First-trimester triglyceride-glucose index and birth weight: a retrospective cohort mediation analysis of preterm birth and gestational complications

Abstract

Background

Insulin resistance during pregnancy, while physiologically adaptive to enhance fetal nutrient supply, becomes pathological when excessive, contributing to low birth weight (LBW). The triglyceride-glucose (TyG) index, a biomarker of insulin resistance, predicts gestational complications, but its pathways to birth weight disparities remain unclear. This study investigates whether and to what extent first-trimester TyG index influences birth weight through preterm birth and gestational complications.

Methods

In this retrospective cohort study, 8,605 singleton pregnancies from a Chinese hospital (2015–2021) were analyzed. TyG index was calculated from first-trimester fasting glucose and triglycerides. Outcomes included gestational diabetes mellitus (GDM), hypertension, preeclampsia, preterm birth, LBW, macrosomia, and small/large-for-gestational-age (SGA/LGA). Logistic/multinomial regression assessed associations, followed by causal mediation analysis (R medflex package) to decompose direct/indirect effects. Models adjusted for maternal age, body mass index, education, parity, and diabetes family history.

Results

A 1-standard deviation TyG index increase was associated with elevated risks of gestational complications (i.e., GDM, gestational hypertension, and preeclampsia). Higher TyG index level also showed positive associations with adverse birth outcomes: preterm birth (OR = 1.20, 95% CI: 1.08–1.34), LBW (OR = 1.11, 95% CI: 1.00-1.24), and LGA (OR = 1.12, 95% CI: 1.05–1.21), but not with macrosomia or SGA. Mediation analysis revealed that individual gestational complications mediated 17.7% (GDM), 11.1% (gestational hypertension), and 18.9% (preeclampsia) of the TyG-LBW association, with a joint mediation effect of 37.5%. Preterm birth alone mediated 89.0% of the TyG index-LBW association. When considering all mediators together (preterm birth and gestational complications), the joint natural indirect effect was 1.12 (95% CI 1.05–1.18) with a null natural direct effect being 1.00 (95% CI 0.90–1.11), indicating full mediation. These mediation patterns were primarily evident among women with normal pre-pregnancy weight. Quartile-based comparisons (4th vs. 1st ) yielded similar results.

Conclusion

Our findings highlight a significant association between the first-trimester TyG index and LBW with preterm birth emerging as the primary mediating pathway and gestational complications contributing partially to this relationship. Future research should explore whether interventions aimed at preventing preterm birth and gestational complications can mitigate the LBW risk.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-025-07885-6.

Introduction

Birth weight is an important indicator of fetal and neonatal health. Inadequate or excessive weight at birth affects neonatal survival, childhood development, and long-term metabolic health [1–3]. One of the modifiable factor for fetal growth in the very beginning of gestation is insulin resistance, which is a physiological adaptation to enhance fetal nutrient supply [4]. Paradoxically, excessive resistance is linked to both macrosomia (> 4000g) [5] and low birth weight (LBW, < 2500g) [6]. While macrosomia aligns with hyperinsulinemia-driven overgrowth, the mechanism connecting insulin resistance to LBW remains unclear. A comprehensive understanding of the association between insulin resistance occurring in the first trimester and abnormal birth weight could enable earlier and more targeted interventions.

The triglyceride-glucose (TyG) index, calculated from fasting triglycerides and glucose, is recognized as a reliable surrogate of insulin resistance for its clinical utility in predicting cardiometabolic risks in general population [7–10]. Among pregnant women, the TyG index has been consistently associated with gestational complications. For instance, recent meta-analysis found the TyG index predicted the gestational diabetes mellitus (GDM) risk, with a 2.25-fold higher odds among those in the highest versus lowest TyG index category [11]. Synthesized evidence from observational studies indicated a positive association between the TyG index and gestational hypertension [12]. Several studies also suggested the TyG index was associated with a higher risk of preeclampsia [5, 13, 14]. Of note, these pregnancy complications have been identified as risk factors of abnormal birth weight [15–18]. These findings together suggests a potential mediating role of gestational complications between the TyG index and abnormal birth weight, yet has not been thoroughly investigated.

Besides gestational complications, preterm birth also potentially plays a role in the pathway leading insulin resistance to LBW since infants are more likely to be LBW due to shorter gestational duration. Higher levels of TyG index and gestational complications increase the risk of preterm birth [5, 15, 16]. Identifying these mediating pathways could provide insights into underlying biological mechanisms and potential targets for intervention. Notably, existing analyses often conflate LBW caused by preterm birth with intrauterine growth restriction (e.g., small-for-gestational-age (SGA) infants), obscuring TyG index’s true biological effects. If accelerates parturition is the case, TyG index would barely increase the risk of SGA.

This large retrospective cohort study addresses these gaps by investigating: (1)TyG index’s associations with preterm birth, LBW/macrosomia, and birth weight accounting for gestational age; and (2) the mediating roles of gestational complications and preterm delivery. By stratifying outcomes by gestational age and employing causal mediation analysis, we clarify whether TyG index directly alters fetal growth or primarily accelerates parturition, providing mechanistic insights into insulin resistance’s dual impacts on birth weight.

Methods

Participants

This retrospective study consisted of pregnant women who underwent regular prenatal checkups and delivered at the Third Affiliated Hospital of Sun Yat-Sen University between January 1, 2015 and June 30, 2021. The inclusion criteria were: (1) Chinese Han ethnicity; (2) singleton pregnancy with delivery after 28 weeks of gestation; (3) established prenatal checkup records at the hospital according to standard perinatal management protocols, and (4) delivery at the hospital with complete medical records. The exclusion criteria were: (1) with type 1 or type 2 diabetes, chronic hypertension, chronic kidney disease, or rheumatic immune diseases before or during pregnancy; and (2) pregnancies complicated by congenital reproductive tract malformations, uterine fibroids, or ovarian cysts. Out of 11,477 pregnant women hospitalized and delivered after 28 weeks during the study period, 8,605 women were included in the final analysis after applying the inclusion and exclusion criteria (see Fig. 1 for the detailed flow chart).

Fig. 1.

Flow chart of participants enrollment

This study was carried out according to the principles of the Declaration of Helsinki and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Ethical approval was obtained from the ​​Human Research Ethics Committee of the Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China​​ (Approval No.: [2021]02–266-01). Given the retrospective nature of this cohort study, which posed no risks to participants and maintained strict confidentiality, the institutional review board granted a waiver for ​​informed consent. All data were anonymized and handled in compliance with institutional and national ethical standards.

Data collection

Relevant data was retrieved from the electronic health records. At the first hospital visit (usually 11 to 13⁺⁶ weeks), demographic factors such as age, pre-pregnancy body mass index, education level, family history of diabetes, and parity were recorded. Fasting plasma glucose and triglycerides were also measured at this visit. The TyG index was calculated using the formula: TyG index = Ln (fasting triglycerides (mg/dL) × fasting plasma glucose (mg/dL)/2).

GDM was diagnosed using the 75g oral glucose tolerance test during gestational weeks 24–28, based on the International Association of Diabetes and Pregnancy Study Groups criteria (fasting plasma glucose ≥ 92 mg/dL, 1-hour ≥ 180 mg/dL, or 2-hour ≥ 153 mg/dL). Gestational hypertension was diagnosed as sustained systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg on two occasions at least 4 h apart after 20 weeks of gestation in previously normotensive women. Preeclampsia was defined as gestational hypertension with proteinuria (≥ 0.3 g/24 hours or ≥ 1 + on dipstick). Preterm birth was defined as delivery prior to 37 weeks of gestation. Birth weight was categorized as normal, LBW, or macrosomia. SGA and large for gestational age (LGA) were defined as a birth weight lower than the 10th percentile and higher than the 10th centile, respectively, of our population for a given gestational age and sex, the rest were defined as appropriate-for-gestational-age [19].

Statistical analysis

Characteristics of the participants were depicted as mean (standard deviation, SD) and frequency (percentage, %) for continuous and categorical variables, respectively. Logistic regression models were performed to examine the associations between the TyG index (both as quartiles and per 1- SD increase) and the risks of GDM, gestational hypertension, preeclampsia, and preterm birth. Multinomial logistic regression model was used to test the association of the TyG index with abnormal birth weight (i.e., LBW and macrosomia) and SGA/LGA. Odds ratio (OR) with 95% confidence interval (CI) were reported. Potential confounders, including maternal age, pre-pregnancy BMI, education level, family history of diabetes, and parity, were adjusted for in all models.

Statistically significant associations (total effect) of TyG index (both as per 1- SD increase and quartiles) with preterm birth and abnormal birth weight found by the logistic/multinomial logistic regression were further decomposed into natural direct effect (NDE) and natural indirect effect (NIE) by causal mediation analysis under the counterfactual framework. We first assessed the mediation effect of individual gestational complications between the TyG index (both as per 1- SD increase and quartiles) and each adverse birth outcome, then we estimated the combined mediation effect of GDM, gestational hypertension, and preeclampsia for each adverse birth outcome. We further tested whether and to what extent preterm birth alone and together with three gestational complications mediated the association between TyG index and LBW. These analyses were performed using the medflex package in R, which could handle path-specific mediation effects with multiple mediators, by adopting the imputation-based approach [20]. We assume no exposure-mediator interaction, and confounding between exposure-mediator, exposure-outcome and mediator-outcome has been addressed [21]. For multiple mediators, we assume no mediator-mediator interaction in influencing the outcome. The 95% CI for the NDE and NIE was calculated using nonparametric bootstrapping for 1000 iterations. The proportion mediated (PM) was obtained by the formula: OR^NDE (OR^NIE−1)/(OR^NDEOR^NIE−1) [22].

Missing values primarily occurred in the data collected during the first trimester, specifically for the TyG index. Since the TyG index serves as the independent variable in this study, no imputation was performed. After excluding this part of the data, no missing values were found in other variables.

Sensitivity and subgroup analyses

A series of subgroup and sensitivity analyses were performed to test the robustness of our findings. We first conducted subgroup (i.e., age: <35 or > = 35 years; BMI: underweight or normal or overweight/obese; education level: <=12 or ~ 16 or > 16 years; family history of diabetes: no or yes; and parity (> = 1): no or yes) analyses for the TyG index-gestational complications, TyG index-adverse birth outcomes, and gestational complications-adverse birth outcomes associations with forest plots were presented. Then we evaluated whether the relationship between the TyG index and birth outcomes differed by the exposure-mediator interaction. Specifically, we expanded the causal mediation model to include all two-way interactions between the TyG index and individual mediators. The significance of these interactions was tested using a likelihood ratio test (LRT), comparing the fit of a model with interaction terms to a nested model without interactions. A statistically significant LRT result (P <.05) indicated that mediation effects varied depending on mediator levels, necessitating stratification in subsequent analyses. Thirdly, we used E-values to quantify the robustness of mediation effects to potential unmeasured confounding in the context of rare outcomes in our study (LBW/preterm birth incidence < 15%) [23, 24]. E-values estimate the minimum strength of association that unmeasured confounders would need to have with both the exposure (TyG index) and outcome to fully explain the observed mediation effects. Lastly, to assess effect modification by pre-pregnancy metabolic status, we stratified participants into three subgroups based on pre-pregnancy BMI: underweight (< 18.5 kg/m²), normal weight (18.5–23.9 kg/m²), and overweight/obese (≥ 24 kg/m²). Causal mediation analyses were repeated within each subgroup to evaluate heterogeneity in pathways.

Statistical analyses were conducted using SPSS version 23.0 (SPSS Inc., Chicago, IL) and R 3.5.1, with a two-sided p-value < 0.05 considered statistically significant.

Results

The study included 8,605 pregnant women with a mean (SD) age of 30.0 (4.1) years (range: 17 to 47). Most participants had a normal pre-pregnancy BMI (61.4%), over 12 years of education (83.6%), no family history of diabetes (94.1%), and were parous (51.1%). During pregnancy, 21.1% developed GDM, 3.5% gestational hypertension, and 1.7% preeclampsia. At delivery, 4.4% of births were preterm, 4.4% resulted in LBW, and 2.9% were macrosomic. SGA cases accounted for 5.3% of all births, while LGA made up 11.4% of the total births (Table 1).

Table 1.

Characteristics of participants (N = 8,605)

	No. (%) or Mean ± SD
Age	30.0 ± 4.1
BMI	21.26 ± 3.07
BMI: Categorical
Underweight	1,568 (18.2)
Normal	5,285 (61.4)
Overweight	1,523 (17.7)
Obese	229 (2.7)
Education (year)
≤ 12	1,411 (16.4)
~ 16	3,874 (45.0)
> 16	3,320 (38.6)
Family history of diabetes	504 (5.9)
Parity (> = 1)	4,396 (51.1)
TyG index	8.39 ± 0.40
TyG index: percentile
25th	8.11
50th	8.36
75th	8.63
GDM	1,812 (21.1)
Gestational hypertension	304 (3.5)
Preeclampsia	142 (1.7)
Preterm birth	376 (4.4)
Low birth weight	377 (4.4)
Macrosomia	252 (2.9)
SGA	458 (5.3)
LGA	977 (11.4)

Associations between TyG index and gestational complications

A 1-SD increase in the TyG index was associated with higher risks of GDM (OR = 1.35, 95% CI: 1.28–1.43), gestational hypertension (OR = 1.22, 95% CI: 1.09–1.37), and preeclampsia (OR = 1.26, 95% CI: 1.07–1.49). Compared with the lowest quartile of TyG index, women in the highest quartile had significantly elevated odds of GDM (OR = 2.03, 95% CI: 1.73–2.39), gestational hypertension (OR = 1.99, 95% CI: 1.33–2.96), and preeclampsia (OR = 1.91, 95% CI: 1.12–3.26) (Table 2). Subgroup analysis found that educational level moderated the association between the TyG index and gestational hypertension. Compared 2nd to 4th TyG index quartiles to 1 st quartile, ORs were 2.65 to 3.85 among women with > 16 years of education, whereas ORs were 0.62 to 1.11 among those with ≤ 12 years of education, interaction P =.015–0.042 (Figure S1).

Table 2.

Association between TyG index and gestational complications^a

	GDM	Gestational hypertension	Preeclampsia
TyG index: 1SD-increase	1.35 (1.28–1.43)***	1.22 (1.09–1.37)**	1.26 (1.07–1.49)**
TyG index: Quartile
Quartile 1	Ref.	Ref.	Ref.
Quartile 2	1.11 (0.94–1.31)	1.35 (0.88–2.07)	0.95 (0.52–1.74)
Quartile 3	1.38 (1.17–1.63)***	1.80 (1.20–2.69)**	1.50 (0.87–2.59)
Quartile 4	2.03 (1.73–2.39)***	1.99 (1.33–2.96)**	2.74 (1.91–3.93)***

a: adjusted for maternal age, BMI, education level, family history of diabetes, and parity. ***: P < 0.001, **: P <.01, *: P <.05, #: P <.10

TyG Index, gestational complications and birth outcomes

A 1-SD increase in TyG index was positively associated with preterm birth (OR = 1.20, 95% CI: 1.08–1.34), LBW (OR = 1.11, 95% CI: 1.00-1.24), and LGA (OR = 1.12, 95% CI: 1.05–1.21). The highest TyG index quartile was associated with increased odds of preterm birth (OR = 1.71, 95% CI: 1.25–2.30), LBW (OR = 1.42, 95% CI: 1.05–1.94), and LGA (OR = 1.29, 95% CI: 1.05–1.57) compared to the lowest quartile. GDM was related to higher risks of preterm birth (OR = 1.76, 95% CI: 1.41–2.19) and LBW (OR = 1.53, 95% CI: 1.22–1.93), but a lower risk of LGA (OR = 0.83, 95% CI: 0.72–0.96). Gestational hypertension and preeclampsia independently associated with higher risks of preterm birth (ORs = 1.89–6.84), LBW (ORs = 2.98–8.06) and SGA (ORs = 2.74–5.53).

Subgroup analysis identified pre-pregnancy BMI as a moderator for three associations: (1) a 1-SD TyG index increase and macrosomia (P for interaction = 0.011, Figure S2): normal-weight women (OR = 1.20, 95% CI: 1.00-1.43) vs. underweight women (OR = 0.65, 95% CI: 0.41–1.04); (2) GDM and preterm birth (P for interaction = 0.019, Figure S4): normal-weight women (OR = 2.10, 95% CI: 1.57–2.82) vs. underweight women (OR = 0.65, 95% CI: 0.29–1.50); and (3) Preeclampsia and SGA (P for interaction = 0.013, Figure S5): Overweight/obese women (OR = 7.70, 95% CI: 4.24–13.96) vs. normal-weight women (OR = 3.91, 95% CI: 1.94–7.86).

Causal mediation analysis

As shown in Table 3, the total effects of the TyG index on LBW and LGA were statistically significant which meets the prerequisites for conducting mediation analyses. However, the associations between gestational complications and LGA showed opposite direction to the total effect, mediation analyses were thus focused on the TyG-LBW pathway. For a 1-SD TyG index increase, individual gestational complications mediated 17.7% (GDM), 11.1% (gestational hypertension), and 18.9% (preeclampsia) of the TyG-LBW association, with a joint mediation effect of 37.5%. Preterm birth alone mediated 89.0% of the TyG index-LBW association. When combined with GDM, gestational hypertension, and preeclampsia, the joint NIE was 1.12 (95% CI: 1.05–1.18), with an NDE of 1.00 (95% CI: 0.90–1.11), indicating full mediation effect. In quartile comparisons (4th vs. 1 st), the PMs for individual gestational complications were 14.2%, 9.5%, and 14.5% for GDM, gestational hypertension, and preeclampsia, respectively. Together, these complications mediated 29.8% of the association. Preterm birth individually accounted for 76.4% of the TyG index-LBW association, increasing to 86.5% when combined with other three gestational complications (Table 4).

Table 3.

Associations of TyG index and gestational complications with adverse birth outcomes^a

	Preterm birth	Low birth weight	Macrosomia	SGA	LGA
TyG index: 1SD-increase	1.20 (1.08–1.34)**	1.11 (1.00-1.24)#	1.08 (0.95–1.23)	0.95 (0.86–1.06)	1.12 (1.05–1.21)**
TyG index: Quartile
Quartile 1	Ref.	Ref.	Ref.	Ref.	Ref.
Quartile 2	0.99 (0.71–1.38)	1.03 (0.76–1.41)	0.93 (0.64–1.35)	0.90 (0.69–1.17)	1.13 (0.93–1.38)
Quartile 3	1.26 (0.92–1.73)	1.07 (0.78–1.45)	0.72 (0.49–1.07)	0.96 (0.73–1.24)	1.15 (0.94–1.40)
Quartile 4	1.71 (1.25–2.30)**	1.42 (1.05–1.94)*	0.99 (0.68–1.43)	0.86 (0.65–1.15)	1.29 (1.05–1.57)*
GDM	1.76 (1.41–2.19)***	1.53 (1.22–1.93)***	0.92 (0.68–1.24)	1.13 (0.91–1.42)	0.83 (0.72–0.96)*
Gestational hypertension	1.89 (1.24–2.89)**	2.98 (2.02-4.39)***	0.75 (0.37–1.51)	2.74 (1.91–3.93)***	0.82 (0.55–1.24)
Preeclampsia	6.84 (4.58–10.22)***	8.06 (5.37–12.09)***	0.89 (0.32–2.48)	5.53 (3.68–8.30)***	0.64 (0.32–1.26)

a: adjusted for maternal age, BMI, education level, family history of diabetes, and parity. ***: P <0.001, **: P <.01, *: P <.05, #: P<.10

Table 4.

Estimates from causal mediation analysis for the associations of TyG index with low birth weight and preterm birth (OR (95%CI))

TyG index	Mediator	LBW			Preterm birth
		NDE	NIE	PM (%)	NDE	NIE	PM (%)
1SD-increase	Individual mediator
	1. GDM	1.09 (0.98–1.22)	1.02 (1.01–1.03) **	17.7	1.18 (1.07–1.31) **	1.02 (1.01–1.04) **	13.8
	2. Gestational hypertension	1.10 (0.98–1.24)	1.01 (1.00-1.02) *	11.1	1.20 (1.09–1.33) ***	1.01 (1.00-1.01) #	3.4
	3. Preeclampsia	1.09 (0.98–1.23)	1.02 (1.00-1.04) **	18.9	1.19 (1.07–1.33) **	1.02 (1.00-1.03) *	9.5
	4. Preterm birth	1.01 (0.91–1.12)	1.10 (1.04–1.16) ***	89.0	/	/	/
	Multiple mediators a	1.07 (0.96–1.20)	1.04 (1.02–1.06) ***	37.5	1.16 (1.04–1.29) **	1.04 (1.02–1.07) ***	23.3
	Multiple mediators b	1.00 (0.90–1.11)	1.12 (1.05–1.18) ***	100.0	/	/	/
4th vs. 1st Quartile	Individual mediator
	1. GDM	1.04 (0.99–1.91) #	1.37 (1.01–1.08) **	14.2	1.63 (1.20–2.20) **	1.06 (1.02–1.09) **	12.9
	2. Gestational hypertension	1.39 (1.02–1.88) *	1.03 (1.00-1.06) *	9.5	1.70 (1.25–2.31) ***	1.01 (1.00-1.03) #	3.5
	3. Preeclampsia	1.37 (0.99–1.88) #	1.05 (1.01–1.09) *	14.5	1.66 (1.20–2.26) **	1.04 (1.00-1.08) *	9.2
	4. Preterm birth	1.10 (0.83–1.44)	1.30 (1.12–1.51) ***	76.4	/	/	/
	Multiple mediators a	1.30 (0.96–1.78) #	1.10 (1.04–1.16) ***	29.8	1.57 (1.14–2.14) **	1.10 (1.04–1.16) ***	21.9
	Multiple mediatorsb	1.07 (0.79–1.41)	1.34 (1.15–1.60) ***	86.5	/	/	/

NDE nature direct effect, NIE nature indirect effect, PM proportion of mediation. a: mediators including GDM, gestational hypertension and preeclampsia. b: mediators in a and preterm birth: ***: P <.001, **: P <.01, *: P <.05, #: P <.10

For the TyG index-preterm birth association, GDM mediated 13.8%, gestational hypertension 3.4%, and preeclampsia 9.5% of the effect (joint 23.3%) per 1-SD TyG increase. In quartile comparisons, the PMs were 12.9% for GDM, 3.5% for gestational hypertension, and 9.2% for preeclampsia, with a combined PM of 21.9% (Table 4).

Sensitivity and subgroup analyses of causal mediation effects

Inclusion of TyG-mediator interaction terms did not improve model fit (all likelihood ratio test P values > 0.05; Table S1). E-value analyses revealed that quartile-based comparisons demonstrated higher E-values than per 1-SD TyG index increase, indicating greater stability of mediation effects existing between extreme high TyG index levels and LBW. Generally, preterm birth exhibited stronger robustness (NIE E-values: 1.43–1.92), requiring substantial unmeasured confounding to fully explain its mediation effects. Notably, GDM demonstrated greater stability in mediating the TyG index-LBW association in quartile comparison (NIE E-value = 2.08; Table S2).

Subgroup analyses by pre-pregnancy BMI revealed that the mediation patterns observed in the primary analysis were only significant among normal-weight women. Among overweight or obese women, two significant indirect paths involving preterm birth were identified; however, the PMs were not calculated due to the opposing directions between the NDE and NIE. (Table S3-S5)

Discussion

This study explores the paradoxical relationship between insulin resistance, a physiological adaptation to support fetal growth, and LBW. Our findings demonstrate that elevated first-trimester TyG index levels were associated with increased risks of gestational complications, preterm birth, and LBW. Mediation analyses revealed that preterm birth mediated approximately 80% of the TyG index-LBW association, while gestational complications collectively accounted for 37.5% of this pathway. The TyG index was positively correlated with LGA infants but showed no significant association with SGA, suggesting that LBW in this context may primarily result from preterm delivery rather than impaired fetal growth.

The dominant mediation role of preterm birth in the TyG index-LBW pathway highlights insulin resistance’s effects on destabilizing placental and uterine homeostasis. Elevated TyG index likely exacerbates oxidative stress and lipid peroxidation [25, 26], processes implicated in premature membrane rupture and inflammatory cascades [27, 28] that trigger labor. Such gestation truncation effect was partially driven by gestational complications. In this study, GDM, gestational hypertension, and preeclampsia jointly mediated nearly 20.0% of the TyG index-preterm birth association.

The modest mediation proportions suggest additional mechanisms beyond overt gestational complications, such as subclinical molecular responses to insulin resistance (e.g., chronic inflammation, adipokine dysregulation) [25, 27, 29], which may impair placental function and fetal development. The absence of a direct TyG index-SGA association underscores that LBW stems primarily from preterm birth rather than fetal growth restriction. Insulin resistance likely disrupts gestational longevity via placental inflammation and oxidative stress, triggering preterm labor, whereas SGA involves distinct mechanisms (e.g., placental insufficiency, nutrient deprivation) [30]. These findings advocate for proactive management of insulin resistance even in pregnancies without diagnosed complications and highlight the need for research into unexplored pathways linking metabolic dysregulation to preterm birth.

Gestational complications, though secondary mediators, provide critical insights into TyG index’s multifactorial impacts. GDM, hypertension, and preeclampsia collectively mediated 37.5% of TyG index’s effect on LBW, aligning with their known roles in placental ischemia and nutrient transfer impairment [31, 32]. However, the modest mediation percentage emphasizes that TyG index’s primary pathway to LBW operates independently of these conditions. Notably, GDM’s paradoxical reduction of LGA risk (OR = 0.83) likely reflects intensified glycemic management in diagnosed cases, illustrating how clinical interventions can modify TyG index’s downstream effects. These findings collectively illustrate that TyG index’s perinatal impacts are not uniform but depend on the interplay between metabolic severity, mediator pathways, and clinical countermeasures.

Notably, while gestational complications mediated the TyG index-outcome relationships in a small proportion, their independent associations with preterm birth and LBW were striking. Preeclampsia, for instance, demonstrated a 6.84-fold higher odds of preterm birth (95% CI: 4.58–10.22) and a 7.97-fold increased risk of LBW (95% CI: 5.36–11.84). ​​Subgroup analysis according to pre-pregnancy BMI category further highlighted that GDM’s association with preterm birth was markedly stronger among normal-weight women compared to those underweight. Similarly, preeclampsia-associated SGA risk was disproportionately elevated among overweight/obese women versus normal-weight women. These robust effects underscore the critical need to prevent gestational complications and reinforce the need for personalized prenatal care that considers metabolic heterogeneity.

The TyG index’s positive association with GDM aligns with insulin resistance’s established role in GDM pathogenesis [33]. Our findings extend this understanding by demonstrating that insulin resistance independently predicts hypertensive disorders and preeclampsia [6, 13, 14]. Mechanistically, compromised insulin signaling and chronic nutrient excess disrupt endothelial function, promoting vasoconstriction, reduced nitric oxide availability, and oxidative stress [25]. Concurrent systemic inflammation and inflammatory adipokine release [27], coupled with oxidative damage to the endothelium [29], likely exacerbate maternal vascular dysfunction, increasing susceptibility to hypertension and preeclampsia. Interestingly, our subgroup analysis found the association between TyG index and gestational hypertension was stronger among higher-educated women than those less educated, which could be partially explained by the component cause model that posits diseases arise from the combined effects of multiple risk factors [34]. In this context, gestational hypertension likely results from a “sufficient cause” comprising insulin resistance (reflected by elevated TyG index) and other contributory factors such as obesity, chronic inflammation, or endothelial dysfunction [35]. Women with higher education levels may exhibit fewer competing risk factors for gestational hypertension (e.g., lower smoking rates, healthier dietary patterns, or better access to prenatal care), allowing the TyG index to emerge as a more prominent driver of hypertensive risk. Conversely, in less-educated women, the cumulative burden of socioeconomic stressors (e.g., limited healthcare access, poorer nutrition, or higher rates of obesity) may dominate the pathway to gestational hypertension, diluting the observable contribution of TyG index. This masking effect parallels findings in dementia research, where alcohol’s impact was more evident in women due to their lower prevalence of other risk factors (e.g., smoking) [36].

The subgroup analyses by pre-pregnancy BMI revealed that insulin resistance’s (TyG index) mediation pathways are influenced by metabolic context: in normal-weight women, gestational complications and preterm birth jointly mediated about 70% of the TyG index-LBW association, highlighting insulin resistance as the primary driver in metabolically uncomplicated pregnancies. By contrast, overweight/obese women exhibited no significant mediation by gestational complications but showed marginal preterm birth mediation, suggesting competing pathways (e.g., chronic inflammation or adipokine dysregulation) surpass TyG index’s effects, potentially compounded by smaller incident cases in subgroup limiting statistical power. Similarly, underweight women demonstrated null mediation, likely due to attenuated TyG index variability, nutritional deficits overriding metabolic influences, and fewer cases reducing the statistical power to detect associations. These findings underscore that TyG index’s perinatal impacts depend on baseline metabolic health and subgroup sample representativeness, advocating for BMI-stratified interventions.

Our findings highlight actionable strategies for reducing adverse pregnancy outcomes at the population level. First, the TyG index, which is calculable early in pregnancy, could serve as a scalable tool for population-wide risk stratification, enabling targeted prenatal interventions (e.g., intensified monitoring for preterm birth) in high-TyG index subgroups. Second, the dominance of preterm birth as a mediator underscores the need for public health initiatives to prioritize gestational longevity, such as community-based programs promoting metabolic health through lifestyle modifications or pharmacological therapies [37–39], to mitigate LBW incidence. Additionally, the modest but significant mediation by gestational complications reinforces the importance of universal screening for hypertension and GDM, even in metabolically healthy women. ​​The interaction between TyG index and education level further suggests that socioeconomic factors should inform risk stratification, as higher-educated women may require tailored counseling on mitigating insulin resistance’s vascular effects.​​ Finally, the limited mediation proportions by gestational complications and residual TyG index effects highlight critical gaps in understanding insulin resistance’s subclinical pathways, urging longitudinal studies to explore biomarkers of placental aging and inflammation.

Strengths and limitations

The strengths of this study include its large sample size and retrospective design, which allowed us to examine the temporal relationships between the TyG index and adverse pregnancy outcomes. However, the study also has several limitations. First, the study was conducted in a single hospital, which may limit the generalizability of the findings. Second, the study did not collect data on other potential confounding variables, such as maternal nutrition and physical activity [40, 41]. While E-value analyses indicated robustness against unmeasured confounding, residual confounding remains possible. Third, the TyG index was calculated only in the first trimester; serial measurements across pregnancy could have provided deeper insight into its dynamic relationship with complications like GDM or preeclampsia. Fourth, while mediation analyses suggested plausible pathways (i.e., preterm birth accounting for 76–89% of the TyG-LBW association), the observational design precludes causal inference, and unmeasured mediators (i.e., placental dysfunction biomarkers) may play additional roles. Future studies should aim to address these limitations and provide further insights into the relationship between the TyG index and adverse pregnancy outcomes.

Conclusions

Our findings highlight a significant association between the first-trimester TyG index and LBW with preterm birth emerging as the primary mediating pathway and gestational complications contributing partially to this relationship. While the TyG index may aid in early identification of high-risk pregnancies, further validation is needed to assess its clinical utility and whether targeted interventions can improve outcomes.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

LBW

Low birth weight

TyG index

Triglyceride-glucose index

GDM

Gestational diabetes mellitus

SGA

Small-for-gestational-age

LGA

Large-for-gestational-age

SD

Standard deviation

OR

Odds ratio

CI

Confidence interval

NDE

Natural direct effect

NIE

Natural indirect effect

PM

Proportion mediated

LRT

Likelihood ratio test

Authors’ contributions

Jinhui Cui and Yanling Wang conceived the study. Hui Jiang, Fei Huang, Mengjun Xie, Ziyi Cui, Xinyuan Chen, Liping OUYang, Ping Li were involved in designing the study and developing the methods. Hui Jiang conducted the statistical analyses. Ping Li and Yanling Wang conducted the content analyses. Jinhui Cui and Hui Jiang drafted the manuscript. All authors critically revised and approved the manuscript.

Funding

This study was supported by research grants from National Natural Science Foundation of China (No. 82072218), and Natural Science Foundation of Guangdong Province (No. 2023A1515010288).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the ethics committee of the Third Affiliated Hospital of Sun Yat-Sen University (serial number: [2021]02–266-01).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.