Gestational diabetes mellitus (GDM) refers to abnormal glucose tolerance with onset or first recognition during pregnancy(Reference Moon and Jang1). Study has shown that the prevalence of GDM has significantly increased two to three times in 10 years(Reference Yang and Wu2), and the total incidence is 14·8 %(Reference Gao, Sun and Lu3). Preeclampsia (PE) is a serious pregnancy complication next only to GDM, with an estimated average incidence of 5 %(Reference Gu, Lin and Hou4). GDM and PE in pregnant women are syndromes associated with substantially increased risk of preterm delivery, low birth weight, fetal growth restriction and caesarean section(Reference Rocha5–Reference Champion, Battarbee and Biggio7). The causes of GDM and PE are multifaceted, involving genetic, physiological and lifestyle-related risk factors. Recently, the role of chronic inflammation has received increasing attention(Reference Phoswa and Khaliq8,Reference McElwain, Tuboly and McCarthy9) . Moreover, studies have shown that diet can regulate the level of inflammation in the body by altering the expression of inflammatory genes and the concentration of inflammatory markers(Reference Bordoni, Danesi and Dardevet10,Reference Gardener, Rainey-Smith and Martins11) . Therefore, recent studies have begun to explore the association between dietary inflammatory potential and risk of GDM and PE.
The dietary inflammatory index (DII) has been developed as the major dietary measurement to evaluate the inflammatory potential of diet, helping to identify the relationship between diet, inflammation and disease. Compared with single food or nutrient intake, DII has been demonstrated to be a reasonable and important indicator for measuring the overall inflammatory potential of diet and an effective tool for studying the relationship between overall diet and diseases(Reference Vicente, Lucio Dos Santos Quaresma and Maria de Melo12). Multiple studies have explored the association between dietary inflammatory potential (evaluated by DII) and the risk of GDM and PE, but the results are inconsistent. A cohort study based on 2639 women in China showed that DII score was positively associated with the risk of GDM(Reference Zhang, Wu and Zhong13). Another case–control study based on 932 women in China suggested that individuals with higher DII score were associated with increased PE risk(Reference Liu, Zheng and Cheng14). However, another Ireland cohort study based on 434 women participants suggested that there was no significant association with GDM or PE(Reference Killeen, Phillips and Delahunt15). To date, only Gao et al.’ s meta-analysis(Reference Gao, Zheng and Jiang16), which primarily focused on the relationship between diet quality and GDM risk, mentioned the association between DII score and GDM risk. However, their meta-analysis only included four studies and missed some eligible publications. Moreover, no comprehensive and systematic meta-analysis has been performed to evaluate the association between DII score and the risk of PE.
It is clearly necessary to provide more convincing evidence to clarify the association of DII score with the risk of GDM and PE. Therefore, we performed a systematic review and meta-analysis based on observational studies to explore the association, in order to provide a theoretical basis for maternal diet management.
Methods
This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines(Reference Moher, Liberati and Tetzlaff17) and registered in the PROSPERO international prospective register of systematic reviews (CRD42023407877).
Search strategy
Electronic databases, including PubMed, Web of Science, Cochrane Library, EMBASE, Scopus, China National Knowledge Infrastructure, Wan Fang Database and VIP Database, were searched to screen for all relevant published studies on the DII score with risk of GDM and PE from database inception to January 2023, using the following combined keywords: (‘dietary inflammatory index’ OR ‘dietary inflammatory’ OR ‘DII’ OR ‘inflammation’ OR ‘diet’) AND (‘gestational diabetes mellitus’ OR ‘gestational diabetes’ OR ‘GDM’ OR ‘preeclampsia’ OR ‘gestational hypertension’ OR ‘pregnancy-induced hypertension’ OR ‘pregnancy hypertension’ OR ‘hypertensive disorder of pregnancy’). Moreover, we also searched the reference lists of pertinent articles for any missing articles.
Inclusion and exclusion criteria
Studies were considered eligible if they met the following criteria: (1) studies reported the association between DII and GDM or PE; (2) studies that reported OR, hazard ratio or risk ratio with 95 % CI and (3) studies with either a cohort study, case–control study or cross-sectional study.
Studies were excluded if they met the following criteria: (1) studies without complete data; (2) studies for which the effect sizes could not be extracted and (3) studies such as systemic reviews, comments, case reports and editorials.
Data extraction
Two researchers independently searched the articles strictly according to the inclusion and exclusion criteria and provided articles with differences to third researcher for analysis to decide whether they should be included. The extracted data included the first author’s name, publication year, country, study design, age, pregnancy, sample sizes, DII/DIP/E-DII components, exposure assessment, outcome examined and adjustment for confounders.
Quality assessment
The Newcastle–Ottawa scale was used to evaluate the quality of the literature in all finally included studies(Reference Stang18). The Newcastle–Ottawa scale consists of three dimensions and eight items, including selection, comparability and outcome (cohort study) or exposure (case–control study). The maximum score for selection, comparability and outcome or exposure was 4, 2 and 3, respectively, for a maximum total score of 9. A total score of 0–3 indicated high risk of bias, 4–6 adequate risk of bias and 7–9 low risk of bias(Reference Fan, Wang and Yang19). Studies with a high risk of bias were excluded from the meta-analysis.
Statistical analysis
All statistical analyses were performed by STATA16.0. OR and 95 % CI were selected for the combined effect size; when both crude and adjusted OR were provided, we used the most fully adjusted OR for all studies. Heterogeneity between studies was assessed and quantified with Cochran’s Q test and the I2 statistic, in which P < 0·1 and I2 > 50 % were defined as statistically significant heterogeneity. When there existed significant heterogeneity, we used a random-effects model; otherwise, a fixed-effects model was used(Reference Zhao, Wu and Gao20). Subgroup analyses were conducted stratifying by study design, exposure assessment, country, pregnancy, sample size, inflammation assessment tool and energy adjustment. Egger test was conducted to assess the publication bias for each outcome and a sensitivity analysis was performed to examine the robustness of the results.
Results
Study selection and characteristics
A total of 568 articles were retrieved in the database according to the established search strategy, and 405 articles were removed due to duplication. Subsequently, during screening through the title and abstract, 121 irrelevant studies, twenty-three systematic reviews and the other three studies unrelated to GDM were excluded. We then further searched the full text of sixteen articles to assess their eligibility, of which four without complete data, one did not use DII, one study was duplication and one study did not provide useful data were excluded. Eventually, we included nine original studies to investigate the association between DII score with the risk of GDM and PE(Reference Zhang, Wu and Zhong13–Reference Killeen, Phillips and Delahunt15,Reference Sen, Rifas-Shiman and Shivappa21–Reference Pajunen, Korkalo and Koivuniemi26) . The flow of the above-described procedures is shown in Fig. 1.
PRISMA flow diagram of study selection process. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; GDM, gestational diabetes mellitus; DII, dietary inflammatory index.
Nine studies were included in the meta-analysis; there were seven(Reference Zhang, Wu and Zhong13,Reference Killeen, Phillips and Delahunt15,Reference Zhao, Yu and Wang22–Reference Pajunen, Korkalo and Koivuniemi26) and three(Reference Liu, Zheng and Cheng14,Reference Killeen, Phillips and Delahunt15,Reference Sen, Rifas-Shiman and Shivappa21) studies evaluating the relationship of DII with GDM and PE. A total of 11 423 participants from various countries were involved, including Finland(Reference Pajunen, Korkalo and Koivuniemi26), China(Reference Zhang, Wu and Zhong13,Reference Liu, Zheng and Cheng14,Reference Zhao, Yu and Wang22,Reference Yang25) , Iran(Reference Shivappa, Hébert and Akhoundan23,Reference Soltani, Aminianfar and Hajianfar24) , USA(Reference Sen, Rifas-Shiman and Shivappa21) and Ireland(Reference Killeen, Phillips and Delahunt15). All nine studies were published in the years from 2016 to 2022 inclusively. With regard to study design, there were seven cohort studies and two case–control studies, and the sample sizes ranged from 336 to 4189 participants. All the studies utilised either FFQ or 3-d food diaries as exposure assessment to calculate dietary intake. Continuous DII score was used to evaluate the relationship between DII score with risk of GDM and PE. Adjusted OR was reported in all nine studies and controlled for different types of confounding factors such as age, BMI, energy intake, pregnancy exercise and education level. Detailed information about the nine studies in the analyses is presented in Table 1. According to the Newcastle–Ottawa scale, three studies scored 6 points, five studies scored 7 points and one study scored 8 points, and the methodological quality of all nine studies was high (Table 2).
Characteristics of the included studies
DII, dietary inflammatory index; PE, preeclampsia; GDM, gestational diabetes mellitus; DIP, dietary inflammatory potential; E-DII, energy-adjusted dietary inflammatory index.
Methodological quality assessment score of the included studies
Association between dietary inflammatory index and the risk of gestational diabetes mellitus
Seven studies evaluated the association between DII score and the risk of GDM, three studies showed that GDM was associated with DII and the rest were considered unrelated; the total sample sizes were 9149. There was moderate heterogeneity among the studies (P = 0·004, I2 = 68·4 %), the meta-analysis result showed that there was a significant association between DII score and risk of GDM (OR = 1·13; 95 % CI 1·02, 1·25) and the difference was statistically significant (Fig. 2).
Forest plot showing the association between the dietary inflammatory index score and the risk of gestational diabetes mellitus.
Association between dietary inflammatory index and the risk of preeclampsia
Three studies evaluated the association between DII score and risk of PE, and one study showed that PE was associated with DII; the total sample sizes were 2708. In the meta-analysis, there was a significant association between DII score and the risk of PE (OR = 1·24; 95 % CI 1·14, 1·35) with heterogeneity (P = 0·125, I2 = 52·0 %) (Fig. 3).
Forest plot showing the association between the dietary inflammatory index score and the risk of preeclampsia.
Subgroup analysis
Subgroup analysis of PE was not performed because only three studies were included. To explore the sources of heterogeneity between DII score and risk of GDM, a total of seven subgroup variables were selected for meta-analysis (Table 3). Subgroup analysis was conducted for study design and exposure assessment, and the results indicated that both were no significant reduction in heterogeneity. A higher DII score predicted a greater risk of GDM, especially in individuals with mid pregnancy (OR = 1·20; 95 % CI 1·07, 1·34, I2 = 36·1 %, P = 0·209). Stratification by country, sample size, inflammation assessment tool and energy adjustment partly reduced the heterogeneity between studies, and the association differed significantly between studies that sample size ≥ 1000 (OR = 1·12; 95 % CI 1·03, 1·20, I2 = 0·0 %, P = 0·908) and those that < 1000 (OR = 1·14; 95 % CI 0·96, 1·35, I2 = 78·3 %, P = 0·001). Furthermore, stratification by energy adjustment partly reduced the heterogeneity between studies, especially for studies that used covariate method to adjust energy intake (I2 = 22·2 %) (online Supplementary Fig. S1–S7). In addition to variables in the subgroup analysis, differences in the dietary parameters used to calculate DII score across studies, with a maximum of 35 and a minimum of only 20, were also a possible reason, which may help explain the moderate heterogeneity (online Supplementary Table 1).
Subgroup analyses of the association between the DII score and GDM (95 % confidence intervals)
DII, dietary inflammatory index; GDM, gestational diabetes mellitus; E-DII, energy-adjusted dietary inflammatory index; DIP, dietary inflammatory potential.
Early pregnancy, pregnancy < 16 weeks; mid pregnancy, pregnancy ≥ 16 weeks. In addition, we combined the instructions of the authors in the original studies, defined pregnancy < 18 weeks as early pregnancy, 13–28 weeks as mid pregnancy.
Sensitivity analysis
Sensitivity analysis was conducted after removing any one of seven studies, to confirm that our results were not determined by a single study. The results showed that removing most of the studies had no significant effect on our results; the only exceptions were found in the removal of the studies by Zhao(Reference Zhao, Yu and Wang22) and Pajunen(Reference Pajunen, Korkalo and Koivuniemi26) changed the relative effect from significant to non-significant in the sensitivity analysis (OR = 1·10; 95 % CI 1·00, 1·20 and OR = 1·11; 95 % CI 1·00, 1·23), as shown in Fig. 4.
Sensitivity analysis of dietary inflammatory index and gestational diabetes mellitus.
Publication bias
Egger test was adopted to evaluate the publication bias included in the study; the Egger test obtained P = 0·202 for GDM and P = 0·538 for PE, indicating that there was no significant publication bias in the included studies.
Discussion
This study provides a systematic review and meta-analysis that evaluated all available observational studies on the association between DII score and risk of GDM and PE and provides the most comprehensive evidence in this research field. We included nine studies with 11 423 participants from five different countries. More importantly, we performed a detailed subgroup analysis, including pregnancy, inflammation assessment tools and type of energy adjustment, thus improving statistical power.
The meta-analysis indicated that a 1-unit increase in the DII score was associated with 13 % higher risk of GDM (OR = 1·13; 95 % CI 1·02, 1·25) and 24 % higher risk of PE (OR = 1·24; 95 % CI 1·14, 1·35). Similarly, in a review by Moslehi et al.(Reference Moslehi, Sakak and Teymoori27), it was concluded that DII was directly associated with increased odds of GDM. In addition, results from our study were in line with other study that reported an association of pro-inflammatory diet indicated by DII, with increased incidence of diabetes mellitus(Reference Tan, Du and Gao28). At present, the relationship between DII score and risk of GDM and PE is controversial, mainly because of the differences of country, sample sizes, energy intake and pregnancy. With regard to country, the pooled results of studies conducted in China(Reference Zhang, Wu and Zhong13,Reference Liu, Zheng and Cheng14,Reference Zhao, Yu and Wang22,Reference Yang25) , including DII score and risk of GDM and PE, showed an association, whereas the other studies indicated no association(Reference Killeen, Phillips and Delahunt15,Reference Sen, Rifas-Shiman and Shivappa21,Reference Shivappa, Hébert and Akhoundan23,Reference Soltani, Aminianfar and Hajianfar24,Reference Pajunen, Korkalo and Koivuniemi26) ; the differences in country between these studies may partly explain the inconsistent findings. The sample sizes considered ≥ 1000 showed an almost strong association between DII score and GDM risk(Reference Zhang, Wu and Zhong13,Reference Yang25) . However, the smaller sample sizes of studies failed to show a significant association between DII score and GDM risk(Reference Killeen, Phillips and Delahunt15,Reference Zhao, Yu and Wang22–Reference Soltani, Aminianfar and Hajianfar24,Reference Pajunen, Korkalo and Koivuniemi26) . Therefore, studies assessing the inflammatory potential of diet on GDM risk need larger sample sizes. In addition, among the variables investigated, adjusted for energy intake was critical for achieving significant results, including covariate method(Reference Zhang, Wu and Zhong13,Reference Zhao, Yu and Wang22,Reference Shivappa, Hébert and Akhoundan23,Reference Yang25) and nutrient-density method(Reference Killeen, Phillips and Delahunt15,Reference Pajunen, Korkalo and Koivuniemi26) , as compared with studies that did not utilise such adjustment(Reference Soltani, Aminianfar and Hajianfar24). The possible reason is that higher energy intake reflects more food intake, which may be variably distributed among pro-inflammatory and anti-inflammatory related foods. The DII score of women in the mid pregnancy(Reference Zhao, Yu and Wang22,Reference Shivappa, Hébert and Akhoundan23,Reference Yang25) was more strongly associated with GDM risk than those in the early pregnancy(Reference Zhang, Wu and Zhong13,Reference Killeen, Phillips and Delahunt15,Reference Soltani, Aminianfar and Hajianfar24,Reference Pajunen, Korkalo and Koivuniemi26) , one possible reason is that compared with the mid pregnancy, women in the early pregnancy intake less due to vomiting or other reasons, and are less likely to intake foods with higher DII score.
Diet can regulate the systemic inflammation level through pro-inflammatory or anti-inflammatory mechanisms of foods, nutrients and dietary patterns(Reference Cavalcante, Lima and Parente29,Reference Menzel, Biemann and Longree30) . Studies have shown that higher adherence to the Mediterranean diet was significantly associated with lower levels of inflammation, including reduced levels of CRP (C-Reactive Protein) IL-6 and fibrinogen(Reference Itsiopoulos, Mayr and Thomas31). The Mediterranean diet is characterised by a high intake of fruits, vegetables, fish and whole grains. It is rich in dietary fibre, vitamin C, vitamin E, folate, carotenoids and PUFA. Previous studies have shown that greater intake of the Mediterranean diet during pregnancy is associated with a reduced risk of GDM and PE(Reference Melero, García de la Torre and Assaf-Balut32–Reference Ou, Zhang and Zhao35). The Western diet, characterised by a greater intake of red meat, processed meat and fried foods, is rich in SFA, nitrosamines and other compounds related to oxidative stress and insulin resistance, which may increase the level of serum inflammatory markers(Reference Quan, Zeng and Jiao36,Reference Asadi, Yaghooti-Khorasani and Ghazizadeh37) . Studies have linked a high intake of Western diet and energy from SFA in early pregnancy may contribute to an increased risk of developing GDM and PE(Reference Mizgier, Jarzabek-Bielecka and Mruczyk38,Reference Herlambang, Puspasari and Maharani39) .
From a mechanistic point of view, the role of diet-related inflammation in GDM and PE can be explained by several mechanisms. First, the pro-inflammatory potential of dietary components may led to an imbalance in the level of inflammation by increasing the level of inflammatory factors, such as CRP, TNF-α and IL-6. High level of inflammatory factors induces phosphorylation of insulin receptor substrate-1, leading to its inactivation, and, in turn, inhibits the insulin receptor from phosphorylating this substrate, thus suppressing insulin receptor signalling(Reference Stephens, Lee and Pilch40–Reference Aleksandrova, Koelman and Rodrigues42). Inflammatory factors also disrupt the expression of insulin responsive GLUT4, thus reducing insulin-dependent glucose transport and peripheral glucose utilisation(Reference Stephens, Lee and Pilch40,Reference Piuri, Basello and Rossi41) . Meanwhile, previous studies have shown an association between cytokines and the risk of PE, with women at higher risk of PE when the pro-inflammatory cytokines TNF-α and IL-6 were elevated and the anti-inflammatory cytokines IL-4 and IL-10 were decreased(Reference Aggarwal, Jain and Mittal43,Reference Nath, Cubro and McCormick44) . The imbalanced level of inflammation causes activation of associated cellular signalling pathways like Toll-like receptor and NF-κB, and the downstream targets of these pathways may further affect the placental immune tolerance by impacting a negative effect on the secretion of trophoblastic microvesicles and endothelial cells disruption, thus leading to several adverse pregnancy complication like PE(Reference Aggarwal, Jain and Mittal43). Another mechanistic hypothesis relies on the gut microbiota. Gut microbiota are essential modulators of immune homoeostasis, and diet can directly or indirectly affect the composition of gut microbiota, which is associated with gut barrier function(Reference Hasain, Mokhtar and Kamaruddin45,Reference Ding, Hu and Fu46) . For example, a long-term high-fat diet decreases the expression of tight junction proteins in gut epithelial cells and increases the permeability of the gut mucosa, which fails to act as an integral barrier to migrating pathogens; thus, a large amount of bacterial lipopolysaccharide is released into the blood, activating the low-grade chronic inflammation of islets, and inflammation can lead to the structural damage and dysfunction of pancreatic islet B cells, promote B cell apoptosis and cause insufficient insulin secretion, which in turn triggers GDM(Reference Ding, Hu and Fu46).
However, several limitations of this study should be considered when interpreting the findings of this study. First, the results of this study were based primarily on observational studies, so causality cannot be firmly established. All nine included studies adjusted for confounding factors, but some studies did not adjust for energy intake, gestational weight gain and family history of diabetes, which may be important risk factors for GDM and PE. Most of the included studies were cohort studies, which had the risk of loss to follow-up, but some studies did not describe it. In addition, the substantial heterogeneity across studies was not fully eliminated with subgroup meta-analyses by common study characteristic. Finally, generalisability of our findings to diverse populations should be taken with caution because most of the participants were of Chinese.
We suggest the following recommendations for future studies. Future studies can be carried out in large samples and multi-centre populations, and at the same time, the race, economic status and BMI of the population should be considered in the sampling. The existing studies focused on the association between DII score and risk of GDM and PE, but there were few discussions on the mechanism, which can be studied in the future. Finally, when possible, adequately powered randomised controlled trials should be conducted to support better causal inferences.
Conclusion
In conclusion, this study conducts a systematic review and meta-analysis of nine included studies, and the results show that pro-inflammatory diet estimated by a higher DII score is independently associated with an increased risk of GDM and PE. The finds prompt interventions of GDM and PE patients from the perspective of dietary inflammation. However, more prospective longitudinal studies with improved methodology are warranted to confirm the current findings.
Acknowledgements
This work was supported by Key Foundation of Social Science and Humanities of Ministry of Education of Anhui Province (No.SK2021ZD0030); Key Training Program of School of Nursing, Anhui Medical University (No. hlzd2020003) and The Seedling Cultivation Program of School of Nursing, Anhui Medical University (No. hlqm12023021).
L. H., L. Y. Z. and M. Z. designed this study. M. Z. obtained the funding. L. H. and L. Y. Z. extracted and confirmed the data. L. H. performed the statistical analysis and drafted the initial manuscript. L. H., L. Y. Z., M. Z., J. R. Z., Y. Q. F. and X. Y. Q. made substantial revisions to the manuscript. All authors read and approved the final manuscript.
There are no conflicts of interest.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114523001678
