FormalPara Key Summary Points

Why carry out this study?

Earlier comparative studies showed that continuous subcutaneous insulin infusion (CSII) in pregnant women with pre-existing type 1 diabetes (pT1D) lowered glycated haemoglobin (HbA1c) to a greater extent than multiple daily insulin injections (MDI) in the first trimester, but a limited number of studies included preconception HbA1c in analysis. At the same time, there are scarce data about glucose variability (GV), a factor that may contribute to glycaemic control, as well as the occurrence of hypoglycaemic episodes, during pregnancy with pT1D on different insulin treatments.

Therefore, the aim of our study was to evaluate the effectiveness of long-term CSII compared with MDI injections for glycaemic control and variability, hypoglycaemic episodes and maternal/neonatal outcomes in pregnant women with pT1D.

What was learned from the study?

We demonstrated the advantages of long-term CSII on glucose control due to the lower level of HbA1c achieved already in preconception and maintained during whole pregnancy. Also, we showed diminished GV and less frequent hypoglycaemic events only in the first trimester on CSII, an aspect of glycaemic control insufficiently evaluated in previous investigations.

In conclusion, treatment with CSII has resulted in favourable reduction of HbA1c in the preconception period and each trimester in pregnancy. Moreover, long-term CSII treatment demonstrated more stable metabolic control with less GV of fasting glycaemia and fewer hypoglycaemic episodes only during early pregnancy.

Introduction

It has been previously shown that achieving and maintaining target glycaemic control in pregnant women with pre-existing type 1 diabetes (pT1D) can be extremely challenging in clinical settings [1]. However, obtaining tight glycaemic control is highly recommended from preconception until delivery [1, 2].

During the first trimester, improved insulin sensitivity increases the risk of hypoglycaemia. However, in late gestation, insulin sensitivity decreases, which requires correction of pronounced postprandial glycaemic spikes, i.e., frequent adjustment of insulin doses [3].

Earlier comparative studies showed that continuous subcutaneous insulin infusion (CSII) in pregnant women with pT1D lowered glycated haemoglobin (HbA1c) to a greater extent than multiple daily insulin injections (MDIs) in the first trimester [4]. Furthermore, recent data suggested superiority of CSII treatment in reducing HbA1c during the whole pregnancy [5]. The latest meta-analysis on the topic revealed better glycaemic control on CSII during the first trimester, with a less pronounced beneficial effect in the second and third trimesters [6].

Previous studies also reported that women with pT1D showed higher glucose variability, with a twofold higher risk of hyperglycaemic excursions during the day, than healthy pregnant women [7]. However, the well-known indicator—HbA1c level—does not reflect glucose variability during pregnancy [8, 9]. In this way, glucose variability is a factor that may contribute to glycaemic control, which is suggested from the recently obtained data based on continuous glucose monitoring (CGM) reports in pregnancy in pT1D, but more evaluation is required with respect to different types of available insulin treatment in pregnancies complicated by pT1D [10]. In this context, it was reported that pregnant women with pT1D using CGM spent more time in target and less time hyperglycaemic, and they had less glycemic variability, than did pregnant control participants, with comparable severe hypoglycaemic episodes and time spent hypoglycaemic [11]. In addition, results from a recently published study obtained in non-pregnant T1D patients suggested stable and long-term improvement in glycaemic parameters with CGM, irrespective of type of insulin delivery [12].

Besides, numerous studies showed an association of increased risk of adverse pregnancy outcome with higher levels of HbA1c during conception [13, 14].

To our knowledge, studies comparing CSII and MDI in pregnancy have had conflicting results due to different study designs, populations followed, initiations of CSII therapy, etc. In this context, some of them have shown an increase in maternal weight, neonatal hypoglycaemia and birth weight with CSII therapy [15,16,17].

A limited number of studies included preconception HbA1c in the analysis, and a minority showed lower HbA1c levels in the CSII group [6]. At the same time, there are scarce data on the occurrence of hypoglycaemic episodes during pregnancy with pT1D on different insulin treatments. Still, the effect of long-term CSII therapy in pregnant women with pT1D on parameters of metabolic control from preconception till delivery and adverse pregnancy events remains to be clarified.

Within this context, the aim of our study was to evaluate the effectiveness of long-term CSII compared with MDI injections for glycaemic control and variability, hypoglycaemic episodes and maternal/neonatal outcomes in pregnant women with pT1D.

Methods

Study Design and Settings, Ethics and Informed Consent

The study was an observational prospective cohort study and included 128 consecutive singleton pregnant women with pT1D who were treated at the Department for metabolic disorders, intensive treatment and cell therapy in diabetes [Clinic for Endocrinology, Diabetes and Metabolic Diseases in Belgrade (public clinic, tertiary care level)] for diabetes from 1 January 2010 to 31 December 2017.

Pregnant women had been completely informed about the study before giving informed consent to participate. All procedures followed were part of the standard medical care and in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and the Helsinki Declaration of 1964, as revised in 2013. Approval of the study by an ethics committee was considered unnecessary because all patient data were extracted from the department’s database and were anonymous, assuring that complete data protection of the participants was provided.

Inclusion and Exclusion Criteria

In general, CSII has been suggested for all women of childbearing potential with pT1D on MDI in case of unsatisfactory metabolic control related to treatment targets. Current treatment of MDI or CSII had been initiated at least 1 year before conception in women with pT1D.

For the present observational study, 80 pregnant women were treated with MDI (MDI group) using a rapid insulin analogue (aspart) before meals and a bedtime basal analogue (detemir), while 48 were treated with CSII (Medtronic Paradigm 722, Medtronic Paradigm Veo, insulin aspart) based on boluses given before meals, and the basal infusion rate was adjusted according to the glycaemic targets for the fasting and preprandial state. Both insulin treatments were fully reimbursed by the National Health Insurance Fund. Women who initiated CSII during actual pregnancy, women with gestational diabetes or type 2 diabetes, pregnancy obtained from assisted reproductive technology, or twin pregnancies as well as women referred from the general hospital or with incomplete medical records were excluded from the study.

Measurements and Definitions of Measurements

In both groups we recorded: age, duration of T1D, current treatment (MDI or CSII) and gestational weight gain (GWG). These data were taken from medical records. Body mass index (BMI) was calculated using the equation BMI = weight (kg)/height (m)2.

HbA1c in preconception and each trimester was measured using electrophoresis. Values for fasting, postprandial glycaemia as well as hypoglycaemia during trimesters were taken from standard logbooks. Hypoglycaemia was defined by glycaemia < 3.9 mmol/l. Based on the number of hypoglycaemia episodes per week per patient, we divided all women into tertiles: rare (0–3 hypoglycaemia), moderate (4–6 hypoglycaemia) and frequent (≥ 7 hypoglycaemia) hypoglycaemia episodes per week per patient. Glucose variability was determined by the coefficient of variation (CV = SD/x̄) of SMBG at different points (fasting or postprandial) during trimesters.

We also measured the total daily insulin requirement and calculated the insulin sensitivity factor (ISF) (using Eq. 100/total daily dose) and carbohydrate ratio (CHR) (using Eq. 500/total daily dose) in each trimester.

Pregnancy Outcomes

Maternal and foetal outcomes, including gestational hypertension, preeclampsia [diagnosed by an obstetrician: blood pressure ≥ 140/90 mmHg with proteinuria (≥ 0.5 g)], preterm delivery (before 37 completed gestational weeks), emergency caesarean section, gestational week at delivery, Apgar score in the first minute, birth weight, newborn large for gestational age (LGA; > 90th percentile according to growth charts for the Serbian population), newborn small for gestational age (SGA) (< 10th percentile according to growth charts for the Serbian population), macrosomia (birth weight > 4000 g), stillbirth, neonatal hypoglycaemia (glycaemia < 1.5 mmol/l or requiring intravenous glucose infusion), congenital malformation, admission to the neonatal intensive care unit (NICU) and neonatal hyperbilirubinaemia, were collected from the obstetrical medical records [16].

Model of Care: Preconception and Pregnancy

Planning pregnancy was defined as having preconception counselling. Standard preconception care for women with pT1D and childbearing potential was applied for all women irrespective of their treatment with MDI or CSII [1]. Pregnancy was confirmed by ultrasound examination and women had regular prenatal obstetrics visits according to the National Guideline for Pregnancy Care at the Clinic for Gynaecology and Obstetrics, Clinical Centre of Serbia [18].

HbA1c was recorded before conception and measured in each trimester. In addition, pregnant women were asked to do self-monitoring of blood glucose (SMBG) at least five times a week (preferably more frequently) during the whole pregnancy. These measurements were taken per day with a glucometer, when fasting, before and 2 h after meals, at bedtime and during the night for a total of at least nine measurements. According to the current recommendations, suggested fasting targets were < 5.3 mmol/l and < 6.8 mmol/l 2 h after meals, respectively [2]. Women were also asked to record hypoglycaemia, which was defined by glycaemia < 3.9 mmol/l. SMBGs were reviewed and treatment adjusted at least once a week during the first trimester and twice a month or more frequently, if necessary, until delivery.

We advised women about optimal timing for prandial insulin (15 min before meals in early pregnancy and 30–40 min before meals in late pregnancy) [19]. In addition, all pregnant women with pT1D were re-educated about carbohydrate counting, and we suggested recording the composition of each meal to better adjust the insulin dose. Sixty-seven per cent of women on CSII used the bolus calculator option during pregnancy.

All instructions listed above regarding monitoring of metabolic control are part of the standard procedures implemented in usual clinical practice in pregnant woman with pT1D at our department.

The blood pressure target was a systolic blood pressure < 130 mmHg and a diastolic blood pressure < 80 mmHg, measured routinely using a sphygmomanometer at each visit.

Statistical Analysis

Data are presented as mean ± SD values and were tested for normal distribution using the Kolmogorov-Smirnov test. The differences were tested for significance using Student’s t test, while the chi-square test and Mann-Whitney were used for parametric and non-parametric data, respectively. The Levine F test was used to access differences between CVs. A two-sided p value of 0.05 was considered statistically significant. Data were analysed using Statistical Package for the Social Sciences (SPSS) software (Advanced Statistics, version 22.0, Chicago, IL USA).

Results

Baseline Characteristics

Clinical characteristics and duration of previous treatment (MDI or CSII) of pregnant women with pT1D are shown in Table 1. There was no significant difference with respect to age and duration of T1D between groups (Table 1). Furthermore, we did not find any significant difference regarding BMI and GWG between the groups.

Table 1 Baseline characteristics of pregnant women with pre-existing type 1 diabetes: MDI vs. CSII group
Full size table

Results Pertaining to Glycaemic Control

In preconception, women on CSII showed lower levels of HbA1c than the MDI group (7.1 ± 0.1 vs. 7.9 ± 0.2%, p = 0.03). Moreover, HbA1c was significantly lower during trimesters in the CSII group compared with MDI [first (6.9 ± 0.1 vs. 7.7 ± 0.2%, p = 0.02), second (6.6 ± 0.1 vs. 7.2 ± 0.1%, p = 0.003) and the third (6.5 ± 0.1 vs. 6.8 ± 0.1%, p = 0.02) trimester] (Fig. 1).

Fig. 1
figure 1

HbA1c in preconception and during trimesters in pregnant women with pT1D: comparison of different intensified insulin regimens: MDI (dark gray) vs. CSII (light gray). Values are mean ± SE. Mann-Whitney test was used to evaluate the differences between groups. There was a statistically significant difference with respect to HbA1c in preconception and during pregnancy between groups. Pregnant women with pT1D on CSII had lower HbA1c levels in preconception (*p = 0.03), the 1st trimester (**p = 0.002) and 2nd trimester (#p = 0.003) as well as the 3rd trimester (##p = 0.02) compared with MDI. Pre-existing type 1 diabetes, pT1D

Full size image

We did not find any significant difference between average fasting plasma glucose levels from SMBG during pregnancy when comparing these two intensified insulin regimens (Fig. 2). Simultaneously, the groups were compared according to average PPG during the whole pregnancy excluding late gestation. In the third trimester, pregnant women with pT1D treated with CSII had a lower average PPG level after breakfast than the MDI group (6.8 ± 1.0 mmol/l vs. 7.7 ± 1.3, p < 0.001) (Fig. 2c).

Fig. 2
figure 2

Average fasting and postprandial glycaemia obtained from SMBG in pregnant women with pT1D: MDI vs. CSII. Data are presented as mean ± SD values. T test was used to evaluate the differences between groups. There was a statistically significant difference in the 3rd trimester in the CSII group due to the lower average PPG level after breakfast comparing with MDI (C; *p < 0.001). FPG fasting plasma glucose, PBG postprandial breakfast glucose, PLG postprandial lunch glucose, PDG postprandial dinner glucose

Full size image

When we analysed the percentage of pregnant women who achieved target PPG levels, there was no difference between the groups (Fig. 3), except in late pregnancy. In the third trimester after breakfast, a significantly higher percentage of pregnant women on CSII reached their target 2-h PPG level (57.4 vs. 18.4%, p < 0.001)(Fig. 3b).

Fig. 3
figure 3

Percentage of pregnant women with pT1D achieved target FPG (a) and PPG (b–d): MDI vs. CSII. Data are presented as  %. Chi-square test was used to evaluate the differences between groups. There was a statistically significant difference in the 3rd trimester among CSII group because of the higher percentage of women reaching their target 2 h after breakfast compared with MDI (b; #p < 0.001)

Full size image

Results Related to Glycaemic Variability

Pregnant women on CSII treatment had lower CV of SMBG measured fasting in the first trimester (17.1 vs. 28.4%, p < 0.001) and after breakfast in the third trimester compared with the MDI group (14.1 vs. 17%, p = 0.012) (Table 2).

Table 2 CV of FPG and PPG obtained from SMBG among pregnant women with pT1D: MDI vs. CSII
Full size table

Results for Hypoglycaemia

Analysing the mean number of hypoglycaemia episodes per week per patient, the CSII group had fewer hypoglycaemia episodes per week in the first trimester compared with MDI (2.0 ± 1.7 vs. 4.8 ± 1.5, p < 0.01), but the difference diminished in the second and third trimester between groups (Fig. 4). At the same time, a higher percentage of women on CSII was in the first tertile (rare) compared with MDI in the first trimester (Fig. 5a) (0–3: 79.1 vs. 29.1%; 4–6: 18.8 vs. 65.8%; ≥ 7: 2.1 vs. 5.1%, p < 0.01, respectively), while there was no difference in the second and third trimester (Fig. 5b, c).

Fig. 4
figure 4

Mean number of hypoglycaemic episodes per week per patient through trimesters in pregnant women with pT1D: MDI vs. CSII. Data are presented as mean ± SD values. Mann-Whitney test was used to evaluate the differences between groups. There was a statistically significant difference in the 1st trimester in the CSII group due to a smaller mean number of hypoglycaemia episodes per week per patient than on MDI (*p < 0.001)

Full size image
Fig. 5
figure 5

Percentage of pregnant women with pT1D distributed in tertiles according to number of hypoglycaemia episodes per week: MDI vs. CSII. Data are presented as % of women. Mann-Whitney test was used to evaluate the differences between groups. There was a statistically significant difference in the first trimester in the CSII group due to the higher percentage of patients in the first tertile (rare) (p < 0.01). There were defined tertiles in relation to the number of hypoglycaemia episodes per week: rare (0–3 hypoglycaemia), moderate (4–6 hypoglycaemia) and frequent (≥ 7 hypoglycaemia)

Full size image

Results Related to Pregnancy Outcomes

We found no difference in the incidence of gestational hypertension between groups, whereas the incidence of preeclampsia and SGA newborn could not be analysed because of a limited number of events. Groups were also comparable in the incidence of preterm delivery, emergency caesarean section, macrosomia, neonatal hypoglycaemia, admission to NICU as well as gestational week at delivery, Apgar score in the first minute and LGA newborns. Simultaneously, women in both groups delivered newborns with comparable neonatal weight, while newborns with neonatal hyperbilirubinaemia were found more often in the MDI group. We did not report congenital malformations in either groups (Table 3).

Table 3 Adverse pregnancy events in pregnant women with pT1D: MDI vs. CSII
Full size table

Table S1 represents the data regarding insulin requirements in both regimen groups, CSII and MDI, as well as the calculated CHR and ISF in each pregnancy trimester. Total insulin daily dose was significantly lower in the CSII compared with the MDI group in the first (34.0 ± 6.4 vs. 41.6 ± 10.3 U, p = 0.001), second (38.5 ± 8.7 vs. 46.3 ± 12.4 U, p = 0.001) and third trimester (42.0 ± 10.6 vs. 52.0 ± 16.8 U, p = 0.001).

Discussion

In our observational study, which encompasses pregnant women with pT1D, we detected better glycaemic control on CSII than on MDI treatment, based on data evaluation from preconception through the trimesters till delivery. Moreover, the majority of pregnant women with pT1D achieved target glycaemic goals already in preconception and maintained them during the trimesters, using long-term CSII.

Data from previous studies focusing on glycaemic control in pregnant women with pT1D using different intensified insulin treatments (CSII or MDI) still remained inconsistent for superiority of either of them [4,5,6], partly because they did not focus on preconception monitoring or sustainability during the whole pregnancy. Moreover, glycaemic fluctuations, GV and hypoglycaemia, were suggested to influence the course of pregnancy in women with long-standing pT1D, but the effectiveness of different insulin treatments for glycaemic control and variability and hypoglycaemic episodes in pregnant women with pT1D has not been elucidated [8].

Some prior studies in this field revealed better glycaemic control with CSII treatment in complicated T1D during pregnancy [20], in line with our study results showing lower HbA1c levels among women with long-standing T1D and CSII.

A recently published meta-analysis comprising 47 predominantly non-interventional studies showed higher HbA1c reduction in early pregnancy, but this advantage was diminished in the second and third trimesters [6].

However, some previous investigations suggested a comparable effect of CSII and MDI on glycaemic control [21]. Nevertheless, most of these available studies included pregnant women in the second trimester or the ones initiating pump treatment during actual pregnancy.

Moreover, a limited number of studies reported better preconception glycaemic control in women with pT1D on CSII compared with MDI treatment [22]. Our results also confirmed a lower HbA1c level in preconception in women on CSII than in those on MDI treatment.

Although our women were not randomized to the MDI or CSII treatment arm, we tried to overcome potential bias by homogenizing groups in respect to age, duration of disease and treatment. In our study, in contrast to other studies on the topic, we followed pregnant women with long-standing T1D and a long-term use of either type of insulin treatment before pregnancy.

During pregnancy, we focused not only on HbA1c, having in mind it might not adequately reflect different aspects of glucose control, but also analysed acute glucose fluctuations or GV, which contribute to adverse pregnancy events [8]. In addition, a better understanding of the pattern of blood glucose fluctuations in all three trimesters of pregnancy could make it easier to optimize glycaemic control in pregnant women with diabetes [7]. In that context, in our study we used CV as an established parameter of GV. Pregnant women with pT1D on CSII treatment had a lower CV of fasting glycaemia in early pregnancy as well as after breakfast in late pregnancy (obtained from SMBG). In that context, our results might imply that assessing CV from SMBG, even without the use of a more sophisticated and recently established standard method of blood glucose measurement, might be useful and clinically relevant as a surrogate marker of GV. Furthermore, in a recently published meta-analysis a significant number of studies, nearly half of those included, used only SMBG as a glucose monitoring method [6].

In addition, our findings are in line with previously published studies showing no difference in mean FPG and PPG between CSII and MDI in the first and second trimesters [23,24,25]. However, according to our results, pregnant women with pT1D using CSII had lower mean PPG levels (after breakfast) in late pregnancy, which is an advantage in clinical settings. Moreover, besides better PPG regulation, a higher percentage of pregnant women with pT1D on CSII in our study reached the target value for the PPG goal in the third trimester after breakfast.

On the other hand, data on hypoglycaemic episodes in pregnant women with pT1D are quite limited. It has been shown that compared with MDI, CSII was associated with a slightly lower HbA1c level and a smaller risk of severe hypoglycaemia [26]. Also, it is important to note that the effort to achieve better glycaemic control together with less GV could predispose to hypoglycaemia. Our results are in agreement with studies showing the superiority of CSII regarding fewer hypoglycaemic episodes during early pregnancy [1, 27], which is an important clinical challenge because pregnant women with pT1D are most vulnerable to hypoglycaemia in that period [28]. As we have shown in this study, in respect to number of hypoglycaemia episodes per week, the majority of women on CSII were in the first tertile (rare hypoglycaemic events), which implies more stable glycaemic control with less GV and hypoglycaemia. In addition, the CSII group had lower TDD compared with MDI, in line with previously published data [6].

Current data on the relationship between glucose variability and adverse foetal outcome are still conflicting. Recent studies using CGM reported a correlation between parameters of glucose variability and foetal growth [29,30,31] in pregnant women with pT1D or gestational diabetes. Although earlier studies had conflicting results [32], in our study we demonstrated an availability of decreased glucose variability on CSII, even in clinical settings without the use of CGM, which might be of significant clinical relevance.

Concerning pregnancy outcomes, we could not demonstrate that CSII treatment induced a lower incidence of adverse pregnancy events compared with MDI therapy. The studies in this field gave conflicting results and, despite some positive results, most of the studies reported that optimal metabolic control during pregnancy does not always guarantee a favourable outcome [32, 33].

In addition, limitations of our study are the observational study design and use of a surrogate marker for glycaemic variability CV obtained by SMBG. Also, the relatively few adverse events are a limitation related to adequate evaluation of the incidence of adverse outcomes and do not allow formulating further conclusions.

Nonetheless, we demonstrated the advantages of long-term CSII on glucose control due to the lower level of HbA1c achieved already in preconception and maintained during the entire pregnancy. Also, we showed a diminished GV and less frequent hypoglycaemic episodes only in the first trimester on CSII, an aspect of glycaemic control insufficiently evaluated in previous investigations.

Conclusions

In conclusion, treatment with CSII resulted in a favourable reduction of HbA1c in the preconception period and each trimester in pregnancy. Moreover, long-term CSII treatment demonstrated more stable metabolic control with less GV of fasting glycaemia and fewer hypoglycemic episodes only during early pregnancy.