sectional study, data were from the 2003–2004 NHANES, undertaken by the National Center for Health Statistics at the Centers for Disease Control and Prevention. The stratified, multistage probability cluster sampling survey was designed to represent the noninstitutionalized civilian population in the 50 states (and the District of Columbia) of the United States. The NHANES methodology has been described in detail previously (13). The study ethics committee of the National Center for Health Statistics approved the study protocol. Study participants provided their written voluntary informed consent. This study was not a clinical trial as per the National Institutes of Health definition. The prespecified outcomes were 3 diabetogenic groups.
Among the 10,122 study participants (unweighted) in the NHANES 2003–2004 data collection cycle, subsets of study participants with fasting blood samples were assayed for HbA1c (n = 6601), FPG (n = 3169), fasting plasma insulin (n = 3136), and 24 individual plasma FA concentrations (n = 1459; Supplemental Figure 1). In our study, the inclusion criteria were: 1) age (≥20 y) and 2) availability of fasting blood samples with assays for FPG, HbA1c, and FAs (n = 1443). Additionally, we excluded any study participants with a nonpositive fasting sampling weight (i.e., zero values; n = 10).
Data collection
Sociodemographic and clinical information (including antidiabetes medications) were collected during household interviews. Approximately 1–2 wk after the household interview, study participants visited a mobile examination center; blood samples and anthropometric measurements were collected. Study participants were instructed to fast for ≥8 h prior to the mobile examination center visit.
Laboratory assays
Blood collection and processing protocols were previously documented in the NHANES Laboratory/Medical Technologists Procedures Manual. Quality assurance and control (including calibrators and bench quality control materials) are detailed in the NHANES laboratory methodology, per the 1988 Clinical Laboratory Improvement Act.
Blood samples were assayed for plasma concentrations of 24 individual FAs (micromoles per liter) by gas chromatography-mass spectrometry, based on a modified version of the Lagerstedt et al. method (14). Further details regarding this laboratory methodology and quality assurance have been documented previously (https://wwwn.cdc.gov/nchs/nhanes/2003–2004/SSFA_C.htm#Description_of_Laboratory_Methodology). In brief, the individual cis-FAs (6 SFAs, 7 MUFAs, 11 PUFAs) were quantified by comparing peak areas of the analyte and calibration curves.
FPG was assessed by a modified hexokinase enzymatic method. Insulin was assayed with a 2-site immunoenzymometric assay (Tosoh AIA-PACK IRI). HbA1c (percentage) was measured by a fully automated glycohemoglobin analyzer (Primus CLC330 and CLC 385; Primus Corporation, Kansas City, MO) at the University of Missouri-Columbia (Diabetes Diagnostic Laboratory). Percentage of HbA1c was calculated from the HbA1c peak area divided by the total hemoglobin peak area.
Definitions
Plasma concentrations of 24 individual FAs were categorized into 6 subgroups, based on chemical structure (Table 1). The subgroups of FA plasma concentrations included: SFAs, MUFAs, highly unsaturated fatty acids, n–3 PUFAs, long-chain n–3 PUFAs, and n–6 PUFAs. Additionally, the plasma concentration of total FAs was considered to be the sum of all 24 individual FAs.
TABLE 1.
Subgroups of plasma FA concentrations1
| SFAs | MUFAs | n–3 PUFAs | Long-chain n–3 PUFAs | n–6 PUFAs | HUFAs |
|---|---|---|---|---|---|
| Myristic (14:0) | Myristoleic (14:1n–5) | α-Linolenic (18:3n–3) | Eicosapentaenoic (20:5n–3) | Linoleic (18:2n–6) | Eicosadienoic (20:2n–6) |
| Palmitic (16:0) | Palmitoleic (16:1n–7) | Eicosapentaenoic (20:5n–3) | Docosahexaenoic (22:6n–3) | γ-Linolenic (18:3n–6) | Homo-γ-linolenic (20:3n–6) |
| Stearic (18:0) | cis-Vaccenic (18:1n–7) | Docosapentaenoic-3 (22:5n–3) | Eicosadienoic (20:2n–6) | Arachidonic (20:4n–6) | |
| Arachidic (20:0) | Oleic (18:1n–9) | Docosahexaenoic (22:6n–3) | Homo-γ-linolenic (20:3n–6) | Docosatetraenoic (22:4n–6) | |
| Docosanoic (22:0) | Eicosenoic (20:1n–9) | Arachidonic (20:4n–6) | Docosapentaenoic-6 (22:5n–6) | ||
| Lignoceric acid (24:0) | Docosenoic (22:1n–9) | Docosatetraenoic (22:4n–6) | Eicosapentaenoic (20:5n–3) | ||
| Nervonic (24:1n–9) | Docosapentaenoic-6 (22:5n–6) | Docosapentaenoic-3 (22:5n–3) | |||
| Docosahexaenoic (22:6n–3) |
1FA subgroups based on chemical structures and nomenclature. These are indicated by lipid numbers (C:D, where C is the number of carbons and D is the number of double bonds) and n–x nomenclature, which indicates the number of double bonds and the location of the double bond closest to the terminal end. FA, fatty acid; HUFA, highly unsaturated fatty acid.
Study participants were categorized into 3 diabetogenic groups, based on the cutoffs of FPG and HbA1c recommended by the American Diabetes Association (Table 2). The quantitative insulin sensitivity check index was calculated as 1/[log FPI (μU/mL) + log FPG (mg/dL)] (15). We stratified analyses based on antidiabetic medications (including any self-reported oral hypoglycemic agents or insulin).
TABLE 2.
Diabetogenic groups based on hyperglycemia and elevated glycated hemoglobin1
| Group | Definitions |
|---|---|
| 1 | FPG ≥126 mg/dL (7.0 mmol/L) or HbA1c ≥6.5% (48 mmol/mol) |
| 2 | FPG 100 to <126 mg/dL (5.6 to <7.0 mmol/L) or HbA1c 5.7 to <6.5% (39 to <48 mmol/mol) |
| 3 | FPG <100 mg/dL (5.6 mmol/L) and HbA1c <5.7% (39 mmol/mol) |
1Definitions based on American Diabetes Association classifications (47) and modified based on available indicators. FPG, fasting plasma glucose; HbA1c, glycated hemoglobin.
In terms of covariates, BMI was calculated as weight divided by height squared (kg/m2) and categorized per the World Health Organization's cutoff values. Race and ethnicity were categorized as non-Hispanic white, non-Hispanic black, Mexican American, other races (including multiracial), and other Hispanic.
Statistical analysis
Descriptive statistics were reported as means ± SEMs and percentages ± SEs. Normality assumptions were assessed for all continuous variables. Group comparisons used the Rao-Scott modified likelihood ratio and F test statistics (and their associated P values). Associations of interest were assessed with univariate and multivariate multinomial logistic regressions with diabetogenic group status as the dependent variable or outcome of interest (Group 3 as the reference group) and FAs as the key independent variables. Potential covariates were determined a priori based on previous literature; the same set was used in regressions to allow for comparability of interpretations. In multivariate regressions, missing indicators were used for covariates with <5% missingness.
Data were analyzed with SAS (version 9.4) survey procedures. All reported values accounted for the fasting subsample weight, cluster, and strata variables. As an exception, study sample and subsample sizes were not reported as weighted values. Two-tailed tests and α values of 0.05 were considered statistically significant. Since a subsample of NHANES study participants provided fasting blood samples, we compared a subset of variables in our analysis with study participants without fasting blood samples and other prior publications. The Cornell Institute for Social and Economic Research reproduced this analysis.
We computed power calculations post hoc, based on the following assumptions: 1) there was a survey design effect, per diabetogenic group status; 2) an α value of 0.05; 3) differences of Group 1 prevalence in the highest versus lowest quartiles of SFAs and MUFAs, respectively; and 4) an analytic sample size of 1433 (which was adjusted based on the design effect). With these assumptions, this study had ≥95% power to detect differences of Group 1 status in the extreme quartiles of SFAs and MUFAs, respectively.
RESULTS
Our final analysis included 1433 study participants (Supplemental Figure 1). Among study participants, 47.7% were male; the mean ± SEM age was 46.2 ± 0.6 y (Table 3). Other sociodemographic and clinical characteristics are in Table 3. All of these sociodemographic and clinical characteristics differed across Groups 1–3 (all P < 0.05), with the exceptions of smoking, race, and ethnicity (Table 3).
TABLE 3.
Sociodemographic characteristics1
| Sociodemographic characteristic | Overall, n = 1433 | Group 12(7.0%), n = 141 | Group 22 (33.3%), n = 500 | Group 32 (59.6%), n = 792 | P 3 |
|---|---|---|---|---|---|
| Age, y | 46.2 ± 0.6 | 59.2 ± (1.5) | 51.8 ± (0.7) | 41.5 ± (0.7) | <0.01 |
| Male, % | 47.7 ± (1.0) | 54.7 ± (5.4) | 55.1 ± (1.5) | 42.7 ± (1.8) | <0.01 |
| Race/ethnicity, % | |||||
| Non-Hispanic white | 72.6 ± (4.6) | 64.8 ± (7.3) | 74.5 ± (5.3) | 72.5 ± (4.4) | 0.11 |
| Non-Hispanic black | 11.2 ± (2.4) | 14.0 ± (4.2) | 9.8 ± (2.6) | 11.7 ± (2.5) | |
| Mexican American | 8.4 ± (2.5) | 13.7 ± (5.1) | 7.5 ± (2.5) | 8.3 ± (2.4) | |
| Other race (including multiracial) | 4.7 ± (1.3) | 3.9 ± (2.5) | 6.4 ± (2.2) | 3.8 ± (1.1) | |
| Other Hispanic | 3.0 ± (1.3) | 3.6 ± (2.5) | 1.8 ± (0.9) | 3.7 ± (1.6) | |
| BMI,4 kg/m2 | 28.2 ± (0.2) | 31.3 ± (0.7) | 29.8 ± (0.4) | 27.0 ± (0.2) | <0.01 |
| <18.5, % | 1.5 ± (0.4) | 0.0 ± (0.0) | 0.8 ± (0.4) | 2.1 ± (0.6) | <0.01 |
| 18.5 to <25, % | 32.0 ± (1.5) | 15.9 ± (3.3) | 21.4 ± (1.8) | 39.7 ± (2.1) | |
| 25 to <30, % | 34.4 ± (1.6) | 31.8 ± (6.3) | 36.8 ± (2.3) | 33.4 ± (2.1) | |
| ≥30, % | 32.1 ± (1.8) | 52.2 ± (5.8) | 41.0 ± (2.3) | 24.8 ± (2.1) | |
| Family history of diabetes,4 % | 51.0 ± (2.0) | 73.5 ± (4.8) | 53.8 ± (2.8) | 46.7 ± (2.5) | <0.01 |
| Smoking,5 % | 49.9 ± (1.8) | 56.0 ± (4.2) | 52.7 ± (3.2) | 47.6 ± (2.2) | 0.09 |
1Values are means ± SEMs or % ± SEs. All values (except sample sizes) account for complex survey design (including fasting subsample weights, strata, cluster variables). FPG, fasting plasma glucose; HbA1c, glycated hemoglobin.
2Per definitions in Table 2, Group 1 included study participants with FPG ≥126 mg/dL, HbA1c ≥6.5%, or both. Group 2 included individuals with FPG 100 to <126 mg/dL, HbA1c 5.7% to <6.5%. Group 3 included study participants with FPG <100 mg/dL and HbA1c <5.7%.
3Based on Rao-Scott modified likelihood ratio tests in univariate logistic regressions [diabetogenic group status as independent variable and respective dependent variables (in rows)].
4Study participants with missing observations included: BMI (n = 21), family history of diabetes (n = 26).
5≥100 cigarettes in lifetime.
Plasma concentrations of FA subgroups
Mean plasma concentrations of FA, including stratifications by quartiles and diabetogenic group status, are presented in Table 4. Mean total FAs, SFAs, MUFAs, n–6 PUFAs, and highly unsaturated fatty acid plasma concentrations differed between Groups 1–3 (all P < 0.05; Table 4). n–3 PUFAs and long-chain n–3 PUFAs plasma concentrations did not differ by diabetogenic group status (both P > 0.05; Table 4).
TABLE 4.
Plasma FA concentrations (μmol/L)1
| Overall,n = 1433 | Group 12 (7.0%), n= 141 | Group 22 (33.3%),n = 500 | Group 32 (59.6%), n = 792 | P3 | |
|---|---|---|---|---|---|
| TFA | 11,510.0 ± 111.1 | 12,657.0 ± 387.7 | 11,931.0 ± 149.5 | 11,140.0 ± 134.1 | <0.01 |
| Quartile 1 (low) | 8437.3 ± 60.7 | 8349.2 ± 227.5 | 8626.9 ± 96.6 | 8369.5 ± 74.5 | <0.01 |
| Quartile 2 | 10,506.0 ± 30.2 | 10,697.0 ± 98.7 | 10,516.0 ± 54.4 | 10,487.0 ± 37.9 | |
| Quartile 3 | 12,238.0 ± 39.1 | 12,328.0 ± 67.7 | 12,196.0 ± 83.1 | 12,253.0 ± 38.8 | |
| Quartile 4 (high) | 15,913.0 ± 183.5 | 17,042.0 ± 653.7 | 15,932.0 ± 231.0 | 15,656.0 ± 251.5 | |
| SFAs | 3835.6 ± 43.4 | 4337.8 ± 136.6 | 4014.2 ± 64.4 | 3676.5 ± 50.5 | <0.01 |
| Quartile 1 (low) | 2685.4 ± 23.9 | 2729.5 ± 65.1 | 2738.6 ± 45.8 | 2662.3 ± 21.9 | <0.01 |
| Quartile 2 | 3389.1 ± 9.2 | 3387.9 ± 49.5 | 3407.3 ± 16.2 | 3379.2 ± 11.4 | |
| Quartile 3 | 4036.0 ± 11.5 | 4131.5 ± 45.8 | 4025.4 ± 21.8 | 4027.2 ± 13.0 | |
| Quartile 4 (high) | 5547.6 ± 65.2 | 6024.1 ± 249.4 | 5530.3 ± 88.5 | 5459.6 ± 87.4 | |
| MUFAs | 2755.4 ± 40.1 | 3216.4 ± 135.5 | 2931.7 ± 55.5 | 2602.5 ± 49.6 | <0.01 |
| Quartile 1 (low) | 1757.6 ± 17.8 | 1784.5 ± 58.3 | 1779.1 ± 33.5 | 1749.3 ± 17.0 | <0.01 |
| Quartile 2 | 2374.8 ± 8.0 | 2320.9 ± 25.8 | 2379.8 ± 22.0 | 2376.4 ± 12.7 | |
| Quartile 3 | 2965.2 ± 10.4 | 3047.0 ± 44.6 | 2971.3 ± 24.7 | 2947.5 ± 15.1 | |
| Quartile 4 (high) | 4371.1 ± 60.4 | 4764.2 ± 255.4 | 4281.7 ± 76.1 | 4359.4 ± 123.5 | |
| n–3 PUFAs | 307.2 ± 7.4 | 327.0 ± 17.0 | 313.3 ± 8.5 | 301.4 ± 8.4 | 0.09 |
| Quartile 1 (low) | 181.3 ± 2.4 | 171.8 ± 12.5 | 183.6 ± 3.7 | 180.9 ± 2.4 | 0.01 |
| Quartile 2 | 254.4 ± 1.4 | 253.4 ± 3.1 | 254.4 ± 2.3 | 254.5 ± 1.7 | |
| Quartile 3 | 324.1 ± 1.4 | 321.9 ± 3.5 | 322.4 ± 2.2 | 325.3 ± 2.1 | |
| Quartile 4 (high) | 495.9 ± 8.7 | 544.7 ± 42.9 | 481.2 ± 13.6 | 500.1 ± 8.9 | |
| Long-chain n–3 PUFAs | 192.7 ± 6.3 | 202.0 ± 12.9 | 194.9 ± 7.1 | 190.4 ± 7.1 | 0.42 |
| Quartile 1 (low) | 103.4 ± 1.6 | 102.0 ± 4.9 | 103.3 ± 2.1 | 103.6 ± 1.8 | 0.03 |
| Quartile 2 | 151.3 ± 0.6 | 151.3 ± 2.9 | 151.4 ± 1.2 | 151.2 ± 0.8 | |
| Quartile 3 | 198.0 ± 1.0 | 194.8 ± 3.2 | 198.5 ± 1.8 | 198.2 ± 1.4 | |
| Quartile 4 (high) | 338.2 ± 8.8 | 357.0 ± 33.8 | 327.4 ± 14.0 | 342.3 ±9.5 | |
| n–6 PUFAs | 4612.3 ± 38.3 | 4775.3 ± 132.9 | 4671.5 ± 49.8 | 4560.0 ± 44.9 | 0.02 |
| Quartile 1 (low) | 3548.0 ± 22.4 | 3460.5 ± 125.5 | 3628.2 ± 26.7 | 3514.0 ± 33.8 | 0.29 |
| Quartile 2 | 4312.9 ± 10.2 | 4349.8 ± 35.9 | 4305.6 ± 16.3 | 4312.9 ± 13.0 | |
| Quartile 3 | 4917.4 ± 16.0 | 4894.8 ± 41.3 | 4920.6 ± 21.4 | 4917.9 ± 18.8 | |
| Quartile 4 (high) | 6006.8 ± 47.9 | 6324.7 ± 214.4 | 6039.8 ± 69.7 | 5930.5 ± 51.4 | |
| HUFAs | 1270.3 ± 14.4 | 1329.5 ± 46.5 | 1295.8 ± 19.9 | 1249.1 ± 16.3 | 0.01 |
| Quartile 1 (low) | 903.5 ± 8.8 | 856.7 ± 36.2 | 900.2 ± 15.0 | 909.4 ± 9.9 | 0.11 |
| Quartile 2 | 1170.5 ± 3.9 | 1191.6 ± 11.2 | 1167.5 ± 6.7 | 1169.7 ± 6.1 | |
| Quartile 3 | 1378.3 ± 3.7 | 1371.9 ± 14.8 | 1379.2 ± 6.8 | 1378.5 ± 4.7 | |
| Quartile 4 (high) | 1759.4 ± 16.1 | 1881.0 ± 84.2 | 1768.4 ± 25.1 | 1734.5 ± 17.3 | |
| DHA:EPA ratio | 192.7 ± 6.3 | 202.0 ± 12.9 | 194.9 ± 7.1 | 190.4 ± 7.1 | 0.42 |
1Values are means ± SEMs. FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; HUFA, highly unsaturated fatty acid; TFA, total fatty acid.
2Per definitions in Table 2, Group 1 included study participants with FPG ≥126 mg/dL, HbA1c ≥6.5%, or both. Group 2 included individuals with FPG 100 to <126 mg/dL, HbA1c 5.7% to <6.5%. Group 3 included study participants with FPG <100 mg/dL and HbA1c <5.7%.
3Based on Rao-Scott modified likelihood ratio tests in univariate logistic regressions [diabetogenic group status as dependent variable and respective independent variables (in rows)].
Elevated FPG and HbA1c
Overall, 7.0% ± 0.8% (% ± SE) of the study participants were categorized in Group 1, 33.3% ± 2.2% were in Group 2, and 59.6% ± 2.3% were in Group 3 (Table 5). A higher proportion of men were in Groups 1 (8.1% ± 1.2%) and 2 (38.5% ± 2.8%) relative to women (6.1% ± 1.0% and 28.6% ± 2.3%, respectively; P < 0.01; Table 5). Mean HbA1c was 5.5% ± <0.1% (SEM; Table 5). Among all study participants, 5.3% ± 0.8% (% ± SE) had HbA1c ≥6.5% and 14.4% ± 1.0% had HbA1c 5.7% to <6.5% (Table 5).
TABLE 5.
Diabetogenic indicators (groups, elevated FPG and HbA1c) and antidiabetes medications1
| Overall (n = 1433) | Male (47.7%) | Female (52.3%) | P 2 | |
|---|---|---|---|---|
| Diabetogenic groups, % | ||||
| 13 | 7.0 ± 0.8 | 8.1 ± 1.2 | 6.1 ± 1.0 | <0.01 |
| 23 | 33.3 ± 2.2 | 38.5 ± 2.8 | 28.6 ± 2.3 | |
| 33 | 59.6 ± 2.3 | 53.4 ± 2.8 | 65.3 ± 2.5 | |
| HbA1c, % | 5.5 ± <0.1 | 5.5 ± <0.1 | 5.4 ± <0.1 | 0.21 |
| ≥6.5% | 5.3 ± 0.8 | 5.6 ± 1.1 | 5.0 ± 0.9 | 0.80 |
| 5.7% to <6.5% | 14.4 ± 1.0 | 14.7 ± 1.4 | 14.1 ± 1.4 | |
| <5.7% | 80.3 ± 1.0 | 79.7 ± 1.5 | 80.9 ± 1.3 | |
| FPG, mg/dL | 99.8 ± 0.7 | 103.2 ± 1.0 | 96.7 ± 0.9 | <0.01 |
| ≥126, % | 5.7 ± 0.6 | 7.4 ± 1.1 | 4.1 ± 0.8 | <0.01 |
| 100 to <126, % | 29.1 ± 2.4 | 34.3 ± 2.7 | 24.4 ± 2.6 | |
| <100, % | 65.2 ± 2.5 | 58.3 ± 2.8 | 71.4 ± 2.7 | |
| Fasting plasma insulin, μU/mL | 10.5 ± 0.5 | 11.5 ± 0.8 | 9.6 ± 0.4 | 0.02 |
| QUICKI | 0.2 ± <0.1 | 0.2 ± <0.1 | 0.2 ± <0.1 | <0.01 |
| Diabetes medications, % | 5.5 ± 0.6 | 5.7 ± 0.8 | 5.3 ± 0.9 | 0.67 |
| Oral hypoglycemic agents4 | 5.2 ± 0.6 | 5.3 ± 0.7 | 5.0 ± 0.9 | 0.73 |
| Insulin | 0.4 ± 0.2 | 0.4 ± 0.2 | 0.4 ± 0.2 | 0.96 |
1Values are means ± SEMs or % ± SEs. All values (except sample sizes) account for complex survey design (including fasting subsample weights, strata, cluster variables). FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; QUICKI, quantitative insulin sensitivity check index.
2For categorical dependent variables, based on Rao-Scott modified likelihood ratio tests in univariate logistic regressions [sex as independent variable and respective dependent variables (in rows)]. For continuous dependent variables, based on F test for independent variable in univariate linear regressions [sex as independent variable and respective dependent variables (in rows)].
3Per definitions in Table 2, Group 1 included study participants with FPG ≥126 mg/dL, HbA1c ≥6.5%, or both. Group 2 included individuals with FPG 100 to <126 mg/dL, HbA1c 5.7% to <6.5%. Group 3 included study participants with FPG <100 mg/dL and HbA1c <5.7%.
4Currently taking any oral hypoglycemic agents to lower blood sugar. Assumed that any people who did not report were not taking any oral hypoglycemic agents.
Associations between FA subgroups and diabetogenic groups
Supplemental Table 5 includes the results from univariate multinomial regression models that assess the associations between FA plasma concentration subgroups as independent variables and diabetogenic group status as the dependent variable (with Group 3 as the reference). Adjusting for covariates in multivariate multinomial regression models, higher MUFA plasma concentrations were positively associated with Group 2 status [highest quartile 4 relative to lowest quartile 1 adjusted OR (aOR): 1.78 (95% CI: 1.11, 2.85); quartile 3 relative to lowest quartile 1 aOR: 1.73 (95% CI: 1.18, 2.54); Table 6]. Additionally, higher SFA plasma concentrations were associated with increased odds of Group 2 status [highest quartile 4 relative to lowest quartile 1 aOR: 1.51 (95% CI: 1.05, 2.18); Table 6]. The other associations that considered plasma concentrations of FA subgroups as independent variables of interest and diabetogenic group status as the dependent variables are reported in Table 6.
TABLE 6.
Multivariate associations between plasma concentrations of FAs and diabetogenic groups1
| Group 12 | Group 22 | Group 32 | ||||
|---|---|---|---|---|---|---|
| aOR3 | 95% CI | aOR3 | 95% CI | aOR3 | 95% CI | |
| TFA | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.57 | 0.25, 1.31 | 0.96 | 0.65, 1.42 | ||
| Quartile 3 | 1.09 | 0.46, 2.59 | 1.20 | 0.82, 1.76 | ||
| Quartile 4 (high) | 1.22 | 0.63, 2.37 | 1.24 | 0.87, 1.76 | ||
| SFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.55 | 0.27, 1.14 | 1.18 | 0.80, 1.74 | ||
| Quartile 3 | 1.31 | 0.68, 2.51 | 1.10 | 0.72, 1.68 | ||
| Quartile 4 (high) | 1.42 | 0.70, 2.87 | 1.51 | 1.05, 2.18 | ||
| MUFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.85 | 0.47, 1.53 | 1.48 | 1.03, 2.12 | ||
| Quartile 3 | 1.64 | 0.77, 3.47 | 1.73 | 1.18, 2.54 | ||
| Quartile 4 (high) | 1.67 | 0.81, 3.44 | 1.78 | 1.11, 2.85 | ||
| n–3 PUFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.59 | 0.63, 4.04 | 1.32 | 0.95, 1.84 | ||
| Quartile 3 | 1.17 | 0.45, 3.04 | 0.80 | 0.51, 1.25 | ||
| Quartile 4 (high) | 0.69 | 0.28, 1.71 | 0.93 | 0.67, 1.29 | ||
| Long-chain n–3 PUFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.27 | 0.55, 2.95 | 1.13 | 0.83, 1.55 | ||
| Quartile 3 | 0.88 | 0.42, 1.84 | 1.00 | 0.70, 1.43 | ||
| Quartile 4 (high) | 0.60 | 0.26, 1.38 | 0.82 | 0.60, 1.10 | ||
| n–6 PUFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.68 | 0.35, 1.33 | 0.77 | 0.52, 1.13 | ||
| Quartile 3 | 0.57 | 0.25, 1.29 | 0.78 | 0.54, 1.13 | ||
| Quartile 4 (high) | 1.07 | 0.57, 1.99 | 1.00 | 0.66, 1.50 | ||
| HUFAs | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.81 | 0.43, 1.51 | 1.05 | 0.74, 1.51 | ||
| Quartile 3 | 0.75 | 0.32, 1.74 | 1.04 | 0.69, 1.58 | ||
| Quartile 4 (high) | 0.78 | 0.36, 1.68 | 0.92 | 0.63, 1.35 | ||
| DHA:EPA ratio | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.27 | 0.55, 2.95 | 1.13 | 0.83, 1.55 | ||
| Quartile 3 | 0.88 | 0.42, 1.84 | 1.00 | 0.70, 1.43 | ||
| Quartile 4 (high) | 0.60 | 0.26, 1.38 | 0.82 | 0.60, 1.10 | ||
1aOR, adjusted OR; FA, fatty acid; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; HUFA, highly unsaturated fatty acid; TFA, total fatty acid.
2Per definitions in Table 2, Group 1 included study participants with FPG ≥126 mg/dL, HbA1c ≥6.5%, or both. Group 2 included individuals with FPG 100 to <126 mg/dL, HbA1c 5.7% to <6.5%. Group 3 included study participants with FPG <100 mg/dL and HbA1c <5.7%.
3Covariates included: age, sex, race/ethnicity, BMI, family history of diabetes, smoking (≥100 cigarettes in lifetime).
Additionally, we stratified the analysis by any self-reported antidiabetes medications (including insulin and oral hypoglycemic agents; Supplemental Table 6). Among study participants without antidiabetes medications, the associations between plasma concentrations of FA subgroups, FPG, and HbA1c were similar (Supplemental Table 6).
Associations between individual FAs and diabetogenic groups
The associations between individual FA plasma concentrations (independent variables) and diabetogenic group status (dependent variable) are presented in Supplemental Table 7 (bivariate) and Table 7 (multivariate). In terms of SFAs, higher plasma concentrations of even-chain SFAs were associated with Group 1 and 2 status (Table 7). Specifically, higher myristic [quartile 4 aOR: 2.27 (95% CI: 1.04, 4.97)] and stearic acid plasma concentrations [quartile 4 aOR: 2.23 (95% CI: 1.05, 4.73)] were associated with Group 1 status (Table 7). Higher palmitic acid plasma concentration was associated with Group 2 status [quartile 4 aOR: 1.68 (95% CI: 1.18, 2.38); Table 7]. Two of the longer-chain SFAs (docosanoic, lignoceric acids) were not associated with diabetogenic group status (P > 0.05; Table 7). Individuals with plasma concentration of arachidic acid in the second quartile were associated with Group 2 status [aOR: 1.59 (95% CI: 1.08, 2.34); Table 7].
TABLE 7.
Multivariate associations between plasma concentrations of individual FAs and diabetogenic groups1
| Group 12 | Group 22 | Group 32 | ||||
|---|---|---|---|---|---|---|
| aOR | 95% CI | aOR | 95% CI | aOR | 95% CI | |
| Myristic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.03 | 0.44, 2.43 | 1.18 | 0.75, 1.86 | ||
| Quartile 3 | 0.67 | 0.28, 1.64 | 0.89 | 0.55, 1.42 | ||
| Quartile 4 (high) | 2.27 | 1.04, 4.97 | 1.57 | 0.95, 2.58 | ||
| Palmitic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.71 | 0.36, 1.41 | 1.29 | 0.95, 1.76 | ||
| Quartile 3 | 0.99 | 0.55, 1.78 | 1.18 | 0.81, 1.72 | ||
| Quartile 4 (high) | 1.79 | 0.94, 3.42 | 1.68 | 1.18, 2.38 | ||
| Stearic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.65 | 0.84, 3.24 | 1.06 | 0.74, 1.50 | ||
| Quartile 3 | 1.43 | 0.65, 3.14 | 1.17 | 0.77, 1.76 | ||
| Quartile 4 (high) | 2.23 | 1.05, 4.73 | 1.44 | 0.97, 2.15 | ||
| Arachidic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.24 | 0.59, 2.57 | 1.59 | 1.08, 2.34 | ||
| Quartile 3 | 0.74 | 0.29, 1.92 | 1.03 | 0.67, 1.58 | ||
| Quartile 4 (high) | 1.10 | 0.58, 2.08 | 1.20 | 0.80, 1.79 | ||
| Docosanoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.81 | 0.34, 1.94 | 0.91 | 0.55, 1.50 | ||
| Quartile 3 | 0.86 | 0.38, 1.92 | 1.08 | 0.71, 1.64 | ||
| Quartile 4 (high) | 1.04 | 0.48, 2.28 | 1.06 | 0.62, 1.80 | ||
| Lignoceric | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.19 | 0.56, 2.51 | 1.22 | 0.76, 1.95 | ||
| Quartile 3 | 0.77 | 0.34, 1.78 | 1.06 | 0.72, 1.56 | ||
| Quartile 4 (high) | 1.35 | 0.62, 2.92 | 1.39 | 0.82, 2.33 | ||
| Myristoleic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.20 | 0.46, 3.14 | 0.94 | 0.55, 1.62 | ||
| Quartile 3 | 0.97 | 0.45, 2.10 | 1.11 | 0.64, 1.93 | ||
| Quartile 4 (high) | 1.58 | 0.70, 3.56 | 1.32 | 0.76, 2.28 | ||
| Palmitoleic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.37 | 0.57, 3.32 | 1.25 | 0.74, 2.10 | ||
| Quartile 3 | 1.22 | 0.53, 2.81 | 1.45 | 0.92, 2.29 | ||
| Quartile 4 (high) | 1.91 | 0.95, 2.84 | 1.69 | 0.97, 2.93 | ||
| cis-Vaccenic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.82 | 0.37, 1.84 | 1.67 | 1.07, 2.59 | ||
| Quartile 3 | 1.08 | 0.46, 2.52 | 1.35 | 0.87, 2.09 | ||
| Quartile 4 (high) | 1.22 | 0.53, 2.80 | 1.60 | 1.04, 2.46 | ||
| Oleic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.83 | 0.46, 1.51 | 1.24 | 0.87, 1.77 | ||
| Quartile 3 | 1.39 | 0.69, 2.81 | 1.64 | 1.13, 2.36 | ||
| Quartile 4 (high) | 1.85 | 0.91, 3.75 | 1.67 | 1.11, 2.49 | ||
| Eicosenoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.83 | 0.43, 1.61 | 1.01 | 0.59, 1.72 | ||
| Quartile 3 | 1.02 | 0.45, 2.31 | 0.97 | 0.68, 1.39 | ||
| Quartile 4 (high) | 2.07 | 1.01, 4.24 | 1.34 | 0.94, 1.90 | ||
| Docosenoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.21 | 0.49, 2.98 | 1.15 | 0.61, 2.16 | ||
| Quartile 3 | 0.83 | 0.37, 1.82 | 0.74 | 0.47, 1.16 | ||
| Quartile 4 (high) | 0.68 | 0.27, 1.76 | 1.02 | 0.60, 1.72 | ||
| Nervonic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.85 | 0.43, 1.70 | 0.75 | 0.51, 1.11 | ||
| Quartile 3 | 0.73 | 0.38, 1.38 | 0.97 | 0.62, 1.53 | ||
| Quartile 4 (high) | 0.26 | 0.12, 0.55 | 0.78 | 0.47, 1.28 | ||
| α-Linolenic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.03 | 0.44, 2.44 | 0.73 | 0.46, 1.14 | ||
| Quartile 3 | 1.14 | 0.40, 3.25 | 1.02 | 0.66, 1.59 | ||
| Quartile 4 (high) | 1.25 | 0.52, 3.04 | 0.91 | 0.57, 1.45 | ||
| Eicosapentaenoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.81 | 0.36, 1.82 | 0.76 | 0.48, 1.20 | ||
| Quartile 3 | 0.58 | 0.25, 1.33 | 1.07 | 0.75, 1.54 | ||
| Quartile 4 (high) | 0.41 | 0.17, 0.95 | 0.75 | 0.47, 1.17 | ||
| Docosapentaenoic-6 | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.97 | 0.32, 2.97 | 1.25 | 0.90, 1.73 | ||
| Quartile 3 | 0.69 | 0.32, 1.48 | 0.91 | 0.61, 1.36 | ||
| Quartile 4 (high) | 1.48 | 0.56, 3.95 | 1.11 | 0.69, 1.76 | ||
| Docosapentaenoic-3 | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.59 | 0.24, 1.48 | 1.03 | 0.70, 1.52 | ||
| Quartile 3 | 0.44 | 0.16, 1.20 | 1.09 | 0.67, 1.77 | ||
| Quartile 4 (high) | 0.47 | 0.22, 1.02 | 0.95 | 0.63, 1.43 | ||
| Docosahexaenoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.03 | 0.46, 2.31 | 1.00 | 0.75, 1.34 | ||
| Quartile 3 | 0.98 | 0.47, 2.03 | 0.85 | 0.57, 1.27 | ||
| Quartile 4 (high) | 0.57 | 0.22, 1.45 | 0.86 | 0.64, 1.15 | ||
| Linoleic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.55 | 0.23, 1.29 | 0.97 | 0.67, 1.40 | ||
| Quartile 3 | 0.71 | 0.32, 1.59 | 1.04 | 0.58, 1.84 | ||
| Quartile 4 (high) | 0.97 | 0.49, 1.89 | 1.09 | 0.73, 1.64 | ||
| γ-Linolenic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.88 | 0.40, 1.97 | 0.83 | 0.49, 1.39 | ||
| Quartile 3 | 1.11 | 0.52, 2.37 | 0.94 | 0.65, 1.35 | ||
| Quartile 4 (high) | 1.01 | 0.47, 2.17 | 1.23 | 0.84, 1.80 | ||
| Eicosadienoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.04 | 0.49, 2.22 | 0.81 | 0.57, 1.14 | ||
| Quartile 3 | 1.06 | 0.49, 2.26 | 1.14 | 0.74, 1.77 | ||
| Quartile 4 (high) | 1.34 | 0.56, 3.19 | 0.92 | 0.63, 1.34 | ||
| homo-γ-Linolenic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 1.16 | 0.52, 2.57 | 0.67 | 0.47, 0.98 | ||
| Quartile 3 | 2.91 | 1.47, 5.75 | 1.01 | 0.62, 1.65 | ||
| Quartile 4 (high) | 1.69 | 0.53, 4.55 | 1.00 | 0.63, 1.59 | ||
| Arachidonic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.66 | 0.27, 1.61 | 1.13 | 0.64, 2.00 | ||
| Quartile 3 | 0.81 | 0.34, 1.89 | 0.98 | 0.73, 1.33 | ||
| Quartile 4 (high) | 0.95 | 0.44, 2.04 | 1.23 | 0.81, 1.86 | ||
| Docosatetraenoic | ||||||
| Quartile 1 (low) | — | — | — | — | reference | |
| Quartile 2 | 0.90 | 0.35, 2.33 | 1.43 | 0.91, 2.26 | ||
| Quartile 3 | 1.28 | 0.58, 2.84 | 1.21 | 0.82, 1.78 | ||
| Quartile 4 (high) | 1.33 | 0.58, 3.08 | 1.25 | 0.82, 1.91 | ||
1aOR, adjusted OR; FA, fatty acid; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin.
2Per definitions in Table 2, Group 1 included study participants with FPG ≥126 mg/dL, HbA1c ≥6.5%, or both. Group 2 included individuals with FPG 100 to <126 mg/dL or HbA1c 5.7% to <6.5%. Group 3 included study participants with FPG <100 mg/dL and HbA1c <5.7%. ORs (and 95% CIs) are from multivariate multinomial regression models with diabetogenic group as the outcome and FA subgroup as the exposure of interest. Covariates included: age, sex, race/ethnicity, BMI, family history of diabetes, smoking (≥100 cigarettes in lifetime).
Among the assessed MUFAs and PUFAs, higher plasma concentration of eicosenoic acid was associated with increased odds of Group 1 status [quartile 4 aOR: 2.07 (95% CI: 1.01, 4.24); Table 7]. Individuals with plasma concentrations of nervonic acid in the highest quartile had decreased odds of Group 1 status (aOR: 0.26 (95% CI: 0.12, 0.55]; Table 7). Higher plasma concentrations of cis-vaccenic and oleic acid were associated with greater odds of Group 2 status (P < 0.05; Table 7). Higher plasma concentration of EPA was associated with decreased odds of Group 1 status [quartile 4 aOR: 0.41 (95% CI: 0.17, 0.95); Table 7]. Higher plasma concentration of homo-γ-linolenic acid was associated with an increased odds of Group 1 status [quartile 3 aOR: 2.91 (95% CI: 1.47, 5.75); Table 7]. Several other n–3 and n–6 PUFAs were not associated with diabetogenic group (P > 0.05; Table 7).
DISCUSSION
Higher plasma concentrations of SFAs and MUFAs (individual FAs, subgroups) were associated with increased odds of elevated fasting blood glucose and HbA1c among a nationally representative sample of the US population. Most of these SFAs (myristic, stearic, palmitic acids) and MUFAs (cis-vaccenic, oleic acids) are considered primary products of the de novo lipogenesis pathway, which highlights the need for further studies to understand the etiological role of endogenously synthesized individual SFAs and MUFAs in the development and progression of diabetes.
Metabolic flexibility and de novo synthesis of FAs
Metabolic inflexibility, including during dysregulation of de novo lipogenesis, has been linked with disease states such as insulin resistance (8, 16, 17). Prior studies have established that metabolic inflexibility occurs during type 2 diabetes, in light of evidence that: 1) glucose and FAs are primary substrates that compete for respiration (18), and 2) insulin affects glucose and lipid metabolism. However, the transition from metabolic flexibility in a healthy individual to metabolic inflexibility (in diabetes) remains unclear. How and to what extent is lipid metabolism affected during prediabetes (19)? Conversely, how do higher concentrations of free FAs contribute to a diabetogenic state (20)? Our results showed that higher quartiles of certain individual SFAs and MUFAs were associated with increased odds of Group 2 status (defined by common prediabetes cutoffs). Therefore, our findings suggest the need to further examine the potential alterations of lipid metabolism, as well as the effects of higher free FA concentrations during impaired glucose tolerance or prediabetes.
In humans, many FAs are endogenously produced through the conversion of excess carbohydrate substrate via glycolysis (8, 21). Prior studies have shown that certain dietary FAs (e.g., the trans isoform of palmitoleate) are associated with lower risk of diabetogenic indicators relative to endogenous FAs (22, 23); however, further studies are necessary to delineate the etiological mechanisms (21), as well as to determine the generalizability of these findings. Regarding the latter, our study found that plasma concentrations of several individual FAs, considered primary products from de novo lipogenesis, were associated with increased odds of diabetogenic indicators.
SFAs and diabetogenic indicators
Our finding that higher plasma concentrations of SFAs were associated with elevated FPG and HbA1c was consistent with previous literature focusing on diabetogenic indicators (24–26). In a prospective case-cohort study among 3737 individuals, those with plasma phospholipid SFA concentrations in the highest quintile had increased odds of diabetes incidence [aOR: 3.76 (95% CI: 2.43, 5.81); P for trend <0.01] relative to the lowest quintile (24). Similarly, in another longitudinal study (Atherosclerosis Risk in Communities Study; n = 2909 adults), diabetes incidence was positively associated with quintiles of SFA (as percentages of total FAs) in plasma cholesterol esters and phospholipids (26). Separately, dietary intake studies have shown positive associations between dietary consumption of SFAs and diabetogenic indicators (including type 2 diabetes, insulin resistance, and FPG) among humans (27, 28) and mice (29).
Our results showed that individual plasma concentrations of several even-chain SFAs (myristic, stearic, palmitic acids) were positively associated with increased odds of elevated FPG and HbA1c, which has been corroborated by other literature (9, 30). Similarly, in the EPIC-InterAct case-cohort study (n = 12,403 cases with type 2 diabetes; n = 16,154 individuals in the general population cohort), myristic, stearic, and palmitic acids were positively associated with type 2 diabetes incidence (30). Among older US adults in the Cardiovascular Health Study, palmitic acid [extreme quintile HR: 1.89 (95% CI: 1.27, 2.83)] and stearic acid [HR: 1.62 (95% CI: 1.09, 2.41)] were associated with greater diabetes risk (9).
MUFAs and diabetogenic indicators
In this study, higher plasma concentrations of MUFAs were associated with increased FPG and HbA1c, which is similar to a prior study (31). Separately, several reviews and meta-analyses considered the effects of dietary MUFAs on diabetogenic indicators. However, findings have been inconsistent, including protective (32, 33), harmful (34), and null associations (33). Among rats, high dietary consumption of MUFAs was associated with severe insulin resistance (29).
In terms of individual MUFA plasma concentrations, a prospective cohort study found that eicosenoic acid [highest compared with lowest tertile aOR: 0.48 (95% CI: 0.27, 0.87)] and vaccenic acid [highest compared with lowest tertile aOR: 0.40 (95% CI: 0.22, 0.72)] were associated with incident type 2 diabetes (35). The significant associations were similar to our study, although our findings showed the opposite directions, which could potentially reflect a bidirectional etiology.
Other FAs and diabetogenic indicators
We found that plasma concentrations of PUFAs were not associated with elevated FPG and HbA1c, which corroborates some studies considering circulating concentrations of PUFAs and diabetogenic indicators (36). In contrast, other studies have shown that circulating n–3 PUFA is inversely associated with insulin sensitivity in rodents (36) and type 2 diabetes (26, 37). Several systematic reviews and meta-analyses focusing on dietary PUFAs and diabetogenic indicators have had inconsistent findings (38), including protective (39–41), harmful (41), and null findings (39, 41–45). Further studies are needed to explain these inconsistencies; one potential explanation is that dietary patterns (or combinations) of FAs may be protective against the diabetogenic state (46).
Strengths and limitations
Strengths of this study include the complex survey design, which allowed for findings to be considered nationally representative and generalizable. Furthermore, data were available on circulating individual FAs, FPG, HbA1c, and potential confounders. However, there are several limitations to this study. First, utilizing FPG and HbA1c as indicators to determine diabetogenic groups could result in misclassification of individuals whose diabetes is well controlled (such as through lifestyle factors, including healthy diet and physical activity). While Group 1 includes study participants with elevated FPG or HBA1c (per common diabetes cutoffs), Groups 2 and 3 may include study participants considered healthy (based on lower FPG and HbA1c) and with well-controlled diabetes. We considered a stratified analysis based on self-reported antidiabetes medications; the results among all study participants were similar to those without antidiabetes medications. Second, selection bias is possible, given that study participants with available fasting blood samples and assays (FA concentrations, FPG, HbA1c) are a subset of the overall study participant population. However, the NHANES methodology was designed for this fasting subsample to be nationally representative; furthermore, our findings for other covariates were similar to the overall NHANES study population during the 2003–2004 data collection cycle. Third, longitudinal studies with repeated measurements of FA blood concentrations are needed in the future, in order to account for intraindividual variability (e.g., FA subgroup absorption, de novo lipogenesis, and baseline status).
In conclusion, higher plasma concentrations of individual SFAs and MUFAs, which are primary products of de novo lipogenesis, were respectively associated with elevated FPG and HbA1c among a nationally representative US study population. Further studies are necessary to understand causal relationships of individual SFAs and MUFAs in the diabetogenic state, as well as implications for prevention and clinical management of prediabetes and diabetes.
Supplementary Material
Acknowledgements
The authors’ responsibilities were as follows—SM and EAY: designed the research; EAY and PJH: analyzed the data; EAY: wrote the first draft of the manuscript; SM: had primary responsibility for final content; and all authors: contributed to critically revising the paper, and read and approved the final manuscript. SM is an unpaid board member of a diagnostic start-up focused on developing assays for low-cost and point-of-care measurement of certain nutrients from a drop of blood using results from his research as a faculty member at Cornell University. The remaining authors have no conflicts of interest related to this study.
Notes
Research reported in this publication was supported by Cornell University (Division of Nutritional Sciences) and the National Institutes of Health (National Institute of Diabetes and Digestive and Kidney Diseases, T32-DK007158 awarded to EAY). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases or the National Institutes of Health.
Supplemental Tables 1–7 and Supplemental Figures 1-3 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: aOR, adjusted OR; FA, fatty acid; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin.
REFERENCES
- 1. World Health Organization. Global report on diabetes. Geneva: World Health Organization; 2016. [Google Scholar]
- 2. International Diabetes Federation. IDF diabetes atlas. 8th ed. Brussels: International Diabetes Federation; 2017. [Google Scholar]
- 3. International Diabetes Federation. IDF clinical practice recommendations for managing type 2 diabetes in primary care. Brussels: International Diabetes Federation; 2017. [Google Scholar]
- 4. American Diabetes Association. 4. Lifestyle management. Diabetes Care. 2017;40(Suppl 1):S33. [DOI] [PubMed] [Google Scholar]
- 5. Saltiel AR, Kahn CR. Insulin signalling and the regulation of glucose and lipid metabolism. Nature. 2001;414:799. [DOI] [PubMed] [Google Scholar]
- 6. Odegaard JI, Chawla A. Pleiotropic actions of insulin resistance and inflammation in metabolic homeostasis. Science. 2013;339(6116):172–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Delarue J, Magnan C. Free fatty acids and insulin resistance. Current Opin Clin Nutri Metab Care. 2007;10(2):142–8. [DOI] [PubMed] [Google Scholar]
- 8. Sanders FWB, Griffin JL. De novo lipogenesis in the liver in health and disease: more than just a shunting yard for glucose. Biol Rev Camb Philos Soc. 2016;91(2):452–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ma W, Wu JH, Wang Q, Lemaitre RN, Mukamal KJ, Djousse L, King IB, Song X, Biggs ML, Delaney JA et al.. Prospective association of fatty acids in the de novo lipogenesis pathway with risk of type 2 diabetes: the Cardiovascular Health Study. Am J Clin Nutr. 2015;101(1):153–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Food and Agriculture Organization of the United Nations. Fats and fatty acids in human nutrition: report of an expert consultation. FAO Food and Nutrition Paper 91. Rome: Food and Agriculture Organization of the United Nation; 2010. [PubMed] [Google Scholar]
- 11. Lammert F, Wang DQH. New insights into the genetic regulation of intestinal cholesterol absorption. Gastroenterology. 2005;129(2):718–34. [DOI] [PubMed] [Google Scholar]
- 12. Baylin A, Campos H. The use of fatty acid biomarkers to reflect dietary intake. Curr Opin Lipidol. 2006;17(1):22–7. [DOI] [PubMed] [Google Scholar]
- 13. National Center for Health Statistics. 2003–2004 National Health and Nutrition Examination Survey (NHANES) [Internet]. [cited 2017 Dec 13]. Available from: http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/current_nhanes_03_04.htm. [Google Scholar]
- 14. Lagerstedt SA, Hinrichs DR, Batt SM, Magera MJ, Rinaldo P, McConnell JP. Quantitative determination of plasma c8–c26 total fatty acids for the biochemical diagnosis of nutritional and metabolic disorders. Mol Genet Metab. 2001;73(1):38–45. [DOI] [PubMed] [Google Scholar]
- 15. Katz A, Nambi SS, Mather K, Baron AD, Follmann DA, Sullivan G, Quon MJ. Quantitative insulin sensitivity check index: a simple, accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab. 2000;85(7):2402–10. [DOI] [PubMed] [Google Scholar]
- 16. Postic C, Girard J. Contribution of de novo fatty acid synthesis to hepatic steatosis and insulin resistance: lessons from genetically engineered mice. J Clin Invest. 2008;118(3):829–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Goodpaster BH, Sparks LM. Metabolic flexibility in health and disease. Cell Metab. 2017;25(5):1027–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Randle PJ. Regulatory interactions between lipids and carbohydrates: the glucose fatty acid cycle after 35 years. Diabetes Metab Rev. 1998;14(4):263–83. [DOI] [PubMed] [Google Scholar]
- 19. Faerch K, Vaag A. Metabolic inflexibility is a common feature of impaired fasting glycaemia and impaired glucose tolerance. Acta Diabetol. 2011;48(4):349–53. [DOI] [PubMed] [Google Scholar]
- 20. Roden M, Price TB, Perseghin G, Petersen KF, Rothman DL, Cline GW, Shulman GI. Mechanism of free fatty acid-induced insulin resistance in humans. J Clin Invest. 1996;97(12):2859–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Yilmaz M, Claiborn KC, Hotamisligil GS. De novo lipogenesis products and endogenous lipokines. Diabetes. 2016;65(7):1800–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Frigolet ME, Gutierrez-Aguilar R. The role of the novel lipokine palmitoleic acid in health and disease. Adv Nutr. 2017;8(1):173s–81s. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Mozaffarian D, Cao H, King IB, Lemaitre RN, Song X, Siscovick DS, Hotamisligil GS. Trans-palmitoleic acid, metabolic risk factors, and new-onset diabetes in U.S. adults: a cohort study. Ann Intern Med. 2010;153(12):790–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Hodge AM, English DR, O'Dea K, Sinclair AJ, Makrides M, Gibson RA, Giles GG. Plasma phospholipid and dietary fatty acids as predictors of type 2 diabetes: interpreting the role of linoleic acid. Am J Clin Nutr. 2007;86(1):189–97. [DOI] [PubMed] [Google Scholar]
- 25. Vessby B, Aro A, Skarfors E, Berglund L, Salminen I, Lithell H. The risk to develop NIDDM is related to the fatty acid composition of the serum cholesterol esters. Diabetes. 1994;43(11):1353–7. [DOI] [PubMed] [Google Scholar]
- 26. Wang L, Folsom AR, Zheng ZJ, Pankow JS, Eckfeldt JH. Plasma fatty acid composition and incidence of diabetes in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Clin Nutr. 2003;78(1):91–8. [DOI] [PubMed] [Google Scholar]
- 27. Deer J, Koska J, Ozias M, Reaven P. Dietary models of insulin resistance. Metabolism. 2015;64(2):163–71. [DOI] [PubMed] [Google Scholar]
- 28. Feskens EJ, Sluik D, van Woudenbergh GJ. Meat consumption, diabetes, and its complications. Curr Diab Rep. 2013;13(2):298–306. [DOI] [PubMed] [Google Scholar]
- 29. Storlien LH, Jenkins AB, Chisholm DJ, Pascoe WS, Khouri S, Kraegen EW. Influence of dietary fat composition on development of insulin resistance in rats: relationship to muscle triglyceride and ω-3 fatty acids in muscle phospholipid. Diabetes. 1991;40(2):280–9. [DOI] [PubMed] [Google Scholar]
- 30. Forouhi NG, Koulman A, Sharp SJ, Imamura F, Kröger J, Schulze MB, Crowe FL, Huerta JM, Guevara M, Beulens JWJ et al.. Differences in the prospective association between individual plasma phospholipid saturated fatty acids and incident type 2 diabetes: the EPIC-InterAct case-cohort study. Lancet Diabetes Endocrinol. 2014;2(10):810–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Krachler B, Norberg M, Eriksson JW, Hallmans G, Johansson I, Vessby B, Weinehall L, Lindahl B. Fatty acid profile of the erythrocyte membrane preceding development of type 2 diabetes mellitus. Nutr Metab Cardiovasc Dis. 2008;18(7):503–10. [DOI] [PubMed] [Google Scholar]
- 32. Qian F, Korat AA, Malik V, Hu FB. Metabolic effects of monounsaturated fatty acid-enriched diets compared with carbohydrate or polyunsaturated fatty acid-enriched diets in patients with type 2 diabetes: a systematic review and meta-analysis of randomized controlled trials. Diabetes Care. 2016;39(8):1448–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Schwingshackl L, Strasser B, Hoffmann G. Effects of monounsaturated fatty acids on glycaemic control in patients with abnormal glucose metabolism: a systematic review and meta-analysis. Ann Nutr Metab. 2011;58(4):290–6. [DOI] [PubMed] [Google Scholar]
- 34. Ha V, Bonner AJ, Jadoo JK, Beyene J, Anand SS, de Souza RJ. The effects of various diets on glycemic outcomes during pregnancy: a systematic review and network meta-analysis. PLoS One. 2017;12(8):e0182095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Patel PS, Sharp SJ, Jansen E, Luben RN, Khaw K-T, Wareham NJ, Forouhi NG. Fatty acids measured in plasma and erythrocyte-membrane phospholipids and derived by food-frequency questionnaire and the risk of new-onset type 2 diabetes: a pilot study in the European Prospective Investigation into Cancer and Nutrition (EPIC)–Norfolk cohort. Am J Clin Nutr. 2010;92(5):1214–22. [DOI] [PubMed] [Google Scholar]
- 36. Lalia AZ, Lanza IR. Insulin-sensitizing effects of omega-3 fatty acids: lost in translation?. Nutrients. 2016;8(6):329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Forouhi NG, Imamura F, Sharp SJ, Koulman A, Schulze MB, Zheng J, Ye Z, Sluijs I, Guevara M, Huerta JM et al.. Association of plasma phospholipid n–3 and n–6 polyunsaturated fatty acids with type 2 diabetes: the EPIC-InterAct case-cohort study. PLoS Med. 2016;13(7):e1002094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Coelho OGL, da Silva BP, Rocha D, Lopes LL, Alfenas RCG. Polyunsaturated fatty acids and type 2 diabetes: impact on the glycemic control mechanism. Crit Rev Food Sci Nutr. 2017;57(17):3614–9. [DOI] [PubMed] [Google Scholar]
- 39. Abbott KA, Burrows TL, Thota RN, Acharya S, Garg ML. Do omega-3 PUFAs affect insulin resistance in a sex-specific manner? A systematic review and meta-analysis of randomized controlled trials. Am J Clin Nutr. 2016;104(5):1470–84. [DOI] [PubMed] [Google Scholar]
- 40. Arnoldussen IA, Kiliaan AJ. Impact of DHA on metabolic diseases from womb to tomb. Mar Drugs. 2014;12(12):6190–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Muley A, Muley P, Shah M. ALA, fatty fish or marine n–3 fatty acids for preventing DM?: a systematic review and meta-analysis. Curr Diabetes Rev. 2014;10(3):158–65. [DOI] [PubMed] [Google Scholar]
- 42. Jeppesen C, Schiller K, Schulze MB. Omega-3 and omega-6 fatty acids and type 2 diabetes. Curr Diab Rep. 2013;13(2):279–88. [DOI] [PubMed] [Google Scholar]
- 43. Buoite Stella A, Gortan Cappellari G, Barazzoni R, Zanetti M. Update on the impact of omega 3 fatty acids on inflammation, insulin resistance and sarcopenia: a review. Int J Mol Sci. 2018;19(1):218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Hartweg J, Perera R, Montori VM, Dinneen SF, Neil AH, Farmer AJ. Omega-3 polyunsaturated fatty acids (PUFA) for type 2 diabetes mellitus. Cochrane Database Syst Rev. 2008;2008(1):CD003205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Jovanovski E, Li D, Thanh Ho HV, Djedovic V, Ruiz Marques AC, Shishtar E, Mejia SB, Sievenpiper JL, de Souza RJ, Duvnjak L et al.. The effect of alpha-linolenic acid on glycemic control in individuals with type 2 diabetes: a systematic review and meta-analysis of randomized controlled clinical trials. Medicine. 2017;96(21):e6531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Imamura F, Sharp SJ, Koulman A, Schulze MB, Kröger J, Griffin JL, Huerta JM, Guevara M, Sluijs I, Agudo A et al.. A combination of plasma phospholipid fatty acids and its association with incidence of type 2 diabetes: the EPIC-InterAct case-cohort study. PLoS Med. 2017;14(10):e1002409. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. American Diabetes Association. 2. Classification and diagnosis of diabetes. Diabetes Care. 2017;40(Supplement 1):S11. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.