In total, 2470 participants were included in our study among which 1109 were diagnosed with NAFLD. We grouped all the participants according to the NAFLD state, namely NAFLD group and non-NAFLD group. Weighted characteristics of the study subjects were summarized in Table 1. There were statistically significant differences in demographic features (gender, age, race), dietary lipid intake (total fat (TTFAT), total monounsaturated fatty acids (TMFA), total saturated fatty acids (TSFA)), BMI, weight grade, waist circumference, hypertension, diabetes, current smoking state, HbA1c, liver function indexes (alanine transaminase, alkaline phosphatase, gamma-glutamyl transferase), uric acid, serum lipid (high-density lipoprotein cholesterol, triglyceride), FPG, current smoking state, CAP and LSM values between the two groups (P < 0.05), with exception of total cholesterol, PFA, low-density lipoprotein cholesterol and aspartate aminotransferase (P > 0.05). NAFLD was more prevalent in middle-aged and elderly men (P < 0.05). In the NAFLD group, the proportion of Mexican American individuals was significantly higher than that in the non-NAFLD group (9.6% vs. 4.9%, P < 0.0001). Compared with non-NAFLD, participants with NAFLD have worse metabolic indicators, including serum lipid, FPG, HbA1c, BMI, uric acid and waist circumference, more metabolic related complications such as hypertension and diabetes, higher alanine transaminase level, more severe steatosis and fibrosis judged by CAP and LSM values. With regard to dietary lipid, we found that the TTFAT, TMFA and TSFA intake in NAFLD group were significantly higher than those in non-NAFLD group (P < 0.05).
Table 1
Weighted characteristics of participants grouped by NAFLD state
|
Non-NAFLD
|
NAFLD
|
P value
|
Gender
|
|
|
< 0.0001
|
Male
|
49.4 (1.5)
|
56.7 (2.1)
|
0.0011
|
Female
|
50.6 (1.5)
|
43.3 (2.1)
|
0.0011
|
Age
|
47.4 (0.7)
|
54.2 (0.6)
|
< 0.0001
|
18-36Y
|
35.2 (2.2)
|
16.3 (1.2)
|
< 0.0001
|
37-54Y
|
25.3 (2.1)
|
27.9 (2.7)
|
0.4662
|
55-66Y
|
21.8 (1.6)
|
30.5 (2)
|
0.0027
|
67-80Y
|
17.6 (1)
|
25.3 (1.6)
|
0.0002
|
Race
|
|
|
0.0003
|
Mexican American
|
4.9 (0.9)
|
9.6 (1.9)
|
< 0.0001
|
Other Hispanic
|
6.4 (1.1)
|
5.3 (0.9)
|
0.3199
|
Non-Hispanic White
|
64.0 (2.9)
|
65.0 (3.3)
|
0.7482
|
Non-Hispanic Black
|
13.7 (1.8)
|
9.5 (1.6)
|
0.0018
|
Non-Hispanic Asian
|
6.7 (1)
|
6.3 (1.2)
|
0.7111
|
Other Race - Including Multi-Racial
|
4.2 (1.2)
|
4.3 (1)
|
0.9326
|
Total fat (gm) and ratios
|
82.4 (1.3)
|
88.7 (1.7)
|
0.0058
|
Total saturated fatty acids
|
26.2 (0.5)
|
29.2 (0.7)
|
0.0029
|
Total monounsaturated fatty acids
|
28.5 (0.5)
|
30.6 (0.6)
|
0.0017
|
Total polyunsaturated fatty acids
|
19.7 (0.4)
|
20.3 (0.4)
|
0.3911
|
RPM
|
0.7 (0)
|
0.7 (0)
|
0.2805
|
RPS
|
0.8 (0)
|
0.8 (0)
|
0.0842
|
RUS
|
1.9 (0)
|
1.8 (0)
|
0.0473
|
RMS
|
1.1 (0)
|
1.1 (0)
|
0.0578
|
BMI
|
26.6(0.3)
|
33.3(0.5)
|
< 0.0001
|
Weight grade
|
|
|
< 0.0001
|
Normal weight
|
42.0 (2.4)
|
6.8 (1.4)
|
< 0.0001
|
Overweight
|
35.3 (1.8)
|
26.7 (3)
|
0.0257
|
Obesity
|
22.7 (2.2)
|
66.5 (3.8)
|
< 0.0001
|
Waist circumference
|
92.9 (0.8)
|
110.8 (1.1)
|
< 0.0001
|
Hypertension
|
29.3 (2.1)
|
56.5 (3.2)
|
< 0.0001
|
Diabetes
|
7.0 (0.8)
|
29.6 (2)
|
< 0.0001
|
Smoking
|
30.0 (3)
|
20.2 (2.6)
|
0.0227
|
HbAlc (%)
|
5.5 (0)
|
6.1 (0)
|
< 0.0001
|
ALT(U/L)
|
19.2 (0.5)
|
25.5 (0.6)
|
< 0.0001
|
AST (U/L)
|
20.7 (0.5)
|
22.0 (0.4)
|
0.1115
|
ALP(IU/L)
|
74.3 (1.3)
|
79.9 (1.3)
|
0.0073
|
GGT(IU/L)
|
21.8 (0.8)
|
31.3 (1)
|
< 0.0001
|
Uric acid (mg/dL)
|
5.3 (0.1)
|
5.8 (0)
|
< 0.0001
|
HDL-c(mg/dL)
|
55.2 (0.8)
|
48.2 (0.8)
|
< 0.0001
|
LDL-c(mg/dL) *
|
112.9 (2.3)
|
110.7 (2.8)
|
0.4342
|
Triglyceride (mg/dL)
|
90.3 (1.9)
|
147.2 (8.9)
|
< 0.0001
|
Total Cholesterol (mg/dL)
|
186.4 (2.9)
|
186.2 (3.3)
|
0.9527
|
Fasting glucose (mg/dL)
|
103.9 (09)
|
124.6 (3.6)
|
0.0001
|
LSM (kPa)
|
4.9 (0.1)
|
6.5 (0.2)
|
< 0.0001
|
CAP (dB/m)
|
220.3 (1.7)
|
322.4 (2.2)
|
< 0.0001
|
Fibrosis level
|
|
|
< 0.0001
|
F0
|
96.1 (0.9)
|
84.0 (1.8)
|
< 0.0001
|
F2
|
2.8 (0.9)
|
10.1 (1.4)
|
0.0056
|
F3
|
0.7 (0.3)
|
3.7 (1)
|
0.001
|
F4
|
0.4 (0.2)
|
2.3 (0.7)
|
0.0135
|
Smoking: current smoking state; * LDL-Cholesterol, Friedewald. Wtsaf4 year was used for weighted analysis of fasting items while Wtmec2 year was used for other items. Survey-weighted linear regression was used to calculate the P values of continuous variables. Survey-weighted Chi-square test was used to calculate P values of categorical variables. |
We further explored the dietary fatty acid intake and related ratios for their relationships with NAFLD via weighted multivariable logistic regression (Table 2). In unadjusted Model 1, a significant positive association was observed between TTFAT, TMFA, TSFA and NAFLD (All the odds ratio (OR) and 95% confidence interval (CI) > 1 and P < 0.05) which indicates the intake of above lipids may increase the risk for NAFLD. For subgroup analysis stratified by sex, OR values were all insignificant except for that of TSFA in males. Then, we use univariate regression analysis to select confounding factors corresponding to each independent variable. In Model 2, which was adjusted for confounders, the significant positive association still survived between TSFA, TMSA and NAFLD. With regard to the ratio of each two kinds of fatty acids, Model 1 showed that only the ratio of unsaturated fatty acids to saturated fatty acids (RUS) was significantly associated with NAFLD. Since statistically significant interaction was noted between gender and the ratio of polyunsaturated fatty acids to saturated fatty acids (RPS) for NAFLD (P for interaction < 0.05), we further performed logistic regression analysis stratified by gender on the ratios and NAFLD. In subgroup analysis, the situation was remarkably changed, significant associations between RPS, RUS, the ratio of monounsaturated fatty acids to saturated fatty acids (RMS), the ratio of polyunsaturated fatty acids to monounsaturated fatty acids (RPM) and NAFLD were observed in male rather than female (P < 0.05). In detail, the unadjusted OR and adjusted OR for NAFLD across the ratios were all less than 1 which indicates male individuals who have higher above ratios may be less likely to develop NAFLD. This relationship still survived in the Model 2. Since OR values of each dietary lipid component were very close to 1, we converted the items (TTFAT, TSFA, TPFA, TMFA) into categorical data according to their quartiles and recalculated the unadjusted OR and 95%CI (Fig. 2). The forest plot visually demonstrated whether the CI included 1. As shown in Fig. 2, participants intaking the highest quartile of TSFA (Q4) presented a noticeably higher risk for NAFLD when compared with those intaking the lowest quartile (Q1) (OR [95% CI] > 1, P < 0.01). Moreover, we conducted weighted logistic regression analysis of dietary lipid components on NAFLD stratified by age stage (divided by quartiles) and weight grade (Table 3). TTFAT, TSFA and TMFA were positively correlated with NAFLD in participants 37–54 years old (all the OR [95%CI] > 1 and P < 0.05). Of note, this age stage is just a turning point. In participants younger than 37 years old, non-NAFLD is significantly more prevalent than NAFLD while in participants over 54 years old, NAFLD is dominant (Table 1). Subgroup analysis stratified by weight grade showed that OR values of TTFAT, TSFA and TMFA for NAFLD were significant only in obese individuals (OR [95% CI] > 1, P < 0.05). We further compared dietary lipid components and ratios between different sexes and discovered that intake of each dietary lipid component was significantly higher in male than female(P < 0.001) while an insignificant difference was observed in ratios between sexes (Table 4).
Table 2
Weighted logistic regression of association between fatty acids, ratios and NAFLD stratified by gender
|
|
OR (95% CI) for Model1
|
OR (95% CI) for Model2
|
TTFAT
|
Male
|
1.0027 (0.9986, 1.0068)
|
1.0024 (0.9972, 1.0077)
|
|
Female
|
1.0041 (0.9996, 1.0085)
|
1.0041 (0.999, 1.0091)
|
|
Total
|
1.004 (1.0015, 1.0065)**
|
1.003 (0.99996, 1.006)
|
TSFA
|
Male
|
1.014 (1.001, 1.028)*
|
1.0139 (0.9997, 1.0283)
|
|
Female
|
1.012 (0.998, 1.027)
|
1.0124 (0.996, 1.0292)
|
|
Total
|
1.0156 (1.0065, 1.0248)**
|
1.0133 (1.0003, 1.0265)*
|
TMFA
|
Male
|
1.006 (0.996, 1.016)
|
1.0056 (0.9948, 1.0164)
|
|
Female
|
1.009 (0.998, 1.02)
|
1.0092 (0.9969, 1.0217)
|
|
Total
|
1.0095 (1.0046, 1.0144)**
|
1.0068 (1.0011, 1.0126)*
|
TPFA
|
Male
|
0.9953 (0.9818, 1.0089)
|
0.994 (0.977, 1.011)
|
|
Female
|
1.013 (0.994, 1.033)
|
1.014 (0.994, 1.034)
|
|
Total
|
1.005 (0.994, 1.016)
|
1.001 (0.988, 1.014)
|
RUS
|
Male
|
0.63 (0.5, 0.8)**
|
0.61 (0.47, 0.79)**
|
|
Female
|
0.92 (0.67, 1.27)
|
0.85 (0.54, 1.34)
|
|
Total
|
0.77 (0.61, 0.97)*
|
0.72 (0.52, 1)*
|
RPS
|
Male
|
0.46 (0.3, 0.7)**
|
0.44 (0.29, 0.68)**
|
|
Female
|
0.95 (0.5, 1.8)
|
0.91 (0.47, 1.76)
|
|
Total
|
0.65 (0.42, 1.02)
|
0.63 (0.4, 1)
|
RMS
|
Male
|
0.47 (0.27, 0.81)*
|
0.48 (0.25, 0.95)*
|
|
Female
|
0.79 (0.48, 1.31)
|
0.8 (0.48, 1.33)
|
|
Total
|
0.62 (0.39, 1.004)
|
0.62 (0.37, 1.07)
|
RPM
|
Male
|
0.43 (0.2, 0.91)*
|
0.43 (0.22, 0.84)*
|
|
Female
|
1.12 (0.36, 3.45)
|
0.91 (0.27, 3.07)
|
|
Total
|
0.64 (0.3, 1.38)
|
0.62 (0.25, 1.58)
|
*P < 0.05, **P < 0.01, ***P < 0.001. |
Model 1: Unadjusted covariates |
Model 2: Race was adjusted for TTFAT, TSFA, TMFA, TPFA and RMS; Race, age, diabetes and hypertension were adjusted for RUS; hypertension, race and age were adjusted for RPS; Race and diabetes were adjusted for RPM. |
Table 3
Subgroup analysis of association between fatty acids and NAFLD stratified by age and weight grade
|
|
TTFAT
|
TSFA
|
TMFA
|
TPFA
|
OR (95% CI)
|
18-36Y
|
1.001 (0.996, 1.007)
|
1.008 (0.996, 1.021)
|
1.004 (0.99, 1.018)
|
0.994 (0.968, 1.02)
|
37-54Y
|
1.011 (1.004, 1.018)**
|
1.043 (1.021, 1.065)**
|
1.024 (1.003, 1.045)*
|
1.021 (0.998, 1.044)
|
55-66Y
|
1.002 (0.997, 1.006)
|
1.011 (0.995, 1.027)
|
1.004 (0.993, 1.016)
|
0.995 (0.977, 1.013)
|
67-80Y
|
1.004 (0.998, 1.01)
|
1.006 (0.99, 1.022)
|
1.009 (0.99, 1.028)
|
1.02 (0.996, 1.048)
|
OR (95% CI)
|
N-weight
|
1.006 (0.995, 1.016)
|
1.015 (0.985, 1.046)
|
1.018 (0.989, 1.047)
|
1.014 (0.981, 1.049)
|
Overweight
|
1 (0.995, 1.004)
|
1.004 (0.992, 1.017)
|
0.997 (0.983, 1.011)
|
0.993 (0.97, 1.016)
|
Obesity
|
1.007 (1.002, 1.011)*
|
1.025 (1.012, 1.038)**
|
1.014 (1.004, 1.025)*
|
1.008 (0.99, 1.027)
|
N-weight: normal weight. Weighted univariate logistic regression was used for analysis. *P < 0.05, **P < 0.01, ***P < 0.001 |
Table 4
Comparison of dietary fatty acids and ratios between gender
|
Male
|
Female
|
TTFAT***
|
96.7 (1.5)
|
72.2 (1.6)
|
TSFA***
|
31.2 (0.6)
|
23.4 (0.6)
|
TMFA***
|
33.6 (0.6)
|
24.7 (0.6)
|
TPFA***
|
22.5 (0.4)
|
17.2 (0.4)
|
RPS
|
0.8 (0)
|
0.8 (0)
|
RPM*
|
0.7 (0)
|
0.7 (0)
|
RMS
|
1.1 (0)
|
1.1 (0)
|
RUS
|
1.9 (0)
|
1.9 (0)
|
Survey-weighted linear regression was used to calculate P values of continuous variables. Chi-square test was used to calculate the P values of categorical variables. *P < 0.05, **P < 0.01, ***P < 0.001. |
Smooth curve fitting exhibited a linear correlation between fatty acids and the risk of NAFLD in males which is non-linear in females (Fig. 3). Except for the intake of TPFA in males (Fig. 3|D), dietary fatty acids were positively correlated with the prevalence of NAFLD in males (Fig. 3|A-C) which is consistent with logistic regression analysis. The slope of TMFA was smaller than those of TTFAT and TSFA. Similar to the above trend, TPFA still presented linear negative relationship with steatosis severity of NAFLD in males judged by CAP values while other fatty acids all showed positive association (Fig. 3|E-H). With regard to ratios, all the ratios exhibited negative associations with NAFLD for males while positive correlations survived in females. (Fig. 3|I-L). As depicted in Fig. 3, all the curves were linear and the slope was greater in males than females. As for the smooth curve fittings between ratios and CAP value, the correlations remained inverse between males and females (Fig. 3|M-P). A wave-shaped declining curve was observed between RUS and CAP value in males while other relationship curves were close to linear whose slope was larger in males than females.