The mediatory role of inflammatory markers on the relationship between the NOVA classification system and obesity phenotypes among obese and overweight adult women: a cross-sectional study

Background Diet and inflammation both play important roles in the occurrence of obesity. We aimed to investigate the role of inflammation in the development of both metabolically healthy obese (MHO) and metabolically unhealthy obese (MUHO) individuals. Methods This cross-sectional study included 221 overweight and obese women aged 18–56 years. The study assessed the metabolic health phenotypes of the participants using the Karelis criterion score. Additionally, dietary intakes were evaluated using a 147-item semi-quantitative questionnaire and the NOVA classification system (comprising 37 food groups and beverages). The study also collected and analyzed the blood parameters, as well as biochemical and anthropometric indices, for all participants. Results Among the women included in the study, 22.9% had MHO phenotypes but 77.1% had MUHO phenotypes. A significant association between the third quartile of the NOVA classification system and the increased likelihood of having the MUHO phenotype was observed (OR = 1.40, 95% CI = 1.09–4.92, p = 0.04). Regarding the potential role of inflammatory markers, high-sensitivity C-reactive protein (hs-CRP) (p = 0.84), transforming growth factor-β (TGF-β) (p = 0.50), monocyte chemoattractant protein-1 (MCP-1) (p = 0.49), plasminogen activator inhibitor-1 (PAI-1) (p = 0.97), and homeostatic model assessment for insulin resistance (HOMA-IR) (p = 0.92) were found to be mediators. Conclusion We observed a significant positive association between ultra-processed food (UPF) consumption and the MUHO phenotype in overweight and obese women. This association appeared to be mediated by some inflammatory markers, such as hs-CRP, TGF-β, MCP-1, PAI-1, and HOMA-IR. Additional studies are needed to validate these findings.


Introduction
Obesity is associated with mild inflammation, and it is a state or condition that actively participates in regulating both physiological and pathological inflammatory processes by releasing pro-inflammatory cytokines, such as C-reactive protein (CRP) (1,2).This inflammatory response can significantly increase the risk of morbidity and mortality (1,(3)(4)(5).The etiology of obesity is influenced by factors such as genetics, gender, lifestyle, and dietary intake (6)(7)(8)(9).It was projected that, by 2030, the number of overweight adults will reach 2.16 billion and the number of obese adults will be approximately 1.12 billion (10).A study conducted on a population of 5,607 individuals reported an overall obesity prevalence of 57.2%, which was higher among women compared to men (11).However, not all obese individuals experience metabolic dysfunction, particularly cardiometabolic issues (12).Recent definitions classify obese individuals into metabolically healthy obese (MHO) and metabolically unhealthy obese (MUHO) groups based on their metabolic status.The MHO group (with a prevalence of 10-34%) is more prevalent in women and comprises individuals with specific metabolic profiles, such as higher insulin sensitivity, a lower incidence of hypertension, and optimal inflammation levels (13)(14)(15)(16).Various indicators and definitions have been used to determine the metabolic health status, and in this study, the Karelis criterion was employed to categorize individuals into either the MHO or MUHO group (14,15,17,18).
Dietary intake plays a crucial role in the development of overweight and obesity, and numerous studies have highlighted the significant impact of dietary patterns on obesity phenotypes (8,(19)(20)(21).Over the past few decades, there has been a notable increase in the consumption of processed foods due to the increasing availability of industrially processed products (22).Consequently, studies have reported associations between the consumption of processed foods, such as sugar (23), sugar-sweetened beverages (SSB) (24), fast food (25), and obesity, as well as unhealthy metabolic phenotypes (26)(27)(28).The energy-dense high-saturated fatty acid (SFA) diet enhances insulin resistance (IR), obesity, and metabolic syndrome (29,30).Literature shows that ultra-processed food (UPF) intake affects inflammatory markers, such as high-sensitivity C-reactive protein (hs-CRP), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) (31, 32).Several studies have been conducted on the relationship between inflammatory markers and obesity and how they activate the innate immune system in the adipose tissue, especially the visceral fat (33)(34)(35).However, regarding the relationship between UPF consumption and inflammatory markers, such as CRP, there are conflicting findings.In a cross-sectional study examining this relationship in both sexes, only a significant relationship was observed in women but not in men (36).
The NOVA classification system is a novel approach to assess the dietary patterns while grouping foods into four categories: unprocessed or minimally processed foods, processed culinary ingredients, processed foods, and UPFs (37).There was a significant association between UPF intake and an increased risk of chronic disorders such as diabetes and cardiovascular disease (CVD) (38)(39)(40)(41).A prospective study in the Brazilian population showed more weight and waist circumference (WC) gains in the highest quartile of UPF consumption (42).Limited studies have reported an association between NOVA and increased inflammatory markers (31), which may be due to the consumption of SFAs, potatoes, hydrogenated fats, snacks, sugar, soft drinks, sweets, and desserts (43)(44)(45)(46).However, there is currently a lack of studies exploring the relationship between UPF intake and inflammatory markers using the NOVA classification system among women.We aimed to investigate the mediating role of inflammatory markers in the relationship between the NOVA classification system and obesity phenotypes among women.

Participants
In this cross-sectional study, 221 overweight and obese women in the age range of 18-56 years participated.Participants with body mass index (BMI) between 25 kg/m 2 and 29.9 kg/m 2 were classified as overweight and those with BMI ≥30 kg/m 2 as obese.Participants were selected from those who attended the health clinics of Tehran University of Medical Sciences (TUMS), Tehran Province, Tehran, Iran.Participants were included in the study after signing a written informed consent.The study protocol entered the practical phase after being approved by the ethics committee of TUMS (Ethics ID: IR.TUMS.MEDICINE.REC.1399.165).Participants having any history of disease, i.e., diabetes mellitus, CVD, hypertension, acute or chronic inflammatory disease, impaired renal and liver function, thyroid disease, autoimmune disease, malignancy, those who were pregnant and lactating, smokers, those with a regular use of drugs, including birth control pills, and those who consumed alcohol or had drug abuse, were excluded from the study.In addition, participants who had an energy intake of less than 800 kcal/day and more than 4,200 kcal/day, as well as those who had 5% weight fluctuations over the last year, were excluded from the study (47).

Assessment of dietary intakes
The dietary intakes were collected using a 147-item semiquantitative food frequency questionnaire (FFQ) by a trained nutritionist.This validated and reliable questionnaire of Iranian adults (48,49) includes a list of common foods consumed by the Iranian community, along with their standard portion sizes.Frequency of intake of each food item was asked and recorded as per daily, weekly, monthly, or annual consumption.After calculating the total energy intake (kcal/day), the data were analyzed using Nutritionist IV software (version 7.0; N-Squared Computing, Salem, OR, United States) (50).The standard portion size and items reported based on the household scale were collected and converted to grams using the home scale guide.Thus, the equivalent of grams consumed for each item and for each subject was calculated (51).

NOVA classification
The NOVA classification was used to assess the amount of UPF intake (52).NOVA classifies a total of 37 food items and beverages obtained from the FFQs as UPFs.These UPFs are further divided into seven distinct food groups as follows (Supplementary Figure S1):

Anthropometry assessment
The body weight of the subjects was measured using a digital scale with a precision of 100 g.They were instructed to wear minimal clothing and no shoes during the measurements.Participant's height was recorded in a standing position with the normal position of the shoulders, without shoes.BMI was calculated by dividing the weight (kg) by the square of the height (m 2 ).We used a multi-frequency bioelectrical impedance analyzer [InBody 770 scanner (InBody Co., Ltd., Seoul, South Korea)] to assess the body composition, including the visceral fat level (VFL), body fat percentage (BF%), fat-free mass index (FFMI) and fat mass index (FMI), WC, waist-to-hip ratio (WHR), fat-free mass (FFM), fat trunk, and fat mass (FM).To record a more accurate result, participants were asked to be present with empty stomachs and bladders, without exercising before the session, and without their shoes, jackets, and coats.The output was printed out after 20 min of measurement.

Assessment of other variables
The assessment of physical activity (PA) levels was conducted using the International Physical Activity Short Form Questionnaire (IPAQ).This questionnaire was administered through interviews, with participants providing oral responses.PA was quantified as metabolic equivalent task minutes per week (MET-minutes/week) (57).Additionally, a demographic questionnaire was utilized to gather information on age, educational background, employment status, marital status, and economic status.

Definition of metabolic health and its components
The assessment of the metabolic health status in individuals was conducted using the Karelis criterion, which considers both insulin sensitivity and inflammatory profiles.According to the Karelis criterion, individuals are considered metabolically healthy if they meet four or more of the following five components: HOMA-IR ≤ 2.7, hs-CRP ≤ 3.0 mg/L, LDL-C ≤ 2.6 mmol/L without any treatment, HDL-C ≥ 1.3 mmol/L without any treatment, and TG ≤ 1.7 mmol/L (14).

Statistical analysis
Data analysis was conducted using SPSS version 26 (SPSS Inc.).The normality of data distribution was assessed using the Kolmogorov-Smirnov test.Continuous variables were reported as mean ± standard deviation (SD), and the categorical variables were reported as numbers and percentages.The mean and SD were reported by a one-way analysis of variance (ANOVA).Analysis of covariance (ANCOVA) was used to examine the relationship between demographic variables (age, height, weight, PA, inflammatory parameters, food groups, macronutrients, and micronutrients) and the NOVA classification system while controlling for confounding variables, including age, BMI, and PA factors.The chi-square test was utilized to analyze the frequency of categorical variables, including education, economic status, marital status, employment status, and the NOVA classification system.Binary logistic regression analysis was employed to evaluate the association between NOVA classification scores and MUHO, with odds ratios (ORs) and 95% confidence intervals (CIs) reported.The analysis was performed in three models: crude, model 1, and model 2. The crude model did not include any confounding factors, with model 1 adjusted for age, BMI, energy intake, and PA and model 2 adjusted for age, BMI, energy intake, PA, economic status, supplement use, educational background, and marital status.In addition, to evaluate the mediatory role, Barrett's model was used, and inflammatory markers including hs-CRP, TGF-β, IL-1β, MCP-1, PAI-1, and HOMA-IR were included in the adjustment 3 Results

Study population and general characteristics
A total of 221 women who were overweight or obese were included in the analysis.The participants had a mean (SD) age of 36

General characteristics across quartiles of NOVA classification
Table 1 represents general characteristics and anthropometric variables across the quartiles of the NOVA classification.A significant difference was observed in mean age (p = 0.03), WHR (p = 0.02), and economic status (p = 0.01) in the crude model.However, after adjusting for confounding factors such as age, BMI, PA, and energy intake, no significant associations were found for these variables (p > 0.05).Other variables did not show any significant relationships among the NOVA quartiles (p > 0.05).Furthermore, after adjusting for confounders, a significant difference was observed in variables such as MCP-1 (p = 0.06), Gal-3 (p = 0.03), PAI-1 (p = 0.024), and supplement use (p = 0.03).

Dietary intake according to the NOVA quartiles
Table 2 displays the dietary intakes of the subjects categorized into NOVA classification system quartiles.The mean differences in protein, carbohydrate, SFA, monounsaturated fatty acid (MUFA), polyunsaturated fatty acid (PUFA), linoleic acid, linolenic acid, vitamin A, vitamin C, vitamin E, vitamin B1, vitamin B2, vitamin B3, vitamin B5, vitamin B9, vitamin B12, biotin, calcium, iron, zinc, copper, manganese, selenium, chromium, fruit consumption, refined grain consumption, dairy consumption, and red meat intake were found to be significant among the NOVA quartiles in the crude model (p < 0.05).In the adjusted model, significant mean differences were observed for vitamin B6, total fiber intake, SSB consumption, processed food consumption, and vegetable consumption (p < 0.05).

Association between quartiles of NOVA classification and MUHO among overweight and obese women
The association between NOVA quartiles and MUHO in both the crude and adjusted models (model 1 and model 2) is represented in Table 3.The results of our study demonstrate a 77% increase in the odds of MUHO in the third quartile compared to the first quartile in the crude model (OR = 1.77, 95% CIs = 1.02, 4.33, p = 0.01).This association remained significant after adjustment in model 1 (controlling for age, BMI, energy intake, and PA) (OR = 1.95, 95% CIs = 1.65, 5.79, p = 0.04) and model 2 (controlling for age, BMI, energy intake, PA, economic status, supplement use, educational background, and marital status) (OR = 1.40, 95% CIs = 1.09, 4.92, p = 0.04).

Association of MUHO across the quartiles of NOVA, mediated by inflammatory markers
The relationship between MUHO and the NOVA quartiles, mediated by inflammatory markers, in overweight and obese women is shown in Table 4.The potential role of inflammatory markers, such as hs-CRP (p = 0.84), TGF-β (p = 0.50), MCP-1 (p = 0.49), PAI-1 (p = 0.97), and HOMA-IR (p = 0.92), was found to be that of mediators.

Discussion
The current study investigated the associations between the mediating role of inflammatory markers and the relationship between the NOVA classification system and obesity phenotypes in obese and overweight adult women.
There was a positive association observed between the third quartile of the NOVA classification system and an increased likelihood of having the MUHO phenotype in women.Inflammatory markers such as hs-CRP, TGF-β, MCP-1, PAI-1, and HOMA-IR showed their mediatory role in Q3.This finding is consistent with a study by Yu et al., which found that the visceral adipose tissue (VAT) is independently associated with various inflammatory markers.Specifically, markers such as white blood cell (WBC) count and hs-CRP showed stronger correlations with VAT compared to other markers such as the neutrophil-lymphocyte ratio (NLR) and plateletlymphocyte ratio (PLR) (58).Furthermore, another study reported a significant negative association between adherence to a healthy plantbased diet index (hPDI) and the MUHO phenotype among overweight and obese Iranian women.This association was found to be mediated by factors such as TGF-β, IL-1β, and MCP-1 (59).Hosseininasab et al. indicated that there is a significant association between UPF consumption and TGF-β, atherogenic coefficient (AC), VFL, and the quantitative insulin sensitivity check index (QUICKI) (60).In a systematic review and meta-analysis of 43 observational studies, there was an association between intake of UPF and an increased risk of obesity, overweight, abdominal obesity, and metabolic syndrome (61).The higher consumption of dairy products, tea, and coffee compared to fast foods decreased the risk of developing an unhealthy phenotype (62).In a systematic review and meta-analysis study, there was an association between increased UPF consumption and a cardiometabolic risk profile (63).Exposure to bisphenol, an industrial chemical used in the plastic packaging of some UPFs, is related to an increased risk of cardiometabolic disease (64).
UPF intake must be cautiously adopted given the high proportion of artificial ingredients in their formulations (37), which are potentially harmful to human health, especially when consumed in excess.Excessive consumption of these foods is related to an increase in the occurrence of NCDs (37,65).Emerging evidence indicates that the consumption of UPF products contributes to an unhealthy dietary pattern (26).Furthermore, additives present in foods such as emulsifiers and non-caloric artificial sweeteners have been linked to various chronic disorders such as systemic inflammation, endothelial dysfunction, and disrupted immune response (66,67).A study conducted in the United States revealed that UPF consumption was associated with increased exposure to phthalates (68), which have been suggested to be linked to obesity (69).Evidence suggests that abdominal obesity's origin can be associated with UPF intake, which delays satiety signals (37).UPF can also increase ad libitum energy intake by approximately 500 kcal/day, leading to higher weight gain compared to a minimally processed diet (70).Carrageenan, one of the commonly used additives, has been implicated in promoting IR and inhibiting insulin signaling in mice (71,72), which may contribute to weight gain (73).On the other hand, a minimally processed diet has been shown to result in higher levels of peptide YY (an appetite suppressant hormone) compared to the UPF diet (70).Certain emulsifiers (such as carboxymethyl cellulose and polysorbate-80) have induced metabolic disturbances, alterations in the gut microbiota, and low-grade inflammation in mice (74).
The present study has its strengths.First, it is the first study to investigate the mediatory role of various inflammatory markers on the relationship between the NOVA classification system and obesity phenotypes among obese and overweight adult women.Second, the assessment of dietary intake was conducted using a validated and reliable questionnaire.Third, confounding factors were adjusted in the statistical models.
However, there are several limitations to be discussed.The observational design of the study restricts the ability to establish causality of associations.Nevertheless, an observational design is the most suitable approach to investigate these types of associations, as conducting longitudinal studies involving interventions targeting UPF intake could pose health risks and ethical concerns.Furthermore, the results cannot be extrapolated to the general population as the participants included in the study were overweight/obese Iranian women.In addition, the assessment of food intake using a FFQ is susceptible to measurement bias.Nevertheless, FFQs have been widely used as a tool in epidemiological studies since the 1990s (75).Finally, despite considering potential confounders, the possibility of residual confounding factors influencing the results cannot be entirely ruled out.

Conclusion
In conclusion, our findings indicate a significant positive association between the consumption of UPFs and the MUHO phenotype in overweight and obese women residing in Iran.This association appears to be mediated by inflammatory markers such as MCP-1, PAI-1, hs-CRP, TGF-β, and HOMA-IR.The results of this study suggest that UPF consumption may have detrimental effects on obesity-related characteristics and, consequently, on NCDs.The role of diet and nutritional status is crucial for improving human health.Identifying unhealthy dietary patterns and their association with obesity is essential for preventing chronic diseases and enhancing public health strategies.Further research is warranted to confirm and expand upon our findings.

TABLE 1
General characteristics of the participants in the study among quartiles of the NOVA classification system in overweight and obese women (N = 221).

TABLE 2
Dietary intakes according to NOVA classification system quartiles in study participants (N = 221).

TABLE 3
Association between NOVA classification score and MUHO among overweight and obese women (N = 221).
The first quartile of the Nova classification score was considered the reference group.Metabolic healthy was considered as the reference group.Model 1 was performed to adjust for potential confounding factors (age, BMI, energy intake, and PA).Model 2 was performed to adjust for potential confounding factors (age, BMI, energy intake, PA, economic status, supplement usage, education status, and marital status).OR: odds ratio; CIs: confidence intervals.p-values of < 0.05 were considered statistically significant, and p-value of 0.

TABLE 4
Association of MUHO across the quartiles of NOVA, mediated by inflammatory markers.The first quartile of the Nova classification score was considered the reference group.Metabolic healthy was considered as the reference group.OR, odds ratio; CIs, confidence intervals; hs-CRP, highsensitive C-reactive protein; TGF-β, transforming growth factor-β; IL-1B, interleukin-1beta; MCP-1, monocyte chemoattractant protein-1; PAI-1, plasminogen activator inhibitor-1; HOMA-IR, homeostatic model assessment for insulin resistance.p-values of < 0.05 were considered statistically significant, and p-value of 0.05, 0.06, and 0.07 were considered marginally significant.