Increased Insulin Resistance in Roma Pregnancies

Abstract

Background: Reduced tissue sensitivity to insulin, as well as the associated increased risk of gestational diabetes mellitus is genetically controlled and often varies racially and geographically. Roma populations constitute a genetically autonomous society with particularities in their type of sociability, and they are reported to have an increased prevalence of type 2 diabetes mellitus, which is pathophysiologically related to insulin resistance. Objectives: The aim of this study was to investigate the level of insulin sensitivity in pregnancies of Roma mothers compared to controls. Methods: A total of 65 pregnancies were studied during the third trimester, divided between 33 Roma mothers (RP) and 32 mothers of European descent to serve as control volunteers (CP). The presence of gestational diabetes was confirmed according to the WHO diagnostic criteria by a 75 mg oral glucose tolerance test and insulin resistance status by the means of HOMA-IR index. Results: The mean fasting insulin levels as well as the mean HOMA-IR index were statistically significantly higher in the Roma population (p = 0.0013) and (p < 0.001), respectively, regardless of the age and BMI of the participants. Gestational Diabetes Mellitus developed in seven women (10.7%), five of whom were Roma (15.1%) and in two controls (6.2%) (p = 0.247). Conclusions: Increased insulin resistance is observed in Roma pregnancies, so it would be beneficial to provide these women with appropriate counseling focused on healthy diet and lifestyle.

1. Introduction

Gestational diabetes mellitus (GDM) is a frequent pregnancy-related disorder, characterized by the first appearance of abnormal glucose levels during gestation [1]. Approximately one in five live births, which corresponds to around 23 million people, are impacted by some type of hyperglycaemia during pregnancy [2]. Even though GDM usually subsides after delivery, it is linked to long-term health risks for both mothers and their children. More specifically, mothers are at an increased risk of developing type 2 diabetes mellitus (T2DM) later in life and possibly to the development of cardiovascular disease (CVD). Generally, the risks of diabetes in pregnancy include spontaneous abortion, fetal anomalies, preeclampsia, fetal demise, macrosomia, neonatal hypoglycemia, neonatal hyperbilirubinemia, and neonatal respiratory distress syndrome [3]. Furthermore, the fetus is at greater risk for childhood obesity, as well as for T2DM, and consequently for CVD in adult life. This phenomenon appears to lead to the transmission of predisposition to obesity and diabetes from generation to generation, which affects the health of the population [4].

During a normal pregnancy, the mother’s body undergoes a series of changes in its physiology to support the demands of the developing fetus. These include adaptations to the cardiovascular, urinary, hematopoietic, and respiratory systems and the metabolism. An important metabolic parameter that varies during pregnancy is insulin sensitivity, which changes depending on the demands of pregnancy. According to the prevailing theory, during early pregnancy, insulin sensitivity remains stable or even increases, promoting glucose uptake into adipose tissue stores in preparation for the energy demands of later gestation [5]. More recent studies, however, suggest that women in early pregnancy may already exhibit reduced insulin sensitivity, showing metabolic features comparable to those seen in polycystic ovary syndrome (PCOS) [6]. In the later gestational ages, increasing doses of different hormones like the estrogens, progesterone, cortisol, leptin, placental lactogen, and placental growth hormone play a part in the onset of insulin resistance by different mechanisms [7]. As a result, blood glucose is slightly elevated, which favors its transport across the placenta to fuel fetal development. This mild state of insulin resistance also promotes endogenous glucose production and the breakdown of fat stores, resulting in further increases in blood glucose and free fatty acid concentrations [8]. The development of GDM, expressed as abnormal hyperglycemia during pregnancy, is the clinical manifestation of a greater degree of insulin resistance. In these pregnancies, due to genetic or other factors, it is likely that lower insulin sensitivity exists both before and following pregnancy [9]. The vast majority (~80%) of GDM cases present as β-cell dysfunction with a background of chronic insulin resistance, which leads the pregnant woman to be unable to adequately cope with the metabolic changes that occur during gestation [10].

Several risk factors for developing GDM have been studied [10]. These include pre-gestational obesity of the mothers, excessive weight gain during pregnancy, unhealthy diet, ethnicity, genetic background, advanced maternal age, low or high maternal birth weight, and personal history of GDM or other insulin resistance conditions, such as polycystic ovary syndrome (PCOS) [8].

The prevalence of GDM also varies worldwide due to different screening methods and diagnostic criteria [2,11,12]. The HAPO study was undertaken in response to the need to establish internationally agreed-upon diagnostic criteria for gestational diabetes based on their predictive value for adverse pregnancy outcomes. An increase in each of the three time-values on the 2 h 75 g oral glucose tolerance test is associated with graded increases in the likelihood of events of pregnancy complication, such as large-for-gestational-age fetuses, need of cesarean section, abnormal fetal insulin levels, and neonatal fat content [13]. A diagnosis of GDM is recommended when any of the following three values on the 2 h 75 g oral glucose tolerance test (OGTT) are met or exceeded: 92 mg/dL at fasting, 180 mg/dL at one hour, 153 mg/dL at two hours. These criteria were also adopted by the World Health Organization (WHO) guidelines in 2013.

While the euglycemic clamp and intravenous glucose tolerance test (IVGTT) are regarded as standard methods for evaluating insulin sensitivity, they are not appropriate for routine clinical application or extensive population studies. The HOMA-IR method is widely used as an indicator of insulin resistance, offering the advantage of being simple and practical to calculate [14]. Fasting insulin and glucose levels are used in the calculations for the homeostasis model assessment (HOMA), specifically HOMA-IR = glucose X insulin/405. This model was first described by Matthews et al. in 1985 [15].

As mentioned above, insulin resistance, or the occurrence of gestational diabetes mellitus, is influenced by racial and ethnic components. A very special and distinct ethnic group is the Roma population, primarily residing in Central and Eastern Europe, with Indian origins. They are defined by a shared culture and various similar Romani languages, and most often they live in segregated communities or on the outskirts of towns and have abandoned the nomadic life. Data provided by social sciences, as well as genetic research, suggest that the 8–10 million Roma (Gypsies) living in Europe today are best described as a society of genetically isolated populations [16]. The relationship between the traditional social structures observed among Roma—where race is the main unit—and the boundaries, demographic history, and biological kinship of their different ancestral communities appears complex and has not been completely analyzed by population genetic studies [17]. The results of Font-Porterias et al. study (2019) suggest that the putative origin of the proto-Roma involves a Punjabi group with low levels of West Eurasian ancestry, and a complex West Eurasian (around 65%) component was identified in the Roma as well, as a result of the admixture events that occurred with non-proto-Roma populations [18]. Recent genetic research has identified several new, or previously known, but rare, conditions caused by common mutations. The limited epidemiological data available suggest a non-random distribution of disease-causing mutations among Roma groups [19].

Examples of such diseases are three neurological disorders, namely the hereditary motor and sensory neuropathies of the Lom and Russe types, and a facial dysmorphic neuropathy syndrome with congenital cataracts [19]. Accordingly, some of the existing studies, specifically those by Enache et al. and Živkovic et al., indicate an increased prevalence of diabetes mellitus in Roma populations [20,21]. Furthermore, based on Piko et al., the Roma population appears to have a greater genetic predisposition to insulin resistance compared to other Caucasians [22]. While some studies indicate that the prevalence of GDM may be greater in Roma than in non-Roma populations, the existing evidence is limited, and there is significant heterogeneity based on lifestyle, socioeconomic status, and methodology. In order to add to this sparse literature, we examined a distinct Greek Roma pregnant population residing in the suburbs of Kalamata, Greece, covered by the health services of the local hospital.

2. Materials and Methods

2.1. Materials

• Our study population were Roma pregnant mothers aged 18 to 35 years old.
• Their BMI at the beginning of pregnancy and after the completion of pregnancy did not exceed the index of 30.
• Mothers with pre-existing diabetes were excluded.
• Samples and data from non-Roma volunteers of European descent with the same age and BMI served as the control group.

Participants were classified as Roma based on multiple, contextually grounded criteria: self-identification, membership in distinct Roma communities, shared familial lineage, and use of the Romani language. These criteria are consistent with prior population-genetic and sociocultural research on Roma groups [22,23]. Non-Roma controls were recruited from the same geographic region and were of European descent, with both parents born in European countries but not in Roma communities. This strategy aligns with established approaches for distinguishing Roma and non-Roma populations in biomedical studies [24,25].

2.2. Methods

• Fasting glucose and insulin samples from the women’s plasma were used to determine the HOMA-IR index, to determine insulin resistance.
• Values from OGTT and plasma glucose during pregnancy were used to diagnose gestational diabetes mellitus, as defined by the WHO diagnostic criteria (2013) [2], and consequently the insulin resistance implied by their pathological values. Women underwent the oral glucose tolerance test (OGTT) at 24–28 weeks of gestation, while fasting glucose, insulin, and HOMA-IR were evaluated during the third trimester (35–39 weeks).
• Data were collected from the medical histories of women and their newborns from the hospital’s archives, and anthropometric data were obtained from women who became pregnant within the study period.

2.3. Statistical Analysis

Quantitative variables were expressed as mean values (standard deviation) and as the median (interquartile range), while categorical variables were expressed as absolute and relative frequencies. For the comparison of proportions, chi-square and Fisher’s exact tests were used. Student’s t-tests and Mann–Whitney tests were used for the comparison of continuous variables between the two groups. Multiple linear regression analysis was used with the dependent variable being women’s HOMA-IR index in the 3rd trimester. The regression equation included terms for age, race, and BMI after pregnancy. Adjusted regression coefficients (β) with standard errors (SE) were computed from the results of the linear regression analysis. All reported p values are two-tailed. Statistical significance was set at p < 0.05 and analyses were conducted using SPSS statistical software (version 27.0).

3. Results

3.1. Demographic Characteristics of the Population (

Table 1. Sample’s demographic characteristics, total and by population group.

Variable	Total Sample (n = 65; 100%)	Controls (n = 32; 49.2%)	ROMA (n = 33; 50.8%)	p
Multiparous (number of Children)
1	27 (41.5)	17 (53.1)	10 (30.3)	0.028
2	26 (40)	13 (40.6)	13 (39.4)
3–4	12 (18.5)	2 (6.3)	10 (30.3)
Abortuses history	9 (13.8)	5 (15.6)	4 (12.1)	0.733
PCOS	1 (1.5)	0 (0)	1 (3)	>0.999
Smoking	29 (44.6)	3 (9.4)	26 (78.8)	<0.001
Cardiovascular disease	3 (4.6)	1 (3.1)	2 (6.1)	>0.999
Chronic lung disease	2 (3.1)	1 (3.1)	1 (3)	>0.999
Chronic liver disease	1 (1.5)	0 (0)	1 (3)	>0.999
Chronic renal disease	0 (0)	0 (0)	0 (0)	-
Immuno-compromised condition	1 (1.5)	0 (0)	1 (3)	>0.999
Neurologic disorder	3 (4.6)	1 (3.1)	2 (6.1)	>0.999
Psychiatric disorder	2 (3.1)	1 (3.1)	1 (3)	>0.999
Autoimmune disorder	3 (4.6)	3 (9.4)	0 (0)	0.114
Age (years), Mean (SD)	26.2 (5.7)	29.9 (4.8)	22.5 (3.9)	<0.001
BMI before pregnancy (kg/m2), Mean (SD)	22.7 (2.2)	22.7 (2.1)	22.7 (2.4)	0.981
BMI after pregnancy (kg/m2), Mean (SD)	24.5 (2.6)	24.2 (2.4)	24.8 (2.9)	0.333
Difference in BMI, Mean (SD)	1.84 (1.5)	1.52 (1.28)	2.15 (1.65)	0.095

SD: Standard Deviation, PCOS: Polycystic Ovary Syndrome.

)

• Data from 65 women were collected, with a mean age of 26.2 years (SD = 5.7 years). Almost half of them (49.2%) were controls, and the rest (50.8%) were Roma mothers.
• Women’s demographic characteristics are presented in Table 1.
• The number of children was significantly greater in the Roma group (p = 0.028), as well as the percentage of smokers (p < 0.001).
• Women in the Roma group were significantly younger than those in the control group (p < 0.001).

3.2. Glucose, Insulin, and HOMA-IR in Roma Pregnancies Compared to Control Pregnancies

Maternal and neonatal characteristics, in total and by race are presented in

Table 2. Maternal and neonatal characteristics, total sample and by population group.

Variable	Total Sample (n = 65; 100%)	Controls (n = 32; 49.2%)	ROMA (n = 33; 50.8%)	p
Glucose (0 min), Mean (SD)	82.3 (7.4)	84.1 (6.6)	80.5 (7.9)	0.050
Glucose (60 min), Mean (SD)	136.4 (28.6)	147.8 (21.8)	125.3 (30.3)	0.001
Glucose (120 min), Mean (SD)	117.4 (23.5)	123.6 (19.7)	111.3 (25.6)	0.034
Fasting Insulin, Mean (SD)	15.11 (8.87)	11.6 (6.49)	18.63 (7.74)	0.0013
HOMA-IR (3rd trimester), Mean (SD)	3.1 (2)	2.4 (1.4)	3.9 (2.3)	0.002
Birth Weight (g), Mean (SD)	3157 (337.5)	3275.5 (323.4)	3042.1 (314.4)	0.004
HBCA1 (1st trimester), Median (IQR)	5 (4.4–5.2)	5 (4.4–5.2)	4.8 (4.3–5.2)	0.782
Pathological glucose curve, n (%)	7 (10.8)	2 (6.3)	5 (15.2)	0.427
Gestational diabetes, n (%)	7 (10.8)	2 (6.3)	5(15.2)	0.427
Gestational hypertension, n (%)	3 (4.6)	0(0)	3 (9.1)	0.238
Preeclampsia, n (%)	2 (3.1)	0 (0)	2 (6.1)	0.492
Gestational age at delivery 35–36 w, n (%)	4 (6.2)	1 (3.1)	3 (9.1)	0.366
Gestational age at delivery 37–38 w, n (%)	10 (15.4)	3 (9.4)	7 (21.2)
Gestational age at delivery 39–40 w, n (%)	35 (53.8)	20 (62.5)	15 (45.5)
Gestational age at delivery > 40 w, n (%)	16 (24.6)	8 (25)	8 (24.2)
Cesarean delivery, n (%)	28 (43.1)	11 (34.4)	17 (51.5)	0.163
Vaginal delivery, n (%)	37 (56.9)	21 (65.6)	16 (48.5)
5 min Apgar score < 4, n(%)	1 (1.5)	0 (0)	1 (3)	>0.999

SD: Standard deviation, IQR: Interquartile Range.

.

• Glucose at 0 min (p = 0.050), at 60 min (p = 0.001) and at 120 min (p = 0.034) was significantly lower in the Roma group.
• On the contrary, the mean fasting insulin levels were significantly higher in the Roma group (p = 0.0013)
• As a result, HOMA-IR in the 3rd trimester was significantly higher in the Roma group.

3.3. Mean Birthweight and Breastfeeding (Table 2)

• Mean birth weight was significantly lower in the Roma group.
• The percentage of breastfeeding in the control group was 90.6% while in the Roma group it was significantly lower, equal to 9.1% (p < 0.001).

3.4. Mean Birthweight Correlated with Smoking Habits (Figure 1)

Mean birthweight for women who smoked was 3050.0 gr (SD = 321.3 gr) while for those who did not smoke it was significantly greater, equal to 3243.2 gr (SD = 329.5 gr), p = 0.020, (Figure 1).

3.5. Association of Maternal Age and BMI with HOMA-IR (

Table 3. Multiple linear regression results with HOMA-IR (3rd trimester) as the dependent variable.

Variable	β (Regression Coefficient)	SE	p-Value
Age (years)	0.05	0.04	0.23
BMI after pregnancy (kg/m2)	−0.01	0.08	0.85
Smoking (yes vs. no)	0.48	0.36	0.19
Group (Roma vs. Controls)	1.92	0.59	0.002

)

• In multivariate linear regression analysis, Roma ethnicity remained significantly associated with higher HOMA-IR values in the third trimester (β = 1.92, SE = 0.59, p = 0.002), even after adjusting for age, BMI after pregnancy, and smoking status (Table 3). Neither age (p = 0.23), BMI (p = 0.85), nor smoking (p = 0.19) were significantly associated with HOMA-IR in the adjusted model.

3.6. Association of Maternal Age and BMI with GDM (

Table 4. Logistic regression analysis with Gestational Diabetes Mellitus (GDM) as the dependent variable.

Variable	Odds Ratio (OR)	95% Cl	p-Value
Age (years)	1.02	0.93–1.12	0.63
BMI after pregnancy (kg/m2)	1.05	0.85–1.31	0.65
Smoking (yes vs. no)	1.40	0.25–7.85	0.70
Group (Roma vs. Controls)	2.70	0.48–15.6	0.25

)

• In logistic regression analysis with GDM as the dependent variable, Roma ethnicity was associated with an increased, though not statistically significant, risk of GDM compared with controls (OR 2.70, 95% CI 0.48–15.6, p = 0.25). Neither age (OR 1.02, 95% CI 0.93–1.12, p = 0.63), BMI after pregnancy (OR 1.05, 95% CI 0.85–1.31, p = 0.65), nor smoking status (OR 1.40, 95% CI 0.25–7.85, p = 0.70) were significantly associated with GDM in the adjusted model (Table 4). These results suggest that the observed difference in GDM prevalence between Roma and control groups was not statistically significant, likely due to limited sample size.

4. Discussion

This study provides evidence for increased insulin resistance among Roma pregnant women in comparison to non-Roma mothers, indicating clear differences between Roma and non-Roma pregnancies, which may reflect a combination of social, environmental, and genetic influences on metabolic processes during pregnancy. This development will add to the existing literature on ethnic disparities in metabolic health and highlights the need for personalized prenatal care regimes.

Pregnant Roma women demonstrate higher insulin resistance when compared with their control peers, despite being adjusted for certain covariates including age, body mass index (BMI), and parity. These results align with previous research showing that ethnic background can independently influence glucose regulation in pregnancy. Retnakaran et al. (2006) found that Asian and South Asian ethnicity independently predicted increased insulin resistance in late pregnancy [26]. The mechanisms behind these disparities are complex and likely involve a combination of genetic predisposition, lifestyle factors, and socioeconomic conditions affecting healthcare access, nutrition, and physical activity [25].

One of the major observations is that the Roma women were significantly younger than the control group; however, they had high fasting insulin and HOMA-IR values. The comment is especially noteworthy in the context of the fact that rising maternal age is characteristically accompanied by a reduction in insulin sensitivity and a rising risk of GDM [27]. Thus, rather than diminishing comparability, this preliminary age difference corroborates our conclusion: despite their young age, the Roma group already had elevated insulin resistance. This suggests the influence of factors beyond age (e.g., genetics, lifestyle options, or rather, socioeconomic circumstances) for the differences identified.

Crucially, although our results show stark differences between Roma and non-Roma participants, these cannot be explained by genetic factors alone. The definition of Roma participants within our study relied on several, contextually grounded criteria, including self-identification, community membership, family descent, and use of the Romani language [23,24]. Such definitions are in line with previous sociocultural and population-genetic studies but are still social in nature and do not entirely reflect underlying genetic variation [17,25]. Socio-environmental factors such as diet, physical activity, and access to healthcare may therefore play a significant role in the differences we noticed. We have therefore amended our interpretation to make it clear that both genetic and environmental factors are likely to be involved and that additional research is required to tease apart these contributions.

Part of the elevated insulin resistance observed in Roma participants may reflect an inherited predisposition. Insulin resistance was found to be higher among Hungarian Roma populations than the general Hungarian population; therefore, genetic predisposition could be put forth as an argument [28]. Insulin resistance may be exacerbated by environmental variables, such as dietary practices, a lack of physical activity, and a higher prevalence of obesity among Roma populations. Pregnancy-related metabolic outcomes are greatly influenced by the interaction of genetic and environmental variables [16]. The interplay between genetic and environmental factors is crucial in shaping metabolic outcomes during pregnancy. Studies have shown that Roma populations often experience socioeconomic disadvantages, which can lead to unhealthy dietary habits and lower levels of physical activity. For instance, a study on the dietary profile and nutritional status of the Hungarian Roma population found significant differences in nutrient intake and body composition compared to the general population, indicating a higher risk of metabolic syndrome [29].

Although studies about GDM in Roma women remain scarce, numerous studies have indicated that exacerbated insulin resistance in pregnancy augments the risk of complications like preeclampsia, macrosomia, and later-in-life metabolic disorders in the offspring. For example, the Mayo Clinic lists obesity or overweightness, inactivity, and a history of gestational diabetes mellitus in prior pregnancies as risk factors for GDM. These elements may raise the risk of problems, including preeclampsia and macrosomia, by causing increased insulin resistance [30].

Social and economic disparities probably play a key role in such inequities. Inhibitors like lower access to prenatal care, low health literacy, and institutionalized differences may cause prolonged diagnosis and prevent proper management of glucose intolerance in Roma mothers [31]. Addressing these inequalities is a key for better outcomes. Programs aimed at community education, enhanced access to prenatal care, and culturally sensitive interventions should be prioritized [32].

Further research should aim to elucidate the genetic and molecular mechanisms underlying increased insulin resistance in Roma pregnancies. Longitudinal studies examining the progression of glucose intolerance from early to late pregnancy in diverse populations could provide valuable insights [22,33]. Moreover, interventional studies focusing on lifestyle modification and healthcare access in Roma communities are warranted to address the disparities identified in this study.

Strengths and Limitations

This study’s strengths include a robust methodology and a well-characterized population of Roma and control pregnant women of European descent. However, there are some limitations to be recognized. Firstly, the low sample size limits generalization to other Roma populations with diverse genetic and environmental influences. We also lacked detailed dietary and physical activity data from the participants, which may greatly affect insulin resistance. Future studies should incorporate detailed assessments of dietary habits, physical activity, and healthcare utilization, since these factors are likely to play a significant role alongside biological predispositions.

In addition, HOMA-IR, while widely used in epidemiological studies, correlates only moderately with dynamic indices of insulin sensitivity derived from OGTT, such as Matsuda, Belfiore, or QUICKI [34,35]. Because insulin concentrations during OGTT were not available in our dataset, these indices could not be calculated. Future studies should include OGTT-derived insulin measurements to provide a more comprehensive evaluation of insulin resistance.

Another limitation is the unexpectedly low prevalence of PCOS in our cohort (0 vs. 1 case), which may reflect under-diagnosis. This information was based on self-reported history and available records, rather than systematic pre-pregnancy clinical or biochemical evaluation. It is important to clarify that especially in the Roma group, which had a higher rate of pregnancy in early adulthood, there was no pre-pregnancy assessment by a gynecologist or an endocrinologist to diagnose PCOS. It is therefore possible that some women with irregular menstrual cycles had undiagnosed PCOS. Future studies should consider standardized endocrine assessments and menstrual history to improve diagnostic accuracy.

Even though we conducted multivariate analyses with adjustment for age, BMI, and smoking, residual confounding from these variables cannot be ruled out. In addition, logistic regression analysis indicated that the elevated odds of GDM in Roma women were not statistically significant, which was probably a result of the small sample size, and our results should thus be considered exploratory and hypothesis-generating. However, despite the relatively limited number of participants in our study, the results of the post hoc power analysis suggested that the sample size was adequate (90% power) to identify a substantial effect size (d ≥ 0.8) concerning third-trimester HOMA-IR, which was the primary outcome of interest.

Additionally, our identification of Roma participants relied on social and cultural markers such as language, community belonging, and familial lineage. While consistent with the approaches used in prior Roma health research [17], these criteria cannot fully separate genetic from socio-environmental influences. Therefore, the observed differences should not be interpreted as solely genetic in origin, but rather as the result of complex interactions between biology, culture, and environment.

5. Conclusions

This research provides evidence that Roma women demonstrate higher levels of insulin resistance during pregnancy than non-Roma counterparts of similar age and BMI. These results point to the necessity of taking both ethnic origin and socio-environmental situation into account in evaluating metabolic well-being during pregnancy. Nevertheless, the differences observed are unlikely to be explained by hereditary factors alone. Social, cultural, and lifestyle factors, such as nutrition, physical activity, smoking, and access to healthcare are also likely to contribute significantly in addition to biological susceptibility.

The small sample size, reliance on HOMA-IR rather than OGTT-based dynamic indices, and potential under-diagnosis of PCOS limit the generalizability of our findings. Future studies in larger and more heterogeneous groups, with comprehensive evaluations of diet, lifestyle, OGTT-based insulin measures, and endocrine function, will be essential for better discerning the genetic versus environmental contributions to gestational insulin resistance in Roma populations.

In addition, our findings highlight the need for targeted prenatal counseling and culturally sensitive interventions aimed at improving dietary habits, lifestyle, and access to healthcare in Roma women, who might represent a population at increased risk of gestational diabetes mellitus due to the combined effect of biological, social, and environmental factors.