Our study showed that ANGPTL8 levels in early pregnancy were significantly associated with risk of subsequent GDM in weeks 24–28 of gestation, independently of conventional risk factors including maternal age, BMI, GGT, positive HBsAg and FPG levels in early pregnancy. In addition, ANGPTL8 had a greater AUC for predicting GDM than maternal BMI, GGT and positive HBsAg. More importantly, our results suggested that incorporating ANGPTL8 into the prediction model including conventional risk factors could significantly improve the ability to predict future GDM.
A75g OGTT performed during the late second trimester is currently regarded as the gold standard of GDM diagnosis, which may expose the infant to adverse intrauterine circumstances for too long. Hence there is a definite need to predict and detect GDM earlier in pregnancy in order to limit the harm that hyperglycaemia can do to the mother and her offspring. The performance of screening strategies based on conventional risk factors has been examined in many studies, but the sensitivity and specificity have not been shown to be satisfactory [12, 13, 23, 24]. Among conventional risk factors, FPG is most commonly measured in early pregnancy. In a large Chinese population, FPG levels at the first prenatal visit were found to be positively associated with risk of GDM diagnosed at 24–28 weeks . Nevertheless, other studies have reported that FPG has poor sensitivity and specificity as a screening test in early pregnancy . In our study, FPG in early pregnancy was independently associated with incidence of GDM. However, in line with the previous studies, the ability to predict GDM using FPG alone is limited, with an AUC of 0.646. Although many studies have demonstrated that increasing maternal BMI is a significant risk factor for the development of GDM [26, 27], a large sample study concluded that there was inadequate evidence to support the use of BMI alone to screen for GDM in early pregnancy . In accordance with that study, we found that BMI in early pregnancy was inferior to FPG and ANGPTL8 in predicting GDM.
As the predictive performance of conventional risk factors to predict GDM is limited, more attention was paid here to other biomarkers. ANGPTL family members have been shown to play a major role in obesity and metabolic diseases [28, 29]. It is well recognised that ANGPTL8 can act as a lipid regulator by inhibiting lipoprotein lipase activity, either directly or indirectly by promoting cleavage of ANGPTL3 [30, 31]. A recent animal study suggested that inhibition of ANGPTL8 might provide a new therapy for the treatment of dyslipidaemia, with beneficial effects on body weight . Clinical epidemiological studies have provided abundant evidence regarding the associations between ANGPTL8 and metabolic diseases including diabetes [19, 33], hypertension , dyslipidaemia [17, 18] and the metabolic syndrome , as well as GDM [20,21,22]. In order to exclude the potential effect of blood pressure on ANGPTL8, we excluded participants with prehypertension (high normal blood pressure: systolic blood pressure [SBP] 120–139 mmHg or diastolic blood pressure [DBP] 80–89 mmHg) from the present study. In case–control studies, participants who developed GDM had significantly higher levels of ANGPTL8 than healthy pregnant women [20, 22]. Although ANGPTL8 levels change dramatically during pregnancy, no study has evaluated the role of ANGPTL8 in early pregnancy in predicting GDM. To the best of our knowledge, our study has reported for the first time that ANGPTL8 levels in early pregnancy are positively and significantly associated with risk of subsequent GDM.
Some researchers argue that relying solely on conventional risk factors is unsatisfactory for predicting GDM; combining new biomarkers into a conventional predictive model may improve the prediction of GDM [36, 37]. Maitland and colleagues reported that introducing adiponectin, measured early in the second trimester, into the model based on routinely measured clinical factors significantly increased predictive ability . Our study demonstrated that ANGPTL8, by itself, showed poor sensitivity for later GDM; however, adding ANGPTL8 to the model based on conventional risk factors in early pregnancy increased the ability to predict GDM.
The potential pathophysiological mechanism linking ANGPTL8 to GDM is not established. Ebert et al found that, during normal pregnancy, maternal ANGPTL8 levels were significantly higher than in the non-pregnant state, and they decreased in postpartum period . The dynamic change in ANGPTL8 levels indicated that ANGPTL8 might participate in maintaining normal pregnancy. Moreover, ANGPTL8 levels have been found to be higher in umbilical cord blood than maternal serum , which suggests a potential role of ANGPTL8 in fetal growth and development. ANGPTL8 might thus offer us new insights into the physiology of normal pregnancy and the pathophysiology of GDM.
Some limitations of the present study should be noted. First, at the first prenatal visit, we did not perform a 75 g OGTT or measure HbA1c; although we excluded women with an FPG ≥7.0 mmol/l, women with higher PPG but normal FPG levels might be misclassified as having normal glucose regulation. Second, the study was also limited by the lack of data on insulin levels; therefore we were unable to assess whether the ability of ANGPTL8 to predict GDM was mediated by insulin resistance or whether ANGPTL8 was merely a biomarker of insulin resistance. Although some researchers have found that ANGPTL8 can promote the suppression of key enzymes involved in gluconeogenic pathways and improving insulin resistance , some clinical studies have demonstrated that the associations between ANGPTL8 levels and metabolic diseases were independent of insulin resistance [18, 35, 39]. Consistent with the studies performed in the general population, the association between ANGPTL8 and GDM still persisted after controlling for HOMA-IR in pregnant women . Third, information on diet and physical activity was also lacking, and therefore we were not able to evaluate the effects of lifestyle change on risk of GDM. Finally, although the ELISA kit for ANGPTL8 is now commercially available and inexpensive, the normal range of ANGPTL8 levels should be defined for future clinical application, so a large population sample is needed to do this.
In conclusion, the present study suggested that ANGPTL8 levels in early pregnancy were significantly and independently predictive of GDM in weeks 24–28 of gestation, and that combining ANGPTL8 levels with conventional risk factors for GDM could improve predictive performance.