Early pregnancy high normal HbA1c: a high risk group?

HbA_1c (glycated haemoglobin) predicts pregnancy-related adverse outcomes in people with pre-diabetes and diabetes.^1–3 Aotearoa New Zealand guidelines recommend measuring HbA_1c with the first antenatal blood tests to identify previously undiagnosed diabetes; where HbA_1c ≤40mmol/mol is normal, 41–49mmol/mol suggests greater risk and ≥50mmol/mol represents probable undiagnosed diabetes.¹ HbA_1c ≥50mmol/mol is diagnostic of diabetes in Aotearoa New Zealand, and those meeting this criteria are referred to secondary services for specialist input during pregnancy.¹ Some people with HbA_1c of 41–49mmol/mol are referred, however this is dependent on local guidelines.⁴

People with pre-diabetes and gestational diabetes (GDM) are at increased risk of adverse perinatal outcomes.^2–7 The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study identified increasing risk of perinatal complications with increasing maternal glycaemia below the threshold for diabetes.⁵ There is limited literature on pregnancy outcomes in people without diabetes with an early pregnancy HbA_1c at the upper limit of normal. In 466 women followed prospectively in Australia, an early pregnancy HbA_1c of ≥38mmol/mol was highly predictive of developing GDM and increased risk for large for gestational age (LGA).⁸ In Aotearoa New Zealand, Hughes et al. (2014) demonstrated that women with an early pregnancy HbA_1c in the pre-diabetes range (41–46mmol/mol) who were not treated for GDM had increased rates of major congenital anomaly, pre-eclampsia, shoulder dystocia and perinatal death compared to women with normal HbA_1c.² There is no data in Aotearoa New Zealand for people who have a booking HbA_1c of <41mmol/mol.

HbA_1c may not be a reliable predictor of glycaemic control at early gestations, falling 4–10mmol/mol by the second trimester.^8,9 Several factors contribute to this fall, including haemodilution and an increase in red cell turnover.¹⁰ Therefore, a person with pre-diabetes based on an HbA_1c of 41–49mmol/mol outside of pregnancy, who has already developed a degree of glucose dysregulation, may have a normal HbA_1c by the time the first antenatal bloods are taken. This means these people are not identified as higher risk—either at all, or until later in pregnancy.

In Aotearoa New Zealand, routine measurement of early pregnancy HbA_1c has enabled identification of previously undiagnosed type 2 diabetes mellitus and more timely interventions. However, when performing screening it is vital to understand the risks associated with “high normal” results of a continuous variable. We hypothesise that women with “high normal” early pregnancy HbA_1c may experience higher rates of adverse perinatal outcomes. The aim of this study was to determine whether early pregnancy HbA_1c of 35–40mmol/mol in people without either pre-existing diabetes or a later diagnosis of GDM was associated with an increased risk of adverse perinatal outcomes compared to people with HbA_1c <35mmol/mol. The second aim was to establish whether the risk of adverse outcomes increases as HbA_1c increases.

Method

A retrospective chart review was performed to look at the relationship between early pregnancy HbA_1c (<20 weeks gestation) and adverse perinatal outcomes in pregnant people from the Wellington region delivering at Wellington Regional Hospital, Kenepuru Maternity Unit or Paraparaumu Maternity Unit between 1 July 2019 to 31 December 2019. This study was approved by the Health and Disability Ethics Committee of Aotearoa New Zealand.

Baseline characteristics and pregnancy outcomes were obtained from the Capital and Coast District Health Board (CCDHB) Patient Information Management System database. HbA_1c results from Wellington Southern Community Laboratory were collected. HbA_1c was quantified using Bio-Rad Variant D-100 Ion Exchange High-performance Liquid Chromatography (HPLC). D-100 has shown reliable analytical performance with good precision and linearity and a CV of <1%.¹¹ Singleton pregnancies with HbA_1c <50mmol/mol at <20 weeks gestation were included. Exclusion criteria were pre-existing diabetes mellitus, developing GDM in the current pregnancy, missing BMI, no HbA_1c at <20 weeks gestation or the first HbA_1c was taken at ≥20 weeks gestation.

Participants were divided into groups: HbA_1c <35mmol/mol, (“nHbA_1c”), HbA_1c 35–40mmol/mol, (“hnHbA_1c”) and HbA_1c 41–49mmol/mol (“pre-diabetes”).

The primary outcome was birth weight (g). An equally important outcome was customised birth weight centiles, which were calculated using the GROW Bulk Centile Calculator version 6.7.8.3 (Perinatal Institute, Birmingham, UK), which adjusts for maternal height, weight, ethnicity, parity, sex and gestational age at delivery. LGA was defined as >90^th customised centile, and small for gestational age (SGA) was defined as <10^th customised centile. Secondary outcomes were mode of delivery: normal vaginal delivery, caesarean delivery, assisted vaginal delivery; shoulder dystocia; perineal tears (third and fourth degree); post-partum haemorrhage (PPH) (estimated blood loss >500ml at delivery); induction of labour; pre-term delivery (<37 weeks); neonatal hypoglycaemia requiring treatment, Neonatal Intensive Care Unit (NICU) admission requiring respiratory support; NICU admission in days and perinatal death. Perinatal death was defined according to the Perinatal and Maternal Mortality Review Committee as foetal death occurring >20 weeks gestation, or ≥400g birth weight, and included neonatal deaths occurring up to 28 days of life.¹²

Power calculations were performed prior to data collection. It was estimated that a sample size of 1,400 was achievable in the study timeframe, which would have at least 80% power to compare HbA_1c groups with respect to birth weight (assuming a difference of 150g, SD=580), and adverse outcomes (assuming a difference of 10% vs 20%), testing at the 5% significance level.

Baseline characteristics and adverse outcomes are presented as mean (Standard Deviation) or median (range), and n (percent) as appropriate. The nHbA_1c and hnHbA_1c groups were compared with respect to age (independent-samples t-Test), BMI (Wilcoxon Rank-Sum Test with continuity correction), ethnicity and parity (Pearson’s Chi-squared test of independence). Dichotomous adverse outcomes were analysed using binomial logistic regression and birth weight was analysed with multiple linear regression. All analyses for adverse outcomes were tested for differences between the nHbA_1c and hnHbA_1c groups, adjusting for age, BMI and ethnicity. In addition, a further analysis was performed for vaginal birth and post-postpartum haemorrhage that controlled for parity. There were few observed events for the adverse events of shoulder dystocia and perinatal death recorded, so no analyses were run. Maternal and neonatal composite adverse outcomes were created separately. Composite outcomes were used that included clinically important outcomes; for neonatal this included birth weight >4,000g, LGA, SGA, shoulder dystocia, pre-term delivery (<37 weeks), admission to NICU, hypoglycaemia requiring treatment and perinatal death. For maternal this included delivery via caesarean section, perineal tears (third and fourth degree), PPH, induction of labour, pre-term delivery (<37 weeks). Secondary analyses were conducted with BMI as the only predictor. No correction for multiple comparisons was applied, p-values less than 0.05 were considered statistically significant and data were analysed in R version 4.2.0 for Windows (Vienna, Austria).

Results

Between 1 July 2019 and 31 December 2019 there were 1,514 singleton births in the Wellington region (Figure 1). Of these, 261 were excluded from analysis, 1,067 recorded nHbA_1c (<35mmol/mol) and 186 recorded hnHbA_1c (HbA_1c 35–40mmol/mol). One person had an HbA_1c in the 41–49mmol/mol range, so this group was not made. Seven people had no BMI recorded and were excluded. The mean HbA_1c of those who developed GDM (and were excluded) was 34.7mmol/mol (SD=6.8).

View Figure 1, Table 1–2.

Baseline characteristics are presented in Table 1. Participants in the hnHbA_1c group had a higher BMI (25.4kg/m² vs 24.4kg/m², p=0.023), were more likely to be Pacific peoples, Indian or Other Asian ethnicity compared to participants in the nHbA_1c group (p <0.05) and multiparous (65.6% hnHbA_1c vs 52.6% nHbA_1c, p=0.001).

Perinatal outcomes are presented in Table 2. There was no difference in birth weight between hnHbA_1c and nHbA_1c groups. There was no significant relationship between pregnancy outcomes, including birth weight, and HbA_1c as a continuous variable. Participants with hnHbA_1c had significantly higher odds of experiencing a normal vaginal delivery than those with nHbA_1c (OR 1.4, 95% CI 1.01–1.97), adjusting for age, BMI and ethnicity. However, after controlling for parity, hnHbA_1c was no longer significantly associated with a normal vaginal delivery (OR 1.33, 95% CI 0.89–1.97).

Participants with hnHbA_1c had significantly lower odds of experiencing a PPH compared to participants with nHbA_1c (OR 0.52, 95% CI 0.35–0.76). After controlling for parity, there was little change (OR 0.56, 95% CI 0.38–0.82). Of the people who had a PPH, 5% in the hnHbA_1c group had a blood loss of >2,000ml compared to 1.5% of people in the nHbA_1c group. No significant differences were found in other pregnancy outcomes, including neonatal outcomes. Five perinatal deaths occurred in the nHbA_1c group—two stillbirths and three neonatal deaths—and none in the hnHbA_1c group.

A statistically significant difference between the hnHbA_1c group and the nHbA_1c group was found in the risk of composite maternal adverse perinatal outcomes (OR 0.64, 95% CI 0.46–0.89), and remained significant after adjusting for parity (OR 0.64, 95% CI 0.45–0.91).

A secondary analysis was performed looking at BMI only as a predictor of adverse perinatal outcomes. Increasing BMI was associated with an increased likelihood of macrosomia (birth weight >4,000g) (OR 1.06, 95% CI 1.04–1.09), caesarean section compared to normal vaginal delivery (OR 1.02, 1.01–1.04), PPH (OR 1.03, 1.01–1.05) and induction of labour (OR 1.04, 1.02–1.06). Maternal and neonatal composite adverse outcomes were significantly increased with increasing BMI.

Of the 91 people who developed GDM, and were excluded, 86.5% had an HbA_1c <41mmol/mol and 13.5% had an HbA_1c ≥41mmol/mol. In those who had an HbA_1c ≥41mmol/mol (n = 12/1,514) (excluding those with pre-existing diabetes), 91.7% went on to develop GDM (11 of 12). Of all the people with high normal HbA_1c (35–40mmol/mol), before excluding for GDM, 13.8% went on to develop GDM (30 of 218).

Discussion

This retrospective review demonstrated that pregnant people with an early pregnancy high normal HbA_1c, without pre-existing diabetes or later development of GDM, have no difference in birth weight compared to people with normal HbA_1c. Those in the hnHbA_1c group did not have an increased risk of adverse perinatal outcomes and were less likely to have a PPH or experience adverse composite outcomes, even after controlling for parity. Increasing BMI, irrespective of HbA_1c, significantly increased the odds of macrosomia, caesarean section, PPH and induction of labour.

There were proportionally more Indian and Pacific peoples in the high normal HbA_1c group compared to the normal HbA_1c group. These ethnicities have the highest rates of gestational diabetes and type 2 diabetes in Aotearoa New Zealand.^13,14 This may be clinically relevant, and future research could explore whether people of these ethnicities should be screened or managed at a lower HbA_1c threshold.

People with high normal HbA_1c were less likely to have a PPH compared to those with normal HbA_1c. The high normal group had more multiparous people and were more likely to experience a vaginal birth, which could have influenced these results. Risk factors identified from previous pregnancies may have resulted in increased use of active management of the third stage of labour, thereby reducing the risk of PPH.

We excluded participants who later developed GDM in order to report on perinatal outcomes independent of any treatment potentially received. Our findings are in keeping with those of Immanuel et al. (2020), who reported that early pregnancy HbA_1c ≥39mmol/mol in obese European women did not predict adverse pregnancy outcomes.¹⁵ Likewise, a recent retrospective cohort study showed no increased risk of adverse outcomes in women with early pregnancy HbA_1c 38.8–46.4mmol/mol.¹⁶ In contrast, Capula et al. (2013) demonstrated that HbA_1c is a strong predictor of negative outcomes in women with GDM, with HbA_1c >34mmol/mol associated with a two-fold increased risk of pregnancy-related hypertension, LGA and neonatal morbidity compared to HbA_1c <34mmol/mol.¹⁷ Poor pregnancy outcomes related to HbA_1c independent of GDM, such as macrosomia, have been reported elsewhere with HbA_1c ≥41mmol/mol¹⁸ and HbA_1c ≥39mmol/mol.¹⁹ Our findings that BMI predicts adverse outcomes are significant in this study and are concordant with international literature.^20–22 Of concern, more people are conceiving with an increased BMI, which, independent of dysglycaemia, increases risk of both maternal and neonatal adverse outcomes.^20–22

The relationship between perinatal outcomes and maternal glycaemia, and the associated role of HbA_1c, has been explored. The HAPO study demonstrated an increased risk of adverse pregnancy outcomes with increasing maternal glycaemia,⁵ and in a sub-group of women, increasing HbA_1c was associated with increased LGA and primary caesarean section.³ Bozkurt et al. (2020) observed beta cell dysfunction and glucose dysregulation when early pregnancy HbA_1c was ≥39mmol/mol and was associated with greater risk for LGA.²³ Comparably, women with pre-pregnancy impaired glucose tolerance had a two-fold increased risk of LGA, demonstrated by Wei et al. (2017).²⁴

Together with the HAPO studies, this evidence supports our hypothesis that women with a degree of glucose dysregulation, below the diagnostic criteria for diabetes, are at increased risk of adverse pregnancy outcomes, including large for gestational age. Although HbA_1c is a useful tool to identify people with undiagnosed pre-existing diabetes, it may be that an alternative assessment for early dysglycaemia is required to reduce adverse perinatal outcomes, or as in the studies reported here, an HbA_1c closer to the pre-diabetes range is required to identify dysglycaemia, which influences perinatal outcomes.

Importantly, some have suggested people who receive treatment for hyperglycaemia or diabetes in early pregnancy have improved outcomes compared to those that do not.²⁵ The TOBOGM Research Group has recently found that early treatment of GDM before 20 weeks gestation improves composite neonatal outcomes, though conversely, treatment did not improve maternal outcomes.²⁵ Also of note, Rowan (2022) demonstrated early treatment of a first antenatal HbA_1c of 41–46mmol/mol reduces the likelihood of LGA, pre-eclampsia and pre-term birth.²⁶ In contrast, the GEMS Study demonstrated that a lower diagnostic threshold for GDM (fasting plasma glucose level of ≥5.1mmol/l, 1-hour level of ≥10.0mmol/l, or a 2-hour level of ≥8.5mmol/l) did not improve pregnancy outcomes, but leads to an increased consumption of healthcare services.²⁷ However, their sub-group analysis showed that women who were treated for “milder” GDM based on the lower glycaemic criteria had a reduced risk of LGA and pre-eclampsia compared to women with similar glucose test results who received no treatment.²⁷

This study has reliably captured HbA_1c for every person who had early pregnancy blood tests during the study time period. All samples were analysed in the same laboratory, reducing potential analytical error. As testing early pregnancy HbA_1c is routine in Aotearoa New Zealand, no additional investigations were required. Moreover, utilising HbA_1c, as opposed to glucose measures, avoids pre-analytical glucose errors. The study was undertaken within the same locality; therefore, management of each person’s pregnancy followed the same guidelines.

Limitations include the short time period (6 months) and the smaller than expected sample size. The difference in birth weight was much smaller than expected, only 42.3g, so we did not have sufficient power to detect the expected difference of 150g. This difference is unlikely to be clinically meaningful; therefore, a larger sample size is required to identify a difference of this magnitude. Seven people had no BMI recorded and were excluded, further reducing sample size. One person had an HbA_1c of 15mmol/mol due to a history of hereditary spherocytosis, so this HbA_1c does not accurately reflect glucose status. It is possible that there were other participants with undiagnosed haemoglobinopathies that may have influenced their HbA_1c results. Additionally, pre-eclampsia is an important adverse outcome associated with hyperglycaemia in pregnancy but was not included as a secondary outcome. There was an unexpectedly low number of participants recorded as having pre-eclampsia, suggesting the data may be incomplete.

In conclusion, there is no evidence of a difference in outcomes of birth weight, neonatal or maternal outcomes in pregnant people who have an early pregnancy high normal HbA_1c or normal HbA_1c. HbA_1c, early in pregnancy, identifies those with pre-diabetes or undiagnosed diabetes, allowing appropriate management of these higher risk groups. There is no evidence that HbA_1c can be used to stratify risk outside of this range. However, given the continuous nature of an HbA_1c measure, and the pregnancy effects on the HbA_1c analysis, it is possible that these diagnostic cut points are not accurate in pregnancy. Further exploration of the appropriate use of HbA_1c in early pregnancy is important if Aotearoa New Zealand is to continue using it as a screening tool.