Gestational Diabetes Mellitus and Heart Failure

Abstract

Background

The extent of the association between gestational diabetes mellitus (GDM) and the incidence of heart failure (HF) largely remains unclear.

Objectives

The aim of the study was to synthesize the evidence on the association of GDM and risk of HF.

Methods

This study is a systematic review and meta-analysis. We searched PubMed and Embase through July 24, 2024, for cohort studies reporting on the GDM and HF association. We pooled adjusted relative risk (RR) estimates of the association of GDM and HF using a random-effects model meta-analysis.

Results

In a meta-analysis of 8 observational studies, a total of 6,371,877 participants (weighted averages—age: 28.7 years, 89.7% White, body mass index 25.6 kg/m², 310,351 with GDM) were assessed and experienced 12,409 incident HF events over ∼8.6 years (weighted average). The pooled adjusted RR for the GDM and HF association was 1.54 (95% CI: 1.24-1.92). There was heterogeneity across the studies (I² = 86.9%, P < 0.001). Sensitivity analyses, excluding the smallest and largest studies, did not appreciably change the significance and magnitude of the overall RR estimate of the risk of HF related to GDM.

Conclusions

The observed independent association of GDM with HF suggests a potential causal role of GDM in adverse myocardial remodeling. A history of GDM should be considered as a risk factor in the efforts to prevent HF.

Central Illustration

Gestational diabetes mellitus (GDM), defined as hyperglycemia that first develops during pregnancy and resolves after birth, is a common complication of pregnancy.¹ GDM is increasingly recognized as an independent risk factor for cardiovascular disease (CVD) in the years after pregnancy,² including heart failure (HF).³ Furthermore, imaging-based studies have shown intra- and post-partum (short- and long-term) myocardial alterations related to GDM^4,5 Mechanistic or laboratory-based studies have described pathogenic features linking GDM and cardiac dysfunction.^6,7

Knowledge of the extent of the association between GDM and the risk of HF can help design preventative strategies among young women. This is important, as recent data suggest stagnation in improvements in heart disease mortality in women, specifically in younger age groups (<55 years), where rates are increasing, possibly because of pregnancy-related factors, among which GDM.⁸

While accruing evidence from population-based studies suggests an independent association between GDM and HF, the magnitude of this association remains largely unclear. The available data have not been adequately synthesized to provide a better understanding of the extent of HF risk associated with GDM. Therefore, we conducted a systematic review and meta-analysis to evaluate the association between GDM and HF.

Methods

Data sources and study selection

We searched PubMed and Embase from inception up to July 24, 2024, using a combination of terms related to GDM and HF, with a restriction to English-language papers. We included cohort studies of the relation between GDM and incident HF (Supplemental Figure 1).

We also manually scanned the reference lists of identified studies and screened the citing references through the ISI Web of Knowledge database for possible additional eligible studies.

We used electronic search using key search terms related to GDM, which are shown in the Supplemental Appendix. The inclusion criteria included the following: 1) patients with a diagnosis of GDM; 2) retrospective or prospective cohort studies with HF as an outcome; and 3) available data on the risk estimate for HF.

Data extraction

Two investigators (E.B. and S.E.) independently abstracted data from eligible studies on study characteristics (setting, period, and design), participant characteristics (demographics [eg, age and parity] and clinical variables [eg, body mass index, history of hypertension, history of pre-eclampsia, and history of diabetes]), duration of follow-up, HF outcome definition, and the relative risk (RR) for the incidence of HF. For each study, we abstracted the minimally and maximally adjusted RR estimates of the GDM and HF association and the adjustment variables.

We resolved discrepancies by consensus, adjudicated by a third reviewer (J.E.T.).

Quality of reporting and risk of bias

We assessed the risk of bias in included studies using the Newcastle-Ottawa Quality Assessment Scale for cohort studies (Supplemental Table 2).⁹ This scale is derived by assigning points to 3 aspects of study design with a maximum total of 10 points: selection of study participants (maximum 5 points), comparability of study groups (maximum 2 points), and ascertainment of the outcome of interest (maximum 3 points). The 3 components of the scoring system related to assessment of the nonexposed cohort and its comparability with the exposed cohort were not relevant to the studies assessing performance of risk scores, making the maximum possible total points 7 (instead of 10) in this review.

Statistical analyses

The characteristics of studies and participants, such as number of participants, age, percent of males, and percent of White, were combined across the studies by calculating the weighted means. The characteristics are presented in tables and summarized as the range of values. We report the weighted average of these study-level characteristics weighted by the appropriate denominators (N).s

We pooled the RR estimate of the GDM and incident HF association across studies using fixed-effects model meta-analysis to provide a single summary estimate after showing a low heterogeneity across studies using the I² statistic (I² >75% indicates high heterogeneity). Given the small number of studies, we did not perform subgroup analyses. The HF event rates were similarly combined using a random-effects meta-analysis model. We performed influence analyses to determine the impact of leverage points on our pooled estimates. We assessed publication bias using the Egger regression test P value and funnel-plot asymmetry (Supplemental Figure 2).¹⁰

Analyses were conducted using Stata version 15 (StataCorp).

We reported the findings according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline.¹¹

This study was based on previously published studies; thus, it did not require an approval by an institutional review board. No written consent was needed for this study because it was based on secondary data that were anonymous.

Results

Characteristics of studies

We included 8 eligible longitudinal studies in our systematic review (Table 1, Supplemental Table 1).^3,12, 13, 14, 15, 16, 17, 18 Across these studies, there were 6,371,877 participants, with an average age of 27 to 54.5 years (weighted average 28.7 years) and 310,351 cases of GDM. In all 8 studies, GDM was defined based on the International Classification of Diseases (ICD) codes. In addition to the ICD codes, only 1 study directly leveraged data on the 50-g glucose challenge test and 75-g oral glucose tolerance test³ for GDM ascertainment, and 1 study defined GDM based on either ICD codes or self-report.¹⁵ The ascertainment of the HF outcome was based on ICD codes for HF hospitalizations in all the studies.

Table 1.

Characteristics of Included Studies

First Author, Year of Publication	Data Source	Country	Study Years	Sample Size	White, %	Average BMI, kg/m2	Average Age, y	Average Follow-Up, y	GDM Cases	HF Identification	Total HF Cases	GDM HF Cases	Adjusted Variables
Savitz et al, 201412	Linkage of all NYC hospitals’ discharge and birth certificate data	USA	1995-2004	849,639	NR	NR	NR	0.5	NR	ICD discharge codes	259	NR	calendar year, maternal age, race/ethnicity, health insurance, gestational HTN, pre-eclampsia, GDM, parity, education, smoking, prenatal care, and pre-pregnancy weight
McKenzie-Sampson et al, 201813	Quebec’s Maintenance and Use of Data for the Study of Hospital Clientele Registry.	Canada	1989-2013	1,108,541	NR	NR	27.0	NR	67,356	ICD discharge code	1,430	151	age at baseline, parity, time period, socioeconomic deprivation, pre-eclampsia
Echouffo et al, 20213	Linkage of Medical Registries in Ontario Province	Canada	2007-2018	906,319	89.5	NR	29.7	7.0	50,193	Hospitalization records	2,248	26	age, ethnicity, neighborhood income quintile, rurality, parity, preterm delivery, pregestational HTN, pre-eclampsia, and CKD
Sun et al, 202114	National Health Insurance Service	Korea	2005-2015	1,500,168	NR	NR	27.0	12.8	159,066	ICD discharge codes	2,705	338	age, parity, household income, preeclampsia or HTN history, PCOS, and dyslipidemia
aLee et al, 202215	UK Biobank	UK	2006-2010	219,330	88.6	28.2	54.5	10.3	1,390	ICD discharge codes	2,248	26	age, race, BMI, smoking, alcohol consumption, pre-existing (HTN, DM, hypercholesterolemia), meds (aspirin, antihypertensives, cholesterol-lowering agents), early menopause, and hysterectomy
Christensen et al, 202216	Linkage of nationwide medical registries	Denmark	1997-2018	700,648	90.9	23.0	28.0	11.1	23,274	ICD discharge codes	977	59	age, parity, pre-existing PCOS/hirsutism, HTN, dyslipidemia, ethnicity, marital status, income, education, occupation, calendar year of delivery, and Charlson Comorbidity Index.
Yu et al, 202217	Linkage of nationwide medical registries	Denmark	1978-2016	1,002,486	NR	NR	27.0	16.2	3,015	ICD discharge codes	3,888	84	historical time period at first delivery (≤1980, 5-year intervals from 1981-2016), age at first delivery, parity, education, smoking during pregnancy, cohabitation, residence, pre-pregnancy obesity, country of origin, maternal CVD history, and paternal CVD history
Ackerman-Banks et al, 202418	Maine, Insurance Claims Database	USA	2007-2019	84,746	NR	NR	27.0	2.0	6,057	ICD discharge codes	139	7	maternal age at delivery, pre-pregnancy depression, pre-pregnancy HTN, obesity, smoking, gravidity, year of delivery, Medicaid coverage during pregnancy, county-level measures, zip code-level measures, HDP, and prenatal depression
Pooled estimate	-	-	-	6,371,877	89.7	25.6	28.7	8.6	310,351	-	12,409	665

BMI = body mass index; CKD = chronic kidney disease; CVD = cardiovascular disease; GDM = gestational diabetes mellitus; HDP = hypertensive disorders of pregnancy; HTN = hypertension; HF = heart failure; ICD = International Classification of Diseases; NR = not reported; NYC = New York City; PCOS = polycystic ovarian syndrome.

^aLee et al, 2022⁸ analyzed data on women with and without a history of GDM and reported prevalent demographic and clinical data at the time of study enrollment (age 40-69 years).

Gestational diabetes and heart failure

Over an average follow-up of 0.5 to 16.2 years (weighted average: 8.6 years), there were 12,409 HF cases, of which 665 occurred in the GDM group (Table 1 and Central Illustration).

Central Illustration.

Investigation of the Relation of Gestational Diabetes and Incidence of Heart Failure

In terms of absolute risk, in the 5 studies that reported HF incidence rates, over an average follow-up of 1.7 to 41 years, the crude incidence rate of HF events among participants with GDM was 3.54 per 10,000 person-years vs 1.87 per 10,000 person-years among those without GDM.

Across all of the 8 studies included, the pooled maximally adjusted RR of HF in GDM vs non-GDM participants was 1.54 (95% CI: 1.24, 1.920) (Central Illustration, Figure 1). There was significant heterogeneity (I² = 86.9%, P < 0.001) across studies. Egger regression test P values for funnel-plot asymmetry for both pooled unadjusted (P = 0.392) and adjusted models (P = 0.787) were not significant.

Figure 1.

Overall Relative Risk of Heart Failure Associated With Gestational Diabetes Mellitus

GDM = gestational diabetes mellitus; NR = not reported; RR = relative risk.

In influence analyses, the pooled maximally adjusted RR for HF was not different when we excluded the smallest study that reported protective associations between GDM and HF,¹⁸ with an overall RR estimate of 1.52 (95% CI: 1.42-1.64). However, when we excluded the study with the largest sample size and GDM cases,¹⁴ there was an increase in the magnitude of the pooled adjusted RR for HF in persons with GDM compared to controls: 1.74 (95% CI: 1.58-1.91). The exclusion of no other single study significantly changed the pooled estimates.

Of note, only 1 study reported on the relation between GDM and peripartum cardiomyopathy, with an estimate of association (OR) being 1.83 (95% CI: 1.45-2.33).³

We explored sources of heterogeneity by performing meta-regression on study-level characteristics, including duration of follow-up, region of the study, and the extent of adjustment for confounders. These variables explained between 3% and 33% of between-study variation in the meta-regression. In subgroup analyses with stratification of studies by duration of follow-up, the pooled HF risk estimate from studies with longer follow-up (>8 years [median of follow-up across studies]–RR: 1.75; 95% CI: 1.30-2.35) was higher than that of studies with a shorter follow-up (<8 years–RR: 1.19; 95% CI: 0.81-1.75). Regarding the extent of adjustment, the studies that accounted for postpartum diabetes and/or postpartum hypertension had a lower pooled estimate of HF risk (RR: 1.49; 95% CI: 1.17-1.89) compared to studies that did not account for these postpartum conditions (RR: 1.54; 95% CI: 1.01-2.34). The pooled estimate of HF risk in studies conducted in North America (RR: 1.39; 95% CI: 1.01-193, 25) was lower than that from studies conducted in other parts of the world (RR: 1.69; 95% CI: 1.19-2.42).

Discussion

Overall, the present study demonstrates that among more than 6 million individuals, GDM was independently associated with a high absolute risk of HF and at least a 50% higher RR of HF after accounting for traditional risk factors. This observation suggests that optimizing cardiometabolic health before, during, and after pregnancy represents an extraordinary opportunity to modify the natural history of HF and can be leveraged to better inform efforts to prevent HF.

Our meta-analysis complements and extends prior studies, including reviews on GDM and HF,¹⁹ by providing more robust and more contemporary estimates of HF risk associated with GDM from a large number of women with and without GDM. The results are important because they reinforce the importance of GDM as a potential screening and therapeutic target to achieve HF prevention. The effect of GDM on incident HF may be mediated by its effect on HF precursors, including cardiovascular risk factors (among which are postpartum type 2 diabetes mellitus [T2DM] or hypertension). Of note, in the years prior to pregnancy, accruing evidence points to disparities in cardiovascular risk factors between women who will develop future GDM and their non-GDM counterparts. Consequently, after pregnancy, these pre-existing disparities in cardiovascular risk factors demonstrate divergently amplified trajectories.²⁰ Thus, pregnancy, a potent cardiometabolic risk factor, aggravates the cardiovascular risk of women who had a pre-existing propensity to developing GDM. However, the persistence of the GDM and HF association after adjustment for these well-known cardiovascular risk factors suggests an intrinsic GDM effect on the myocardium.

The possible mechanistic pathways linking GDM and adverse cardiac remodeling are incompletely understood. GDM can induce endothelial dysfunction and microvascular damage,²¹ a precursor of HF, with other putative pathways linking GDM and myocardial remodeling, including abnormal calcium handling⁶ and altered myocardial energetics.⁷ Other possible pathways are similar to that observed with T2DM, including glucotoxicity, lipotoxicity, and hyperinsulinemia on the myocardium; increased myocardial deposition of advanced glycation end products and systemic oxidative stress leading to myocardial injury and fibrosis; and mitochondrial dysfunction.²² Moreover, concomitant risk factors, which can be present among women with GDM, such as obesity,²³ dyslipidemia²⁴ and elevated blood pressure,²⁵ may also act synergistically to contribute to adverse cardiac remodeling, the substrate for future HF.

Our findings point to the need for additional investigations to assess the role of GDM in HF risk stratification and whether interventions to address GDM during pregnancy and postpartum can become part of our armamentarium for preventing future HF. Such interventions may include pharmacotherapies, as well as pre- and post-partum lifestyle changes to alter the natural history of GDM. In addition, our findings support the American College of Obstetricians and Gynecologists recommendations for ongoing comprehensive individualized postpartum care, including a CVD risk assessment and long-term CVD prevention in women with GDM.²⁶

Our study has strengths, including a comprehensive examination of GDM and HF risk across different studies, which improved the statistical power to detect smaller effects, as well as the consideration of a wide spectrum of HF outcomes, including postpartum cardiomyopathy and long-term HF.

Study Limitations

Our meta-analysis had limitations. The diagnostic criteria for GDM and HF and the extent of adjustment for potential confounders varied across studies, which can account for differential HF risk estimates. The GDM and HF ascertainment was based on ICD codes or self-report, hence the possibility of underestimating the extent of the associations. We did not have data on oral glucose tolerance testing and criteria for diagnosing GDM, which may vary from 1 setting to the other,^1,27 and explain the heterogeneity in the GDM-related HF risk estimates. The variable extent of adjustment for confounders across studies may have led to overestimation of the HF risk, especially given the possibility of residual confounding. Of note, the studies did not always account for intercurrent postpartum T2DM or coronary heart disease. The studies were mainly based in North America or Europe, thus mainly including White individuals, whereas GDM is more frequent in minoritized populations. Furthermore, the studies did not include the HF subtypes (HF with preserved ejection fraction and HF with reduced ejection fraction), and only 1 study distinguished between peripartum cardiomyopathy and long-term HF. Future studies of GDM and incident HF will need to examine the HF subtypes. The number and design of the studies limited our ability to conduct relevant subgroup analyses by race/ethnicity, gestational hypertension status, intercurrent diabetes or hypertension, body mass index, or other comorbidities.

Conclusions

Our data show a significantly higher HF risk among individuals with GDM compared to those without a GDM, independent of traditional cardiovascular risk factors. Further studies are needed to better define the observed link, including with HF subtypes, and the potential clinical and therapeutic implications of these findings. A history of GDM should most probably be considered as a risk factor in the efforts to prevent HF.

Perspective.

COMPETENCY IN MEDICAL KNOWLEDGE: The extent of the association between gestational diabetes and the risk of heart failure is largely unknown. In a systematic review and meta-analysis of 8 observational studies involving 6,371,877 individuals (among whom 310,351 cases had gestational diabetes), gestational diabetes was associated with a 54% higher relative risk of heart failure compared to non-gestational diabetes.

TRANSLATIONAL OUTLOOK: The results of this study support the potential utility of monitoring cardiac function in women with a history of gestational diabetes in routine clinical practice for preventing heart failure.

Funding support and author disclosures

Dr Echouffo-Tcheugui was supported by NIH/NHLBI grant K23 HL153774. Dr Erqou is supported by the Department of Veterans Affairs, Veterans Health Administration, and VISN 1 Career Development Award. Dr Retnakaran holds the Boehringer Ingelheim Chair in Beta-Cell Preservation, Function and Regeneration at Mount Sinai Hospital (Toronto, Ontario, Canada), and his research program is supported by the Sun Life Financial Program to Prevent Diabetes in Women. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.