Maternal third trimester hyperglycaemic excursions predict large-for-gestational-age infants in type 1 diabetic pregnancy

1. Introduction 

Maternal hyperglycaemia is an important determinant of foetal macrosomia [1]. However, despite improved glycaemic control in pregnant women with type 1 diabetes, the incidence of macrosomia is still high [2], [3]. A number of studies aiming at the relation between foetal macrosomia and maternal glycosylated haemoglobin have variably reported no correlation [4], [5], predictive value only for third trimester HbA1c [3], [6], [7], main influence of first or second trimester HbA1c [8], [9], or greater relation to preconception HbA1c [10].

It appears that true maternal glucose levels should better predict foetal macrosomia, since a wide range of glucose values are associated with normal glycosylated haemoglobin [11]. Among different possible glucose determinations during pregnancy in women with diabetes, overall mean glucose levels [11], [12], postprandial capillary glucose [13] and specifically third trimester postprandial glucose [14], [15] have been associated with the incidence of large-for-gestational-age (LGA) infants. Episodic hyperglycaemia [16] and recurrent episodes of hypoglycaemia [17] have also been related to excessive foetal growth in pregnant women with type 1 diabetes.

The purpose of this study was to evaluate which maternal glycaemic parameter, from preconception throughout pregnancy, better predicts LGA infants in women with type 1 diabetes.

2. Research design and methods 

This study was performed at the Hospital Universitario La Paz, Madrid, Spain. Ethical approval for this research was provided by the Hospital Ethical Committee and all mothers gave their informed consent.

From January 1999 to March 2003 a total of 161 women with pregestational type 1 diabetes were attended at the Diabetes and Pregnancy Unit. We selected 73 mother–infant pairs, who met the following inclusion criteria: preconception management at the hospital, singleton pregnancy and live-born infants with no evidence of intra-uterine growth retardation.

Patients were seen on an outpatient basis, every 4 weeks during preconception care and every 1–2 weeks throughout pregnancy. All women were requested to perform home blood glucose monitoring six times daily (three preprandial; three 2-h postprandial) using the same type of glucose meter (One Touch profile; LifeScan, Milpitas, CA, USA) and glucose readings stored in the meter were downloaded to a computer (Diabetes management software: In Touch; LifeScan, Milpitas, CA, USA) at each visit. The glucose targets were: preprandial 3.9–5.6mmol/l (70–100mg/dl) and postprandial 5.6–7.8mmol/l (100–140mg/dl). Women were on flexible basal bolus regimens and were trained to adjust their insulin dose to achieve the glucose targets. 71 women used multiple injections of regular and NPH insulin and 2 women used continuous subcutaneous insulin infusion.

Mean glucose levels were obtained from the glucose readings for the 28 days prior to conception (preconception), first trimester (from conception to 12 complete gestational weeks), second trimester (from 13 gestational weeks to 24 complete gestational weeks) and third trimester (from 25 gestational weeks to 36 complete gestational weeks). The mean number of glucose readings/day was 4.9 for preconception, 4.7 for first trimester, 5.8 for second trimester and 5.6 for third trimester. Preprandial weighted mean glucose and postprandial weighted mean glucose were calculated for each period. The percentage of glucose values above the glucose goal and the percentage of glucose values below the glucose goal were obtained from the glucose records for each period. Maternal HbA1c was measured every month using high-performance liquid chromatography (BioRad, Richmond, RA). The HbA1c levels included in this study were the last measurement during preconception (nearest to conception) and the last measurement of each trimester (nearest to 12, 24 and 36 weeks’ gestation).

Gestational age was based on the last menstrual period and on foetal ultrasonography performed between 8 and 12 weeks’ gestation. Patients’ age, diabetes duration, vascular complications (retinopathy or nephropathy), parity, smoking status, prepregnancy weight, height, weight gain during pregnancy, diagnosis of pre-eclampsia, gestational age at delivery, route of delivery and infants’ sex and birth weight were obtained from clinical records. The rate of maternal weight gain was calculated as: maternal weight gain—infant birth weight/total weeks’ gestation. Neonates>90th percentile using the Battaglia and Lubchenco growth standards were considered LGA [18]. The rate of macrosomia was also calculated using a Spanish neonatal growth chart [19].

Statistical analyses were conducted using SPSS version 8.0 statistical software (SPSS Inc., Chicago, IL, USA). To compare mean values among quantitative variables, the independent-samples Student's t-test was used. To compare categorical variables, the χ2 and Fischer exact tests were used. Forward stepwise multivariate logistic regression analysis was performed to select significant factors when variables were considered jointly. Results are reported as mean S.D. or as percentages. A p-value<0.05 was considered significant.

3. Results 

Of the 73 neonates, 36 were appropriate for gestational age (AGA) and 37 were LGA. The rate of macrosomia was 47.9% using 1998 Spanish birth weight standards. Mean birth weight of AGA infants was 3139±387 and mean birth weight of LGA infants was 3830±387. The incidence of congenital malformations, which were all minor anomalies, was similar for AGA and LGA neonates (5.4% versus 8.1%; p=0.850). Other neonatal outcomes were also similar for AGA and LGA infants: birth trauma (8.3% versus 5.4%; p=0.647), hypoglycaemia (25% versus 29.7%; p=0.651), hypocalcemia (2.8% versus 5.4%; p=1.000), hyperbilirubinemia (19.4% versus 37.8%; p=0.083) and respiratory distress syndrome (13.9% versus 2.7%; p=0.107). Maternal characteristics associated with neonates who were AGA and those who were LGA were similar, with the exception of the proportion of smokers and history of macrosomia in a previous pregnancy (Table 1).

Table 1. Maternal characteristics associated with LGA and AGA neonates

		LGA	AGA	p-Value
	Number	37	36
	Maternal age (years)	31.9±2.4	32.0±3.4	0.814
	BMI (kg/m2)	24.9±3.4	24.0±3.2	0.219
	Diabetes duration (years)	15.0±7.9	13.4±7.2	0.356
	Rate of maternal weight gain (kg/week)	0.26±0.11	0.24±0.11	0.294
	First trimester daily insulin dose (U/kg)	0.66±0.14	0.63±0.17	0.409
	Second trimester daily insulin dose (U/kg)	0.81±0.18	0.76±0.20	0.266
	Third trimester daily insulin dose (U/kg)	1.03±0.22	0.95±0.31	0.226
	Number of pregnancies	1.8±1.0	1.8±0.9	0.983
	Macrosomia in previous pregnancy (%)	24.3	5.6	0.025
	Retinopathy (%)	16.2	27.8	0.269
	Nephropathy (%)	2.7	5.6	0.615
	Smoking (%)	18.9	44.4	0.024
	Pre-eclampsia (%)	5.4	13.9	0.261
	Gestational age at delivery (weeks)	37.4±1.4	37.5±1.6	0.796
	Caesarean section (%)	56.8	52.8	0.816

Mean glucose and the percentage of glucose readings above glucose targets were significantly higher in mothers of LGA babies in all three trimesters. Mean preprandial glucose was significantly higher in mothers of LGA babies in the third trimester and mean postprandial glucose was significantly higher in mothers of LGA babies in the second and third trimesters. HbA1c level was significantly higher in mothers of LGA babies only in the third trimester measurement. No significant differences in glycaemic parameters were seen between the mothers of AGA and LGA neonates in the preconception period (Table 2).

Table 2. Comparison of glycaemic parameters between mothers of LGA and AGA infants during preconception and each trimester of pregnancy

		28 days prior to conception		First trimester		Second trimester		Third trimester
		LGA	AGA	LGA	AGA	LGA	AGA	LGA	AGA
	Hb A1c (%)	6.8±0.7	6.5±0.6	6.5±0.7	6.2±0.5	6.1±0.6	5.9±0.6	6.2±0.5*	5.9±0.5
	Mean glucose
	mmol/l	8.0±1.0	7.6±1.0	7.8±0.8*	7.3±0.9	7.5±0.6*	7.2±0.8	7.4±0.6**	6.9±0.7
	mg/dl	143.7±17.2	136.8±17.9	140.2±14.7	131.7±15.5	135.8±11.5	128.9±14.4	133.8±11.1	124.8±12.8
	Mean preprandial glucose
	mmol/l	7.4±1.2	7.2±1.2	7.2±0.9	6.9±1.0	6.7±0.6	6.4±0.7	6.6±0.6**	6.2±0.8
	mg/dl	134.1±20.7	130.5±20.8	130.3±16.7	123.5±17.2	120.5±10.0	115.5±13.2	119.6±10.5	111.3±14.1
	Mean postprandial glucose
	mmol/l	8.8±1.2	8.4±1.5	8.6±1.0	8.1±1.1	8.6±0.9*	8.1±1.0	8.4±0.9*	7.9±0.9
	mg/dl	158.0±21.4	150.7±27.1	154.2±18.8	145.4±20.1	154.8±16.3	145.7±18.7	151.8±17.0	141.9±15.8
	% Glucose values>goal	44.3±10.6	40.9±12.3	43.9±7.9*	38.9±10.2	43.0±8.1*	37.6±10.2	42.1±8.3**	35.0±10.3
	% Glucose values<goal	16.5±6.3	18.2±6.8	17.2±3.9	17.8±6.5	17.2±4.4	17.6±5.3	12.9±4.4	15.1±5.1

*p<0.05; **p<0.01.

The predictive value of glycaemic parameters (mean glucose, mean preprandial glucose, mean postprandial glucose, percentage of glucose values above glycaemic goal) during each trimester of pregnancy, adjusted for smoking and history of macrosomia, was analysed using multiple logistic regression analysis. For each glycaemic variable, only the third trimester value was a significant predictor of LGA infants (Table 3). When all third trimester variables (mean glucose, mean preprandial, mean postprandial glucose, percentage of glucose values above glycaemic goal and HbA1c) were included in the logistic regression model, only the third trimester percentage of glucose values above glycaemic goal increased the risk of LGA babies.

Table 3. Adjusted odds ratio for LGA infants

		OR	95% CI	p-Value
	Third trimester mean glucose (mmol/l)	3.35	1.47–7.68	0.004
	Third trimester mean preprandial glucose (mmol/l)	2.97	1.31–6.73	0.009
	Third trimester mean postprandial glucose (mmol/l)	2.03	1.14–3.60	0.016
	Third trimester percentage glucose values>goal	1.09	1.02–1.15	0.005

4. Discussion 

The incidence of LGA infants in this group of women with type 1 diabetes, with overall good glycaemic control, who had attended the hospital for preconception care was still very high (50.7% or 47.9%, depending on the reference growth chart used), in concordance with that reported in other studies [2], [3].

This study shows that in women with type 1 diabetes mean glucose values during each trimester of pregnancy were higher in mothers of LGA infants than in mothers of AGA infants, while differences in maternal HbA1c could only be detected for the third trimester measurement. These data are in concordance with the fact that maternal glycaemic control measured by HbA1c is less sensitive to discriminate excessive foetal growth than true glycaemic levels [11]. Aside from glycaemic control, many factors have been related to macrosomia in women with type 1 diabetes. In our study, only non-smoking and a history of previous macrosomia were more frequent among mothers of LGA babies, in agreement with other reports [7], [20], therefore, the predictive value of glycaemic parameters was adjusted for these factors.

Some studies have suggested that tighter control at conception and in the first trimester of pregnancy have greater influence on birth weight than does glycaemic control later in pregnancy [8], [10], [21]. However, other studies indicate that glycaemic control in late pregnancy was the strongest determinant of birth weight [3], [6], [7], [14], [15]. In our study there were no differences in glycaemic parameters during the preconception period between mothers of LGA and AGA neonates. Moreover, when we analysed the predictive power of each trimester mean glucose values, only mean third trimester glucose was a risk indicator of LGA neonates. Again, when we analysed the effect of mean preprandial glucose or mean postprandial glucose during each trimester, only third trimester mean preprandial glucose or third trimester mean postprandial glucose were predictive of LGA babies. Our data confirm that third trimester glycaemic parameters of mothers with type 1 diabetes are more powerful predictors of foetal growth than glycaemic parameters earlier in pregnancy.

It has been proposed that intermittent maternal hyperglycaemia will have effects on foetal growth [16] and on perinatal outcome [22]. This study indicates that the overall percentage of glucose values above glycaemic target during each trimester of pregnancy is higher in mothers of LGA infants than in mothers of AGA infants, supporting the importance of episodic hyperglycaemia on foetal size. Interestingly, our data show that of all third trimester glycaemic parameters, the percentage of glucose values above glycaemic target is the most powerful predictor of LGA infants. Maternal third trimester postprandial glucose levels in women with type 1 diabetes have been found to be the most closely correlated with infant birth weight [14], [15]. Our data allow to add that any glycaemic excursion, either preprandial or postprandial, increases the risk of excessive foetal growth.

High incidence of macrosomia in type 1 diabetes pregnancies with apparent good glycaemic control, as observed in our population (mean HbA1c throughout pregnancy≤6.5%), is a well-known fact [2], [3], [4], [15], [18], [23]. The high rate of readings above glucose goals, which are not reflected in the HbA1c values [16], may account for the very high rate of macrosomia.

We conclude that in women with type 1 diabetes third trimester glycaemic parameters are more predictive of LGA infants than glycaemic parameters earlier in pregnancy or at preconception. Episodic hyperglycaemia during the third trimester of pregnancy is the strongest predictor of LGA infants.