Assessment of the optimal cutoff value of fasting plasma glucose to establish diagnosis of gestational diabetes mellitus in Chinese women

Our aim is to assess the optimal cutoff value of fasting plasma glucose (FPG) in Chinese women at 24–28 weeks’ gestation by performing oral glucose tolerance test (OGTT) to improve diagnostic rate of gestational diabetes mellitus (GDM). Data were derived from the Medical Birth Registry of Xiamen. A FPG cutoff value of 5.1 mmol/L confirmed the diagnosis of GDM in 4,794 (6.10%) pregnant women. However, a FPG cutoff value of 4.5 mmol/L should rule out the diagnosis of GDM in 35,932 (45.73%) pregnant women. If we use this cutoff value, the diagnosis of GDM to about 27.3% of pregnant women will be missed. Additionally, a 75-g OGTT was performed in pregnant women with FPG values between 4.5 and 5.1 mmol/L, avoiding the performance of formal 75-g OGTT in about 50.37% pregnant women. Meanwhile, according to maternal age and pre-pregnancy BMI categories, with FPG values between 4.5 mmol/L and 5.1 mmol/L, which had high sensitivity, to improve the diagnostic rate of GDM in all groups. Further researches are needed to present stronger evidences for the screening value of FPG in establishing the diagnosis of GDM in pregnant women.

three strategies in establishing the diagnosis of GDM: 100-g OGTT, 75-g OGTT, and FPG test of the OGTT. Consequently, the research revealed that performing FPG test was the ideal method 12 . Therefore, performing FPG test is recommended to establish the diagnosis of GDM at 24-28 weeks' gestation. Additionally, initially performing an FPG test prevents the need of performing many subsequent OGTTs, hence reducing the health burden of the Chinese societies and families.
The advantages of performing FPG test to establish the diagnosis of GDM include the following: the test is cheap, reproducible, and reliable and has no vomiting response, response that is usually evident when performing OGTT or the glucose challenge test. FPG test can be performed in pregnant women who were unable to tolerate glucose-containing drinks. FPG test was performed to establish the diagnosis of GDM 13 . Zhu et al. reported that during the first prenatal visit, increased FPG level was strongly associated with GDM at 24-28 weeks' gestation in China 14 . However, there were only two similar studies that assessed the sensitivity and specificity of FPG value in establishing the diagnosis of GDM at 24-28 weeks' gestation in China, stressing that if the FPG cutoff value was between 4.4 mmol/L and 5.1 mmol/L, the pregnant women should undergo OGTT 14,15 . Whereas, one of the disadvantages in the above studies was that the data from the above study were derived from a hospital that consisted of a small sample size due to the following reason: rural residents rarely visited to the general hospital.
Chinese women tend to get pregnant at young ages 16 is different from the other countries or healthcare systems, especially in developed countries, where women get pregnancy at older ages 17 . As well, pre-pregnancy body mass index (BMI) is a known risk factor for GDM 18,19 . In addition, studies revealed pre-pregnancy BMI might be as a predictor for GDM. Whereas, compared to population in other races, the BMI levels of Asians might be generally lower, which was contributed to ethnic differences 20,21 .
Therefore, this study aimed to assess the sensitivity and specificity of FPG value to establish the diagnosis of GDM based on maternal age and pre-pregnancy BMI categories in China with a registered data and to improve the diagnostic rate of GDM, hence avoiding the performance of a number of OGTTs and consequently reducing the health burden of the Chinese societies and families.

Materials and Methods
Study design. The Medical Birth Registry of Xiamen (MBRX) recorded the results of a 75-g OGTT for pregnant women and implemented the International Association of Diabetes and Pregnancy Study Group (IADPSG) criteria (one-step approach). Over 7 years, from March 1, 2011, to March30, 2018, pregnant women at 24-28 weeks' gestation who were registered at the MBRX underwent the 75-g OGTT in Xiamen. According to the GDM diagnostic criteria, the venous plasma glucose values were recorded and analyzed. We created the receiver operating characteristic (ROC) curves to detect the FPG value and subsequently to establish the diagnosis of GDM based on the result of the 75-g OGTT, which is the standard method, and each point as a screening node was analyzed.
Definition. The diagnosis of GDM could be established by the Ministry of Health in China according to IADPSG 22 when any of the following FPG values were met or exceeded: 0 h, greater than or equal to5.1 mmol/L; 1 h, greater than or equal to 10.0 mmol/L; and 2 h, greater than or equal to 8.5 mmol/L. BMI was calculated on account of self-reported weight and measured height. According to World Health Organization recommendations for Asian population 23 , the pregnant women were classified into four groups: underweight, BMI <18.5 kg/ m 2 ; normal weight, 18.5-23.9 kg/m 2 ; overweight, 24.0-27.9 kg/m 2 ; and obesity, ≥28 kg/m 2 . Maternal age was classified into four groups: ≤25 years; 26-30 years; 31-35 years; and >35 years.
Informed consent and ethics statements. The ethics committee of the First Affiliated Hospital of Xiamen University approved our study waived the need for informed consent, which composed and worked in accordance with the Chinese GCP and relevant regulations. The application number was KYH2018-007. In addition, this study was carried out in accordance with the rules of the Declaration of Helsinki of 1975, revised in 2013.

Statistical analysis.
The characteristics of study population were analyzed by SPSS version 17.0 statistical software (SPSS Inc., IL, USA). Continuous variables were showed as Median (min-mix). Discontinuous variables were expressed as n (%), which were analyzed by Chi-square (χ 2 ) test. This study compared FPG cutoff values across pre-pregnancy BMI and maternal age categories. Receiver operating characteristic (ROC) curve analysis was conducted to identify the diagnostic power of FPG value of OGTT at pre-pregnancy in predicting development of GDM. The level of statistical significance is set at 0.05.

Results
Characteristics of study population. A total of 78,572 pregnant women (age range, 18-53 years old) registered at the MBRX underwent a 75-g OGTT. Most of the pregnant women (51,600) had education levels higher than 9 years. Additionally, the parity of the 36,869 (53.00%) pregnant women was more than two times (Table 1). A total of 13,658 pregnant women were diagnosed to have GDM on account of maternal age and pre-pregnancy BMI categories based on the IADPSG criteria. The results of performing a 75-g OGTT in order to establish the diagnosis of GDM revealed that the following pregnant women have met or exceeded the FPG values: 0 h (≥5.1 mmol/L), 2,753 pregnant women; 1 h (≥10.0 mmol/L), 3,088 pregnant women; and 2 h (≥8.5 mmol/L), 3,257 pregnant women. The diagnostic rates of GDM were 18.8%, 21.07%, and 22.2%, respectively. Furthermore, the diagnostic rate of GDM was 35.7% following the FPG criteria in the OGTT. To assess the FPG value of OGTT to establish the diagnosis of GDM at 24-28 weeks' gestation, the ROC curves are shown in Figure 1. The area under the ROC curve was 0.752 (95% CI, 0.747-0.757; SE, 0.003; P < 0.001).
www.nature.com/scientificreports www.nature.com/scientificreports/ The overall association between FPG cutoff value and GDM diagnosis. FPG value was more than 5.1 mmol/L, the diagnosis of GDM (6.10%) in pregnant women was confirmed. As shown in Table 2, with the cutoff FPG value of 4.5 mmol/L, 72.7% of pregnant women were diagnosed to have GDM with specificity of 0.600. If the FPG value of 4.5 mmol/L was the cutoff value to identify who should undergo the 75-g OGTT, then 50.37% (44.27% with values were less than 4.5 mmol/L plus 6.10% with values were greater than 5.1 mmol/L) of pregnant women could avoid the performance of a 75-g OGTT with the probability that 27.3% of pregnant women with GDM may miss to undergo a 75-g OGTT. If the FPG value of 4.4 mmol/L or 4.3 mmol/L serves as the cutoff value, the missed percentage of pregnant women who should undergo a 75-g OGTT would be 20.2% or 14.2% with the specificity was 0.474 and 0.350 respectively.
The association between FPG cutoff value and GDM diagnosis according to maternal age categories. 1  www.nature.com/scientificreports www.nature.com/scientificreports/ at 24-28 weeks' gestation based on different maternal age, the ROC curves are shown in Table 3. The area under the ROC curve of different maternal age was 0.76, 0.74, 0.74, and 0.75, respectively for maternal age ≤ 25, 26-30, 31-35, and >35 years groups.
As presented in Table 4, with the cutoff FPG value of 4.5 mmol/L, 70.4% of pregnant women aged less than 25 years were diagnosed to have GDM with specificity of 0.658. As well, with the FPG cutoff value of 4.5 mmol/L, 70.6% of pregnant women aged among 26 to 30 years were diagnosed as GDM with specificity of 0.601, and 74.9% of pregnant women aged among 31-35 years were diagnosed as GDM with specificity of 0.542. Besides, 77.6% of pregnant women aged more than 35 years were diagnosed as GDM with the specificity of 0.497.
As presented in Table 6, with the FPG cutoff value of 4.5mmom/L, 54.9% of pregnant women with pre-pregnancy BMI less than 18.5 kg/m 2 (underweight) were diagnosed as GDM with the specificity of 0.692. Meanwhile, with the cutoff FPG value of 4.5 mmol/L, 70.8% of pregnant women with pre-pregnancy BMI among 18.5 to 23.9 kg/m 2 (normal) were diagnosed to have GDM with specificity of 0.597. In addition, with the cutoff FPG value of 4.5 mmol/L, 84.7% or 89.3% pregnant women with pre-pregnancy BMI among 24-27.9 kg/m 2 (overweight) or ≥28.0 kg/m 2 (obesity) were diagnosed as GDM with specificity of 0.485 and 0.391 respectively.

Discussion
Although it had been reported that FPG was a poor predictor for GDM later in pregnancy based on the FPG level decreases at the end of the first trimester and with a low sensitivity or poor specificity 24,25 , in this study, ROC curve analysis presented that FPG ≥4.5 mmol/L was the optimal threshold for predicting GDM, with a sensitivity of 72.7% and a specificity of 60.0%.
Our study showed that the diagnostic rate of GDM using 5.1 mmol/L as the cutoff FPG value based on the IADPSG criteria was only 35.7%. However, another study indicated that the IADPSG criteria diagnosed GDM using the FPG value (55%) alone for the complete Hyperglycemia and Adverse Pregnancy Outcome (HAPO) cohort study. Interestingly, the diagnostic rate of GDM using the FPG value (5.1 mmol/L) was 47% in Singapore 26 . Compared with these results, the diagnostic rate is lower in this study. It is clear that the FPG cutoff value of   www.nature.com/scientificreports www.nature.com/scientificreports/ 5.1 mmol/L was to establish the diagnosis of GDM based on the IADPSG criteria was not possible in China. If we use these criteria, there is a bigger possibility that the diagnosis of GDM in some pregnant women will be missed. Any local adaptations of these criteria will have to be dependent on the exact local data of the population under consideration.
Additionally, the Ministry of Health in China published the GDM diagnostic criteria in 2011, which stated that if FPG value was greater than or equal to 5.1 mmol/L, immediately, the pregnant women can be diagnosed as having GDM and if FPG value was greater than or equal to 4.4 mmol/L and less than 5.1 mmol/L, the 75-g OGTT must be performed in pregnant women 22 . Our research found that if FPG level was greater than or equal to 5.1 mmol/L,     immediately, pregnant women can be diagnosed as having GDM with a specificity of 100%. Additionally, if FPG value was less than or equal to 4.5 mmol/L, pregnant women can be ruled out in the diagnosis of GDM with a sensitivity of 72.7%. In view of the above results, we conclude that with the FPG values between 4.5 and 5.1 mmol/L, an optimal threshold to rule out or identify GDM in China. This threshold range will reduce the performance of a number of OGTTs by about 50.37%. Interestingly, one study reported that an FPG value of 4.4 mmol/L should be used as the optimal cutoff value, which could reduce the performance of a number of OGTTs by about 50.3% with the probability that the diagnosis of GDM to about 12.2% of pregnant women may be missed 15 . Although the threshold of FPG value is different, the proportion of avoiding performance of OGTT is similar. The study had some advantages. The major advantage was that its data were derived from a registered system with a large sample size, which registered Chinese pregnant women who came from the countryside and city to avoid selection bias. Moreover, it was the first research to evaluate the sensitivity and specificity of FPG value to establish the diagnosis of GDM in China with registered data, which had strong public health significance.
The study also had some limitations. Firstly, this study is a retrospective design that included unavoidable selection bias. Secondly, all data were from the MBRX, and pre-pregnancy BMI of Chinese differs from other population; hence, there was a lack of data from other regions. Thirdly, the effects of other factors on incidence of GDM are unavailable, such as physical activity or dietary habit.
As a conclusion, FPG test, with a high sensitivity and specificity, is used as an ideal GDM screening tool for low-resource countryside that simplifies the demanding algorithm for the establishment of the diagnosis of GDM. Moreover, for maternal age and pre-pregnancy BMI categories, FPG ≥4.5 mmol/L could be as an optimal predictor for GDM in all groups with a high sensitivity, which could improve the diagnostic rate and reduce the health burden of the Chinese societies and families.

Data availability
Data are available upon request. Please contact Xue-jun Li, professor, xmlixuejun@163.com.