Pregnant women recruited from Sidra Medicine (ISRCTN Registration number ISRCTN16448440) were recruited to conduct a randomized double-blinded clinical trial. The MIGDM study design and population characteristics were published previously (15). Briefly, pregnant women attending the antenatal care clinic at Sidra Medicine before 16 weeks of gestation and older than 18 years were considered for participation in the study after receiving comprehensive written and oral information in both English and Arabic.

Randomization was done via computer-generated numbers to assign participants to MI or Placebo (PLA) arm, the details of which were previously published (15). Briefly, MI and PLA were arranged at the source with identical packaging and provided through the hospital pharmacy. The MI pack contained 2 g of Myo-inositol, whereas the placebo option contained a pharmacologically passive substrate. Participants in both arms received twice-daily dosing and completed at least 12 weeks of intervention prior to undergoing the Oral Glucose Tolerance Test (OGTT). The OGTT was performed at 24–28 weeks’ gestation for routine screening for GDM. Women were instructed to continue consuming the trial packs and OGTT results should not have an impact on the intake. Patient demographics and clinical parameters were collected, including age at the recruitment, pre-pregnancy weight and BMI categories (underweight <18.5 kg/m², normal weight 18.5–24.9 kg/m², overweight 25–29.9 kg/m², obese ≥30 kg/m²), overall weight gain, fasting glucose, fasting insulin, 1-HR and 2-HR glucose post-OGTT, Homeostasis model assessment of beta cell function (HOMA-B), and Homeostasis model assessment of insulin resistance (HOMA-IR). Pregnancy and delivery outcomes were collected by the research coordinator.

The dietary consumption was assessed through the 24-h dietary (24HR) recalls and using Food Frequency Questionnaires (FFQ) during recruitment (baseline) and the treatment at all trimesters of pregnancy. Nutrient intake was calculated using dedicated software (Nutritionist Pro, Axxya). Physical activity and Lifestyle questionnaires were administered at the same time points by a trained dietitian, including questions on walking times/week and vigorous and moderate exercise times/week, screen time/week, and sleeping hours/day.

Dietary pattern analysis was done using two scores. The first one is the healthy eating index (HEI), which reflects the diet’s alignment with key dietary recommendations. HEI utilizes 13 Food components that are weighted at 5 or 10 points according to the food type. A higher total HEI score indicates a diet that aligns better with dietary recommendations (16). The second dietary score is the dietary inflammatory index (DII), which categorizes diets based on their inflammatory effect. It is calculated by using 28 food parameters that are weighted based on the world average and standard deviation. The higher total DII score indicates that the participant is following a pro-inflammatory dietary intake and vice versa (17).

Proportions and frequencies were utilized to report qualitative results, while means, median, standard deviation and interquartile range (IQR) were used to report quantitative variables. Normality tests have been performed on continuous variables; nonparametric tests have been used with not-normally distributed data, while t-test and one-way ANOVA have been applied for normally distributed variables. Chi-squared analysis was used to assess the differences between qualitative variables. Also, analyzing the data using logistic regression analysis to assess odds ratios (ORs) and to study the predictive effect of all risk factors. 95% confidence intervals of ORs for related factors were estimated. All statistical assessments were performed on SPSS package version 29 (IBM) and Prism Graph Pad 10, using two-sided measurements and considered significant at a p-value < 0.05.

The patients who completed the study procedures (n = 72) were included in the analysis for both MI (n = 37) and Placebo (n = 35) groups. Clinical parameters recorded from patients include age, pre-pregnancy BMI, overall weight gain, fasting glucose, 1-h and 2-h post-OGTT glycemia, fasting insulin, HOMA-IR, and HOMA-B measured at 24–28 weeks of gestation. The values for each arm of the intervention are summarized in Table 1. The comparison analysis revealed only a minor difference in fasting glucose in OGTT between the two groups.

Gestational weight increased steadily across trimesters (Figure 1A), peaking in the third trimester as expected, but showing a significant difference between MI and PLA groups and a notable weight increase in both GDM-PLA and non-GDM-PLA groups in the third trimester (p = 0.0385) (Figure 1B). According to the American College of Obstetricians and Gynecologists, weight gain should be evaluated based on the pre-pregnancy BMI categories (underweight, normal, overweight, and obese) (18). We measured the overall weight gain of each subject by the end of the pregnancy and compared it between the two intervention groups. In the normal-weight category, the mean weight gain in the MI group was 10.2 kg, slightly below the recommended range of 11.5–16 kg. In contrast, the placebo group had a mean weight gain of 11.9 kg, which falls within the recommended range. For the overweight category, both groups exhibited mean weight gains below the lower limit of the recommended range (7–11.5 kg). The MI group had a mean weight gain of 5.5 kg, while the placebo group had a mean of 3.5 kg. In the obese category, the MI group showed a mean weight gain of 2.8 kg, which is notably below the recommended range of 5–9 kg. Conversely, the placebo group had a mean weight gain of 16.7 kg, far exceeding the upper limit of the recommended range. However, the standard deviation shows great interindividual variability, impacting the results’ significance (Table 1).

Our investigation of the participants’ lifestyle habits showed no differences in physical activity at the baseline, except for a relatively shorter sleeping time (Supplementary Table 1). During the intervention period, we recorded a significant difference in walking time (minutes/day) (p = 0.004), with the MI group (both GDM and non-GDM) being more active than PLA. Additionally, screen time (hours/day) showed significant differences between trimesters (p = 0.046), reflecting behavioral changes over time and between the two intervention groups (Figure 2 and Supplementary Table 2). It’s worth mentioning that, overall, most of the participants performed insignificant physical activity and did not adhere to the recommended sleeping time across groups (Supplementary Table 2).

Diet indexes were used to assess the quality of diet at the baseline and across trimesters, with the HEI addressing adherence to the dietary recommendations and the DII measuring the inflammatory effect of certain dietary habits. The HEI score was similar at the baseline to show then a significant improvement in the quality of diet during the three trimesters (p = 0.0328), with a major difference between 1st and 3rd trimesters (p = 0.0268). The DII follows a similar trend with an high pro-inflammatory diet at the baseline and across pregnancy, which reduced going toward the end of pregnancy (p = 0.0462), and in particular when comparing 1st vs. 3rd trimester (p = 0.0354) (Figures 3A,B and Supplementary Table 3). We observed that the GDM-MI group adhered more closely to an anti-inflammatory diet compared to the placebo group, but without reaching a statistical difference (Figure 3D), even if no improvement was observed in the HEI (Figure 3C).

The analysis of the dietary habits did not show any differences in the frequency of food consumption at the baseline (Supplementary Table 3), whether significant differences were identified across pregnancy and between the intervention groups. A larger amount of poultry was consumed daily in trimester three (p = 0.013, Figure 4A) along with an increase in the preference for chocolate, cookies, and cakes (p = 0.011, Figure 4B). On the contrary, the consumption of added salt decreased in trimester three (p = 0.005, Figure 4D) and at the same time, the women started consuming skimmed milk (p = 0.032, Figure 4C).

Most importantly, changes in food preferences were also observed between the two treatment arms. In particular, the daily intake of legumes, beans, and tubers was higher in the MI group (p = 0.008 and p = 0.036, respectively, Figures 5A,B), whereas pizza consumption was higher in the PLA group, both GDM and non-GDM (p = 0.019, Figure 5C). Moreover, the chi-square analysis revealed differences in the daily consumption of fresh juices, added sugar, chocolates, and cookies. Participants in the MI group exhibited lower consumption of freshly pressed fruit/vegetable juice (45.2%) compared to the PLA group (66.7%, p = 0.044) and sugar (40.5% vs. 66.7%, p = 0.014, Figures 5D,E). In contrast, the same group consumed significantly more chocolate, cookies, and cake (64.3%) than those in the placebo group (42.2%, p = 0.038, Figure 5F). All these results together suggest that the MI groups follow a healthier diet, with increased consumption of recommended foods (legumes, tubers) and reduced intake of high sugar and fat foods (added sugar, fruit juices, and pizza), except for sweets.

The daily nutrient intake was computed from the 24HR recalls and compared between MI and PLA groups. We first analyzed the modification in nutrient intake at the baseline and during pregnancy. No differences emerge in nutrient intake between the GDM and non-GDM groups at the baseline (Supplementary Table 4). During pregnancy, instead, the total dietary fiber intake appears to decrease in the second trimester to recover in trimester three (p = 0.018, Figure 6A). This is reflected by changes in the subtype of fibers. The intake of crude fiber peaks at trimester three (p = 0.005, Figure 6B), whereas the soluble fiber gradually decreases across trimesters (p = 0.001, Figure 6C), and insoluble follows shows an opposite behavior, with the highest intake at trimester three (p < 0.0001, Figure 6D). A similar trend is shown in the intake of fructose and glucose (p = 0.037 and p = 0.024, Figures 6E,F). This trend suggests a marked shift in dietary behavior and nutritional requirements as pregnancy progresses, potentially reflecting an increased emphasis on fiber-rich foods during the later stages of pregnancy. The intake of all amino acids, many minerals and vitamins also change during the course of pregnancy (Supplementary Table 5).

The nutrient intake also differs between the two intervention groups. Carbohydrate, glucose, fructose and soluble fiber intake are all significantly greater in the MI group compared to the PLA group (p = 0.047, p = 0.029, p = 0.030, and p = 0.036, respectively) as shown in Figures 7A–D. The polyunsaturated fatty acid PFA22:5, eicosapentaenoic acid (EPA), showed a significantly higher intake in the non-GDM-MI group compared to both GDM-MI and GDM-PLA and non-GDM groups (p = 0.017; Figure 7E).