Dietary inflammatory index, inflammation biomarkers and preeclampsia risk: a hospital-based case–control study

Abstract

This study evaluated the association between inflammatory diets as measured by the Dietary Inflammatory index (DII), inflammation biomarkers and the development of preeclampsia among the Chinese population. We followed the reporting guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology statement for observational studies. A total of 466 preeclampsia cases aged over 18 years were recruited between March 2016 and June 2019, and 466 healthy controls were 1:1 ratio matched by age (±3 years), week of gestation (±1 week) and gestational diabetes mellitus. The energy-adjusted DII (E-DII) was computed based on dietary intake assessed using a seventy-nine item semiquantitative FFQ. Inflammatory biomarkers were analysed by ELISA kits. The mean E-DII scores were −0·65 ± 1·58 for cases and −1·19 ± 1·47 for controls (P value < 0·001). E-DII scores positively correlated with interferon-γ (r _s = 0·194, P value = 0·001) and IL-4 (r _s = 0·135, P value = 0·021). After multivariable adjustment, E-DII scores were positively related to preeclampsia risk (P _trend < 0·001). The highest tertile of E-DII was 2·18 times the lowest tertiles (95 % CI = 1·52, 3·13). The odds of preeclampsia increased by 30 % (95 % CI = 18 %, 43 %, P value < 0·001) for each E-DII score increase. The preeclampsia risk was positively associated with IL-2 (OR = 1·07, 95 % CI = 1·03, 1·11), IL-4 (OR = 1·26, 95 % CI = 1·03, 1·54) and transforming growth factor beta (TGF-β) (OR = 1·17, 95 % CI = 1·06, 1·29). Therefore, proinflammatory diets, corresponding to higher IL-2, IL-4 and TGF-β levels, were associated with increased preeclampsia risk.

Preeclampsia is one of the gestational complications that leads to maternal and neonatal morbidity and mortality⁽¹⁾. It has been reported that approximately 2–8 % of first pregnancies are diagnosed with preeclampsia⁽²⁾. Pregnant woman with preeclampsia generally has a higher risk of adverse outcomes. For instance, respiratory distress syndrome and pneumonia in newborn infants have been reported in a Chinese large-prospective cohort study⁽³⁾. In addition, there were about 2·57 times of odds of preterm birth in women with preeclampsia⁽⁴⁾.

The aetiology of preeclampsia has not yet been fully clarified; however, the hypothesis is that a deficiency of restoring uterine spiral arteries from incomplete placental implantation is the primary causative factor of this disorder⁽²⁾. This disruption promotes placental ischaemia and contributes to more cellular oxidative stress, which is reflected by the release of soluble fms-like tyrosine kinase-1 and proinflammatory cytokine such as TNF-α. Moreover, studies found that the shift towards M1 phenotype leads to strong synthesis of proinflammatory cytokines that would enhance trophoblastic apoptosis⁽⁵⁾. The overexpression of Th1 proinflammatory cytokines such as TNF-α, interferon gamma (IFN-γ) and IL-2 is associated with preterm delivery and intrauterine growth retardation^(6,7) . The upregulation of other proinflammatory cytokines, such as IFN-γ, IL-2 and IL-17, and the downregulation of anti-inflammatory cytokines, including IL-4, IL-10 and transforming growth factor beta (TGF-β), are eventually detected in preeclampsia^(8–10).

More recent studies have focused on some interactions between nutrients and the possible impact of overall dietary habits on chronic systematic inflammation. Higher consumption of dietary fibre, vitamins (e.g. vitamin C, vitamin A and vitamin D), minerals (e.g. Ca, phosphorus, potassium and Mg) and MUFA and PUFA are examples of antioxidant nutrients⁽¹¹⁾, while SFA and cholesterol have some inflammation-promoting properties⁽¹²⁾. Unfortunately, since we eat foods as a mixture of numerous nutrients, only focusing on some particular nutrients and diseases may miss much crucial information about their casual relationships. In fact, a systematic review and meta-analysis found that adequate vegetable and fruit intake could lower the risk of preeclampsia and reduce proteinuria⁽¹³⁾.

Based on the inflammatory properties of nutrients, the Dietary Inflammatory Index (DII) was first developed in 2014 by Shivappa et al. ⁽¹⁴⁾ to evaluate the inflammatory properties of the diet from maximally anti- to proinflammatory by using a variety of dietary assessment tools such as 24-h recall, FFQ or 3-d food records^(15,16) . SFA, retinol, processed meat, full-fat dairy, refined grain, sweets, desserts, carbonated beverages and sugar-sweetened beverages are examples of proinflammatory foods with a higher DII score^(17,18) . Nuts, fruits, vegetables, whole grains, fish and olive oil are characterised as anti-inflammatory foods⁽¹⁷⁾. Higher DII scores and proinflammatory diets were reported to be associated with 2·12 times higher odds of miscarriage, with a corresponding increase in IL-6⁽¹⁹⁾.

Recent studies have been published on the association between the DII and the risk of inflammatory-mediated noncommunicable chronic diseases as well as mortality⁽²⁰⁾. Based on our knowledge, however, no study has evaluated the potential of diets that are classified by the overall energy-adjusted dietary inflammatory index (E-DII) on the risk of preeclampsia. Therefore, this study aimed to explore the association between inflammatory diets as measured by the E-DII and the development of preeclampsia in a hospital-based population among the Chinese population in Henan province.

Methods

Study population and study design

This study was performed between March 2016 and June 2019 at the First Affiliated Hospital of Zhengzhou University, China. Full details about the study design have been published in previous studies^(21–23). In particular, a total of 466 cases with a ratio of 1:1 matched for case control were enrolled. The inclusion criteria were women aged over 18 years with at least 28 weeks of gestation with a singleton pregnancy. The presence of preeclampsia among the cases was defined on China’s ‘Diagnosis and treatment guideline of hypertensive disorders in pregnancy, 2015’ guideline⁽²⁴⁾. According to this guideline, cases of incident preeclampsia were identified as having systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg after the 20th week of pregnancy and met any of the following criteria: (1) urinary protein ≥ 0·3 g/24 h, urinary protein/creatinine ratio ≥ 0·3 or random urinary protein ≥ (+) (as the examination for quantitative urine protein cannot be carried out in pregnant women) and (2) nonalbuminuria but with damage to organs or systems such as the heart, lung, liver, kidney and other important organs, or abnormal changes in the blood system, digestive system, nervous system and placental–fetal involvement.

The controls recruited corresponding to the cases were women who were preparing for delivery, had never been diagnosed with hypertension or albuminuria and were matched with the cases for age (±3 years), week of gestation (±1 week) and gestational diabetes mellitus. The following exclusion criteria were applied to both groups: (1) being diagnosed with heart disease, malignancy, hyperthyroidism, immune system diseases, chronic renal insufficiency and other endocrine system diseases and (2) having epilepsy, depression and other mental or cognitive dysfunction.

Ethics

The study protocol including other study-related documents was run under the Declaration of Helsinki guidelines and was approved by the Ethics Committee of Scientific Research and Clinical Trials of the First Affiliated Hospital of Zhengzhou University. The informed consent form for participation in the study was signed by all participants before any epidemiological data and biological specimens were obtained. Data collection followed the reporting guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology statement for observational studies.

Dietary assessment

A validate and reproducible seventy-nine item semiquantitative FFQ was used to collect information about the participants diet during the past 3 months prior to the study period⁽²⁵⁾. The definition of the food groups has been described in detail in a previous study⁽²⁵⁾. Each food item asked participants to report how often, on average, they had consumed it for frequency range from never, per day, per week and per month. A photobook with different food portion sizes was provided for more precisely estimating the quantity of food intake. Nutrients and energy analyses were derived by the China Food Composition Tables 2009, which includes the nutrient portion and energy of each food item⁽²⁶⁾.

Dietary inflammatory index

To compute the DII score, we used the method described in Shivappa et al. ⁽¹⁴⁾. Briefly, the literature published from 1950 to 2010 with 1943 articles determined the association of forty-five food parameters in total for at least one of six inflammatory factors. Scores of –1, +1 and 0 were assigned if the effects were significantly decreased, increased or nonsignificant changes, respectively, of inflammatory biomarkers including IL-1β, IL-6, TNF or C-reactive protein. E-DII scores were calculated using the nutrient density method in which all food parameters were converted to per 1000 kcal of nutrients. In this study, thirty-five available food parameters from the FFQ were used to calculate the E-DII score and included the following nutrients: energy, carbohydrate, protein, total fat, cholesterol, fatty acids, SFA, MUFA, PUFA, n-3 fats, n-6 fats, soluble fibre, carotene, β-carotene, total carotenoids, thiamin, riboflavin, niacin, vitamin C, vitamin A, vitamin E, vitamin D, vitamin B₆, vitamin B₁₂, folic acid, Mg, Fe, Zn, Se, total anthocyanidins, total flavonoids, total flavan-3-ol, total flavanones, total flavones and total flavonols. Higher E-DII scores indicate a more proinflammatory diet, whereas lower scores represent a more anti-inflammatory diet.

Laboratory evaluation

A 12-h overnight fasting blood sample was collected from all participants. The serum was obtained by centrifugation at 3000 rpm at 4°C for 15 min and stored at –80°C for further analysis⁽²⁷⁾. Inflammatory biomarkers were measured in this study using ELISA kits, including TNF-α, IFN-γ, TGF-α and IL -2, -4, -10 and -17A. The Th1/Th2 and Th17/Treg ratios were also determined.

Assessment of covariates

Demographic characteristics included age, gestational age (weeks), education level (primary school or less, secondary/high school, college/university or above) and average monthly household income (≤ 2000, 2000–4000, 4000–6000 and > 6000 RMB). Lifestyle habits included passive smoking (yes or no), alcohol drinking status and dietary supplements used during the 3 months before pregnancy (yes or no) and physical activity (MET-h/d). Medical conditions, including pre-pregnancy BMI (kg/m²) and blood pressure (systolic blood pressure and diastolic blood pressure, mmHg), were collected via face-to-face interview of each case or control.

Statistical analysis

Descriptive statistics were used to summarise the characteristics of the participants. Continuous normally distributed variables are expressed as the mean ± sd . The Kolmogorov–Smirnoff test was used to test the normality of the data. Skewed distribution continuous variables are expressed as medians and interquartile ranges. Categorical variables are expressed as frequencies or percentages. The general characteristics of the cases and controls were compared using a paired t test for continuous variables with a normal distribution, whereas the Wilcoxon signed-ranks test was used for nonnormally distribution. The Pearson χ ² test was applied to compare all categorical data. Tertiles of E-DII scores were calculated based on the distribution of E-DII among controls. For every distribution of the food groups, energy was adjusted prior to comparison across tertiles of the E-DII and the Kruskal–Wallis H test was used. The correlation among nutrients, E-DII and inflammatory biomarkers was performed by Spearman’s correlation coefficient. Binary logistic regression analysis was performed to estimate the association between inflammatory markers and preeclampsia risk. To determine the risk of preeclampsia, the E-DII was analysed as both a categorical and a continuous variable, and the OR and 95 % CI were estimated using conditional logistic regression with different models. The first model was adjusted for age. The second model was fully adjusted for age, gestational age (weeks), pre-pregnancy BMI (kg/m²), passive smoking (yes or no), alcohol drinking status and dietary supplements used during the 3 months before pregnancy (yes or no), physical activity (MET-h/d) and energy intake (kcal/d).

All tests were two-sided, and a P value < 0·05 was considered statistically significant. Data were analysed using SPSS version 2.0 (IBM Corp).

Results

The mean age was 30·88 ± 5·00 and 31·00 ± 4·85 in the case and control groups, respectively (Table 1). Cases had significantly higher E-DII scores than controls, of which the means and SD were −0·65 ± 1·58 and −1·19 ± 1·47, respectively (P value < 0·001) (Table 1). Cases with preeclampsia were statistically more likely to have a higher pre-pregnancy BMI, systolic pressure and diastolic pressure. Furthermore, compared with the controls, cases with preeclampsia were more likely to have primary school or less and were also passive smokers and active alcohol drinkers (Table 1).

Table 1. Characteristics of cases and controls

(Mean values and standard deviations)

Test for differences of case and control using a paired t test for normally distributed continuous variables; Wilcoxon signed-ranks test for skewed distributed continuous variables; χ ² test for categorical variables; P value < 0·05 indicates significant difference.

Values are presented as mean (sd) for continuous variables; number (%) for categorical variables.

† Drinking status refers to alcohol consumption during the 3 months before pregnancy.

‡ Physical activities included daily occupational, leisure time and household chores, evaluated by metabolic equivalent (MET) hours per day.

With respect to the controls, subjects with preeclampsia had significantly higher IL-4 (2·05 ± 1·51 v. 1·64 ± 1·09, P value =0·008) and TGF-β (12·43 ± 7·77 v. 9·13 ± 4·72, P value = 0·004) levels (Table 2). Moreover, the Th1/Th2 ratios in preeclampsia were significantly higher than those in healthy pregnancy for both TNF-α/IL-10 ratios (1·99 ± 2·11 v. 0·68 ± 9·25, P value =0·039) and IL2/IL-10 ratios (1·95 ± 1·91 v. 0·59 ± 6·12, P value =0·035) (Table 2). The levels of proinflammatory cytokines, including TNF-α, IFN-γ, IL-2 and IL-17, were not significantly different between preeclampsia cases and controls (Table 2).

Table 2. Comparison of inflammatory biomarkers between cases and controls

(Mean values and standard deviations)

IQR, interquartile range; Th, helper T cells; Treg, regulatory T cells; IFN-γ, interferon gamma; TGF-β, transforming growth factor beta.

Test for differences of case and control using Wilcoxon signed-ranks test; P value < 0·05 indicates significant difference.

While the consumption of total grains significantly increased across tertiles, the consumption of leafy vegetables, starchy vegetables and fruits as well as fish and seafood was significantly decreased in the preeclampsia cases (P value < 0·001) (Table 3). Whole-grain consumption also decreased across the tertiles of E-DII scores; nevertheless, no significant difference was observed. Similar trends in the consumption of total grains across tertiles were found in the controls (P value < 0·001) (Table 2). In contrast to preeclampsia cases, with respect to the first tertiles of the E-DII scores, healthy pregnancy in the third tertiles of the E-DII scores had significantly lower whole-grain, poultry and red meat intake (P value < 0·001) (online Supplementary Table 1).

Table 3. Distribution of food groups across tertiles of energy-adjusted dietary inflammatory index (E-DII) for preeclampsia cases

(Median values and interquartile range, n 466)

IQR, interquartile range.

Comparison of food groups across tertiles of energy adjusted-dietary inflammatory index (E-DII) by using Kruskal–Wallis H test; P value < 0·05 indicates significant difference.

All foods groups were calculated by edible portions; total grains, whole grains, nuts and seeds were calculated by dry uncooked weights.

Dairy and products included fresh whole and skim cow’s milk, yogurt, butter, cheese and ice cream.

Regarding the controls, subjects with preeclampsia had significantly greater consumption of tubers, whereas consumption of leafy and starchy vegetables, nuts and seeds, fruits, dairy products, eggs, fish and seafood, poultry and red meat was significantly lower (all P values < 0·05) (online Supplementary Table 2). In addition, the lower levels of soluble fibre, protein, cholesterol, Ca, Mg, Fe, Zn, Se, β-carotene, vitamins A, C, D, E, B₆ and B₁₂, as well as folate and total flavonoids in preeclampsia were extremely significantly different from those in healthy pregnancies (all P values < 0·05) (online Supplementary Table 2).

The correlations among the E-DII score, inflammatory biomarkers, food groups and nutrients are shown in Supplementary Tables 3 and 4. E-DII scores showed a significant positive correlation with IFN-γ (r _s = 0·194, P value = 0·001) and IL-4 (r _s = 0·135, P value = 0·021). The correlation between E-DII scores and food groups revealed that while increasing intake of total grains provided significantly higher E-DII scores (P value < 0·001), whole grains, leaves and starchy vegetables, nuts and seeds, fruits, eggs, fish and seafood, poultry and red meat were significantly inversely correlated with the E-DII scores (P value < 0·001). In particular, total fat, SFA, MUFA, PUFA, Se and vitamin E presented significantly higher E-DII scores (all P values < 0·05). Th1 cytokine profiles such as TNF-α, IFN-γ, IL-2 and IL-17A were negatively correlated with leafy vegetables, fish and seafood, poultry, soluble fibre, Mg, β-carotene, vitamins A, C, B6, folate, total anthocyanidins and total flavonoids at a significant level (all P values < 0·05).

Participants whose E-DII belonged to the highest tertile, the most proinflammatory group, were 2·10 times more likely to have preeclampsia than those in the lowest tertiles, the most anti-inflammatory group (age-adjusted OR = 2·10, 95 % CI = 1·53, 2·89, P _trend < 0·001) (Table 4). Similarly, when further adjusting was conducted for all possible covariances, there was a 2·18-fold risk of being preeclampsia for those in the highest tertile of E-DII compared with those in the lowest tertiles (full-adjusted OR = 2·18, 95 % CI = 1·52, 3·13, P _trend < 0·001) (Table 4). Significant positive associations between the continuous E-DII score and preeclampsia risk in the age-adjusted model and full-adjusted model were simultaneously observed (age-adjusted OR = 1·26, 95 % CI = 1·16, 1·38, P value < 0·001; full-adjusted OR = 1·30, 95 % CI = 1·18, 1·43, P value < 0·001) (Table 4).

Table 4. OR and 95 % CI for association between the energy-adjusted dietary inflammatory index (E-DII) and the risk of preeclampsia

(Odd ratio and 95 % confidence intervals)

Crude and adjusted OR and 95 %CI were obtained from conditional logistic regression model by entering method; P value or P _trend < 0·05 indicates significant difference.

Model 1 = Adjusted for age.

Model 2 = Fully adjusted for age, gestational age (weeks), pre-pregnancy BMI (kg/m²), education level (primary school or less, secondary/high school, college/university or above), passive smoking (yes or no), drinking status and dietary supplements used (e.g. vitamin D, Ca, Ca plus vitamin D, Fe, folic acid and multivitamin supplements) during the 3 months before pregnancy (yes or no), physical activity (MET-h/d) and energy intake (kcal/d).

The OR, 95 % CI and P values for the preeclampsia risk of inflammatory biomarkers are given in Table 5. After full adjustment for confounding factors, for every one-unit (mg/dl) increase in IL-2, IL-4 and TGF-β, there was a significant corresponding increase in the risk of developing preeclampsia by 7 % (95 % CI = 3 %, 11 %, P value = 0·001), 26 % (95 % CI = 3 %, 54 %, P value = 0·024) and 17 % (95 % CI = 6 %, 29 %, P value =0·002), respectively. However, TNF-α, IFN-γ, IL-10 and IL-17A were not associated with preeclampsia risk.

Table 5. Binary logistic regression analysis according to the association between the inflammatory biomarkers and preeclampsia

(Odd ratio and 95 % confidence intervals, n 466)

β, beta estimate; se, standard error; IFN-γ, interferon gamma; TGF-β, transforming growth factor beta; P value < 0·05 indicates significant difference.

All values adjusted for age, gestational age (weeks), pre-pregnancy BMI (kg/m²), education level (primary school or less, secondary/high school, college/university or above), passive smoking (yes or no), drinking status and dietary supplements used (e.g. vitamin D, Ca, Ca plus vitamin D, Fe, folic acid and multivitamin supplements) during the 3 months before pregnancy (yes or no), physical activity (MET-h/d) and energy intake (kcal/d).

The β estimate represents the change of each outcome per each 1-unit increase in inflammatory biomarkers.

Discussion

Our case–control study indicated a positive association between the E-DII, as well as inflammation markers (e.g. IL-2, IL-4 and TGF-β), and the risk of preeclampsia. Considering that the E-DII is a sensitive index for inflammatory markers, it can be inferred that lowering the intake of proinflammatory foods with high E-DII scores and increasing the consumption of anti-inflammatory foods with low-EDII scores would be negatively associated with the development of preeclampsia.

Recent studies found some relationship of different food types and intakes amount to pregnancy-related adverse outcomes⁽²⁰⁾. Our study is consistent with others that found when compared with the lowest quartile, participants who were more adherent to an unhealthy dietary pattern had a higher risk of preeclampsia, while those in the top quartile of adherence to a healthy dietary pattern had a decreased preeclampsia risk⁽²⁸⁾. In this study, an unhealthy dietary pattern was defined as consuming high amounts of processed meat, high GI foods, potatoes, legumes, high-fat dairy, whole-grain and soft drinks, whereas the healthy dietary pattern included more vegetables, poultry, red meat, eggs and unsaturated fat⁽²⁸⁾. A systematic review and meta-analysis showed that frequently consuming vegetables, fruits, legumes and whole-grains significantly reduced adverse pregnancy outcomes⁽²⁹⁾. Similarly, high plant-based food and vegetable oil consumption decreased preeclampsia risk, while processed meat, salty snacks and sweet drinks were positively correlated with the odds of preeclampsia in another study⁽³⁰⁾. Nevertheless, a cluster randomised controlled trial conducted in northwestern China concluded that only vegetable-type dietary patterns were inversely related to preeclampsia⁽¹³⁾.

All of these publications emphasised the relationship between diet and the development of preeclampsia through a crucial mediator called inflammation. Levels of serum inflammatory biomarkers such as C-reactive protein, IL-2, IL-4, IL-10 and IL-17A, TNF-α, transforming growth factor beta (TGF-β), interferon (IFN-γ), adiponectin, etc., are able to indicate the level of systematic inflammation and explain the pathogenesis of diseases⁽³¹⁾.

The aetiology of preeclampsia is hypothesised to involve chronic immune activation or oxidative stress, which is attributed to impaired implantation, leading to an imbalance in the production of proinflammatory and regulatory cytokines. This imbalanced secretion during the gestational period might be the primary causative factor of preeclampsia. In general, regulatory T cell (Treg) proliferation triggered by TGF-β is induced by natural killer cells, resulting in increased maternal tolerance towards the fetus⁽³²⁾. In addition, M2 phenotype macrophages promote the releases of the Th2 cytokines TGF-β, IL-4 and IL-10, which occurs in successful pregnancy⁽³²⁾. In contrast, pregnancies with preeclampsia have limited trophoblast invasion that leads to narrow blood vessels, developing ischaemia and hypoxia in the placenta. Moreover, the activation of Th1 cells through the production of TNF-α, IL-6 and IFN-γ also occurs in preeclampsia⁽³²⁾. Apart from that, the expansion of Th1 and Th17 cells is responsible for triggering inflammation. Among women with a healthy pregnancy, Th1/Th17 activity is suppressed by the proliferation of Tregs; however, this activity seems to be higher in preeclampsia⁽³³⁾.

Previous studies, including a systematic and meta-analysis study, have observed associations between cytokines and the risk of preeclampsia, indicating women are at a higher risk of preeclampsia when their proinflammatory cytokines TNF-α, IFN-γ, IL-2 and IL-17 are elevated and their anti-inflammatory cytokine IL-4, IL-10 and TGF-β are depressed^(34,35) . These results might be attributed to the depletion of Th2 cells and Tregs in preeclampsia cases⁽³⁶⁾. The decrease in Treg induces apoptosis in trophoblast cells, which limit the invasion of trophoblast cells and remodel the maternal spiral artery⁽³⁵⁾. Placental ischaemia and hypoxia increase systematic oxidative stress and stimulate pro-inflammatory cytokine secretion as mentioned above.

Moreover, immune cell differentiation and proliferation are influenced by the presence of cytokines, for instance, the presence of IL-6 promoted Th17 cell differentiation while inhibiting Treg cells⁽³⁷⁾. At the same time, the IL-17 cytokine produced by Th17 cells could increase the production of IL-6 cytokines⁽³⁸⁾. These findings are partly in line with our study, which also showed a significant positive relationship between IL-2 and preeclampsia. However, the association between IL-4 and TGF-β and preeclampsia found in this study was inconsistent with other studies: increasing IL-4 and TGF-β significantly corresponded to an increase in preeclampsia. Since the serum levels of cytokines vary, depending on the condition of the pathology⁽³⁹⁾, these contradictory reports are probably due to differences in gestational age at sampling, differences in sample size, the severity of the condition, population diversities among studies and variations in blood collection and laboratory processing. Moreover, the half-life of the cytokines could also contribute to discrepancies among studies⁽⁴⁰⁾.

To provide a further explanation for these findings, DII scores were adapted to explain the underlying strategies, since DII scores are based on dietary data sets collected from a dietary survey among the participants in that study, hence they could reflect the activation of oxidative stress and the inflammatory immune response related to specific food parameters⁽⁴¹⁾. The production of TNF-α, IFN-γ, IL-2 and IL-6 is mediated by proinflammatory diets high E-DII foods, corresponding with higher intake of refined grains, dairy products, red meats, soft drinks, sweets and desserts and lower consumption of whole grains, vegetables and fruits^(42–44). Nevertheless, the consumption of red meat across tertiles E-DII among cases in this study was parallel. The effects of red meat and dairy products on inflammation are still controversial. A systematic review and meta-analysis of randomised controlled trials found that high dairy product intake significantly decreased TNF-α and IL-6 levels⁽¹¹⁾. Moreover, vegetables and fruits, low E-DII foods, contain antioxidant compounds such as β-carotene, polyphenols and vitamins C and E. High antioxidants content in these serum and placenta prevents hypoperfusion, thus protecting against preeclampsia development⁽⁴⁵⁾. In addition, flavonoids and β-carotene inhibit the progression of inflammation through the reduction of free oxidative agents and IL-6 and TNF-α production^(46,47) .

Some limitations need to be taken into account in the interpretation of our results. First, the present study is a case–control study, and a FFQ was used for dietary assessment in this study, so recall bias might exist. Second, since this study is a case–control study, selection bias is inevitable; however, we could reduce differences in socio-demographic information between two groups, ensure the comparability between two groups and reduce the risk of selection bias by recruiting cases and controls from the same referral hospital. Another limitation is that smaller samples with available inflammatory biomarkers were provided in our study; therefore, there is insufficient information about whether inflammation specifically mediates the association between a proinflammatory diet during pregnancy and preeclampsia. Our results indicate that E-DII scores were associated with greater IFN-γ, which supports the hypothesis that diets induce preeclampsia via systematic inflammation. However, the association between the E-DII and preeclampsia risk remains to be confirmed in prospective analysis.

Conclusion

Individuals with higher E-DII scores and higher IL-2, IL-4 and TGF-β cytokines were associated with increased preeclampsia risk. Additional prospective studies are recommended to confirm our results.