Macrosomia is a risk factor for incident maternal chronic kidney disease

Chronic kidney disease (CKD) is an issue known as a considerable cause to increase morbidity and mortality with a prevalence of about 13.4% across the world [1]; the prevalence rate for CKD in Iran compared to the global rate is higher with a rate of 18.9% [2]. Moreover, we previously reported that more than 2% of the Iranian population develops CKD each year [3]. Impaired glucose tolerance, high blood pressure, and high body-mass index (BMI) have been associated with a higher risk of CKD development [4]. Tackling CKD and its risk factors could be deemed an obligation due to its insidious nature and significant burden on global health.

Gestational diabetes mellitus (GDM) is one of the most common complications of pregnancy that is a glucose intolerance status with onset or first diagnosis during pregnancy [5]. A global perspective study in 2016 reported that the Middle East and North Africa (MENA) region had the highest rank in the prevalence of GDM [6]. Results from a meta-analysis estimated the prevalence of GDM in Iran to be around 3.41% with a range of 1.3 to 18.6% [7]. GDM is increasing in prevalence among the Iranian population in tandem with the dramatic increase in the prevalence of obesity, physical inactivity, and glucose intolerance status including both prediabetes and type 2 diabetes mellitus (T2DM) [8‐10]. Although most women with GDM will be normoglycemic after delivery [11], a history of GDM is known to increase the risk of subsequent T2DM, hypertension, metabolic syndrome, cardiovascular diseases (CVD), and endothelial dysfunction [12‐18].

Macrosomia is defined as birth weight > 4000 g by the American College of Obstetricians and Gynecologists and has a prevalence of about 9% worldwide [19, 20]. It is reported that GDM and pre-GDM were associated with a higher prevalence of macrosomia [21] which is likely to increase the risk of future maternal T2DM [22, 23].

The question that whether GDM can independently increase the risk of CKD was poorly addressed in limited studies with inconsistent results [24‐30]; most of them were based on a cross-sectional or retrospective design. A recent meta-analysis reported that no significant association between GDM and CKD was observed. However, black women with GDM had a higher risk of developing CKD [31]. Furthermore, to the best of our knowledge, there is no study investigating the impact of macrosomia on maternal CKD incidence.

Regarding the high burden of both GDM and CKD on the MENA region [4, 6] and relatively high prevalence of macrosomia [20], we aimed to study the relationship between macrosomia with or without GDM and incident CKD in the context of the oldest cohort of the region called the Tehran Lipid and Glucose Study (TLGS).

The TLGS is a population-based longitudinal study carried out on individuals aged ≥3 years living in the urban area of metropolitan city of Tehran. This study aimed to determine the prevalence and incidence of non-communicable diseases and their related risk factors.

The TLGS recruitment was done in two phases (the first, 1999–2001 and the second, 2002–2005). Data collection is planned to continue for at least 20 years at approximately 3-year intervals (i.e. third phase: 2005–2008, fourth phase: 2009–2011, fifth phase: 2012–2015). The design and registration of the TLGS have been previously described [32].

Of all 2669 participants, 204 women (7.64%) had a history of GDM/macrosomia, 113 subjects (4.23%) were incident cases of NDM and 2352 participants (88.13%) were neither with T2DM nor had a history of GDM/macrosomia. The baseline characteristics of these three groups are described in Table 1. The mean age (SD) and BMI of all participants at baseline were 36.2 (7.3) years and 27.7 (4.6) Kg/m², respectively. Generally, incident cases of NDM showed a less favorable situation for all baseline characteristics compared to other groups, excluding incident CKD and obesity, which were more prevalent in the GDM/macrosomia group (all P values < 0.05). We observed no significant difference in smoking status and marital status between study groups. Moreover, the frequency of incident CKD cases among GDM/macrosomia were higher compared with incident cases of NDM and T2DM-free individuals (33.8 vs. 30.1 vs. 21.7%, respectively).

During a median follow-up (interquartile range) of 11.9 years (8.3–13.2), 613 incident cases of CKD were identified with the corresponding incidence rate of 21.6 (95% CI: 20–23.4) per 1000 person-year. Figure 2 shows the Kaplan-Meier survival curve for the cumulative CKD incidence in the three study groups (log-rank test P-value < 0.001).

Crud, age and multivariable adjusted hazard ratio (HR) of GDM/macrosomia and incident NDM with incident CKD is shown in Table 2. Accordingly, only GDM/macrosomia showed 33% significant risk for incident CKD in the age-adjusted analysis [HR: 1.33 (95% CI: 1.03–1.71)]; the risk still remained significant after further adjustment for potential risk factors [HR: 1.32 (95% CI: 1.02–1.72), P = 0.032]. In the multivariable analysis, aging [HR: 1.06 (95% CI: 1.04–1.08)], current smoker [HR: 1.42 (95% CI: 1.00–1.99)], level of eGFR [HR: 0.91 (95% CI: 0.90–0.92)] and SBP [HR: 1.01 (95% CI: 1.00–1.02)] also remained independent risk factors. The results remained essentially unchanged when we replaced the definition of GDM/macrosomia with a history of macrosomia alone as shown in Table 3. Accordingly, having a history of macrosomia was associated with a higher risk for incident CKD of 36% [HR: 1.36 (95% CI: 1.04–1.78)]. However, when we replaced the definition of GDM/macrosomia with a history of GDM alone no association was found between GDM and incident CKD [HR: 0.92 (95% CI: 0.34–2.46)] (Table 4). In stratified analyses as shown in Fig. 3, we did not find a significant interaction between different study groups and general obesity (P = 0.59), central adiposity (P = 0.14), hypercholesterolemia (P = 0.22), and low HDL-C (P = 0.13). However, the data analysis showed a significant interaction between hypertension status and different study groups (P = 0.046). Hence, only among the non-hypertensive population, GDM/macrosomia group showed a significant risk for incident CKD [HR: 1.41 (95% CI: 1.07–1.85)]. Moreover, unexpectedly we found that the incident NDM group among the hypertensive population had a lower risk for CKD; however, considering only 6 events in this group, the effect size is unstable [HR: 0.42 (95% CI: 0.18–0.97)].

As a sensitivity analysis, the inverse probability weighting of the multinomial propensity score [36] was used to balance characteristics among three exposure groups (i.e. those without GDM/macrosomia and diabetes, with GDM/Macrosomia, and with undiagnosed diabetes). The analysis adjusted for baseline age, BMI, WC, number of parities, smoking status, educational level, gestational hypertension, eGFR, SBP, DBP, anti-hypertensive medication, and FH-DM. The result remained unchanged, the HRs (95% CI) for GDM/macrosomia and NDM groups were [HR: 1.34 (95% CI: 1.03–1.74)] and 1.05 (95% CI: 0.75–1.50)], respectively (data not shown).

For more than a decade of follow-ups among women aged 18–50 years in Tehran, we found that a history of GDM/macrosomia was associated with more than 30% higher risk of CKD, independent of a large set of important potential confounders including age, general obesity, central adiposity, parity, educational level, gestational hypertension, smoking status, FH-DM, SBP, DBP, using anti-hypertensive medication, and eGFR. This correlation also applied to women with a history of macrosomia alone. Moreover, we also found that in women with a history of GDM/macrosomia, non-hypertensive participants had a higher risk of CKD incidence compared to hypertensive participants.

The results of other studies examining the relationship between GDM and CKD are incompatible for comparison with each other, due to the different approaches applied for diagnosis of GDM, different sample sizes, various study designs (cross-sectional, retrospective, or prospective), the selection method of participants (hospital or population-based study), and number and type of confounders. Moreover, all of the studies in this field were confined to the United States (US), European and Israeli women, and no population-based study examined this relationship in the MENA region [24‐30].

The question regarding the relationship between GDM and subsequent impaired kidney function was first addressed by Friedman et al. [24] in a cross-sectional study of 72 Israeli women with a history of GDM, and as a result, a higher risk of microalbuminuria was observed. In another cross-sectional study with larger sample size, US women with GDM alone had similar odds of CKD stages 1–2 (odds ratio: 1.54) comparable with women with overt diabetes (odds ratio: 1.68) [26]. These findings were supported by another retrospective hospital-based cohort study in Israel which observed that GDM was a significant risk factor for long-term renal morbidity [27]. To the best of our knowledge, only 3 prospective cohorts have examined the relationship between GDM and incident CKD. The first study was conducted in 2015 with a sample size of 320 Canadian women of whom 100 cases had a history of GDM, confirmed by oral glucose tolerance test, and after 3 years of follow-up, they reported that current glucose intolerance, but not GDM, was associated with microalbuminuria. Importantly, in contrast to previous studies, the authors considered current glucose intolerance and eGFR as important confounders in their data analysis [28]. In the Coronary Artery Risk Development in Young Adults (CARDIA) Study, the presence of the self-reported GDM among black women, was associated with incident CKD, mainly attributable to albuminuria. This association did not change after adjustment for a large set of pre-pregnancy risk factors for both GDM and CKD including eGFR and fasting glucose concentrations [30]. In a prospective study of 607 Danish women with GDM with a mean follow-up of 13 years, a higher rate of increased eGFR levels was seen compared to the control group, indicating an early stage of renal damage. Moreover, the authors also showed that renal damage defined as increased urine albumin-creatinine ratio (UACR), was limited to GDM cases with subsequent development of T2DM [29]. In a recent meta-analysis including 2 studies, it was shown that history of GDM was associated with a 20% increased risk of CKD, the values which did not reach the significance level [risk ratio: 1.19 (95% CI: 0.72–1.97), I²: 66%] [31].

Interestingly, in the current study, we showed for the first time that a history of macrosomia alone in the absence of a history of self-reported GDM was also significantly associated with incident CKD. The association between the history of macrosomia and the development of T2DM and CVD had been previously reported [22, 23, 37]. Studies have also shown a strong association between pre-pregnancy maternal obesity, decreased insulin sensitivity, and increased insulin resistance with the delivery of macrosomic babies [38, 39]. Hence, the significant relationship between the history of macrosomia and incident CKD among Iranian women might be attributable to the high prevalence of pre-pregnancy maternal obesity and metabolic syndrome, two of the main risk factors for incident CKD, among Iranian women [4, 10, 40].

Although the precise mechanism underlying the higher risk of CKD among GDM cases observed in some studies is unknown, several potential mechanisms may explain the relationship. Interleukin-6, C-reactive protein, and plasminogen activator inhibitor-1 which are inflammatory factors that can anticipate future CVD and CKD [41‐44] are significantly higher in women with a history of GDM compared to those with no prior history [45, 46]. Moreover, a lower plasma adiponectin concentration, an anti-inflammatory factor, has been observed in women with GDM history [47]. Vascular endothelial dysfunction is associated with incident CKD [48], which is more prevalent in GDM women compared to the control group, in terms of increased wall stiffness [49], increased peripheral vascular resistance [46], impaired endothelium-dependent vasodilatation [50], and impaired acetylcholine-induced skin vasodilatation [49, 51]. Subclinical inflammation and vascular endothelial dysfunction in GDM, in the absence of overt T2DM, may suggest a possible link between GDM and subsequent CKD.

The strength of this study lies in the long-term follow-up of participants in a population-based study with a high burden of CVD risk factors and CKD [2, 52]. Moreover, the higher risk of CKD was also shown in the presence of well-known CKD risk factors including age, smoking status, eGFR, and SBP [4]. Importantly we also considered incident T2DM as another important confounder in our data analysis. Furthermore, most confounders were precisely measured using standardized protocols rather than relying on self-report data. We also for the first time showed the significant relationship between the history of macrosomia and incident CKD.

Our study has important limitations that should be considered. First, GDM was defined based on self-reported data; an approach applied in multiple other studies [30]. Second, we defined CKD according to a single estimate of eGFR and accepted the possibility of overestimation of CKD cases. Estimated GFR shows intra-individual biological variability and preferably needs a second creatinine measurement to accurately represent kidney function. The use of successive eGFR, had they been accessible, would have most likely reduced the incidence of CKD stage 3–4 but might not have affected the relationship between the 3 study groups (GDM/macrosomia, incident NDM, and reference group) and CKD. Despite this, most of the population-based studies focusing on CKD as the outcome used single serum creatinine measurements [29]. Furthermore, a recently published systematic review and meta-analyses aimed to assess the predictive role of obesity, overweight, and BMI on the risk of new-onset CKD, showed the association between obesity and low eGFR did not affect by the number of eGFR measurements [53]. Third, we did not validate the CKD-EPI equation in a local population, and this could also lead to an overestimation of the incidence of CKD. Fourth, the definition of CKD was built only upon eGFR criteria, since no valid data for urine protein excretion were available in the TLGS. Fifth, this study was conducted in the metropolitan city of Tehran, hence other studies particularly in the rural area should examine this association.

We concluded that Iranian women with a history of GDM/macrosomia or even macrosomia alone are at higher risk for subsequent CKD, independent of traditional CKD risk factors as well as incident T2DM. Given the high incidence of CKD among the Iranian population [3], pregnancy may provide a unique situation to identify high-risk women at risk for CKD that could benefit from regular monitoring of kidney function and providing risk modifying strategies.