Journal Pre-proof
The intermediary effect of inflammation on the associations between dietary patterns and non-alcoholic fatty liver disease Yang Xia , Qing Zhang , Li Liu , Ge Meng , Hongmei Wu , Xue Bao , Yeqing Gu , Shaomei Sun , Xing Wang , Ming Zhou , Qiyu Jia , Kun Song , Qijun Wu , Kaijun Niu , Yuhong Zhao PII: DOI: Reference:
S0899-9007(18)31079-7 https://doi.org/10.1016/j.nut.2019.110562 NUT 110562
To appear in:
Nutrition
Received date: Accepted date:
21 September 2018 24 July 2019
Please cite this article as: Yang Xia , Qing Zhang , Li Liu , Ge Meng , Hongmei Wu , Xue Bao , Yeqing Gu , Shaomei Sun , Xing Wang , Ming Zhou , Qiyu Jia , Kun Song , Qijun Wu , Kaijun Niu , Yuhong Zhao , The intermediary effect of inflammation on the associations between dietary patterns and non-alcoholic fatty liver disease, Nutrition (2019), doi: https://doi.org/10.1016/j.nut.2019.110562
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.
Highlights
Animal foods pattern is positively associated with the prevalence of NAFLD. Inflammatory dietary pattern is positively associated with the prevalence of NAFLD. Inflammation may mediate the associations between dietary patterns and NAFLD independent of BMI.
1
The intermediary effect of inflammation on the associations between dietary patterns and non-alcoholic fatty liver disease Yang Xia 1, Qing Zhang 2, Li Liu 2, Ge Meng 3, Hongmei Wu 3, Xue Bao 3, Yeqing Gu 3, Shaomei Sun 2, Xing Wang 2, Ming Zhou 2, Qiyu Jia 2, Kun Song 2, Qijun Wu 1, Kaijun Niu 2,3†, and Yuhong Zhao 1† †Address for correspondence to: Kaijun Niu, M.D., Ph.D. or Yuhong Zhao, M.D., Ph.D. Names of department(s) and institution(s): 1. Department of Clinical Epidemiology, Shengjing Hospital of China Medical University, Shenyang, China. 2. Health Management Centre, Tianjin Medical University General Hospital, Tianjin, China. 3. Nutritional Epidemiology Institute and School of Public Health, Tianjin Medical University, Tianjin, China. Corresponding author information: 1. Kaijun Niu, M.D., Ph.D. Nutritional Epidemiology Institute and School of Public Health, Tianjin Medical University, 22 Qixiangtai Road, Heping District, Tianjin 300070, China. Tel: +86-22-83336613. E-mail address:
[email protected] or
[email protected] 2. Yuhong Zhao M.D., Ph.D. Department of Clinical Epidemiology, Shengjing Hospital of China Medical University, No. 36, San Hao Street, Shenyang, Liaoning, 110004, China. Tel: +86 024-96615-10012. E-mail:
[email protected] Running title: Dietary patterns and NAFLD
2
1
Abstract
2
Objective: Previous studies demonstrated that nutritional status was associated with
3
non-alcoholic fatty liver disease (NAFLD). Meanwhile, subclinical inflammation is
4
associated with the prevalence of NAFLD. But no study has investigated the
5
intermediary effect of inflammation on the association between dietary pattern and
6
NAFLD. Thus, we designed this case-control study to explore the intermediary effect
7
of inflammation on the associations between dietary patterns and NAFLD.
8
Research Methods & Procedures: 2043 cases and 2043 controls were generated
9
using the propensity score matching method. Dietary intake was assessed using a
10
valid self-administered food frequency questionnaire. Major dietary patterns in the
11
population were derived by factor analysis. Reduced rank regression with leukocyte
12
count as response variable was used to derive an inflammatory pattern. NAFLD was
13
diagnosed by liver ultrasonography and drinking history. The associations between
14
dietary patterns and NAFLD were tested using multiple conditional logistic regression
15
analysis.
16
Results: Three major dietary patterns were derived by factor analysis: sweet pattern,
17
animal foods pattern, and traditional pattern. Compared with the participants in the
18
lowest quartile of animal foods pattern, the odds ratio (OR) (95% confidence interval;
19
95% CI) of NAFLD in the highest quartile was 1.30 (1.09, 1.55). After adjustment of
20
inflammation status, the OR (95% CI) was weaker (OR, 1.23; 95% CI, 1.03-1.48),
21
albeit significant. Compared with the participants in the lowest quartile of
22
inflammatory pattern, the OR (95% CI) of NAFLD in the highest quartile was 1.52
23
(1.28, 1.81). 3
24
Conclusion: Our data suggest that inflammation may mediate the associations
25
between dietary patterns and NAFLD.
26
Key words: dietary patterns; non-alcoholic fatty liver disease; inflammation; OR
4
27
Introduction
28
Non-alcoholic fatty liver disease (NAFLD) is an emerging problem in hepatology
29
clinics [1], it develops without alcohol abuse and is the main cause of liver disease [2].
30
Recently, a meta-analysis included 86 studies and 8,515,431 participants from 22
31
countries found the global prevalence of NAFLD is 25% with highest prevalence in
32
Middle East and South America and lowest in Africa [3]. There is lacking efficacy and
33
safety profiles of pharmacotherapies aimed at treating NAFLD in nowadays and
34
lifestyle management, including sustained weight loss, health dietary, and increased
35
physical activity (PA), is still an important approach in treating NAFLD [4, 5].
36
Regular diets consist of complex combinations of foods and nutrients ingested
37
together that may act independently or may interact with one another [6]. A large
38
amount of these foods and nutrients have pro-inflammatory and anti-inflammatory
39
effect which could modulate inflammation status [7]. A previous review suggested
40
that subclinical inflammation is associated with NAFLD [8]. High-fat diets promote
41
an increased uptake and storage of free fatty acids and triglycerides in hepatocytes,
42
which initiates steatosis and induces inflammation. Activation and signaling of
43
Toll-like receptor 4 by free fatty acids induces inflammation evident in NAFLD [9].
44
Thus, inflammation may partly mediate the associations between dietary patterns and
45
NAFLD.
46
Previous studies have demonstrated that dietary patterns are associated with the
47
prevalence of NAFLD [2, 10-16]. High intake of Western dietary patterns and fruit
48
pattern [14] are associated with higher prevalence of NAFLD [10, 11] while high 5
49
intake of healthy dietary patterns are associated with lower prevalence of NAFLD [13,
50
15]. However, to the best of our knowledge, no study has explored the association
51
between dietary pattern and NAFLD by incorporating the pro-inflammation effect of
52
dietary pattern into analyses independent of BMI. We, therefore, conducted this
53
case-control study 1) to derive major dietary patterns in Chinese using factor analysis;
54
2) to identify an inflammatory pattern by reduced rank regression; 3) to explore the
55
intermediary effect of inflammation on the associations between dietary patterns and
56
the prevalence of NAFLD.
6
57
Materials and Methods
58
Participants
59
This case-control study was based on the Tianjin Chronic Low-grade Systemic
60
Inflammation and Health (TCLSIHealth) Cohort Study, which is a large prospective
61
dynamic cohort study focusing on the associations between chronic low-grade
62
systemic inflammation and the health status of a population living in Tianjin, China [2,
63
17, 18]. Participants were recruited while having their annual health examinations at
64
the Tianjin Medical University General Hospital-Health Management Center and
65
community management centres in Tianjin. This dynamic cohort study was launched
66
in 2007. Moreover, a detailed lifestyle questionnaire covering family income, marital
67
status, employment status, educational level, physical activity (PA), sleep habits,
68
dietary habits, overall computer/mobile device usage time, television time, history of
69
prior infections and use of medicines as well as physical performance tests were
70
administered to about 70–80 % randomly selected subjects from this population since
71
May 2013. The present study used data of the lifestyle questionnaires of participants
72
collected from May 2013 to December 2016 and data of annual health examinations
73
of participants collected from January 2007 to December 2016.
74
23063 participants without acute inflammatory disease completed a comprehensive
75
health examination (including evaluation of anthropometric parameters and
76
biochemical blood examination etc.) and a study questionnaire reporting personal
77
information, dietary intake, lifestyles and health condition. We excluded participants
78
who changed their lifestyles, including lifestyles of diet, drinking, smoking, activity, 7
79
and sleeping, in last 5 years (n=5883), or those with a history of cardiovascular
80
disease (n=1052) or cancer (n=197). We also excluded participants who have a history
81
of NAFLD (n=2463). The final study population comprised 13468 (3008 cases and
82
10460 controls) participants for propensity score matching. The protocol of this study
83
was approved by the Institutional Review Board of the Tianjin Medical University
84
and participants gave written informed consent before participation in the study.
85
Propensity score matching
86
Propensity score were calculated using a logistic regression model and the following
87
covariates: sex, age, body mass index (BMI), PA, energy intake, education level,
88
household income, smoking status, drinking status, employment status, metabolic
89
syndrome status, and family history of cardiovascular disease, hypertension, and
90
diabetes. Using these propensity scores, cases were individually matched by controls
91
using the nearest matching method within a caliper distance, which selects for
92
matching a control subject whose propensity score is closest to that of the case subject
93
(nearest neighbor matching approach) with the further restriction that the absolute
94
difference in the propensity scores of matched subjects must be below some
95
pre-specified threshould (the caliper distance) [19]. Thus, participants for whom the
96
propensity score could not be matched because of a greater caliper distance were
97
excluded from further analysis. As suggested by Austin [19] , a caliper of width equal
98
to 0.2 of the standard deviation of the logit of the propensity score was used, as this
99
value minimized the mean squared error of the estimated treatment effect in several
100
scenarios. To better match cases and controls, we used the 1:1 ratio matching method. 8
101
If a case subject could not be matched to any control subject, then the case subject
102
was discarded. Finally, 2043 cases and 2043 controls were generated using this
103
propensity score matching method.
104
Identification of dietary pattern
105
Dietary intake was assessed using a modified version of the food frequency
106
questionnaire (FFQ) that included 100 food items (the initial version of the FFQ
107
included 81 food items [2]) with specified serving sizes. The FFQ included 7
108
frequency categories ranging from ‘almost never eat’ to ‘twice or more per day’ for
109
foods and 8 frequency categories ranging from ‘almost never drink’ to ‘four or more
110
times per day’ for beverages. The mean daily intake of nutrients was calculated by
111
using an ad hoc computer program developed to analyze the questionnaire. The
112
Chinese food composition tables [20] were used as the nutrient database. The
113
reproducibility and validity of the questionnaire were assessed in a random sample of
114
150 participants and living in Tianjin by comparing the data from the questionnaire
115
with the data from 2 dietary questionnaires collected approximately 3 months apart
116
and 4-day weighed dietary records (WDRs). Spearman rank correlation coefficient for
117
energy intake between 2 food frequency questionnaires administered 3 months apart
118
was 0.68. Correlation coefficients for food items (fruits, vegetables, fish, meat, and
119
beverages) between 2 food frequency questionnaires administered 3 months apart
120
ranged from 0.62 to 0.79 Spearman’s rank correlation coefficient for energy intake by
121
the WDRs and the FFQ was 0.49. Correlation coefficients for nutrients (vitamin C,
122
vitamin E, polyunsaturated fats, saturated fats, carbohydrate and calcium) by the 9
123
WDRs and the FFQ ranged from 0.35 to 0.54. Similar food items were further
124
collapsed into 25 food groups based on the characteristics of food items.
125
We applied factor analysis in order to generate major dietary patterns and factor
126
loadings on all 25 food groups. After evaluation of eigenvalues (> 1.0) and the scree
127
test, these factors were determined. Food groups with a factor loading > |0.30| were
128
the main contributors to dietary pattern and representative of the character of each
129
factor. Factors were named descriptively according to the food groups showing high
130
loading (absolute value) with respect to each dietary pattern as follows: sweet pattern,
131
animal foods pattern, and traditional pattern.
132
The inflammatory pattern was assessed using reduced rank regression which is a
133
statistical method determining linear functions of predictors (the dietary pattern) by
134
maximizing the explained variation in responses [21]. Leukocyte count is a simple,
135
widely available, inexpensive, and well-standardized biomarker of inflammation. A
136
previous study demonstrated that leukocyte count was independently associated with
137
the presence of NAFLD regardless of classical cardiovascular risk factors and other
138
components of metabolic syndrome [22]. Another study conducted in Chinese found
139
that leukocyte count was a significant factor associated with incident NAFLD in Han
140
Chinese [23]. Thus, we selected leukocyte count as responses variable in order that
141
the final derived dietary pattern could explain the inflammation status and be
142
associated with the prevalence of NAFLD.
143
Liver ultrasonography and definitions of NAFLD
144
Liver ultrasonography was conducted by trained sonographers using a TOSHIBA 10
145
SSA-660A ultrasound machine (Toshiba, Tokyo, Japan), with a 2-5-MHz curved array
146
probe. According to the revised definition and treatment guidelines for NAFLD by the
147
Chinese Hepatology Association in February 2006 [24], we defined ‘heavy drinking’
148
as >140 g alcohol intake per week in men and >70 g per week in women. Total
149
alcohol intake in the past week was assessed by the FFQ. Participants were diagnosed
150
as having NAFLD using abdominal ultrasonography (brightness of liver and a
151
diffusely echogenic change in the liver parenchyma) and no history of heavy drinking.
152
Assessment of other variables
153
The sociodemographic variables, which include sex, age, education, employment,
154
smoking status, drinking status, and household income, were also assessed by
155
questionnaire. The educational level was assessed by asking the question ‘what is the
156
highest degree you earned?’ and was divided into 2 categories:
157
≥College graduate. Employment status was classified as either Senior Officials and
158
Managers or Professionals. Information on the smoking (‘never,’ ‘former,’ and
159
‘current smoking’) and drinking (‘never,’ ‘former,’ ‘current drinking everyday’, and
160
‘current drinking sometime’) status of the participants was obtained from a
161
questionnaire survey. PA in the most recent week was assessed using the short form of
162
the International Physical Activity Questionnaire (IPAQ) [25]. The questionnaire
163
asked whether subjects had performed any activities from the following categories
164
during the previous week: walking; moderate activity (household activity or child
165
care); vigorous activity (running, swimming, or other sports activities). Metabolic
166
equivalent (MET) hours per week were calculated using corresponding MET 11
167
coefficients (3.3, 4.0 and 8.0, respectively) according to the following formula: MET
168
coefficient of activity × duration (hours) × frequency (days). Total PA levels were
169
assessed by combining separate scores for different activities.
170
BMI was calculated as weight in kilograms divided by the square of height in meters
171
(kg/m2). Fasting blood samples were taken by venipuncture of the cubital vein and
172
immediately mixed with EDTA. Leukocyte and its differential counts were carried out
173
using the automated hematology analyzer XE-2100 (Sysmex, Kobe, Japan) and
174
expressed as ×1,000 cells/mm3. The test for blanks was ≤0.2 × 109 cells/L; the intra-
175
and interassay coefficients of variation (CV) were ≤2.0%, and the cross-contamination
176
rate was ≤0.5%. Waist circumference was measured at the umbilical level with
177
participants standing and breathing normally. Blood pressure (BP) was measured
178
twice from the upper left arm using a TM-2655P automatic device (A&D CO., Tokyo,
179
Japan) after 5 minutes of rest in a seated position. The mean of these 2 measurements
180
was taken as the BP value. Blood samples for the analysis of fasting blood glucose
181
(FBG) and lipids were collected in siliconized vacuum plastic tubes. FBS was
182
measured by the glucose oxidase method, triglycerides (TG) were measured by
183
enzymatic methods, and high-density lipoprotein cholesterol (HDL) was measured by
184
the chemical precipitation method using reagents from Roche Diagnostics on an
185
automatic biochemistry analyzer (Roche Cobas 8000 modular analyzer, Mannheim,
186
Germany). Metabolic syndrome was defined in accordance with the criteria of the
187
American Heart Association scientific statement of 2009 [26].
188
Statistical analysis 12
189
In order to characteristics of participants according to NAFLD status, descriptive data
190
have been presented as the least square mean (with 95% confidence interval, CI) or as
191
percentages and examined using analysis of variance and chi-square test for
192
categorical variables. Quartiles were categorized across the scores of inflammatory
193
dietary pattern based on the distribution of the scores for all the participants and used
194
for further analyses. Association between quartile categories of inflammatory dietary
195
pattern scores and NAFLD status was examined using conditional logistic regression
196
analysis. Odds ratios (OR) and 95% CI were calculated. A linear trend cross
197
increasing quartiles was tested using the median value of each quartile as a continuous
198
variable based on linear regression. Model 1 was used to calculate the crude OR, and
199
model 2 was adjusted for scores of other dietary patterns. Model 3 further adjusted for
200
leukocyte count. All analyses were performed using the Statistical Analysis System
201
9.3 edition for Windows (SAS Institute Inc., Cary, NC, USA) and STATA (version
202
12.1; Stata Corp LP, College Station, TX, USA). All P-values were two-tailed and
203
difference was defined to be significant when P<0.05.
204
Results
205
Characteristics of participants
206
Characteristics of participants according to NAFLD status before and after propensity
207
score matching are shown in Table 1 and Table 2, respectively.
208
Among 13468 participants who were available to be analyzed before propensity score
209
matching, 22.3% were classified as newly diagnosed NAFLD. As shown in Table 1,
210
participants with NAFLD trended to be men (P<0.0001), older (P<0.0001), current 13
211
smoker (P<0.0001), ex-smoker (P<0.0001), and current drinker (P<0.0001), who also
212
had metabolic syndrome (P<0.0001), higher daily energy intake (P=0.02), lower
213
education level (P<0.0001), less likely be employed as managers (P<0.0001), family
214
history of diabetes (P<0.0001). After propensity score matching, 2400 cases and 2400
215
controls were generated and showed no significant baseline differences in any
216
character (Table 2).
217
Dietary patterns
218
Factor analysis revealed three major dietary patterns (Table 3), which accounted for
219
39.3 % of the variance in total food intake. According to the contribution to the total
220
variance, the three dietary patterns were: factor 1 was defined as the sweet pattern and
221
characterized by high intake of fruits, cakes and ice cream; factor 2, the animal foods
222
pattern, was typified by intake of animal organs, animal blood and meat products;
223
factor 3, identified as the traditional pattern and included intake of whole grain,
224
refined grain, vegetables, eggs, and legume.
225
We identified the inflammatory pattern by reduced rank regression which explained
226
2.6 % and 4.6 % of the total variation of the response variable (leukocyte count) and
227
dependent variables (food groups), respectively. The inflammatory pattern (Table 3)
228
was characterized by high intake of sugar-containing beverages, tea and tea beverages,
229
ice cream and candy, meat, and animal organs.
230
Dietary patterns and NAFLD
231
The associations between major dietary patterns and NAFLD were shown in Table 4.
232
The associations between the sweets pattern and NAFLD demonstrated an “U” type 14
233
curve. The ORs across quartiles were 1 (reference), 0.88 (0.47, 1.05), 0.80 (0.67,
234
0.95), and 1.00 (0.84, 1.20) before adjustment of leukocyte count. After adjustment of
235
leukocyte count the ORs across quartiles were 1 (reference), 0.89 (0.74, 1.07), 0.81
236
(0.68, 0.98), and 1.01 (0.84, 1.22). Participants with the higher intake of animal foods
237
pattern were associated with higher prevalence of NAFLD before (P for trend < 0.01)
238
and after (P for trend = 0.03) adjustment of leukocyte count. Compared with the
239
participants in the lowest quartile, the OR (95% CI) of NAFLD in the highest quartile
240
of animal foods pattern was 1.30 (1.09, 1.55) before adjustment of leukocyte count.
241
After adjustment of leukocyte count, the associations between consumption of animal
242
foods pattern and NAFLD turned to be weaker, but still significant. Compared with
243
the participants in the lowest quartile, the OR (95% CI) of NAFLD in the highest
244
quartile of animal foods pattern was 1.23 (1.03, 1.48).
245
The associations between inflammatory pattern and NAFLD were presented in Table
246
5. Higher adherence to inflammatory pattern was associated with higher prevalence of
247
NAFLD (P for trend<0.0001). Compared with the participants in the lowest quartile,
248
the OR (95% CI) of NAFLD in the highest quartile of inflammatory pattern was 1.52
249
(1.28, 1.81).
250
Discussion
251
In this case-control study, we derived three major dietary patterns and an
252
inflammatory pattern in the population and explored the associations between these
253
dietary patterns and the prevalence of NAFLD. The associations between these
254
dietary pattern scores and NAFLD were independent of confounding factors as the 15
255
propensity scores matching approach reduced the differences between case group and
256
control group. The major dietary patterns and the inflammatory dietary pattern
257
accounted for 39.3 % and 4.6 % of the variance in total food intake, respectively. The
258
inflammatory pattern explained 2.6 % of the total variation of the response variable
259
which is similar to former studies [27, 28].
260
Previous studies have examined the associations between nutrition status and NAFLD.
261
Consumption of fructose [29], soft drinks [30], and red meat [30] was demonstrated
262
associated with higher prevalence of NAFLD. In contrast, intake of n-3 [31], n-6 fatty
263
acid [32], and coffee [33] appeared to have a favorable effect on NAFLD.
264
Considering of the regular diets consist of complex combinations of foods and
265
nutrients ingested together that may act independently or may interact with one
266
another, examination of dietary patterns, which assess the effects of overall diet,
267
would more closely parallel the real world [6]. A few studies have indicated that
268
dietary patterns were associated with NAFLD [2, 11-16].
269
In the present study, we demonstrated an “U” type curve between the sweets pattern
270
and NAFLD. Moderate intake of the sweets pattern was negatively associated with the
271
prevalence of NAFLD. However, the association turned to be non-significant in the
272
fourth quartile. Our former study found that participants in the highest quartile of
273
high-carbohydrate/sweet pattern, which was characterized by high intakes of fruits,
274
cakes and sugared beverages, had a 2.19-fold greater risk of developing NAFLD than
275
those in the lowest quartile after adjustment of confounding variables [2]. However,
276
we did not exclude participants who has a history of NAFLD or had changed their 16
277
lifestyles in last 5 years in the former study [2]. Thus, the reverse causation may be
278
existed (i.e. participants with NAFLD changing their diet to increase fruits intake for
279
health reasons). A plausible reason for this “U” type curve may be that the sweets
280
pattern in the present study was not only high in fruits intake but also with high intake
281
of ice cream and cakes. Even though that consumption of fruit may be associated with
282
low prevalence of NAFLD because of the anti-inflammation effect of fruit
283
polyphenols [34]. Too much intake of the sweets pattern also results in a large amount
284
of sugar intake which has been associated with the pathophysiology of NAFLD [35].
285
The results were also partly in line with previous studies which indicated that dietary
286
patterns rich in animal foods were associated with higher prevalence of NAFLD. A
287
cohort study using exploratory factor analysis in Australian found that the Western
288
dietary pattern, which was typically by high intake of soft drinks, fat, refined grains,
289
red meat, and take-away foods, at 14 years in a general population sample was
290
associated with an increased risk of NAFLD at 17 years [11]. Another study
291
conducted in China found that ‘animal food’ dietary pattern was associated with an
292
increased risk of NAFLD [12]. In the present study, we also found that the
293
associations between animal foods pattern and NAFLD may mediated by
294
inflammation status by different adjustment models.
295
Previous studies, however, used the exploratory factor analysis method to derive
296
dietary patterns in local population. Even though this method reflects the real and
297
complete nutrition background of local population, but it is not able to include any
298
evidence about predictors of targeted disease risk [36]. It means if a dietary pattern 17
299
obtained by this method and turns out to be a risk factor of NAFLD, but a plausible
300
explanation is hard to establish. Although we know which food groups substantially
301
contribute to the factor by looking for high factor loadings, it remains unclear why
302
these food groups are important in the incidence of disease [21]. However, the RRR
303
method could test dietary hypotheses based on etiology by set up response variables.
304
Different from previous approaches (diet-quality scores, principal component analysis,
305
and exploratory factor analysis) have been used to derive dietary patterns, the RRR is
306
a “hybrid” method of the a priori and a posteriori method and is a modern statistical
307
method to derive dietary patterns that can be used to test specific hypothesis on
308
pathways from diet to development of a disease [37]. A few studies have used this
309
method to explore the associations between dietary patterns and diseases [38, 39]. To
310
the best of our knowledge, no study has examined the association between
311
inflammatory dietary pattern and NAFLD.
312
Although the etiology of NAFLD is multifactorial and remains largely enigmatic, it is
313
well accepted that inflammation is central component of NAFLD pathogenesis [40].
314
Inflammation and hepatocyte injury and death are the hallmarks of nonalcoholic
315
steatohepatitis (the progressive form of NAFLD) [41]. Duarte N. suggested
316
subclinical inflammation plays a prominent role in the development of NAFLD and
317
avoiding subclinical inflammation is important in treating with NAFLD [8]. In this
318
study, we chose the total leukocyte count as the response variable since total
319
leukocyte count is an important biomarker of inflammation [42]. Furthermore, we
320
found that the associations between dietary patterns and NAFLD may partly mediated 18
321
by inflammation status by different adjustment models. Thus, we derived an
322
inflammatory pattern by reduced rank regression and explored the associations
323
between it and NAFLD.
324
In this study, the inflammatory pattern was positively associated with NAFLD. The
325
important character of the inflammatory pattern was high intake of sugar-containing
326
beverages (such as cola), ice cream and candy. This finding was in line with previous
327
studies [2, 11]. The mechanism underlying this association may be through the large
328
amounts of sugar [11]. High sugar intake has been found associated with development
329
of skeletal muscle insulin resistance and inflammation in mice [43] and a
330
fat-cholesterol-sugar diet model induced the full spectrum of liver pathophysiologic
331
changes, which characterizes the progression of NAFLD in humans, in mice [44].
332
Meanwhile, we found that the inflammatory pattern had high factor loadings on tea in
333
this study. That is perhaps because of that tea is an important character in Chinese
334
diets. Previous studies suggested that consumption of tea extract could be beneficial to
335
inflammation in rats [45] and mice [46]. But the beneficial effect of tea on
336
inflammation may be covered up by consumption of sugared beverages, ice cream,
337
and animal foods in the present study. The inflammatory pattern in this study also
338
included high intake of animal foods (meat and animal organs) which has been
339
reported associated with NAFLD [12]. In recent years, several studies have reported a
340
positive association between consumption of animal foods and inflammation [47, 48].
341
Animal foods is high in saturated fatty acids and may cause an activation of the
342
immune system, most likely by an excessive production of pro-inflammatory 19
343
cytokines associated with a reduced production of anti-inflammatory cytokines [49].
344
In conclusion, the inflammatory pattern derived in the present study may be
345
associated with higher prevalence of NAFLD by shifting the balance of immune
346
system.
347
The present study has some limitations. First, due to the nature of the self-reporting
348
questionnaire, recall bias exists and the food intake maybe not exact. Second, we
349
excluded participants for reason of health conditions and the final sample may not be
350
representative of the population. Third, we cannot rule out the possibility that
351
unmeasured factors might contribute to the association observed. Finally, we only
352
used the leukocyte count to explain the inflammation status.
353
Conclusion
354
Despite the limitations, our data suggest that animal foods pattern and inflammatory
355
pattern are positively associated with the prevalence of NAFLD. And inflammation
356
may partly mediate the associations between dietary patterns and NAFLD.
357
Acknowledgments
358
We gratefully thank all of the participants in the study and Tianjin Medical University
359
General Hospital-Health Management Center for the opportunity to perform the study.
360
Declaration of Source of Funding
361
This study was supported by grants from the National Natural Science Foundation of
362
China (grant numbers 81673166, 81372118, 81372467 and 81302422).
363
Conflicts of Interest
364
There is no potential conflict of interest that relates to the manuscript. 20
21
References 1.
Karim, M. F., M. Al-Mahtab, S. Rahman, and C. R. Debnath. "Non-Alcoholic Fatty Liver Disease (Nafld)--a Review." Mymensingh Med J 24, no. 4 (2015): 873-80.
2.
Jia, Q., Y. Xia, Q. Zhang, H. Wu, H. Du, L. Liu, C. Wang, H. Shi, X. Guo, X. Liu, C. Li, S. Sun, X. Wang, H. Zhao, K. Song, G. Huang, Y. Wu, N. Cui, and K. Niu. "Dietary Patterns Are Associated with Prevalence of Fatty Liver Disease in Adults." Eur J Clin Nutr 69, no. 8 (2015): 914-21.
3.
Younossi, Z. M., A. B. Koenig, D. Abdelatif, Y. Fazel, L. Henry, and M. Wymer. "Global Epidemiology of Non-Alcoholic Fatty Liver Disease-Meta-Analytic Assessment of Prevalence, Incidence and Outcomes." Hepatology (2015).
4.
Malhotra, N., and M. D. Beaton. "Management of Non-Alcoholic Fatty Liver Disease in 2015." World J Hepatol 7, no. 30 (2015): 2962-7.
5.
Bellentani, S., R. Dalle Grave, A. Suppini, G. Marchesini, and Network Fatty Liver Italian. "Behavior Therapy for Nonalcoholic Fatty Liver Disease: The Need for a Multidisciplinary Approach." Hepatology 47, no. 2 (2008): 746-54.
6.
Hu, F. B. "Dietary Pattern Analysis: A New Direction in Nutritional Epidemiology." Curr Opin Lipidol 13, no. 1 (2002): 3-9.
7.
Minihane, A. M., S. Vinoy, W. R. Russell, A. Baka, H. M. Roche, K. M. Tuohy, J. L. Teeling, E. E. Blaak, M. Fenech, D. Vauzour, H. J. McArdle, B. H. Kremer, L. Sterkman, K. Vafeiadou, M. M. Benedetti, C. M. Williams, and P. C. Calder. "Low-Grade Inflammation, Diet Composition and Health: Current Research Evidence and Its Translation." Br J Nutr 114, no. 7 (2015): 999-1012.
8.
Duarte, N., I. C. Coelho, R. S. Patarrao, J. I. Almeida, C. Penha-Goncalves, and M. P. Macedo. "How Inflammation Impinges on Nafld: A Role for Kupffer Cells." Biomed Res Int 2015 (2015): 984578.
9.
Patton, A., T. Church, C. Wilson, J. Thuma, D. J. Goetz, D. E. Berryman, E. O. List, F. Schwartz, and K. D. McCall. "Phenylmethimazole Abrogates Diet-Induced Inflammation, Glucose Intolerance and Nafld." J Endocrinol 237, no. 3 (2018): 337-51.
10.
Liu, X., Y. Peng, S. Chen, and Q. Sun. "An Observational Study on the Association between Major Dietary Patterns and Non-Alcoholic Fatty Liver Disease in Chinese Adolescents." Medicine (Baltimore) 97, no. 17 (2018): e0576.
11.
Oddy, W. H., C. E. Herbison, P. Jacoby, G. L. Ambrosini, T. A. O'Sullivan, O. T. Ayonrinde, J. K. Olynyk, L. J. Black, L. J. Beilin, T. A. Mori, B. P. Hands, and L. A. Adams. "The Western Dietary Pattern Is Prospectively Associated with Nonalcoholic Fatty Liver Disease in Adolescence." Am J Gastroenterol 108, no. 5 (2013): 778-85.
12.
Yang, C. Q., L. Shu, S. Wang, J. J. Wang, Y. Zhou, Y. J. Xuan, and S. F. Wang. "Dietary Patterns Modulate the Risk of Non-Alcoholic Fatty Liver Disease in Chinese Adults." Nutrients 7, no. 6 (2015): 4778-91.
13.
Adriano, L. S., H. A. Sampaio, S. P. Arruda, C. L. Portela, M. L. P. de Melo, A. A. Carioca, and N. T. Soares. "Healthy Dietary Pattern Is Inversely Associated with Non-Alcoholic Fatty Liver Disease in Elderly." Br J Nutr 115, no. 12 (2016): 2189-95.
14.
Fakhoury-Sayegh, N., H. Younes, Gnha Heraoui, and R. Sayegh. "Nutritional Profile and Dietary Patterns of Lebanese Non-Alcoholic Fatty Liver Disease Patients: A Case-Control Study." Nutrients 9, no. 11 (2017). 22
15.
Shim, P., D. Choi, and Y. Park. "Association of Blood Fatty Acid Composition and Dietary Pattern with the Risk of Non-Alcoholic Fatty Liver Disease in Patients Who Underwent Cholecystectomy." Ann Nutr Metab 70, no. 4 (2017): 303-11.
16.
Zelber-Sagi, S., F. Salomone, and L. Mlynarsky. "The Mediterranean Dietary Pattern as the Diet of Choice for Non-Alcoholic Fatty Liver Disease: Evidence and Plausible Mechanisms." Liver Int 37, no. 7 (2017): 936-49.
17.
Song, K., H. Du, Q. Zhang, C. Wang, Y. Guo, H. Wu, L. Liu, Q. Jia, X. Wang, H. Shi, S. Sun, and K. Niu. "Serum Immunoglobulin M Concentration Is Positively Related to Metabolic Syndrome in an Adult Population: Tianjin Chronic Low-Grade Systemic Inflammation and Health (Tclsih) Cohort Study." PLoS One 9, no. 2 (2014): e88701.
18.
Sun, S., H. Wu, Q. Zhang, C. Wang, Y. Guo, H. Du, L. Liu, Q. Jia, X. Wang, K. Song, and K. Niu. "Subnormal Peripheral Blood Leukocyte Counts Are Related to the Lowest Prevalence and Incidence of Metabolic Syndrome: Tianjin Chronic Low-Grade Systemic Inflammation and Health Cohort Study." Mediators Inflamm 2014 (2014): 412386.
19.
Austin, P. C. "An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies." Multivariate Behav Res 46, no. 3 (2011): 399-424.
20.
Yuexin Y., Yaguang W. China Food Composition. 2rd Ed. Beijing, China: Peking University Medical Press, 2009.
21.
Hoffmann, K., M. B. Schulze, A. Schienkiewitz, U. Nothlings, and H. Boeing. "Application of a New Statistical Method to Derive Dietary Patterns in Nutritional Epidemiology." Am J Epidemiol 159, no. 10 (2004): 935-44.
22.
Lee, Y. J., H. R. Lee, J. Y. Shim, B. S. Moon, J. H. Lee, and J. K. Kim. "Relationship between White Blood Cell Count and Nonalcoholic Fatty Liver Disease." Dig Liver Dis 42, no. 12 (2010): 888-94.
23.
Wang, S., C. Zhang, G. Zhang, Z. Yuan, Y. Liu, L. Ding, X. Sun, H. Jia, and F. Xue. "Association between White Blood Cell Count and Non-Alcoholic Fatty Liver Disease in Urban Han Chinese: A Prospective Cohort Study." BMJ Open 6, no. 6 (2016): e010342.
24.
Zeng, M. D., J. G. Fan, L. G. Lu, Y. M. Li, C. W. Chen, B. Y. Wang, Y. M. Mao, and Disease Chinese National Consensus Workshop on Nonalcoholic Fatty Liver. "Guidelines for the Diagnosis and Treatment of Nonalcoholic Fatty Liver Diseases." J Dig Dis 9, no. 2 (2008): 108-12.
25.
Craig, C. L., A. L. Marshall, M. Sjostrom, A. E. Bauman, M. L. Booth, B. E. Ainsworth, M. Pratt, U. Ekelund, A. Yngve, J. F. Sallis, and P. Oja. "International Physical Activity Questionnaire: 12-Country Reliability and Validity." Med Sci Sports Exerc 35, no. 8 (2003): 1381-95.
26.
Alberti, K. G., R. H. Eckel, S. M. Grundy, P. Z. Zimmet, J. I. Cleeman, K. A. Donato, J. C. Fruchart, W. P. James, C. M. Loria, S. C. Smith, Jr., Epidemiology International Diabetes Federation Task Force on, Prevention, Lung Hational Heart, Institute Blood, Association American Heart, Federation World Heart, Society International Atherosclerosis, and Obesity International Association for the Study of. "Harmonizing the Metabolic Syndrome: A Joint Interim Statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity." Circulation 120, no. 16 (2009): 1640-5.
27.
Vermeulen, E., I. A. Brouwer, K. Stronks, S. Bandinelli, L. Ferrucci, M. Visser, and M. Nicolaou. 23
"Inflammatory Dietary Patterns and Depressive Symptoms in Italian Older Adults." Brain Behav Immun 67 (2018): 290-98. 28.
Ozawa, M., M. Shipley, M. Kivimaki, A. Singh-Manoux, and E. J. Brunner. "Dietary Pattern, Inflammation and Cognitive Decline: The Whitehall Ii Prospective Cohort Study." Clin Nutr 36, no. 2 (2017): 506-12.
29.
Vos, M. B., and J. E. Lavine. "Dietary Fructose in Nonalcoholic Fatty Liver Disease." Hepatology 57, no. 6 (2013): 2525-31.
30.
Zelber-Sagi, S., D. Nitzan-Kaluski, R. Goldsmith, M. Webb, L. Blendis, Z. Halpern, and R. Oren. "Long Term Nutritional Intake and the Risk for Non-Alcoholic Fatty Liver Disease (Nafld): A Population Based Study." J Hepatol 47, no. 5 (2007): 711-7.
31.
Di Minno, M. N., A. Russolillo, R. Lupoli, P. Ambrosino, A. Di Minno, and G. Tarantino. "Omega-3 Fatty Acids for the Treatment of Non-Alcoholic Fatty Liver Disease." World J Gastroenterol 18, no. 41 (2012): 5839-47.
32.
Petit, J. M., B. Guiu, L. Duvillard, V. Jooste, M. C. Brindisi, A. Athias, B. Bouillet, M. Habchi, V. Cottet, P. Gambert, P. Hillon, J. P. Cercueil, and B. Verges. "Increased Erythrocytes N-3 and N-6 Polyunsaturated Fatty Acids Is Significantly Associated with a Lower Prevalence of Steatosis in Patients with Type 2 Diabetes." Clin Nutr 31, no. 4 (2012): 520-5.
33.
Molloy, J. W., C. J. Calcagno, C. D. Williams, F. J. Jones, D. M. Torres, and S. A. Harrison. "Association of Coffee and Caffeine Consumption with Fatty Liver Disease, Nonalcoholic Steatohepatitis, and Degree of Hepatic Fibrosis." Hepatology 55, no. 2 (2012): 429-36.
34.
Gonzalez-Gallego, J., M. V. Garcia-Mediavilla, S. Sanchez-Campos, and M. J. Tunon. "Fruit Polyphenols, Immunity and Inflammation." Br J Nutr 104 Suppl 3 (2010): S15-27.
35.
Lim, J. S., M. Mietus-Snyder, A. Valente, J. M. Schwarz, and R. H. Lustig. "The Role of Fructose in the Pathogenesis of Nafld and the Metabolic Syndrome." Nat Rev Gastroenterol Hepatol 7, no. 5 (2010): 251-64.
36.
Ozawa, M., M. Shipley, M. Kivimaki, A. Singh-Manoux, and E. J. Brunner. "Dietary Pattern, Inflammation and Cognitive Decline: The Whitehall Ii Prospective Cohort Study." Clin Nutr (2016).
37.
Weikert, C., and M. B. Schulze. "Evaluating Dietary Patterns: The Role of Reduced Rank Regression." Curr Opin Clin Nutr Metab Care (2016).
38.
Yang, T. C., L. S. Aucott, G. G. Duthie, and H. M. Macdonald. "An Application of Partial Least Squares for Identifying Dietary Patterns in Bone Health." Arch Osteoporos 12, no. 1 (2017): 63.
39.
Jessri, M., R. D. Wolfinger, W. Y. Lou, and M. R. L'Abbe. "Identification of Dietary Patterns Associated with Obesity in a Nationally Representative Survey of Canadian Adults: Application of a Priori, Hybrid, and Simplified Dietary Pattern Techniques." Am J Clin Nutr 105, no. 3 (2017): 669-84.
40.
Peverill, W., L. W. Powell, and R. Skoien. "Evolving Concepts in the Pathogenesis of Nash: Beyond Steatosis and Inflammation." Int J Mol Sci 15, no. 5 (2014): 8591-638.
41.
Arrese, M., D. Cabrera, A. M. Kalergis, and A. E. Feldstein. "Innate Immunity and Inflammation in Nafld/Nash." Dig Dis Sci (2016).
42.
Pecht, T., A. Gutman-Tirosh, N. Bashan, and A. Rudich. "Peripheral Blood Leucocyte Subclasses as Potential Biomarkers of Adipose Tissue Inflammation and Obesity Subphenotypes in Humans." Obes Rev 15, no. 4 (2014): 322-37. 24
43.
Benetti, E., R. Mastrocola, M. Rogazzo, F. Chiazza, M. Aragno, R. Fantozzi, M. Collino, and M. A. Minetto. "High Sugar Intake and Development of Skeletal Muscle Insulin Resistance and Inflammation in Mice: A Protective Role for Ppar- Delta Agonism." Mediators Inflamm 2013 (2013): 509502.
44.
Ganz, M., T. N. Bukong, T. Csak, B. Saha, J. K. Park, A. Ambade, K. Kodys, and G. Szabo. "Progression of Non-Alcoholic Steatosis to Steatohepatitis and Fibrosis Parallels Cumulative Accumulation of Danger Signals That Promote Inflammation and Liver Tumors in a High Fat-Cholesterol-Sugar Diet Model in Mice." J Transl Med 13 (2015): 193.
45.
Qin, Y. J., K. O. Chu, Y. W. Yip, W. Y. Li, Y. P. Yang, K. P. Chan, J. L. Ren, S. O. Chan, and C. P. Pang. "Green Tea Extract Treatment Alleviates Ocular Inflammation in a Rat Model of Endotoxin-Induced Uveitis." PLoS One 9, no. 8 (2014): e103995.
46.
Okuda, M. H., J. C. Zemdegs, A. A. de Santana, A. B. Santamarina, M. F. Moreno, A. C. Hachul, B. dos Santos, C. M. do Nascimento, E. B. Ribeiro, and L. M. Oyama. "Green Tea Extract Improves High Fat Diet-Induced Hypothalamic Inflammation, without Affecting the Serotoninergic System." J Nutr Biochem 25, no. 10 (2014): 1084-9.
47.
Ley, S. H., Q. Sun, W. C. Willett, A. H. Eliassen, K. Wu, A. Pan, F. Grodstein, and F. B. Hu. "Associations between Red Meat Intake and Biomarkers of Inflammation and Glucose Metabolism in Women." Am J Clin Nutr 99, no. 2 (2014): 352-60.
48.
Montonen, J., H. Boeing, A. Fritsche, E. Schleicher, H. G. Joost, M. B. Schulze, A. Steffen, and T. Pischon. "Consumption of Red Meat and Whole-Grain Bread in Relation to Biomarkers of Obesity, Inflammation, Glucose Metabolism and Oxidative Stress." Eur J Nutr 52, no. 1 (2013): 337-45.
49.
Giugliano, D., A. Ceriello, and K. Esposito. "The Effects of Diet on Inflammation: Emphasis on the Metabolic Syndrome." J Am Coll Cardiol 48, no. 4 (2006): 677-85.
25
Table 1. Participant characteristics by NAFLD status before matching a. Characteristics Sex (male %)
NAFLD status No (n = 10460) 41.5 c
Yes (n = 3008)
P value b
73.8
<0.0001
44.5 (44.1, 44.9)
<0.0001
Age (y)
40.8 (40.6, 41.0)
BMI
22.9 (22.8, 23.0)
27.8 (27.7, 27.9)
<0.0001
11.6
58.1
<0.0001
9.3 (9.1, 9.6)
9.6 (9.1, 10.1)
0.33
8198.8 (8154.9, 8243.2)
8310.2 (8227.3, 8393.9)
0.02
Education (≥College graduate, %)
64.3
51.3
<0.0001
Household income (≥10,000 Yuan, %)
35.1
34.8
0.79
Smoker
16.0
33.1
<0.0001
Ex-smoker
3.7
7.3
<0.0001
Non-smoker
80.4
59.6
<0.0001
Metabolic syndromes (%) Physical activity (Mets × hours/week) Energy intake (kJ/d)
Smoking status (%)
Drinker (%) Everyday
4.3
9.2
<0.0001
Sometime
53.6
60.9
<0.0001
Ex-drinker
8.5
8.5
0.95
Non-drinker
33.6
21.5
<0.0001
Managers
43.8
37.7
<0.0001
Professionals
17.0
17.2
0.83
Other
39.2
45.1
<0.0001
CVD
29.8
27.0
<0.01
Hypertension
48.5
50.8
0.03
Diabetes
22.6
26.2
<0.0001
Employment status (%)
Family history of diseases (%)
a
NAFLD, non-alcoholic fatty liver disease; CVD, cardiovascular disease. BMI, body mass index.
b
Analysis of variance or chi-square test.
c
Least square mean (95% confidence interval) (all such values).
26
Table 2. Participant characteristics by NAFLD status after matching a. Characteristics Sex (male %)
NAFLD status No (n = 2043) 68.6 c
Yes (n = 2043)
P value b
69.6
0.48
44.6 (44.1, 45.1)
0.22
Age (y)
45.1 (44.5, 45.6)
BMI
26.5 (26.4, 26.6)
26.6 (26.5, 26.7)
0.19
42.4
44.0
0.31
10.2 (9.6, 10.8)
10.1 (9.5, 10.7)
0.83
8247.0 (8149.0, 8346.6)
8252.4 (8154.0, 8352.0)
0.94
Education (≥College graduate, %)
51.6
53.5
0.22
Household income (≥10,000 Yuan, %)
33.8
34.7
0.54
Smoker
29.0
29.1
0.94
Ex-smoker
7.6
7.6
1.00
Non-smoker
63.4
63.3
0.95
Metabolic syndromes (%) Physical activity (Mets × hours/week) Energy intake (kJ/d)
Smoking status (%)
Drinker (%) Everyday
8.4
9.1
0.46
Sometime
58.8
59.9
0.49
Ex-drinker
7.9
8.1
0.76
Non-drinker
24.9
22.9
0.14
Managers
39.7
40.2
0.79
Professionals
17.6
17.0
0.61
Other
42.7
42.9
0.90
CVD
28.3
27.8
0.68
Hypertension
50.4
49.9
0.73
Diabetes
23.4
23.5
0.91
Employment status (%)
Family history of diseases (%)
a
NAFLD, non-alcoholic fatty liver disease; CVD, cardiovascular disease. BMI, body mass index.
b
Analysis of variance or chi-square test.
c
Least square mean (95% confidence interval) (all such values).
27
Table 3. The factor loadings of food groups of dietary patterns. Food groups EFA a RRR b Sweet pattern Animal foods pattern Traditional pattern Inflammatory pattern Refined grain 0.01 c 0.18 0.05 0.54 Whole grain 0.06 -0.16 -0.39 0.55 Dairy 0.11 0.22 0.38 -0.26 Meat -0.04 0.28 0.38 0.34 Meat products 0.05 0.20 0.16 0.57 Animal blood 0.12 0.18 0.10 0.55 Animal organs 0.05 0.30 0.59 0.23 Fish 0.15 0.51 0.35 0.01 Egg 0.03 -0.04 -0.08 0.55 Preserved egg 0.12 0.16 0.15 0.58 Fruit 0.09 0.10 0.05 0.80 Vegetable 0.60 0.15 -0.06 0.53 Tubers 0.61 0.02 0.31 -0.02 Legume and legume products 0.59 -0.03 -0.05 0.44 Pickled foods 0.15 0.07 0.12 0.75 Western-style cake, cookie 0.17 -0.13 0.03 0.76 Chinese cake 0.17 -0.14 0.05 0.75 Ginger and garlic 0.49 -0.08 0.38 -0.08 Ice cream and candy 0.21 -0.17 0.72 0.28 Nuts 0.68 0.03 0.08 -0.12 a EFA, exploratory factor analysis. b RRR, reduced rank regression. c Factor loadings represent the relative contribution of each food group to the dietary pattern. The five food groups with highest factor loadings in each dietary pattern are shown in bold characters.
28
Table 3. The factor loadings of food groups of dietary patterns (continued). Food groups EFA a RRR b Sweet pattern Animal foods pattern Plant foods pattern Inflammatory pattern Tea and tea beverages 0.03 c 0.18 0.20 0.35 Coffee 0.07 0.41 -0.18 0.12 Sugar-containing beverages 0.05 -0.17 0.51 0.50 Fruits or vegetables juice 0.11 0.43 -0.06 0.11 Alcohol and alcoholic beverages -0.02 0.32 0.05 0.10 Explained variation in food groups (%) 19.0 10.6 9.6 4.6 Explained variation in BMI and leukocyte count 2.6 (%) a EFA, exploratory factor analysis. b RRR, reduced rank regression. c Factor loadings represent the relative contribution of each food group to the dietary pattern. The five food groups with highest factor loadings in each dietary pattern are shown in bold characters.
29
Table 4. Association between major dietary patterns and NAFLD *. Dietary patterns Quartiles of factor scores (range, n=4086) Sweet pattern Level 1 (-2.23, -0.43) Level 2 (-0.43, -0.19) No. of NAFLD 537 504 Model 1 c Ref 0.88 (0.74, 1.05) b Model 2 d Ref 0.88 (0.47, 1.05) Model 3 e Ref 0.89 (0.74, 1.07) Animal foods pattern Level 1 (-2.55, -0.53) Level 2 (-0.53, -0.21) No. of NAFLD 480 511 Model 1 c Ref 1.13 (0.95, 1.35) Model 2 d Ref 1.12 (0.94, 1.34) Model 3 e Ref 1.09 (0.91, 1.31) Traditional pattern Level 1 (-5.16, -0.63) Level 2 (-0.63, -0.12) No. of NAFLD 518 524 Model 1 c Ref 1.02 (0.86, 1.22) Model 2 d Ref 1.04 (0.88, 1.24) Model 3 e Ref 1.01 (0.85, 1.21) * NAFLD, non-alcoholic fatty liver disease; BMI, body mass index. a Multiple conditional logistic regression analysis. b Odds ratios (95% confidence interval) (all such values). c Crude model. d Adjusted for the scores of other two dietary patterns. e Further adjusted for leukocyte count based on Model 2.
30
Level 3 (-0.19, 0.17) 471 0.77 (0.65, 0.92) 0.80 (0.67, 0.95) 0.81 (0.68, 0.98) Level 3 (-0.21, 0.25) 502 1.09 (0.92, 1.30) 1.08 (0.90, 1.29) 1.04 (0.86, 1.25) Level 3 (-0.12, 0.46) 503 0.95 (0.80, 1.12) 0.97 (0.82, 1.16) 0.97 (0.81, 1.16)
P for trend a Level 4 (0.17, 10.77) 531 0.97 (0.82, 1.16) 1.00 (0.84, 1.20) 1.01 (0.84, 1.22) Level 4 (0.25, 13.75) 550 1.31 (1.10, 1.56) 1.30 (1.09, 1.55) 1.23 (1.03, 1.48) Level 4 (0.46, 6.59) 498 0.92 (0.78, 1.10) 0.93 (0.78, 1.11) 0.92 (0.77, 1.10)
0.73 0.99 0.87
<0.01 <0.01 0.03
0.25 0.27 0.38
Table 5. Association between inflammatory pattern and NAFLD *. Dietary patterns Quartiles of inflammatory pattern scores (range, n=4086) Inflammatory pattern Level 1 (-4.23, -0.50) No. of NAFLD 461 Model 1 c Ref * NAFLD, non-alcoholic fatty liver disease. a Conditional logistic regression analysis. b Odds ratios (95% confidence interval) (all such values). c Crude model.
Level 2 (-0.50, -0.04) 493 1.13 (0.95, 1.35) b
31
Level 3 (-0.04, 0.44) 520 1.26 (1.06, 1.49)
Level 4 (0.44, 7.70) 569 1.52 (1.28, 1.81)
P for trend a
<0.0001