The Effect of 6-months Fruit-rich Diet on Liver Steatosis, Liver Enzymes, Insulin Resistance, and Lipid Prole in Patients With Non-alcoholic Fatty Liver Disease: a Randomized Clinical Trial

TG: Triglyceride; LDL-c: low-density lipoprotein; HDL-c,: high-density lipoprotein; HOMA-IR: Homeostatic Model Assessment for Insulin Resistance; QUICKI: Quantitative Insulin-Sensitivity Check Index; MUFA: fatty acids; PUFA: fatty acids; fatty acids; TFA: fatty index; carbohydrate-responsive


Introduction
Non-Alcoholic Fatty Liver Disease (NAFLD) is characterized by the accumulation of fat in the hepatic parenchymal hepatocytes more than 5% of the liver weight, without a history of high amounts of alcohol consumption. NAFLD can lead to a variety of histological problems, from steatosis to in ammation, necrosis, brosis, cirrhosis, and eventually liver cancer [1]. Epidemiologic studies show that NAFLD is the most prevalent liver disease worldwide with an estimated prevalence of 25% in total population currently, with the highest rate in South America and the Middle East [2]. In recent years, the mortality from chronic liver disease increased and in 2019 it was 10th cause of death worldwide [3]. It has been reported that the NAFLD along with brosis increases the mortality rate by 30% [4]. Currently, the tissue biopsy is the gold standard of the NAFLD diagnosis. Since the liver biopsy is an invasive and expensive procedure, it is not suitable for general screening, and ultrasonography, Computed Tomography (CT), and Magnetic Resonance (MRI) may be used to evaluate the amount of liver fat [5]. Ultrasound is a tool for early detection of fatty liver disease, which is less sensitive and speci c for grade 1 steatosis than grade 2 and 3 non-alcoholic fatty liver [6]. The sensitivity and speci city of ultrasound to detect hepatic fat content decreases with increasing of body mass index (BMI) and increases with the high degree of fat penetration in the liver and BMI between 18.5 to 30 kg/m 2 and at least 33% of steatosis is optimal for the diagnosis of NAFLD by ultrasonography [7].
The pathology of NAFLD has not been yet well understood and molecular mechanisms are currently being investigated. Macro-vesicular steatosis is the result of increased intake or hepatic synthesis of fatty acids [1]. Impaired regulation of fatty acids and consequent steatosis is mainly caused by elevated levels of insulin, which can make the liver more vulnerable to oxidative damage [8]. Patients with non-alcoholic fatty liver disease often have other conditions such as hypertriglyceridemia, hypertension, and other factors of insulin resistance syndrome [9]. The nutritional risk factors for fatty liver include high intake of saturated fatty acids (SFA), trans fatty acids (TFA), simple carbohydrates (CHO), sweetened beverages, and fructose [10].
Medication for NAFLD is very limited and their long-term effects have not been well understood. Various dietary approaches have been recommended to improve the disease, including the Mediterranean diet [11], replacing SFA with mono-unsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA) [12], reducing fructose and simple sugars [13], and increasing dietary ber intake [14]. The effect of fruit, which in most cases is a part of a healthy diet, is not clear in the case of NAFLD. The high content of ber, antioxidants, avonoids, carotenoids, vitamins, especially vitamin C, and prebiotic properties of some fruits could have bene cial effects on liver health [15]. On the other hand, the high amount of fructose in fruits has raised concerns about lipogenic properties and its complications, including hepatic steatosis [16]. Observational studies have found con icting results in relationship between fatty liver and fruit consumption [17,18]. To the best of our knowledge, no clinical trial has studied the effect of a fruit-rich diet (FRD) on liver function in patients with NAFLD. This study aimed to evaluate the effect of FRD on liver steatosis, liver enzymes, insulin resistance, and lipid pro le in patients with NAFLD.

Materials And Methods
Study design and participants (Recruitment and eligibility screening) A randomized controlled trial was performed to evaluate the effect of the FRD for 6 months on NAFLD outcomes. The sample size was calculated according to the study of Cantero et al. [19] using the following formula: The α and 1-β were considered equals to 0.05 and 0.90, respectively. The δ were considered as 10 units of change in the ALT levels in intervention group compared to the placebo group. The σ at the baseline and end of the study was considered as 16.5 (the mean of baseline ALT SD of two interventional groups) and 8.0 (the mean of post-intervention ALT SD of two interventional groups), respectively, and σ diff was calculated to be equal to 11.18. This equation estimated 28 cases in each group, and taking into account 30% of the drop-out, the sample size of 40 people for each group was considered. Eighty subjects were recruited between October 2020 to March 2021 from patients with NAFLD referred to the gastrointestinal and liver clinic in Imam Khomeini University Hospital in Urmia., Iran. All participants gave their written informed consent before entering the study. The written informed consent was signed by all study subjects. This study registered in Iranian randomized clinical trial website with IRCT registration no.
IRCT20201010048982N1. This study was approved by Ethics committee at the Urmia University of Medical Sciences (Ethic number: IR.UMSU.REC.1398.535, Date: 02/03/2020).Inclusion criteria were de ned as age older than 18 years, BMI between 18.5 to 29.9 kg/m 2 , and presence of grade 2 or 3 of NAFLD con rmed by gastroenterology and liver specialist. Individuals with viral hepatitis, diabetes mellitus, mental disorders, not-treated hypothyroidism, renal diseases, heart failures, bone diseases, gastrointestinal diseases (such as celiac), α1-antitrypsin de ciency, history of alcohol consumption,; using of nonsteroidal anti-in ammatory drugs (NSAIDs), cholesterol-lowering drugs (such as statins), phenytoin, carbamazepine, and barbiturates (such as phenobarbital); following a certain diet; pregnant, lactating, and menopause women; and smokers (smoking more than 5 cigarettes/week), were not included to the study. The present study was conducted following the deceleration of Helsinki and the Ethics Committee in Urmia University of Medical Sciences approved the protocol of the study.

Randomization and intervention
The owchart of participants' enrollment was presented in Figure 1. The Strati ed Blocked Randomization was performed by an independent statistician by the grade of NAFLD, age, and gender. A blinded person to the aims of the study and patients' baseline status assigned participants to the two groups using sealed envelopes. Patients based on inclusion and exclusion criteria were assigned to the FRD and control groups. Subjects in the FRD group were recommended to consume at least 4 servings of fruits per day and the control group was asked not to consume more than 2 servings of fruit per day. The category of fruits was based on: 1) colored fruits 2) dried fruits 3) and other fruits. To eliminate the effect of pesticides on NAFLD, participants were recommended to unpeel fruits or if they want to eat fruit with the peel, to consume after 20-30 minutes soaking in water. For the same consumption of other food groups, both groups were advised to follow the recommendations of the Food and Agriculture Organization (FAO) for Iranians [20].

Procedures
At the baseline, the data including gender, age, level of education, family size, duration of NAFLD, physical activity, energy intake, type and dose of medication, herbal medicines and dietary supplements, marital status, place of residence, income, other chronic disease histories, and familial history of the disease was obtained using a general questionnaire. Anthropometric measurements and ultrasonography were performed at the start and end of the study. Moreover, 5 mm of venous blood samples were collected at the baseline and after the intervention to conduct biochemical assessments. After the assignment, participants were asked not to change their physical activity, and medications during the study. To ensure the consumption of fruits within the recommended range, as well as assessing of other food groups consumption, three 24-hours food recalls (two non-consecutive days and one day off) were taken from individuals each month (totally 18 food recalls). In addition, the physical activity was assessed using a metabolic equivalents (MET) questionnaire every month [21]. Patients were called every week and the necessary reminders were made. Those who received less than 4 servings of fruits in the intervention group or more than 2 servings of fruits in the control group were excluded from the study.

Secondary outcomes
Secondary outcomes included weight, body mass index (BMI), and waist circumference (WC).

Biochemical assessments
The blood samples were collected at the baseline and end of the study between 7:00 to 9:00 am, after 12 hours of fasting. Blood samples were centrifuged at 4000 rpm for 10 minutes and the isolated serums

Liver steatosis
The liver condition was evaluated under at least 6 hours of fasting by an experienced radiologist. To assess the severity of steatosis the ultrasonography (Siemens ACUSON S2000 Siemens Healthcare, Erlangen, Germany) was performed with previously described methodology [22]. The amount of fat accumulation is associated with an increase in the degree of echogenicity in ultrasound. Accordingly, steatosis was divided into 4 degrees: grade 0 with normal echogenicity, grade 1 or mild in which the echogenicity of the liver increases and the ability to see blood vessels and sound penetration in the liver tissue is normal, grade 2 or moderate in that the vascular wall are seen vaguely and the sound penetration is reduced, and grade 3 or severe, in which the arteries are di cult to see and the sound penetration is very limited. Due to a lower sensitivity and speci city of ultrasonography in diagnosis of grade 1 steatosis, in the present study the subjects with grades 2 and 3 were only recruited. As mentioned earlier, this method is most accurate at BMI between 18.5 and 30, so the participants were recruited in the same range. The size of the liver was also divided into large and normal by the radiologist based on its appearance.

Anthropometric measurements
A digital scale and stadiometer were used to assess the weight and height of the patients with a precision of 100 gr and 0.1 cm, respectively. Measurements were performed with the minimal dress and without shoes. To calculate the BMI, the weight (kg) was divided by the square of height (m 2 ). WC was measured using a exible tape at the midpoint of the lowest rib and the iliac crest hip bone. All measurements were repeated 3 times, and the mean of measurements was used to establish the re-test reliability.

Statistical Methods
Quantitative and qualitative variables were presented as mean ± SD and frequency (%), respectively. To calculate the change of dietary intakes, baseline values were subtracted from mean intakes of each food groups throughout the 6 months. The normality of the quantitative variables was evaluated using the Kolmogorov-Smirnov test. The independent sample t-test was used to compare quantitative variables (or their log-transformed) between groups. Also, the paired sample t-test was used to compare the values before and after the study. Moreover, the repeated measure ANOVA was used to compare the change in dietary intake and physical activity in different time frames (baseline, 1st, 2nd, 3rd, 4th, 5th, and 6th months). To adjust the effect of change in BMI and intake of energy and other food groups, the analysis of covariance (ANCOVA) was used. The Chi-square test was used to compare the frequency of qualitative variables between two groups. Statistical analyses were conducted using SPSS software version 25 (IBM Corp. IBM SPSS Statistics for Windows, Armonk, NY). The P-value < 0.05 was considered statistically signi cant.

General characteristics
The baseline general characteristics of the participants were presented in Table 1. Totally, 32 males (16 FRD and 16 controls) and 40 females (20 FRD and 20 controls) with a mean age of 46.25 ± 9.80 years participated in the study. Two participants lost to follow-up. In addition, 2 of them discontinued participating in the study. Four participants were excluded from the study due to low compliance. No signi cant difference was observed between age, education status, family size, duration of disease, gender, and marital status between the intervention and control groups (P>0.05). However, subjects in the control group had signi cantly higher income compared to the FRD group. Data are presented as mean (SD) for quantitative and frequency (%) for qualitative variables. 1 FRD, fruits rich diet; 2 Calculated using independent sample t-test or chi-square.
Dietary intake and physical activity Table 2 shows the dietary intake and physical activity of participants during the study. During the study, the mean ± SD intake of fruits in the FRD and control group was 6.96±0.61 and 1.65±0.17 serving/day, respectively. At the end of the study, there was a signi cant increase in fruits (P<0.001), bread and cereals (P<0.001), meats (P=0.002), vegetables (P=0.01), dairies (P=0.001), fats and oils (P<0.001), and energy intake (P<0.001) and a signi cant decrease in sugars intake (P=0.001) compared to the baseline in the FRD group. In the control group, a signi cant decrease in fruit intake (P<0.001) and increase in the intake of bread and cereals (P<0.001), meats (P=0.015), vegetables (P<0.001), dairies (P<0.001), sugars (P<0.001), fats and oils (P<0.001), and energy intake (P<0.001) was observed after 6 months compared to the baseline. Between-group analysis in the change of dietary intake during the study showed that the FRD group compared to control group increased daily servings of fruit intake (+3.59 vs. -0.95, respectively, P <0.001) and decreased sugar intakes (-1.93 vs. +0.46, respectively, P<0.001). In contrast, a higher intake of vegetables was observed in the control group, compared to the FRD group (+2.29 vs. +0.75, respectively, P<0.001). The mean change of other food groups and energy were not signi cantly different between two groups. There was no difference in physical activity change between two groups during the study (P=0.792).  3 Calculated using independent sample t-test. 4 The difference between baseline and mean of six values during study 5 Calculated using repeated measure ANOVA to compare intakes during six months. Table 3 compares the mean ± SD of the liver enzymes between two groups at the baseline and end of the study. According to the paired t-test, at the end of the study, there was a signi cant increase in the serum levels of ALT, AST, ALP, GGT compared to the baseline in the FRD group (P<0.001). In contrast, there was a signi cant decrease in all liver enzymes in the control group during the study (P<0.001). After 6 months, the FRD group had higher serum levels of ALT, AST, ALP, and GGT compared to the control group.

Liver enzymes
Adjustments for the effect of change in BMI, energy, bread and cereals, meats, vegetables, dairies, sugars, fats, and oils intake in the ANCOVA models did not change these ndings.  2 Data are presented as mean (SD). 3 Calculated using independent sample t-test. 4 Calculated using ANCOVA, adjusted for the effect change in energy intake. 5 Calculated using ANCOVA, adjusted for the effect of change in bread and cereals, meats, vegetables, dairies, and oils intake. 6 Calculated using ANCOVA, adjusted for the effect of BMI change. 7 Calculated using paired sample t-test. * Logtransformed were entered into the analysis  2 Data are presented as mean (SD). 3 Calculated using independent sample t-test. 4 Calculated using ANCOVA, adjusted for the effect change in energy intake. 5 Calculated using ANCOVA, adjusted for the effect of change in bread and cereals, meats, vegetables, dairies, and oils intake. 6 Calculated using ANCOVA, adjusted for the effect of BMI change. 7 Calculated using paired sample t-test. * Logtransformed were entered into the analysis Following 6 months of intervention, the FRD group had a higher FBS, serum insulin, and HOMA-IR and a lower QUCKI compared to the control group. Nevertheless, the between-groups difference in the FBS was not statistically signi cant after adjusting for the effect of BMI change (P=0.06). Other ndings were not changed after adjustment of the effect of changes in energy and dietary intakes and BMI.

Anthropometric measures
The results showed a signi cant increase in weight, BMI, and WC in the FRD group after 6 months of intervention (P<0.001). The analysis in the control group showed a signi cant decrease in all of these variables (P<0.001). At the baseline, there was no difference between the two groups in weight (P=0.82), BMI (P=0. 35), and WC (P=0.10). However, at the end of the study the FRD group had a higher weight (P<0.001), BMI (P<0.001), and WC (P<0.001).
Liver sonography Figure 2 shows the frequency of subjects with a mild, moderate, or severe grade of steatosis in two groups. Before study (2A) there was no difference between groups in grade of steatosis (P=1.00). After 6 months (2B) the frequency of severe and moderate steatosis was signi cantly higher in the FRD group (P<0.001).
As shown in gure 3A, there was no signi cant difference in the size of the liver before the study. At the end of the study (3B), most of the participants in the FRD group had a large liver, but the size of the liver in the control group was normal (P<0.001).

Discussion
The present study investigated the effect of a FRD compared to the low-fruit diet on liver steatosis, lipid pro le, and glycemic control in NAFLD. Surprisingly, after 6 months of intervention, exacerbation of steatosis, dyslipidemia, and glycemic control were observed in the FRD group. In contrast, patients in the low fruit diet gruop had an improvement in their conditions.
There are limited studies on the relationship between fatty liver and fruit consumption. Randomized clinical trials (RCT) are even more scarce. Cantero, I. et al. [23] showed that calorie restriction along with fruit ber intake (≥8.8 g/day) improved fatty liver index, hepatic steatosis index, and serum levels of GGT, ALT, and AST in obese subjects with NAFLD. In the mentioned study, in addition to intervention with fruit ber intake, the energy intake (-30% of subject's requirement) and the distribution of macronutrients of total caloric value (40% carbohydrate, 30% protein, and 30% lipids in intervention group vs. 55% carbohydrate, 15% protein, and 30% lipids in control group) were altered, each of which could have an independent effect on fatty liver. In addition, the dietary habits were changed, with at least 7 meals/day in the intervention group, compared to the 5 meals/day in the control group. Therefore, the observed changes cannot be attributed only to the intake of fruit ber.
There are other reports of an improvement in hepatic function or lipids metabolism due to intake of speci c fruits or compounds that naturally occur in fruits. Previous studies have shown the hepatoprotective effect of antioxidants including polyphenols, carotenoids, glucosinolates, and bers [24,25]. Among them, resveratrol, which is found in the family of plums and grapes, can increase the oxidation of fatty acids [26]. Quercetin is a avonoid found in a variety of plants, including berries, whose antioxidant activity has been well established [27]. Moreover, anthocyanins found in many fruits have shown some anti-liver damage activity in experimental studies [28]. Carotenoids are other substances that generally accumulate in the liver where they attach to lipoproteins. Dietary carotenoids can purify physiologically active oxygen species, which can prevent liver damage. Also, due to the role of carotenoids in regulating the polarization activity of macrophages, they can prevent the formation and progression of nonalcoholic steatohepatitis (NASH) [29]. Despite this evidence, contradictory results have also been obtained in some studies. Fakhoury-Sayegh et al.
[18] Showed in a case-control study that a fruit-rich dietary pattern (more than 2-3 serving/day of fruits and >20 gr/day of fructose) was directly related to NAFLD. Earlier, Kobayashi et al. [30] reported that people with fatty liver were even more likely to eat fruits and sweets than people with diabetes. In addition, Xia et al. [31] found in a relatively large study (with a sample size of more than 27,000 people) that consuming oranges seven times a week was associated with an increased chance of fatty liver.
In our study, no calorie restriction was considered, and also an increase in energy intake was observed in both groups. Since the weight loss is one of the rst approaches in controlling fatty liver [2], it is probably recommended to the patients in the treatment process. So, it is important to consider the weight reduction of participants in addition to the other procedures or treatments to control NAFLD. In the present study, there was an increase in the BMI of the FRD group and a decrease in the control group. However, further analysis in the present study showed that the ndings are independent from changes in the BMI, energy or other food groups intake. A cross-sectional study showed that controlling for the effect of BMI eliminates the association between fruit intake and NAFLD [32]. Therefore, more clinical trials should investigate the interaction between fruit consumption and weight changes on the consequences of NAFLD during other treatments. Although, Some observational studies found a lower intake of fruits in patients with NAFLD [33], moreover, dietary habits and eating behaviors are other important factors in NAFLD patients [34]. It is also important to consider the intake of other food groups. In the present study, an increase of more than 2 servings/day of vegetable and about 0.5 serving of sugars and a decrease of about 1 serving/day of fruits were observed in the control group. In contrast, in the FRD group the intake of sugars decreased about 2 servings/day and an increase was observed in the consumption of fruits and vegetables 3.6 and 0.75 servings/day, respectively. Although some bene cial effects of reduced fruit diet could attributed to increased intake of vegetables, [35] however, fruit consumption may play a more important role in the accumulation of fats in the liver in FRD group. The reason for these observations could be traced to the lipogenic potential of fructose compared to the glucose. There is an evidence that fructose leads to a greater increase in liver fat content than glucose [36]. It has been suggested that lipogenic effect of fructose is due to downregulation of fatty acids oxidation rather than its production [36]. Lactate and glucose are two metabolites of fructose that in the skeletal muscles spares fatty acids from oxidation. Decreased fatty acid oxidation in skeletal muscle induces the free fatty acids ux to the liver, thereby increasing the hepatic fat deposition [37]. On the other hand, fructose may increase hepatic fat content through de novo lipogenesis from acetate [38]. After absorption, glucose is mainly metabolized by peripheral tissues, while fructose is transported directly to the liver. Due to the lack of feedback control, fructose is metabolized faster and enters the path of lipogenesis compared to the glucose [39]. Also, fructose induces lipogenesis more e ciently than glucose through upregulation of carbohydrate-responsive element-binding protein (ChREBP) and sterol regulatory element-binding protein 1c (SREBP1c) signaling pathways in the hepatocytes [40]. In addition, fructose could intensify bacterial growth in the small intestine, which increases endotoxin levels in the portal vein and can lead to in ammation in the NASH [41]. Studies suggest that fructose restriction decreases steatosis and serum levels of hepatic enzymes [42]. However, the hypothesis of an increased odds of NAFLD as a result of high fructose intake was rejected in a cross-sectional study, and an inverse association between NAFLD and fructose intake was reported [17].
To the best of our knowledge, limited studies investigated the effect of fruit intake on NAFLD outcomes.
Clinical trial design, strati ed randomization, including only grade 2 and 3 of NAFLD and limiting participants to a range of BMI between 18.5 to 29.9 kg/m 2 (Which eliminates the diagnostic bias of ultrasound), and controlling for the effect of change in BMI, energy, and dietary intake with 18 food recalls are the strengths of the present study. However, some limitations should be noted. Given the lack of differences between the two groups in terms of changes in energy intake and physical activity, it should be determined what factor led to weight loss in the control group. Also, perhaps setting a speci c range and limiting the maximum amount of fruit intake could help patients improve their condition. In addition, it may be better to determine participants' fruit daily servings based on the individual energy requirement in future studies.

Conclusion
In the present study, 6 months of intervention with FRD exacerbated steatosis, dyslipidemia, and glycemic control of NAFLD patients. It seems that excessive fruit consumption (about 7 servings per day) makes worse the condition of patients with fatty liver. According to the ndings of the study, fruits intake increases the fat content of the hepatocyte probably through lipogenic effect of fructose. To clarify the issue, more studies specifying a range for fruit intake (with minimum and maximum values) and considering the energy requirements are warranted. The CONSORT ow diagram of the study participants. The grade of steatosis according to sonography in two groups before (A) and after (B) study. The P-value of difference between groups were 1.000 and <0.001 at the baseline and after study, respectively.