Possibilities of using different indices of insulin resistance in various subtypes of gestational diabetes mellitus

Introduction

The prevalence of various carbohydrate metabolism disorders is currently steadily increasing both among the general population and among pregnant women. Modern research is increasingly more focused on gestational diabetes mellitus (GDM), which is induced by a number of factors. GDM accounts for up to 84% of all cases of hyperglycemia during pregnancy. Various types of hyperglycemia have been recorded in 20 million women, which is 16% of live births. Consequently, every 6th birth results from a pregnancy that occurred against the background of GDM¹.

Although women with GDM are not at increased risk of having children with birth defects, the carbohydrate metabolism disorder poses a serious threat to the health of the mother and fetus due to its high association with the development of short-term and long-term complications. Well-known short-term GDM complications include an increased risk of preeclampsia, large fetus, birth injuries, neonatal hypoglycemia and jaundice, respiratory distress syndrome, and stillbirth. In addition, GDM is a predisposing factor for the development of obesity, type 2 diabetes mellitus (T2DM), and cardiovascular diseases in the mother and offspring in the future. Such complications are usually called long-term. Taking into account the above, GDM is considered to be an urgent problem in modern medicine².

Algorithms for managing pregnant women with GDM are well established and focus on achieving normoglycemia and preventing excessive fetal growth and the development of other complications. Currently, non-pharmacological and pharmacological treatments for GDM are distinguished. The first group includes lifestyle modification (diet correction, increased physical activity, weight gain control), and regular self-monitoring of glycemia. The second group includes insulin therapy with insulins approved for use during pregnancy. Despite lifestyle modification, up to 30% of patients require pharmacological treatment [1–3].

The choice of GDM therapy depends on a number of factors and is decided individually in each case. One of the factors determining the efficiency of non-pharmacological therapy and the need for pharmacological correction may be associated with the pathophysiological aspects of the formation of hyperglycemia during pregnancy [4–8].

Based on the opinion that the main factors in T2DM development in non-pregnant women include a defect in insulin secretion (a defect in the β-cell function) or insulin sensitivity, it was assumed that these factors had a similar effect on the development of GDM in pregnant women. Depending on the prevalence of the β-cell defect or insulin resistance (IR) in pregnant women, various GDM subtypes were defined.

Further research has shown that different pathogenetic subtypes of GDM are associated with increased risks of developing various complications [5][9]. The study “Heterogeneous Contribution of Insulin Sensitivity and Secretion Defects to Gestational Diabetes Mellitus” (2016) by Powe et al. has demonstrated that GDM can be based on both a β-cell defect with normal insulin sensitivity and IR with hyperinsulinemia [9]. At the same time, pregnant women with predominant IR had an altered adipokine profile, bigger children at birth, and a higher risk of GDM-associated complications even in comparison with women with normal glucose tolerance (NGT). In women with GDM against the background of a β-cell defect, the indicators of body mass index (BMI), fasting glucose, child's birth weight, and the risk of complications did not differ from the indicators of pregnant women with normal carbohydrate metabolism [9].

Three GDM subtypes were identified by Liueta in 2018 [4] and Feghali et al. in 2019 [10] with a predominance of β-cell dysfunction, as well as with a predominance of IR, and a mixed type (where both features were expressed equally). According to the authors, pregnant women with GDM and IR combined with β-cell dysfunction had the highest frequency of adverse perinatal outcomes compared to women with normal carbohydrate metabolism. These included higher BMI before pregnancy, blood glucose levels, the birth of a large child, and a higher incidence of GDM-associated complications, including adverse neonatal outcomes [4][10]. In 2019, Benhalima et al. also demonstrated that women with GDM and high IR had a more unfavorable metabolic profile (high hyperglycemia, BMI, blood pressure, and lipid levels) and a higher risk of adverse pregnancy outcomes compared to women with NGT and women with GDM and lower IR. In addition, women with GDM with severe IR had a higher degree of hyperglycemia in both early and late pregnancy compared to women with GDM without IR [5].

Taking into account the above, the GDM subtypes should be determined to correct treatment measures and predict the occurrence of GDM-associated pregnancy complications. Since the prevailing criterion for verifying the GDM subtypes is the IR presence and severity, it is important to consider the methods of its diagnosis.

IR is defined as an impaired insulin action that results in decreased glucose uptake by muscle and increased glucose production by the liver and adipose tissue due to increased lipolysis [11]. Thus, in patients, the degree of IR is assessed by measuring the metabolic effects of insulin, i.e., peripheral glucose uptake and glucose production, during fasting or insulin-stimulated conditions [11].

Currently, there are no clear normative indicators that could unambiguously define IR presence. However, it is known that individuals with reduced sensitivity to insulin more often develop excess body weight and obesity, arterial hypertension, disorders of carbohydrate and lipid metabolism, and the coagulation system [4]. Wide variations in insulin sensitivity can also be found in healthy individuals [12–14].

The “gold standard” for measuring peripheral IR is the hyperinsulinemic euglycemic clamp (HEC) [11][15]. However, this method is not widely used due to its labor intensity, time and financial costs, but has found its application as a standard for the development of other methods.

The leading role in IR diagnostics is currently occupied by mathematical models based on the relationship between the concentration of insulin, glucose, triglycerides, and other parameters obtained either on an empty stomach or during an oral glucose tolerance test (OGTT) with 75 g of glucose. Two groups of indicators are defined. One group consists of IR indices calculated during the fasting blood sugar test (the so-called “fasting indices”). These include the homeostasis model assessment (HOMA-IR), the quantitative insulin sensitivity check index (QUICKI), and the McAuley index.

The second group includes indices taking into account glucose and insulin data obtained during 120 minutes of standard OGTT (the so-called “OGTT indices”). The Matsuda, Belfiore, Cederholm, Stumvoll, and Avignon indices are included in this group.

According to the HEC data, among the best predictors of peripheral IR are fasting plasma glucose and insulin concentrations, low-density lipoprotein cholesterol, BMI, and waist circumference [16].

The most widely used index is the HOMA-IR, which is based on fasting plasma glucose and insulin levels. HOMA is a model of glucose-insulin dynamics that predicts steady-state fasting glucose and insulin concentrations for a wide range of possible combinations of IR and β-cell function. Insulin levels depend on the response of pancreatic β-cells to glucose concentrations, while glucose concentrations are regulated by insulin-mediated glucose production in the liver. Therefore, inadequate β-cell function will reflect a reduced β-cell response to glucose-stimulated insulin secretion. Resistance will be reflected by a reduced inhibitory effect of insulin on hepatic glucose production.

The HOMA model is a reliable clinical and epidemiological tool for IR assessment. Thus, the research by Stern et al. detected IR at HOMA‐IR > 4.65, BMI > 28.9 kg/m² or HOMA‐IR > 3.60 and BMI > 27.5 kg/m² [18]. Tam et al., having analyzed white-skin participants, determined the presence of IR at HOMA‐IR > 2.8 and HDL < 51 mg/dl [17]. Isokuortti et al. demonstrated that HOMA‐IR > 2.0 allowed identifying participants with non‐alcoholic fatty liver disease [18]. The studies on the general population involving both Caucasian [19] and Asian [20] participants also determined IR by means of a low threshold for HOMA‐IR. However, the reference values ​​for HOMA‐IR still require revision.

QUICKI is essentially a HOMA-IR version, but takes into account insulin asymmetry using a logarithmic transformation [21]. Otten et al. in their mathematical analysis demonstrated that among fasting IR/sensitivity indices, the revised QUICKI (using the concentration of non-esterified fatty acids) had the strongest correlation with insulin sensitivity measured by HEC [22].

Another well-known index, the McAuley index, uses triglyceride levels rather than glucose as a surrogate and predicts resistance in normoglycemic individuals [23].

The Matsuda and Stumvoll indices are among the most widely used OGTT indices [24] which have also been confirmed in populations with different characteristics and ethnic backgrounds [25][26]. The Matsuda index was proposed by Matsuda and DeFronzo and uses data from both the entire three-hour OGTT and its first two hours. This is a composite index for calculating the insulin sensitivity of the entire body, combining the insulin sensitivity of both the liver and peripheral tissues.

The Matsuda and Belfiore indices can also be used for OGTTs lasting longer than 2 hours or for mixed meal tolerance testing. As meta-analysis showed that the Matsuda, Stumvoll, and Gutt indices had also demonstrated the strongest correlation with insulin sensitivity measured by HEC [22].

It has been established that various indices can be used to assess IR. As a rule, HOMA-IR, QUICKI, and Matsuda indices are suitable for clinical purposes, while HEC, Belfiore, Cederholm, Stumvoll, and Avignon indices are suitable for research and epidemiological purposes [27]. The correct use of IR indices is important for diagnosing carbohydrate metabolism disorders [16].

Materials and methods

This research was performed at the clinical sites of the Department of Internal Medicine No. 3 of the Research Institute of Obstetrics and Pediatrics of the Rostov State Medical University of the Ministry of Health of Russia.

In order to assess the changes in IR indices in different GDM subtypes, 130 pregnant women aged 18 years and older were examined, regardless of the risk factors for GDM. The research did not include pregnant women who had resorted to assisted reproductive technologies, suffered from any carbohydrate metabolism disorders before pregnancy, or were taking hypoglycemic drugs.

All women underwent a full clinical examination, including collection of complaints and anamnesis and general checkup. The dynamics of weight and blood pressure were determined using the standard method, hereditary anamnesis was assessed (first and second degrees of kinship), and anamnesis of previous pregnancies and births was collected.

Laboratory studies included assessment of carbohydrate metabolism, IR determination, assessment of β-cell function, and lipid metabolism parameters.

Carbohydrate metabolism was assessed by studying fasting venous plasma glucose, OGTT with 75 g of glucose with additional determination of fasting and post-exercise insulin values, and lipid metabolism was assessed by performing a lipidogram.

The function of β-cells was assessed using the HOMA-B index, and IR was detected using such indices as homeostasis model assessment – HOMA-IR, quantitative insulin sensitivity test index (QUICKI), Matsuda index, McAuley index, Belfiore index, Gutt index, Stumvoll index, and Avignon index [16, 17, 18]. When the Matsuda index was less than the 50th percentile, the values ​​of pregnant women without GDM indicated the prevalence of IR processes, and when it was more than the 50th percentile, it indicated β-cell dysfunction.

Based on the predominant pathogenetic mechanism and using the Matsuda index, the following groups of patients were formed: group I comprised 45 pregnant women with GDM and β-cell dysfunction, group II – 43 pregnant women with GDM and IR, and group III – 42 pregnant women without GDM (control group).

All laboratory parameters were determined in venous blood, and indices were calculated.

Statistical analysis of the research results was performed using R (version 3.2, RFoundation for Statistical Computing, Vienna, Austria). Quantitative indicators in groups were compared using the Kruskal-Wallis test (pairwise post hoc comparisons were performed using the Nemenyi method), and frequencies were compared using Fisher's exact test and Holm's correction for multiple comparison. Data were considered statistically significant at p<0.05. Data were presented as medians and interquartile ranges of quantitative indicators in groups.

Results

During the study, various IR indices in the selected groups were analyzed, and their differences were identified.

Comparative characteristics of IR indices in different subgroups are presented in Table 1.

The HOMA-IR index values ​​differed between all subgroups. Thus, in group II, it was higher relative to groups I and III [ 2.33 (1.76; 4.23) versus 1.13 (0.85; 1.34) and versus 1.25 (1.01; 2.43), p_{I – II}<0.001 and p_{II – III}<0.001, respectively]. However, in group 1, HOMA-IR was lower than in group 3 [ 1.13 (0.85; 1.34) versus 1.25 (1.01; 2.43), pI – II=0.01]. Consequently, pregnant women with different GDM subtypes differed in the HOMA-IR level. Moreover, in the group of pregnant women with IR, it was expressed maximally, which was expected. Herewith, in patients with GDM and β-cell dysfunction, IR according to HOMA-IR was less pronounced than in healthy pregnant women.

The QUICKI index also differed among all subgroups. The index value was the highest in subgroup I [ 0.38 (0.37; 0.39)] and differed from subgroup II [ 0.34 (0.31; 0.35); p_I–II<0.001] and subgroup III [ 0.37 (0.33; 0.38); p_I–III =0.01]. In group II, the QUICKI index was significantly lower than in group III [0.34 (0.31; 0.35) versus 0.37 (0.33; 0.38); p_II–III<0.001]. Thus, the QUICKI index was the lowest in pregnant women with GDM and IR compared to patients from other subgroups. The QUICKI index values ​​among the subgroups without severe IR were similar, although statistically differences were revealed (p=0.01).

Multiple differences were revealed between the groups when assessing the Matsuda index values. In group I, the index was higher than in groups II and III [ 0.0078 (0.0063; 0.0110) versus 0.0016 (0.0008; 0.0026) and versus 0.0047 (0.0023; 0.0080); p_I–II<0.001 and p_I-III<0.001, respectively]. At the same time, the Matsuda index in group II was lower relative to group III [ 0.0016 (0.0008; 0.0026) versus 0.0047 (0.0023; 0.0080); p_II–III<0.001]. Thus, the Matsuda IR index was the highest among pregnant women with GDM and β-cell dysfunction, and the lowest among pregnant women with IR (determined by HOMA-IR). An intermediate value of the Matsuda index was recorded among healthy pregnant women.

When analyzing the McAuley index values ​​in the subgroups, a similar picture was revealed. The highest index value was recorded in subgroup I, which significantly exceeded the value in group II [ 6.81 (6.25; 7.53) versus 5.08 (4.65; 5.97); p_I–II<0.001] and group III [ 6.81 (6.25; 7.53) versus 6.2 (5.37; 7.8); p_I–III=0.04]. In group III, the McAuley index was also higher than in group II [ 6.2 (5.37; 7.8) versus 5.08 (4.65; 5.97); p_II–III<0.006]. In pregnant women with GDM and β-cell dysfunction, the McAuley index had the highest value, and in pregnant women with IR (according to HOMA-IR data), the index value was the lowest one. In pregnant women with normal carbohydrate metabolism, it occupied an intermediate position.

The Belfiore index in group II was lower than in group I [ 0.0055 (0.0037; 0.0082) versus 0.014 (0.010; 0.017); p_I–II<0.001] and group III [ 0.0055 (0.0037; 0.0082) versus 0.010 (0.008; 0.015); p_II–III<0.001]. The indices were statistically similar between groups I and III (p=0.05).

In pregnant women with GDM and IR, the Belfiore index was lower than in pregnant women with GDM and β-cell dysfunction, as well as in healthy women. At the same time, the index in pregnant women with β-cell dysfunction and in healthy women was similar. The Belfiore index values demonstrated that the IR severity in pregnant women with GDM and IR was the highest compared to pregnant women with GDM and β-cell dysfunction, as well as compared to healthy women. Herewith, the IR severity was similar among the other two subgroups of pregnant women.

Analysis of the distribution of the Gutt insulin sensitivity index values ​​showed the following data. In group II, the index value was significantly lower than in groups I and III [ 2.68 (2.21; 3.46) versus 3.85 (3.48; 4.45) and 2.68 (2.21; 3.46) versus 4.01 (3.64; 4.38); p_I–II<0.001 and p_II–III<0.001, respectively]. At the same time, the index values ​​in groups I and III were similar.

The Gutt index value was lower in pregnant women with GDM and IR compared to pregnant women with GDM and β-cell dysfunction and healthy individuals. Herewith, the index value in women with GDM and β-cell dysfunction and normal carbohydrate metabolism was similar. Thus, according to the Gutt index data, it can be assumed that the IR severity in pregnant women with GDM with IR revealed by HOMA-IR was higher than that of pregnant women with GDM and β-cell dysfunction and without carbohydrate metabolism disorder. At the same time, the IR severity in the last two categories of patients was similar.

The Stumvoll index was also examined in this research. Once again, the lowest index results were obtained in subgroup II [ 0.10 (0.10; 0.11)], which differed from the index values in group I [ 0.12 (0.11; 0.12); p_I–II<0.001] and group III [ 0.12 (0.11; 0.13); p_II–III<0.001, respectively]. The values ​​of the calculated index in patients from groups I and III did not differ.

Pregnant women with GDM and IR had lower Stumvoll index values compared to both pregnant women with GDM and β-cell dysfunction and healthy individuals. Consequently, according to the described index, pregnant women with IR (revealed by the HOMA-IR index) had the highest IR severity compared to other groups of women. Herewith, the Stumvoll index values, and therefore the IR severity in pregnant women with β-cell dysfunction and pregnant women with normal carbohydrate metabolism were similar.

Multiple differences were obtained between the subgroups when analyzing the Avignon insulin sensitivity index. The index was higher in subgroup I compared to subgroup II [ 469 (358; 650) versus 169 (89.8; 239); p_I–II<0.001] and subgroup III [ 469 (358; 650) versus 328 (226; 488); p_I–III=0.005]. Herewith, the Avignon index in subgroup II was lower than in subgroup III [ 169 (89.8; 239) versus 328 (226; 488); p_II–III<0.001].

The Avignon index values revealed different IR levels across all subgroups: the highest index value was found among pregnant women with GDM and β-cell dysfunction, and the lowest index value was defined among pregnant women from the GDM group with IR determined by the HOMA-IR index; pregnant women with normal metabolism occupied an intermediate position.

In addition to IR indices, β-cell function was determined using the HOMA-B homeostasis model. In subgroup I, the HOMA-B index was significantly lower than in subgroups II and III [ 15.3 (11.1; 18.0) versus 36.9 (19.4; 57.0) and versus 25.9 (20.4; 59.9); p_I–II<0.001 and p_II–III<0.001, respectively]. The values ​​in subgroups II and III were similar.

In pregnant women with GDM and β-cell dysfunction, the HOMA-B index values were the lowest, and therefore the β-cell dysfunction was the highest among all categories of pregnant women. Herewith, the HOMA-B index in pregnant women with GDM and IR and in women with normal carbohydrate metabolism did not differ, the β-cell function was similar.

Discussion

The obtained data demonstrated the difference in the actual values ​​of IR indices in different GDM subtypes. IR in pregnant women with GDM and IR revealed by the HOMA-IR index was more pronounced compared to other categories of pregnant women. Herewith, IR was minimal in pregnant women with GDM and β-cell dysfunction, and healthy women occupied an intermediate position.

The results of calculating other indices (QUICKI, Matsuda, McAuley, Belfiore, Gutt, Stumvoll, and Avignon) also showed significant differences between patients in the subgroups. The greatest difference was found between patients from subgroups I and II, and patients from group III (with NGT) either occupied an intermediate position ​​or did not differ from patients in group I (with GDM and without pronounced IR).

When analyzing the results of the HOMA-B index calculation, the lowest index value was defined in patients from group I compared to patients from group II [ 15.3 (11.1; 18.0) versus 36.9 (19.4; 57.0), p<0.001]. The index values ​​of patients from the GDM group with β-cell dysfunction were statistically significantly lower than the values ​​of patients from group III with NGT [ 15.3 (11.1; 18.0) versus 25.9 (20.4; 59.9)]. Herewith, no difference in index values was found between patients from groups II and III.

Conclusion

Currently, the definition of different GDM subtypes may be of extreme clinical importance, since the choice of tactics and timeliness of pharmacological treatment for patients and, as a result, an improvement in the prognosis for mothers and newborns may depend on the mechanisms underlying the pathogenesis of carbohydrate metabolism disorders.

The features of IR indices in various GDM subtypes have been studied and numerous differences among patients from different pathogenetic subgroups have been identified, which demonstrated the heterogeneity of such patients. Reliably significant differences between patients with different GDM subtypes were obtained when calculating all the selected indices. For the diagnosis of the GDM subtype with severe IR, the HOMA-IR, QUICKI, Matsuda, McAuley, Belfiore, Gutt, Stumvoll, and Avignon indices can potentially be used. The HOMA-IR index is not only reliable, but also the most accessible calculation method for diagnostics in clinical practice.

In order to determine the GDM subtype of with β-cell dysfunction, the HOMA-B index can be used, which, according to the results of the present study, allows reliably identifying patients with impaired insulin secretion.

Additional high-quality and well-designed studies with a larger number of participants are needed to determine the threshold values ​​of the indices allowing differentiation of patients with different GDM subtypes.