Association Between HbA1c Levels on Adverse Pregnancy Outcomes During Pregnancy in Patients With Type 1 Diabetes

Abstract Context Despite optimization of metabolic balance during pregnancy in type 1 diabetes (T1D), maternal–fetal complications remain higher than in the background population. Objective We examined whether there is an association between glycated hemoglobin (HbA1c) levels and these complications. Methods Retrospective study of pregnancies in 678 T1D subjects at Lille Hospital (1997-2019). The association between variations in HbA1c levels and complications was examined. The composite criterion (CC) was defined as having at least 1 of the following complications: prematurity, pre-eclampsia, large for gestational age (LGA), small for gestational age (SGA), or cesarean section. Results Among the 678 births, median preconception HbA1c was 7.2% (55 mmol/mol), 361 were LGA (56%), 29 were SGA (4.5%), and 504 were births without preterm delivery (76.1%). The CC occurred in 81.8%. Higher HbA1c during the first trimester was associated with the CC (OR 1.04; 95% CI 1.02-1.06 per 0.1% increase; P < .001). Higher HbA1c during the third trimester was associated with the CC (OR 1.07; 95% CI 1.03-1.10 per 0.1% increase; P < .001). The group defined by a first trimester Hba1c >6.5% (48 mmol/mol) and a third trimester HbA1c <6% was associated with an increased rate of the CC (OR 2.81; 95% CI 1.01-7.86) and an increased rate of LGA (OR 2.20; 95% CI 1.01- 4.78). Conclusion Elevated HbA1c is associated with maternal–fetal complications. Despite optimization of metabolic balance during the third trimester, for patients with early glycemic imbalance the risk of LGA persists.

A c c e p t e d M a n u s c r i p t 5 above values since HbA1c decreases during the first and second trimesters, linked to pathophysiological changes [9]. O'Connor et al. defined a normal range for HbA1c in pregnant Caucasian women as <5.4% (36 mmol/mol) in the first trimester, <5.4% (36 mmol/mol) in the second trimester and <5.7% (39 mmol/mol) in the third trimester [10].
The aim of our study was to examine whether there is an association between HbA1c levels and maternal-fetal complications in T1D who were followed in the same tertiary obstetric care center by the same multidisciplinary team.

RESEARCH DESIGN AND METHODS:
This single-center observational study was performed at the University hospital of Lille, France based on electronic records, including the metabolic and obstetric data that are routinely collected at delivery for every birth. Under French law, care-related data may be used for research purposes unless the patient opposes such use. Data were analyzed anonymously, and our database was declared to the French Committee for computerized data (CNIL). In this observational cohort, we included all women with pregestational diabetes who gave birth between 1997 and 2019. All pregnancies were analyzed but only pregnancies of women with T1D were included. Patients were excluded if they were under 18 years old or had other types of diabetes, including type 2 diabetes, monogenic diabetes, syndromic diabetes, or secondary diabetes. Additional exclusion criteria included lack of data or consent, persistent doubt regarding diagnosis, lost to follow-up or twin/multiple pregnancies since these had a higher risk of adverse outcomes. Patients were treated with short-acting insulin analogues before meals and long-acting insulin analogues in the morning and/or at bedtime, or with continuous subcutaneous insulin infusion (CSII). We have followed the French guidelines which recommended self blood glucose monitoring with a glucose target <100 mg/dL before meals and <140 mg/dL after meals [11].
Age, height, and body weight were recorded and BMI was calculated in kg/m 2 . Blood pressure was measured; with hypertension defined as >140/90 mmHg or the use of an antihypertensive drug before pregnancy. Diabetes history was recorded including: duration of diabetes, therapy used (multiple daily injections (MDI) or Continuous subcutaneous insulin infusion (CSII), pre-conception HbA1c and vascular complications: history of nephropathy (albuminuria ≥ 30 mg/24h or renal insufficiency), history and status of retinopathy.
Obstetric history was assessed: parity, gravidity, date of pregnancy, history of macrosomia, hypertension, preeclampsia, miscarriage or stillbirth. Administration of a daily dose of 5 mg of folic acid has been started as soon as conception had been planned, and continued until week 12 weeks' gestation. Losses of pregnancy were recorded: miscarriage was defined as the loss of pregnancy before 24 weeks' gestation. Stillbirth was defined as fetal loss occurring after 24 weeks' gestation. Hypertension was defined as the appearance or aggravation of hypertension. Preeclampsia was defined as association of systolic blood pressure >140 mmHg or diastolic blood pressure >90 mmHg and proteinuria greater than or equal to 300 mg/24 hours after 20 weeks amenorrhea. A c c e p t e d M a n u s c r i p t 7 Prematurity was defined as birth prior to 37 weeks amenorrhea. Delivery modality was recorded (vaginal or cesarean section). Induction of labour is systematically performed between 38-39 weeks' pregnancy, in accordance with the recommendations of the CGNOF (French Society of Obstetrics).
Birthweight (BW) was used to define macrosomia as BW superior or equal to 4000 grams, Birthweight was adjusted for neonatal sex, and gestational age for singleton pregnancies using customized percentiles with Large for gestational age (LGA) defined as birthweight centile above the 90th percentile and small for gestational age (SGA) defined as below 10 th percentile (AUDIPOG curves) [12].
A composite criterion (CC) which associated preterm delivery, preeclampsia, LGA, SGA and cesarean section, was defined to estimate the proportion of maternal-fetal morbidity in our population. This criterion was considered positive if at least one component was present.
HbA1c was measured monthly using automated high pressure liquid chromatography in the period 1997-2015. After 2015, capillary electrophoresis was performed (Capillaris Tera SEBIA, normal range: 4.0-6.0% (20-42 mmol/mol); coefficient of variation <3%). Assay performance was certified by Bio Rad. HbA1c was measured during the first month then during 1 st (< 15 weeks' gestation), 2 nd (< 28 weeks' gestation), and 3rd trimesters (<41 weeks' gestation). For the statistical analysis, we did a mean of each HbA1c done every trimester. Delta HbA1c 1st-3rd trimester was defined as the difference in the means of HbA1c in the first trimester compared to means in the 3rd trimester. We compared the CC and most frequent individual adverse outcomes (LGA, prematurity, cesarean section) between four glycemic control subgroups defined by HbA1c levels in 1st and 3rd trimesters : A c c e p t e d M a n u s c r i p t 8 group I) <6.5% in the 1st trimester and <6.0% in the 3rd trimester, group II) ≥6.5% in the 1st trimester and <6.0% in the 3rd trimester, group III) <6.5 % in the 1st trimester and >6.0% in the 3rd trimester, group IV) and ≥6.5% in the 1st trimester and ≥6.0% in the 3rd trimester.

Statistical analysis
Statistical analyses were conducted using SAS software (SAS Institute 9.4, Cary, USA).
Categorical variables were reported as numbers (percentage). Quantitative variables were described by means ± standard deviation, in case of Gaussian distribution, or otherwise by median (interquartile range (IQR)). Normality of numerical variables was checked graphically and tested using the Kolmogorov-Smirnov test. We assessed the association of HbA1c (assessed at the 1 st and 3 rd trimesters, as well as the difference between 1 st and 3 rd trimester values) with pregnancy outcomes (CC and individual adverse events) using logistic regression models before and after adjustment for years of the date of pregnancy, or for treatment. Comparisons were made using logistical regression models before and after adjustment on period of date of pregnancy, using subgroup I as reference. All results were expressed in odds ratios (OR) and their 95% confidence intervals (CI). For CC outcome, the predictive ability of HbA1c, assessed in 1 st and 3 rd trimesters, was evaluated by receiver operating characteristic (ROC) curve analysis, by calculating the area under the ROC curve (AUC) and its 95% confidence interval. Statistical testing was two-tailed with p<0.05 accepted as significant.

Demographic characteristics of type 1 diabetic population:
During the study period, we have included 1587 pregnancies with maternal diabetes. We have excluded 861 pregnancies: 734 with DT2, 51 with other forms of diabetes, 76 T1D women because data missing (n=44), lost to follow-up (n=28) and twin pregnancy (n=4). So A c c e p t e d M a n u s c r i p t 9 in this study, 726 pregnancies, among 510 patients, were included in our study: 348 with 1 pregnancy, 119 with 2 pregnancies, 32 with 3 pregnancies, and 11 with 4 or more pregnancies. 48 pregnancies were excluded due to miscarriage or stillbirth so the data were analyzed in 678 live birth (see Figure 1).
Concerning fetal complications, the median gestational age at delivery was 38 weeks (IQR, 37 to 38.2). The prematurity rate was 24% (n=159). Mean birthweight was 3484 (± 675. 6) grams. The rate of macrosomia was 19.5% and of LGA was 56%. The rate of SGA was 4.5%. Shoulder dystocia was described in 9.6% of children and 8.7% of children were A c c e p t e d M a n u s c r i p t 10 admitted in NICU ( Neonatal Intensive Care Unit). 5.9% of children showed a neonatal malformation. The CC, defined by at least one complication among preeclampsia, LGA, SGA, cesarean section and prematurity, was present in 81% of cases.
Four glycemic control subgroups, defined by HbA1c levels in 1 st and 3 rd trimesters, were created in order to define the period during which HbA1c was a better predictor of maternalfetal morbidity: Group I was optimal balance throughout pregnancy included 21.6% of the cohort and was considered the reference group, Group II had early imbalance but progressive correction during pregnancy represented 7.6% of the population, Group III had a late glycemic imbalance 3rd trimester) representing 18.3% of the population and Group IV had glycemic imbalance throughout pregnancy making up 52.5% of the population.
Associations of HbA1c with CC and most individual adverse events, according to these glycemic control levels , are presented in Table 3 between these groups and prematurity or cesarean section was observed. Excluding cesarean sections from the CC did not modify the results (data not shown).

DISCUSSION:
The aim of this study was to define if there is an association between HbA1c levels and maternal fetal complications in T1D. In this large cohort, we found that HbA1c was associated with complications and that early HbA1c was able to predict several adverse outcomes including LGA, SGA, preeclampsia, and preterm delivery. Interestingly, results showed that improvement in HbA1c levels between first and third trimester is not sufficient to have the same rate of LGA as the background population.
Before conception, our large cohort had similar characteristics to other published cohorts in terms of age, BMI, duration of diabetes and the presence of diabetic vascular complications.
Similar to other studies, only a quarter of our population had HbA1c levels in the target range at conception [14]. Indeed, The American Diabetes Association recommends an HbA1c less than or equal to 6.5% (48 mmol/mol) before pregnancy or <6% (42 mmol/mol) in the absence of hypoglycemia [7]. The small number of women with ideal HbA1c levels at conception may suggest a failure of, or non-compliance with, pre-conception counseling and programming of pregnancy. Publications have confirmed the impact of good metabolic control prior to pregnancy on maternal-fetal morbidity [15]. As expected, conception levels of HbA1c gradually improved over time in our study. In our cohort, 40.8 % of women had an A c c e p t e d M a n u s c r i p t 13 HbA1c level less than or equal to 7% (53 mmol/mol) in early pregnancy, allowing us to evaluate this point.
Despite a gradual improvement in glycemic balance, as shown by HbA1c, many complications are still reported in pregnancy. In our study, we have demonstrated that the rate of composite criterion was above 80%. There is currently no consensus regarding the composite criterion of choice for assessing maternal fetal morbidity. We focused on the complications that seemed most important either in terms of frequency or severity. A recent study on 488 pregnant women with T1D reported good outcomes in 44% of pregnancies, but included only uncomplicated deliveries of normal infants, non-LGA after spontaneous labor, and no perinatal complications [16]. This study also showed a continuous association between good perinatal outcomes and HbA1c at delivery, suggesting that the lower the HbA1c the better the perinatal outcome. They reported that HbA1c <6.0% (42 mmol/mol) at delivery, which was achieved in 41% of women, was identified by ROC analysis as the best threshold for predicting good outcomes. However, this threshold does not allow accurate identification of infants at low risk of perinatal complications since an adverse perinatal outcome was observed in 40% of women with HbA1c <6.0% (42 mmol/mol) at delivery and, conversely, good perinatal outcomes were noted in 32% of women with HbA1c of >6.0% (42 mmol/mol) at delivery.
Our results suggest that when all complications are combined, the composite criterion was associated with HbA1c in the 1st trimester and the 3rd trimester, with a greater risk of onset when the HbA1c level was high. We confirm that the tighter the glycemic control, the lower A c c e p t e d M a n u s c r i p t 14 delivery, cesarean section and preeclampsia [5]. This is consistent with other published data [17][18][19]. However, they reported a cut-off of HbA1c >6.8% (51 mmol/mol) that could predict this comorbidity, which is higher than the value we found, while also finding an equally weak sensitivity/specificity. Despite the large cohort, ROC curves in our study did not show sufficiently sensitivity and specificity to predict complications of pregnancy. It seems likely that the variables included in our composite criterion may explain this result. A composite criterion must be defined which includes complications associated with a higher risk for the mother and the fetus. To develop this criterion, we carried out the same analyses while excluding cesarean section, and this did not modify our results, with an added risk of maternal-fetal complications still found when there was an early glycemic imbalance, despite this being corrected during pregnancy. The impact on maternal-fetal complications of HbA1c levels in the 1st trimester has been infrequently described in the literature [9,20], and even less so in women with T1D [21,22]. The data are weak and sometimes contradictory, and it is important to note that the distinction between type 1 and type 2 diabetes is rarely made. A 5-year cohort study based on type 1 and type 2 diabetic subjects highlighted a cut-off for HbA1c in the 3 rd trimester of >6.6% as an independent risk factor for perinatal mortality [23].
Similarly, the impact of HbA1c in the 3 rd trimester has been little described. However, a recent Qatari study reported different results; similar to our results they highlighted an association between LGA and HbA1c in the 3rd trimester but also a reduction in the risk of LGA which was greater when there was significant change in HbA1c between 1 st and 3 rd trimesters. The impact of ethnicity and initial glycemic imbalance (more severe with an A c c e p t e d M a n u s c r i p t 15 limiting the risk of maternal hypoglycemia [7]. We conducted the same analyses in our population with these ideal targets; the results were similar ( see supplementary data 3: https://doi.org/10.5061/dryad.2fqz612q2 [13]). Unfortunately, the retrospective nature of the study did not allow us to collect data on maternal hypoglycemia, especially severe hypoglycemia. However, in our cohort, the variation of HbA1c between 1 st and 3 rd trimesters was not associated with this criterion, which led us to evaluate each element of this composite criterion independently in each period, with LGA, prematurity, and cesarean section being associated with HbA1c levels in the first and also the third trimester. Only preeclampsia was not significantly associated with HbA1c. In addition to the low number of cases of preeclampsia, our data differs from most published data [26]. During this first stage, we noted the absence of an association between its complications and 1 st -3 rd variation of HbA1c, suggesting that correction of glycemic balance after the 1st trimester would have less impact on this morbidity.
Our study has several strengths including the large sample size of T1D subjects and their evaluation by the same multidisciplinary team, their clinical characteristics being in accordance with the literature. This provides statistical power and constitutes the main strong point of the study. In addition, a single laboratory performing HbA1c measurements provides a robust data set. In our clinical practice, we take an HbA1c test every month during pregnancy. So taking HbA1c each month of each quarter allowed us to obtain medians of HbA1c during each trimester, thus limiting the risk of error induced by a single assay which would be erroneous for different reasons. However, the collection of monocentric data, the particularly long duration and the retrospective design of this study with some missing data could generate a lack of power in statistical comparisons.
Furthermore, some potential limitations should be discussed. Firstly, hemoglobin and mean corpuscular volume (MCV) levels in our patients were unknown, thus we were unable to exclude the presence of hemoglobinopathy, iron deficiency or anemia, which can impact the accuracy of HbA1c assessments during pregnancy [27]. However we have a lower rate of In conclusion, HbA1c, even if little used in clinical practice, could be a marker for monitoring glycemic balance during pregnancy. Elevated HbA1c is associated with numerous maternalfetal complications during the 1st and also during the 3rd trimester. Improving glycemic balance seems to only partially reduce this risk, without eliminating it, suggesting the involvement of other associated mechanisms, for example BMI [28], immune mechanisms [29], and glycemic variability [30].

DATA AVAILABILITY:
Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.  Table 1: Baseline maternal characteristics.
 Table 2: Maternal and fetal adverse pregnancy outcomes.
   A c c e p t e d M a n u s c r i p t 24