Analysis glycemic variability in pregnant women with various type of hyperglycemia

Abstract

Objective

The study primarily aims to compare alterations in the daily patterns of glucose fluctuations across individuals with different kinds of diabetes in pregnancy and secondly investigate influencing factors that may react with glucose variations.

Methods

We conducted a retrospective cohort study of 776 pregnant women in Shanghai General Hospital. We grouped participants who were exposed to gestational hyperglycemia into 5 sub-groups [Type 1 diabetes (T1DM), Type 2 diabetes (T2DM), Overt diabetes, Gestational diabetes (GDMA1 and GDMA2). Demographic variables and GV parameters were compared among 5 groups through ANOVA-test and Chi-square test. We estimated odd ratios (ORs) for the association between glucose coefficient of variation (CV) and possible influencing variables.

Results

A final total of 776 pregnant women were analyzed. The proportion of pregnant women with pre-gestational diabetes was 31.83% (T1DM: 3.35%,T2DM: 28.48%), ODM 26.68%, and GDM was 41.49% (GDMA1:18.04%, GDMA2: 23.45%). T1DM group performed greatest glucose fluctuations with a CV value 35.02% whereas the number in all the other groups was no more than 22.82% (ODM group). In terms of achieving glycemic control target, only 57.70% participants hit the goal while all the other groups achieved the standard with at least a percentage of 94.20% (ODM group). Other parameters (GMI < 6.0%, GA < 15.70% and HbA1c < 6.0%) showed similar trends in each group. On multivariate logistic regression analysis of possible factors influencing CV, only body mass index (BMI) (OR: 0.754, 95% CI: 0.585–0.971; P = 0.029), HOMA- β (OR:0.969, 95%CI: 0.959–0.976; P = 0.037) and fasting plasma glucose (FPG) (OR: 1.832, 95% CI: 1.170–2.870; P = 0.008) reached statistical significance.

Conclusions

Pregnant women with type 1 or type 2 diabetes exhibit significantly greater glycemic variability compared to those with gestational diabetes, with the ODM group showing intermediate variability, and BMI, HOMA-β, and FPG identified as independent risk factors for unstable glucose variability.

Introduction

In China, Gestational diabetes mellitus (GDM) causes multiple complications and affects 14% of pregnancies globally with a GDM incidence of 14·8%, which continues to increase [1, 2]. GDM and other forms of gestational hyperglycemia, such as overt diabetes mellitus (ODM) and pregestational diabetes (type 1 and type 2), are associated with numerous adverse maternal and fetal outcomes [3]. These complications include preeclampsia, macrosomia, preterm birth, and an increased risk of developing type 2 diabetes later in life. For the fetus, the risks are equally concerning, including birth defects, intrauterine growth restriction, and neonatal hypoglycemia [4, 5]. Given these potential risks, improveing management of GDM in China is of paramount importance to reduce maternal and fetal morbidity.

According to the International Association of Diabetes in Pregnancy Study Groups (IADPSG) 2010 criteria, gestational hyperglycemia mainly includes gestational diabetes mellitus (GDM), overt diabetes mellitus (ODM) during pregnancy, and pregestational diabetes [type 1 diabetes diagnosed before pregnancy, type 2 diabetes mellitus (T2DM), or special types of diabetes] [6, 7].

Glycemic variability (GV) is a crucial aspect of glucose homeostasis and has garnered increasing attention in recent years. Studies suggest that excessive fluctuations in glucose levels may contribute to the development of diabetes-related complications by promoting oxidative stress, endothelial dysfunction, and inflammatory pathways, ultimately resulting in microvascular and macrovascular damage [10, 11]. These effects are particularly concerning in pregnant women, as elevated GV can increase the risk of adverse maternal and fetal outcomes, including preeclampsia, macrosomia, and preterm birth. Despite its clinical importance, GV remains underexplored in gestational hyperglycemia, especially in distinguishing between different types of diabetes during pregnancy [12, 13]. High variability in glucose levels can lead to elevated levels of free radicals and endothelial dysfunction, which can trigger pathological pathways, resulting in tissue damage including but not limited to micro‐ and macrovascular damage, leading to the development of diabetes-related embryopathy [14]. Diabetes management has identified the reduction of GV as a therapeutic objective [10].

Traditional methods for glucose monitoring, such as self-monitoring of blood glucose (SMBG) and HbA1c, have been widely used in clinical practice. However, these approaches have limitations in reflecting short-term glucose fluctuations, which are crucial in assessing overall glucose control, especially during pregnancy [15,16,17]. HbA1c, although considered a reliable marker for long-term glucose control, is less effective in capturing acute glucose fluctuations, particularly in pregnant women. Physiological changes during pregnancy, such as altered red blood cell turnover, can lower HbA1c levels, leading to an inaccurate assessment of glucose control. Similarly, SMBG provides intermittent snapshots of glucose levels and fails to account for variability throughout the day. As a result, these methods may not fully reflect the true glucose control status, particularly in gestational diabetes [18, 19]. Recent attention has hence shifted to evaluate the role of continuous glucose monitoring (CGM) data in pregnancy [20].

The enhanced accuracy of sensors, the greater convenience, and the coverage of medical insurance have all contributed to the increasing adoption of continuous glucose monitoring system (CGMS) in pregnancy. CGMS gives greater insight into glucose levels throughout the day, revealing the magnitude, frequency, and duration of glucose excursions that conventional glucose self-monitoring can not provide. Yet, the successful integration of CGMS technology into everyday clinical practice is still relatively low. This could be attributed, at least in part, to the absence of clearly defined and universally accepted glycemic targets that can be targeted by both healthcare providers and individuals with diabetes. Despite the establishment of unified recommendations for utilizing key CGM metrics for pregnancy women in a few articles [21, 22], there is still a lack of continuously monitored data detailing the glycemic characteristics of women with different types of gestational hyperglycemia nor acknowledged variables acting on GV formally endorsed by diabetes professional organizations.

Thus, the current study (CGM study) primarily aims to compare alterations in the daily patterns of glucose fluctuations, calculate time in range (TIR), above range (TAR) and below range (TBR) across individuals with different kinds of diabetes in pregnancy (T1DM, T2DM, ODM, GDMA1 and GDMA2). Based on above discovery, finally we investigate influencing factors that may relate with GV. By exploring the potential factors influencing GV, such as body mass index (BMI) and fasting plasma glucose (FPG), our study provides valuable data that could help clinicians optimize diabetes management strategies, reduce complications, and improve both maternal and fetal outcomes.

Material and methods

Study design and population

This study was designed as a retrospective cohort, observational, single-center study conducted at Shanghai General Hospital. This study was conducted between May 2018 and May 2021 at Shanghai General Hospital. Inclusion criteria were: (1) Pregnant women aged ≥ 18 years; (2) Diagnosis of gestational hyperglycemia or pregestational diabetes based on IADPSG criteria or a confirmed diagnosis of pregestational diabetes (T1DM, T2DM, ODM, GDMA1, GDMA2); (3) Completion of at least 3 consecutive days (72 h) of CGM examination with data availability. (4) Complete clinical data required for this study. The final cohort comprised 776 women after excluding 24 cases that did not meet the criteria.

Demographic and parity information for included women was obtained from the electronic maternity medical record system in Shanghai General Hospital, into which routine pregnancy, birth, and complications data can be entered by all health care providers.

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Shanghai General Hospital ([2021]045), which waived the requirement for informed consent due to the retrospective nature of the study. The data used in this study were extracted from electronic medical records and CGM reports, and all identifiable patient information was removed to ensure anonymity. The study adhered to the principles of the Declaration of Helsinki and relevant national guidelines for the ethical conduct of retrospective research.

CGM protocols in this study

This study utilized retrospective CGM technology. All patients were subjected to CGM monitoring via Medtronic MiniMed for a period of 3 days. Real-time glucose values are not displayed by the CGM; Instead, the data is stored in the recorder. The system, which consists of a single-use glucose detection sensor and a lightweight glucose recorder, is a miniature wearable device system that continuously measures and records glucose concentration in patient interstitial fluid through glucose detection sensors. Every 5 min, glucose values are recorded, leading to up to 288 measurement values per day. Pregnant women experience a gradual increase in abdominal circumference as their gestational weeks increase, making the abdomen unsuitable for sensors. Therefore, the sensor is attached to the upper arm of the patient in this experiment.

Since the value monitored by CGM technology is through interstitial fluid, not via venous blood or capillary blood, it is necessary to compare the CGM value with the blood glucose index to ensure the accuracy of CGM detection. Blood glucose in fingertip capillaries was measured 4 times a day and input into the CGM system for correction, as well as corresponding event markers such as meal time.

Definitions

Screening for and diagnosis of GDM (GDMA1 and GDMA2)

The diagnosis of GDM was based on IADPSG criteria [23]. Individuals who had not been previously diagnosed were tested for GDM using a 75-g OGTT between 24–28 weeks of pregnancy. The definition of GDM is any degree of glucose intolerance that was first noticed during pregnancy. More specifically, when any of the plasma glucose values listed below are met or exceeded, GDM is diagnosed, namely 92 mg/dL (5.1 mmol/L) at fasting, the level of 180 mg/dL (10.0 mmol/L) at 1 h and 153 mg/dL (8.5 mmol/L) after 2 h. GDM can be further classified into diet-controlled GDM (GDMA1) and treated GDM (GDMA2) [8, 9]. GDMA1 type refers to that GDM can be controlled by diet, while GDMA2 type needs to be controlled by drugs to maintain normal blood glucose levels.

ODM

When and how to define women with ODM during pregnancy (not previously diagnosed) were referred to the IADPSG Consensus Panel [6]. One of below standards must be met to identify the patient as having overt diabetes in pregnancy: (1) FPG ≥ 7.0 mmol/l (126 mg/dl); (2) A1 C ≥ 6.5%; (3) Random plasma glucose ≥ 11.1 mmol/l (200 mg/dl) + confirmation.

PGDM

PGDM (pregestational diabetes mellitus) refers to type 1, 2 or special type of diabetes that has been diagnosed before pregnancy. In this study, T1DM refers to type 1 diabetes has been diagnosed prior pregnancy and similarly T2DM for type 2 diabetes.

CGM metrics included the following

TIR, defined as a percentage of all time with CGM glucose values within the pregnancy-specific target range of 3.5–7.8 mmol/L (63–140 mg/dL); TAR 7.8 mmol (> 140 mg/dL); TBR 3.5 mmol/L (< 63 mg/dL); mean glucose; glucose SD, Standard difference of blood glucose value throughout the day; CV, Coefficient of blood glucose variation, calculated with ratio of standard deviation to mean; Glucose management indicator (GMI), the average A1c level that would be expected based on mean glucose measured in a large number of individuals with diabetes. MAGE, the average difference between successive glycemic peaks and nadirs is used to calculate MAGE, but only if changes in glycemic values are greater than 1 standard deviation; LAGE, the difference between maximum and minimum blood glucose values;

CGM targets were based on the consensus TIR targets proposed for T1DM pregnancy [24]. The TIR target range 3.5–7.8 mmol/L (63–140 mg/dL) was > 70%, TAR target < 25%, TBR target < 4%, and CV target < 36%. GMI targets were based on the HbA1c targets, < 6.5% in early and < 6.1% in mid and late pregnancy. The calculation of the duration of blood glucose is based on data monitored by CGM continuously for 3 days.

Laboratory measurements

In the morning after an overnight fast, venous blood samples were collected. Ethylenediaminetetraacetic acid (EDTA) tubes were used to collect blood samples from the women when being fasted, one hour, and two hours after drinking 300 ml of water that contained 75 g of anhydrous glucose, respectively. Before 3 glucose samples were drawn, blood samples were stored at room temperature. Upon delivery of the samples to the laboratory, centrifugation was performed to measure plasma glucose using hexokinase technique with a Beckman Coulter AU5800 automatic analyzer. INS were determined by chemiluminescence immunoassay. HbA1c was detected by HPLC. The enzymatic colorimetric GPO-PAP method (Siemens Healthcare diagnostics Inc) was utilized to measure serum TG levels. Serum cholesterol levels were determined using the automatic analysis method (ADVIA Chemistry XPT) with the use of the enzymatic endpoint (CHOD-PAP). The homogeneous enzyme-based colorimetric assay was used on an automatic analyzer (ADVIA Chemistry XPT) to directly measure serum levels of low density lipoprotein-cholesterol (LDL-C) and high density lipoprotein-cholesterol (HDL-C).

Insulin resistance and Pancreatic β cell function were assessed by HOMA Homeostasis model assessment (HOMA):

$$\text{HOMA}-\text{IR}=\text{ FPG }(\text{mmol}/\text{L}) \times \text{FINS }(\text{uIU}/\text{mL}) /22.5;$$

$$\text{HOMI}-\upbeta =20\times \text{FINS }(\text{uIU}/\text{mL}) /[\text{PFG}-3.5) (\text{mmol}/\text{L})].$$

Statistical analysis

We performed a sample size calculation assuming a 10% difference in glycemic variability (CV) between groups, an alpha level of 0.05, and a power of 80%, which indicated a requirement of at least 750 participants. The power calculation was performed for one-way ANOVA comparing CV across the five groups. The assumed standard deviation (variance) for CV was 7.2%, based on previous studies and pilot data. The final sample size of 776 participants provides an actual power of 85%, calculated using G*Power 3.1. For comparing continuous variables (e.g., age, BMI, glycemic variability (CV), fasting plasma glucose (FPG), insulin resistance index (HOMA-IR), etc.) across the five groups (T1DM, T2DM, ODM, GDMA1, GDMA2), we used one-way ANOVA if the data followed a normal distribution (tested using the Shapiro–Wilk test). If normality was violated, we applied the Kruskal–Wallis test, a non-parametric alternative to ANOVA. For post-hoc pairwise comparisons, we used Tukey’s Honestly Significant Difference (HSD) test for ANOVA and Dunn’s test with Bonferroni correction for the Kruskal–Wallis test.

To identify independent factors associated with glycemic variability (CV ≥ 36%), we performed a multivariate logistic regression analysis. The dependent variable was CV ≥ 36% (unstable glucose variability) vs. CV < 36% (stable glucose variability). Confounder adjustment in regression models included BMI, HOMA-IR, HOMA-β, fasting plasma glucose (FPG), and age, selected based on clinical relevance and prior literature. Results were reported as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). Since multiple comparisons were conducted among five groups, we applied Benjamini–Hochberg False Discovery Rate (FDR) correction to control for type I error inflation. For post-hoc analyses, we used Bonferroni correction where necessary. All hypothesis tests in the manuscript were conducted as two-sided tests, unless otherwise specified. All statistical analyses were conducted using SPSS 25.0 (IBM, Chicago, IL, USA). Figures were generated using GraphPad Prism 9.0.

Results

Characteristics of participants during CGM study according to mother’s hyperglycemia status

A final total of 776 pregnant women diagnosed with various types of diabetes were included in the analysis. Among them, 31.83% had pre-gestational diabetes, which was further categorized into T1DM (3.35%) and T2DM (28.48%). Additionally, 26.68% were diagnosed with overt diabetes in pregnancy (ODM), and 41.49% had gestational diabetes mellitus (GDM), which was further divided into GDMA1 (18.04%) and GDMA2 (23.45%). The participants had an average age of 31.93 years. There were no differences in age, parity, height, systolic pressure (SBP), diastolic pressure (DBP), and mean arterial pressure (MAP) among these groups. The women with T2DM had a higher BMI (28.23 kg/m²) than that in the other groups (22.38 kg/m² for T1DM, 26.61 kg/m² for ODM, 26.84 kg/m² for GDMA1 and 27.99 kg/m² for GDMA2). The gestational age at the time of OGTT testing differed significantly among the groups, among which ODM group took blood glucose test earliest (15.01 weeks) while GDMA1 and GDMA2 group took OGTT test at 23.75 and 24.88 weeks respectively.

The mean fasting plasma glucose (FPG) was 5.80 mmol/L for all participants (Table 1). FPG was highest (7.23 mmol/L) in T1DM group whereas GDMA1 category remained 4.46 mmol/L, the other groups (6.61 mmol/L for T2DM, 6.79 mmol/L for ODM, and 4.53 mmol/L for GDMA2) ranking in the middle. Owing to a high FPG occurred in ODM or previously diagnosed diabetes and considering the subsequent risk of hyperglycemia, these participants were exempted 75 g-OGTT, therefore, 1 h-PG and 2 h-PG for these 3 groups were missing. HbA1c and glycated albumin (GA) presented the same trend in the five sub-groups. The average HbA1c and GA for T1DM group were 6.07% and 17.95% respectively, which were both highest compared with the other groups (5.69% and 16.64% for T2DM; 5.8% and 17.54% for ODM; 5.04% and 13.98% for GDMA1; 5.36% and 14.02% for GDMA2) (Table 1).

Table 1 Characteristics of mothers during GCM study according to maternal hyperglycemia status

There were also statistical differences in insulin secretion at fasting (INS- 0 h) among these groups. As presented in Table 1, the mean value of insulin of T2DM and ODM groups reached 16.15µU/mL and 24.18uU/ML at fasting while the average number in the other groups were no more than 11.43uU/ML. Similarly, same tendency occurred in insulin at 2 h and homa insulin-resistance (HOMA-IR). Whereas, about index of Homeostasis Model Assessment β-cell Function Index (HOMA-β), these groups performed almost opposite trend. As depicted in Table 1, GDMA2 group reached 254.02 in this term while T1DM group only got 44.73, still, the other groups remaining in the middle in this study. To evaluate the situation of serum lipid in participants, we focused analyses on total cholesterol (TC), total triglycerides (TG), high density lipoprotein cholesterol (HDL-C), and low density lipoprotein cholesterol (LDL-C). About these parameters, as presented in Table 1, women diagnosed with GDM exhibited a higher concentration compared with the other groups even in HDL-C term.

Comparison of CGM parameters according to maternal hyperglycemia status.

To further compare continuous glucose monitoring parameters according to maternal hyperglycemia status, we conducted comparison analyses on CGM parameters among these five groups.

As observed in Table 2, in traditional parameters assessment of plasma glucose fluctuation HbA1c and GA, the group of T1DM has the highest value compared with the other groups and ODM group came second while GDM group remained as the lowest. CGM parameters in this study enrolled GMI, SD, CV, MAGE and LAGE. These CGMS data therefore reveal significant differences among these five kinds of diabetes in both hyperglycemia and hypoglycemia and the changes in these variables over pregnancy. Generally speaking, based on all the items that could reflect blood sugar fluctuations, it is striking that blood sugar levels of women with pre-pregnancy diabetes tend to fluctuate more widely than GDM group participants. Detailed information are outlined in Table 2. Also, CV is considered to be the best indicator of blood glucose variability. 36% is the cut-off value of stable and unstable blood sugar levels in diabetes patients. Therefore, the study selected the ratio of whether CV is below 36% as a reference to represent blood glucose variability [24]. Based on our statistics, T1DM group performed greatest glucose fluctuations with only 57.70% participants hit the goal while all the other groups achieve the standard with at least a percentage of 94.20% (ODM group). Other parameters (GMI < 6.0%, GA < 15.70% and HbA1c < 6.0%) showed similar trends in each group.

Table 2 Comparison of continuous glucose monitoring parameters according to maternal hyperglycemia status

Euglycemia

The duration of time spent within the euglycemic range for pregnancy increased from PGDM to ODM to GDM (Table 2). The proportion of time in the euglycemic range rose from an average of 19.20% in T1DM to 96.40% in GDM1 group. Since the pregnancy-specific target of TIR during pregnancy is no less than 70% [25], we calculated the time proportion of TIR ≥ 70% in these participants. According to our results, GDM group had the highest proportion achieving the target (94.5% and 96.4% respectively) of in this study, but only 19.2% of population achieve this goal in T1DM, 66.2% for ODM and 73.30% for T2DM.

Hyperglycemia and hypoglycemia

The time spent hyperglycemic decreased from women with PGDM to GDM. Women with type 1 diabetes spent nearly one-third (32.00%) of the amount of time hyperglycemic compared with women with type 2 diabetes (19.27%), overt diabetes (23.79%), GDMA1 (5.08%) and GDMA2 (9.29%) (Table 2 & Fig. 1). The same patterns were seen at the percentage of TBR less than 25%. The average of TBR < 25% was 22.4% while that in PGDM1 group hit 61.50% and 29.40% for T2DM group. In contrast, GDMA1 group only has 3.60% population did not meet the standards and 4.90% for GDMA2 division (Table 2). Intriguingly, in GDM group, the proportion below target of time spent hypoglycemic presented a sharp increase with a percentage of 26.40% for GDMA1 and 28.00% for GDMA2 group. This phenomenon was less pronounced in the other groups. As presented in Table 2, the difference between below target of time spent hypoglycemic and hyperglycemic is not striking.

Fig. 1

CGM-based targets for different diabetes populations

Exploring factors affecting glycemic variability in gestational hyperglycemia

CV is considered to be the best indicator of blood glucose variability. 36% is the cut-off value of stable and unstable blood sugar levels in diabetes patients. Therefore, the study selected the ratio of whether CV is below 36% as a reference to represent blood glucose variability [24]. Patients are divided into conforming and non-conforming groups at this demarcation point. In the logistic regression model, whether the CV reaches 36% is labeled as a cause variable, while CV < 36% is considered a reference category and possible potential influencing factors are listed as argument.

Univariate analysis of CV in CGM study showed that maternal weight, BMI, FPG, GA, HOMA-IR and HOMA-β are factors inflecting the degree of CV (Fig. 2), while on multivariate logistic regression analysis, only BMI (OR: 0.754, 95% CI: 0.585–0.971; P = 0.029), HOMA- β (OR:0.969, 95%CI: 0.959–0.976; P = 0.037) and FPG (OR: 1.832, 95% CI: 1.170–2.870; P = 0.008) reached statistical significance (Fig. 3). For further analyzing the exact effects of BMI and FPG on CV degree, BMI was deeply classified into four groups: < 18.5 kg/m², 18.5–25 kg/m², 25–29.9 kg/m², ≥ 30.0 kg/m². Based on classification criteria for diagnosis of diabetes, FPG was stratified into four layers: < 5.10 mmol/L, 5.1–6.1 mmol/L, 6.1–7.0 mmol/L and ≥ 7.0 mmol/L. BMI < 18.5 kg/m² and FPG < 5.10 mmol/L were set as reference group. ORs for the other groups were then calculated. As shown in Fig. 4a, BMI was negatively associated with elevated CV in pregnancy. When women with a BMI ≥ 30.0 kg/m², the chance of CV ≥ 36% is only 0.052 times compared with those with a BMI < 18.5 kg/m². Whereas FPG was found to be positively correlated with CV in pregnancy. As presented in Fig. 4b, for those with a FPG ≥ 7.0 mmol/L, the likelihood to develop CV ≥ 36% was about 12.494 times in comparison with women with FPG < 5.10 mmol/L.

Fig. 2

Univariate analysis of CV in CGM study

Fig. 3

Multivariate analysis of CV in CGM study

Fig. 4

a BMI category multivariate analysis of CV in CGM study. b FBG category multivariate analysis of CV in CGM study

Discussion

This study provides the first comprehensive evaluation of glycemic variability (GV) in pregnant women with different types of diabetes using CGMS data. We observed significantly greater GV in patients with pre-gestational diabetes (PGDM, including T1DM and T2DM) compared with those with gestational diabetes mellitus (GDM). Among the subgroups, T1DM patients exhibited the worst glycemic profiles, with the lowest time in range (TIR) and the highest time above range (TAR) and coefficient of variation (CV). Conversely, women with GDM demonstrated relatively better glycemic control. In addition, logistic regression analysis revealed that fasting plasma glucose (FPG) and body mass index (BMI) were independent risk factors for elevated GV (CV ≥ 36%), highlighting the importance of these variables in managing diabetes during pregnancy.

These findings are in line with some previous studies, yet the specific factors contributing to these differences between our study and others require further discussion. One key factor that differentiates our study from previous studies is that our study includes a diverse group of pregnant women with different types of diabetes, categorized into T1DM, Type1, T2DM, ODM, and GDM. This wide spectrum of conditions, including GDMA1 and GDMA2, may account for the variability in outcomes observed. For example, T1DM patients in our cohort exhibited the highest levels of glycemic variability, consistent with other studies, but we also found GDM patients achieving relatively better glycemic control. In contrast, previous studies have not always included such a comprehensive range of diabetes subtypes, which may explain why the outcomes in those studies were less nuanced. The regional and demographic characteristics of our study population in Shanghai may also contribute to the differences observed. Studies conducted in different geographical locations may yield results influenced by cultural, dietary, and environmental factors that impact glucose metabolism. Furthermore, the ethnic diversity of the study populations could also introduce variability in the results. Our cohort primarily consisted of Chinese women, and ethnic differences in insulin sensitivity, β-cell function, and the risk of diabetes may explain some of the differences observed between our study and international research.

Regardless of the severity of hyperglycemia, GDM was once classified as any degree of glucose intolerance initially identified during pregnancy. This criteria has significant drawbacks, on one hand, the best data to date indicates that a large number of GDM cases are actually preexisting hyperglycemia found during pregnancy, as routine screening is not commonly carried out in individuals of reproductive age who are not pregnant. Clinical significance pertains to the degree of hyperglycemia in relation to the hazards to the mother and fetus in the short- and long-term [23]. Therefore, in our present research, we firstly introduced the definition of ODM into CGM study, which partly took into account the existence of preexisting hyperglycemia and severity of gestational hyperglycemia.

Consistent with findings in previous studies, group of PGDM showed greater levels of all glucose variability indicators accompanied by higher glucose level compared to the other groups (p < 0.05) [26]. T1DM patients in our study had the lowest TIR, whereas the greatest TAR, TBR, and GV. Compared to T1DM patients, T2DM, ODM and GDM patients exhibited superior glycemic control. In 2020, Zhang et al. reported that GV index determined by self-monitoring of blood glucose (SMBG) is related to the function of pancreatic islet β cells in T2DM [27]. Therefore, impaired islet β function may increase glycemic variability in individuals with diabetes. In a multicenter cross-sectional trial, Chinese researchers Zhang et al. enrolled 510 adult patients with T1DM, 227 patients with T2DM, and 105 individuals with latent autoimmune diabetes. According to their results, T1DM subjects had a higher GV while T2DM patients had a lower one, which is also in line with the results of ours [28]. Research has demonstrated that improved islet beta cell function can account for 10%–27% of individual glycemic variability and is linked to decreased glycemic variability (CV, SD, MAGE, high glycemic index, and instability index) [29]. Numerous studies have demonstrated a relationship between blood glucose fluctuation and islet beta cell function, regardless of the presence of pre-diabetes or other kinds of diabetes. Long-term hyperglycemia in T1DM and T2DM patients will worsen insulin resistance, which will eventually deplete islet beta cells and lessen their sensitivity to glucose and thus leading to a greater level GV, which partly explains why GDM subjects exhibited better glycemic control in this regard [30].

Since the time range spent on euglycemia is critical for evaluating outcomes of mothers and fetus, therefor we investigated TIR, TBR and TAR among different types diabetes population. In the current study, the mean TIR for all participants is 80.83%, which is considerably higher than previously described in T2DM pregnancies (66–68%) in an Austria study [31]. The hyperglycemia metrics in our patient population were 16%, lower than those reported in T2DM pregnancies in second trimester (26–28%) [31]. Though, in the current study, TAR in T1DM group was around 32%, which was in agreement with reported in T1DM subjects in third trimester [32], and also in accordance with findings of another research [33]. Similar with TBR (10–12%) in other researches in T2DM pregnancies, the index in our study was about 11.85% [31, 34].

In our present research, it's intriguing to notice that reduced BMI in diabetes individuals was complicated with higher CV. Logistic regression analysis with a single factor and multifactor analysis both show that CV is negatively impacted by BMI. Previous researches have revealed an inverse relationship between BMI and the HOMA-IR assessment of insulin resistance [35, 36]. The β-cell mass of obese participants was found to be larger than that of thin subjects, according to early observations on the carcasses of normal glucose-resistant subjects [37]. A cross-sectional study was carried out in 2012 by Saisho et al., they enrolled 2000 T2DM patients to evaluate the effect of obesity on the loss in beta cell activity following T2DM diagnosis. During hospitalization, the levels of serum and urine C were tested both on an empty stomach and following a meal. The study discovered that those with higher BMI also had higher levels of C-carotene, which was substantially negative throughout the duration of diabetes, suggesting a progressive decline in insulin beta cell function as the course of T2DM increased. However, this decrease was more pronounced in obese subjects compared to thin ones. The present study's sample of participants had diabetes for a short history, which could account for the lower BMI in DM individuals with greater CVs. Smaller fluctuations in blood glucose in higher BMI individuals is probably due to they can secrete more endogenous insulin due to a shorter illness cycle [38]. We firstly analyze the exact effects of BMI on CV degree through stratifying BMI as four layers, and we identified that when women with a BMI > 29.9 kg/m², the chance of developing unstable glucose control is only 0.052 times compared with those with a BMI ≤ 18.5 kg/m², suggesting that being overweight is not always detrimental.

As the CV provides a more detailed picture of glycemic excursions than SD alone, we adopted it as a measure of GV in our investigation. While our study identified a mean CV of 21.23%, individuals with impaired β-cell function had greater CVs, highlighting the necessity to address this issue when optimizing medication in low HOMA-β patients. Stable glucose levels have been described as a CV < 36% [39]. Numerous cross-sectional investigations revealed that β-cell malfunction may be the cause of the elevated glycemic variability linked to diabetes [15, 40]. In this study, we identified HOMA-β was negatively corelated with CV in pregnancy population with hyperglycemia, as assessed by the CV of fasting glucose, which is in accordance with several earlier studies [29, 41]. We also found for those with a FPG ≥ 7.0 mmol/L, the likelihood to develop CV > 36% was about 12.494 times in comparison with women with FPG < 5.10 mmol/L, which highlights the critical role of fasting glucose and β-cell function. Surprisingly, glycemic variability in early type 2 diabetes is a variable that can be decreased by enhancing β-cell activity with short-term intensive insulin therapy [42]. Therefore, it becomes clear that glucose variability in the early stages of diabetes is a changeable characteristic for which intervention may reduce the likelihood of unfavorable outcomes in the future.

CGMS data can provide clinicians with more accurate information on glucose control, potentially guiding new personalized treatment strategies, especially in the management of gestational diabetes. This study will fill the existing gap in the literature regarding glycemic variability (GV) in different types of gestational diabetes and will provide the first comprehensive data on GV, which will be of significant reference value for future research and the development of diabetes treatment strategies.

Several limitations should be taken into account when interpreting our findings. Firstly, as this was a retrospective cohort observational study, future large-scale prospective long-term investigations are mandatory to the explore GV in different type of gestational diabetes problems and thus generate consensus on GV control. Secondly, it is widely acknowledged the hyperinsulinemia euglycemic clamp approach as the most sensitive and accurate way to measure islet beta cell function, which is intrusive and cannot be used in a large sample study, thus HOMA-β was chosen to evaluate β cell function. Lastly, the generalizability of our findings is uncertain as our study was limited to public patients in a single Chinese hospital over twelve months and coincided with the first and second year of the COVID- 19 pandemic in China.

Conclusions

In summary, we assessed glucose variability in pregnancy complicated by diabetes, finding a greater glucose variability in pregnant women with type 1 or 2 diabetes than in cases of GDM controls, and the former had a worse interstitial glucose profile than the other two groups. Given there is a paucity of data of GV in various kinds of gestational hyperglycemia, our findings have important public health implications. Also we found ODM group comes second in terms of parameters of glucose variability. In the logistic regression model, we found BMI, HOMA-β and FPG remained an independent risk factor for the CV ≥ 36% group even after correcting other possible potential factors. This study provides a reasonable estimate of the risk of unstable blood glucose based on islet β function, BMI and FPG, consequently help women who are suffering gestational hyperglycemia and health care professional to guide treatment decisions.

Specifically, our study highlights the differences in glucose variability among pregnant women with various types of diabetes. By utilizing CGMS, clinicians can provide more accurate data on glucose fluctuations, which can aid in the development of personalized treatment strategies.