Higher early pregnancy plasma myo‐inositol associates with increased postprandial glycaemia later in pregnancy: Secondary analyses of the NiPPeR randomized controlled trial

Myo‐inositol supplementation from ~13 weeks' gestation reportedly improves glycaemia regulation in metabolically at‐risk women, with speculation that earlier supplementation might bring further improvement. However, the NiPPeR trial of a myo‐inositol‐containing supplement starting preconception did not lower gestational glycaemia in generally healthy women. We postulated that the earlier timing of supplementation influences the maternal metabolic adaptation for gestational glycaemia regulation.


| INTRODUCTION
Higher gestational glycaemia is linked to higher clinical risk across a continuum for both the woman and her offspring, including hypertensive disorders of pregnancy, caesarean section delivery, foetal macrosomia, shoulder dystocia and neonatal hypoglycaemia. 1 In-utero exposure to increasing maternal glycaemia is also associated with long-term adverse cardiometabolic health in offspring. 2[5] One approach is through prenatal supplementation with myo-inositol, an endogenously synthesized polyol, which is also enriched in dietary grains, fruit and vegetables. 6It is a precursor of phosphoinositides and other key secondary messengers for hormone signal transduction, including that of insulin, and is derivatized with different compounds to regulate many cellular functions and form signalling agents, including insulin-mimetics. 7veral trials of antenatal myo-inositol supplementation in White Italian women with risk factors for gestational diabetes [e.g.family history of diabetes, 8,9 high body mass index (BMI), 10,11 polycystic ovary syndrome 12 ] have reported reduced gestational glycaemia.A meta-analysis of six Italian studies (n = 995 women), which started myo-inositol supplementation mostly at the end of the first trimester, reported glycaemia reductions at all three time points of a classic 2-h 75 g oral glucose tolerance test (OGTT) conducted at 24-28 weeks' gestation. 13In addition, these trials reported accompanying reductions in gestational diabetes incidence and insulin resistance [homeostasis model assessment for insulin resistance (HOMA-IR)].With these trials showing that myo-inositol supplementation at a daily dose of 4 g is safe and tolerable in pregnancy, some have advocated starting myo-inositol supplementation earlier in gestation or even preconception to improve gestational glycaemia regulation further.
However, these trials lack generalizability to other populations.Indeed, some trials conducted elsewhere reported different results.
An Irish trial in predominantly White women with a family history of diabetes reported that a combined myo-inositol and D-chiro-inositol supplement made no difference to gestational glycaemia. 14Another trial, the NiPPeR (Nutritional Intervention Preconception and During Pregnancy to Maintain Healthy Glucose Metabolism and Offspring Health) study, reported no reduction in the primary outcome of gestational glycaemia overall, with a marginal increase in 2 h glycaemia among women with overweight/obesity. 15NiPPeR differed from the other trials in several ways: it was double-blinded, supplementation commenced preconception, recruited many women with no metabolic risk factors, included different ethnicities from three continents with a significant proportion of Asians, and administered myo-inositol in a preparation enriched with other micronutrients and probiotics.
Here we have utilized data from the NiPPeR study to explore possible reasons for the inconsistent findings observed among trials of antenatal myo-inositol supplementation in relation to gestational glycaemia.If myo-inositol supplementation is to be considered more widely for optimizing glycaemia regulation in pregnancy, it is important to understand which subpopulations may derive the most benefit, the potential impact of co-supplementation with other micronutrients, and the optimal timing for the commencement of supplementation.
We postulated that an early pregnancy timing of supplementation may unfavourably alter the maternal metabolic adaptation for gestational glycaemia regulation.This study aimed to investigate the associations of plasma myo-inositol concentrations at 7 weeks (early pregnancy) and 28 weeks (late pregnancy) gestation with glycaemia at 28 weeks, as well as explore the potential influences of ethnicity, pre-existing metabolic risk factors, variations in inositol metabolism and excretion, and co-variation in other supplement components.Besides gestational diabetes risk, we also assessed associations with the related outcomes of birthweight 16 and cord C-peptide, 21 serving as indicators of clinical impact on the offspring.Exploratory analyses were conducted to elucidate potential underlying mechanisms of effect, with in-silico modelling to mimic conditions in other trials to investigate if increasing myo-inositol later in pregnancy could still benefit glycaemia regulation.

| Study design and participants
Approval was granted by research ethics services at each site and written informed consent obtained from participants. 20This study is a secondary analysis of data collected in the NiPPeR international multicentre, double-blind randomized controlled trial (ClinicalTrials.govNCT02509988); it recruited 1729 women from the community in the UK, Singapore and New Zealand, aged 18-38 years who were planning conception in 2015-2017.Women were randomized into control and intervention arms (1:1 ratio), with supplements commenced preconception and continued throughout pregnancy.Supplements for both arms contained folic acid, iron, calcium, iodine and β-carotene; the intervention additionally included myo-inositol (4 g daily), vitamin D, riboflavin, vitamin B6, vitamin B12, zinc and probiotics (Lactobacillus rhamnosus and Bifidobacterium animalis sp.lactis). 15,17llowing preconception randomization, women were given up to a year to conceive before withdrawal from the study.The current substudy included 585 participants who conceived and provided myoinositol and gestational glycaemia data (Figure S1).Adherence was determined by supplement counting; 96.6% reported rates above the pre-specified good compliance threshold of 60% averaged from recruitment to delivery. 15

| Procedures and laboratory analyses
At recruitment, height, weight, waist circumference and blood pressure were measured and questionnaires ascertained maternal age, ethnicity, household income, parity, previous history of gestational diabetes and family history of diabetes.Preconception BMI was calculated.During pregnancy, participants reported their smoking status.
Offspring birthweights were extracted from medical records and birthweight centiles standardized for gestational age and sex. 1875 g OGTT was conducted at preconception baseline (fasting, 30 and 120 min) and again at 28 weeks' gestation (range 24-32 weeks; fasting, 30, 60, 90 and 120 min), with plasma glucose (PG) processed using a standardized protocol across the sites. 17Gestational diabetes was defined by the International Association of Diabetes and Pregnancy Study Group (IADPSG) criteria. 19Serum insulin concentrations at these same OGTT time points, and fasting plasma triglycerides and high-density lipoprotein cholesterol were batchanalysed (Roche Cobas).The Stumvoll first phase index, 20 the HOMA2-IR (www.OCDEM.ox.ac.uk) 21and Matsuda index (http:// mmatsuda.diabetes-smc.jp/xpoints.html) 22were calculated.

| Statistical analysis
Log e transformation of skewed plasma myo-inositol and glycaemia data achieved approximately normal distributions.Linear regressions were conducted in a single combined group of control and intervention cases to associate plasma myo-inositol concentrations (7 and 28 weeks) with gestational glycaemia (fasting, and 1 h and 2 h PG at 28 weeks), adjusting for recruitment site, and relevant covariates based on literature, including ethnicity (non-Asian/Asian), preconception BMI (continuous), parity (nulliparous/parous), maternal age (continuous), household income level (decile for country), family history of diabetes and smoking during pregnancy (none/passive/active).
Resulting β coefficients [with 95% confidence interval (CI)] represent percentage change in glycaemia for each percentage increase in myoinositol; β equivalents in mmol/L per μmol/L were calculated using the anti-log e mean of glycaemia and myo-inositol of the combined control-intervention group.The predicted effect of 7-week myo-inositol on 28-week glycaemia was represented by residuals computed using linear regression.Pearson's correlation was used to evaluate relations between log e plasma myo-inositol and 28-week glycaemia (adjusted for the above covariates).
As specific preconception maternal risk factors are known to relate to gestational glycaemia, we stratified the study population by these factors to detect potentially different associations between myo-inositol and glycaemia, reporting any statistical interactions.Preconception baseline risk factors studied were ethnicity (non-Asian/ Asian), family history of type 2 diabetes (present/absent), and metabolic risk using the five criteria defined by the International Diabetes Federation (IDF; low: no risk factors, moderate: with risk factors but did not fulfil metabolic syndrome criteria, and high: metabolic syndrome having central obesity with two other risk factors; Table S1).
In sensitivity analyses, we additionally adjusted for the following covariates in regression models: (a) plasma concentrations of other supplement components (folate, and vitamins B12 and D, which are thought to influence gestational glycaemia 24,25 ), and (b) inherent variations in baseline inositol processing represented by inositol metabolism (plasma scyllo-inositol/myo-inositol ratio) and urinary excretion (urine/plasma myo-inositol ratio).
Causal mediation analysis examined if plasma myo-inositol concentration could explain the overall NiPPeR intervention effect on gestational glycaemia, adjusting for covariates listed above.Analysis was performed under the assumption of sequential ignorability, reporting the average causal mediation effect (mediation package of statsmodels; Python).To elucidate a potential underlying mechanism, we examined associations between plasma myo-inositol concentra- There was no imputation of missing data.Statistical analyses were carried out using STATA 15 (Stata Corp.) and Python version 3.9.7.All statistical tests were two sided, with p < .05considered statistically significant.

| In-silico modelling
To investigate if previous myo-inositol trial results could be replicated within a similar subpopulation of the NiPPeR dataset, we used Monte Carlo simulation with bootstrapping (100 iterations; i.e. resampling with replacement for 100 times with each time ran with a selection of 90% of the population).First, we only used data from a population similar to previous myo-inositol trials, specifically White women with a family history of diabetes or with BMI ≥25 kg/m 2 .During resampling, similar ratios of UK and NZ participants, and of control and intervention were maintained.For each resampled test, the association of late-pregnancy myo-inositol (representing supplementation after the first trimester) with 28-week glycaemia was examined, with adjustment for previously listed covariates, the predicted effect of 7-week myo-inositol on 28-week glycaemia (to control for impact of early myo-inositol supplementation) and intervention group (to account for all other intervention components).If the CI of the pooled iteration results did not cross 0, it was considered statistically significant. 26We also conducted similar in-silico modelling in other subpopulations to identify potentially different effects.

| RESULTS
Of the 585 women in this study, 290 took the control supplement and 295 the intervention supplement containing myo-inositol.Plasma myo-inositol concentrations in early and late pregnancy were higher in the intervention group than in controls (Table 1).Overall, the population mean age was 30.3 years, with average BMI 23.7 kg/m 2 (Table 1).Non-Asian participants comprised 64.3%, mostly White (92.2%;59.3% of total cohort).The majority of Asian women were Chinese (87.4%; 24.8% of total cohort).Most were nulliparous (63.4%), did not smoke during pregnancy (96.6%), came from highincome households (66.2% from top two quintiles) and did not have a family history of type 2 diabetes (76.9%).The population was generally metabolically healthy at preconception baseline: median fasting glucose 4.9 mmol/L, mean glycated haemoglobin 5.2%, and median HOMA2-IR 0.9.
Adjusting for the predicted effect of 7-week myo-inositol, 28-week myo-inositol still showed no association with 28-week glycaemia (Figure 1C).The association between 7-week myo-inositol and 2 h PG were similar in control and intervention groups (myo-inositol Â group interaction p = .967).These associations were confirmed by positive correlations between 7-week myo-inositol and adjusted 28-week 1 h PG (Figure 1D) and 2 h PG (Figure 1E) concentrations.Linear relationships held across the 7-week myo-inositol range, where a substantial overlap between control and intervention groups was noted.
In exploratory analyses to identify a possible mechanism for the 7-week myo-inositol-glycaemia association, higher 7-week myo- In-silico modelling mimicking circumstances of previous myoinositol trials showed that among White women with a family history of diabetes or a high BMI (n = 168), a higher 28-week myo-inositol associated with lower fasting, and 1 h and 2 h PG at 28 weeks, equivalent to reductions of 0.09, 0.07 and 0.10 mmol/L, respectively, for an average increase in 28-week myo-inositol of 10.1 μmol/L with 4 g daily myo-inositol supplementation (Table 2).Glycaemia reductions were also observed in subpopulations of White women with and without IDF risk factors, and in non-Asian women.However, among Asian women, a higher 28-week myo-inositol associated with overall increased 28-week glycaemia, particularly among those with an IDF risk factor, although effect sizes were small (Table 2).

| DISCUSSION
Secondary analyses of the NiPPeR trial data showed that a higher plasma myo-inositol concentration very early in gestation ($7 weeks) was associated with higher postprandial glucose concentrations later in pregnancy ($28 weeks), regardless of ethnicity and in those with and without risk factors for hyperglycaemia.Additional analyses suggest a lowered insulin secretory capacity as a possible underlying mechanism.However, in silico modelling indicated that, accounting for the glucose-raising influence of early pregnancy myo-inositol, higher plasma myo-inositol concentrations later in pregnancy may still be associated with slightly reduced gestational glycaemia in White women but not in Asian women.Our findings may explain why the NiPPeR supplement containing myo-inositol did not reduce gestational glycaemia like previous trials, with demonstration that the 7-week myo-inositol concentration could account for the nonstatistically significant modest increase in 2 h PG in the intervention group compared with control.We conclude that myo-inositol supplementation that increases plasma myo-inositol in the early first trimester of pregnancy may impair later pregnancy postprandial glycaemic regulation.
NiPPeR was the only trial of myo-inositol supplementation aimed at optimizing gestational glycaemia that started supplementation preconception and continued through pregnancy, resulting in higher plasma myo-inositol concentrations from the very beginning of pregnancy.Other trials have not reported plasma myo-inositol levels to allow comparison with our findings.We speculate that in previous trials an early pregnancy plasma myo-inositol increase would not have happened as supplementation commenced typically at 8-13 weeks' gestation, 10,11,27 with one stretching to 26 weeks, 28 thus they unintentionally avoided the unfavourable impact on glycaemia.In an observational study, higher urinary concentrations of myo-inositol and D-chiro-inositol in the first trimester, probably reflecting correspondingly higher plasma inositol concentrations, were predictive of later gestational diabetes development, 29 supporting our findings.
We found an association between a higher 7-week myo-inositol and a lower 28-week Stumvoll index of acute insulin secretion.One possible mechanism for the observed increase in gestational glycaemia could therefore be through myo-inositol limiting beta-cell expansion in early pregnancy (Figure 3).1][32] Studies in cultured rat pancreatic islets suggest that myo-inositol promotes beta-cell responses to a glucose load, but high levels of myo-inositol suppressed beta-cell proliferation. 33It is thus possible that higher plasma myo-inositol levels in early pregnancy in humans could limit physiological beta-cell expansion, resulting in a lasting defect for the remainder of the pregnancy in the insulin secretory capacity in response to glucose challenge.This postulation is also consistent with the lack of association between 7-week myo-inositol and 28-week fasting glycaemia.
This postulation may also explain the NiPPeR subpopulation differences in myo-inositol-glycaemia associations (Figure 3).The only subgroup displaying a possible reduction in postprandial glycaemia in association with higher 7-week myo-inositol were those who met criteria for metabolic syndrome preconception.This result on a small sample needs to be interpreted with caution, but we speculate that such women may already be near their maximal beta-cell secretory capacity preconception 34 ; hence, a higher 7-week myo-inositol has little impact on beta-cell expansion, if any.Instead, the reduction in 28-week 2 h PG may simply reflect the insulin-sensitizing action of a higher myo-inositol concentration during pregnancy.
Overall, the magnitude of glycaemia increase associated with early pregnancy myo-inositol supplementation was small, with no apparent short-term clinical consequences (no associated increases in gestational diabetes, birthweight, or cord C-peptide concentrations).
However, given the documented continuum of risk with increasing gestational glycaemia across the glycaemia range, 1 there may be more subtle impacts on offspring, which may emerge with our ongoing follow-up.For some reassurance, we previously showed that higher placental inositol may suppress the adiposity-generating effects of maternal glucose in neonates, 35 with multiple actions of inositol potentially acting collectively to neutralize maternal glycaemiaassociated alterations in the offspring.
While most other trials either provided supplements containing a combination of inositol isomers or myo-inositol with folic acid, the NiPPeR intervention also contained other micronutrients and probiotics.Our analyses accounting for folate, and vitamins B12 and D did not alter the 7-week myo-inositol-glycaemia association.However, we cannot exclude the possibility that other supplement components may weaken 28-week myo-inositol-glycaemia associations and underlie the relatively modest in-silico-estimated supplement-induced In-silico bootstrapping model of the association between late pregnancy plasma myo-inositol and gestational glycaemia at 28 weeks in different subpopulations.Application of each β coefficient to the general average increase in 28-week plasma myo-inositol with 4 g daily myo-insoitol supplementation of 10.1 ± 0.17 μmol/L (calculated as the mean difference in anti-log e plasma myo-inositol between control and intervention at 28 weeks) to derive an estimated difference in glycaemia with supplementation.
reductions in fasting and 1 h and 2 h PG of 0.09, 0.07 and 0.10 mmol/L, respectively, in White women, being less than the 0.23, 0.49 and 0.48 mmol/L equivalents reported in a meta-analysis of six Italian trials. 13like White women, a higher 28-week myo-inositol was associated with slightly increased gestational glycaemia in Asian women in our in-silico modelling.This suggests that genetic or lifestyle factors may influence myo-inositol action or myo-inositol-micronutrient interactions in glycaemia regulation.Indeed, gestational hyperglycaemia in Asians, particularly East Asians (the predominant Asian ethnic group in the NiPPeR trial), is more heavily driven by pancreatic beta-cell/ insulin insufficiency than peripheral insulin resistance. 36As we postulate that myo-inositol limits early pregnancy beta-cell expansion, Asian women could therefore be disproportionately affected by a higher early pregnancy plasma myo-inositol concentration (Figure 3).
A strength of our study is the robust measurement of the myoinositol isomer in plasma, which confirmed that higher concentrations were achieved in early and late gestation with overall good adherence to the NiPPeR supplement. 15Blood sample collection and processing were strictly standardized and batch-analysed (except glucose) in accredited laboratories, minimizing technical variabilities and imprecision.However, only free myo-inositol was quantified, while conjugated myo-inositol, including the inositol-phosphoglycans, which are insulin-mimetics, were not measured.Without a full representation of myo-inositol and its derivatives in the circulation, interpretation is limited.In addition, with the lack of myo-inositol measurements between 7 and 28 weeks' gestation, we could not determine more precisely the optimal gestational timing for the commencement of myo-inositol supplementation to achieve good glycaemia outcomes.
Gold-standard hyperinsulinaemic-euglycaemic glucose clamp studies would be needed to confirm our postulation of a myo-inositol-induced impairment in insulin secretory capacity, but this has not been done.
While the NiPPeR supplement containing myo-inositol yielded some benefits, including lower risks of preterm birth and postpartum haemorrhage as secondary outcomes, 15  Early pregnancy (7 weeks) myo-inositol may limit physiological beta-cell expansion in early pregnancy (reducing beta-cell hyperplasia alongside inducing a more efficient beta-cell response that reduces the stimulus for physiological beta-cell expansion) leaving a lasting defect, which compromises pancreatic response to a glucose challenge later in pregnancy.However, in women with metabolic syndrome (red lines) where beta-cell capacity is already near maximal preconception, there is limited potential for further beta-cell expansion early in pregnancy anyway, hence a muted impact of early pregnancy myo-inositol.Late pregnancy (28-week) myo-inositol promotes the beta-cell response (i.e.insulin secretion) to a glucose load as well as increases peripheral insulin sensitivity to promote good glycaemia regulation.In Asian women (blue lines) where beta-cell insufficiency is a greater contributor to poor glycaemia regulation, 36 early pregnancy myo-inositol-induced suppression of beta-cell expansion would have a disproportionately greater adverse effect on glycaemia regulation than in White women (black lines) where peripheral insulin resistance is the more predominant contributor to impaired glycaemia control.
Urinary myoinositol was corrected for dilution using urinary creatinine.Plasma concentrations of folate and vitamin B12 (cobalamin) were measured by microbiological assay, and vitamin D (25(OH)D) by LC-MS/MS (Bevital).Umbilical cord venous serum samples collected at delivery were batch-analysed for C-peptide concentrations (electrochemiluminescence immunoassay; Roche Cobas).
tions and the three insulin parameters of Stumvoll first phase index (approximate measure of insulin secretion in response to a glucose load), HOMA2-IR (insulin resistance) and Matsuda index (insulin sensitivity) at 28 weeks.The adjusted association between 7-week myoinositol and gestational diabetes was assessed by Poisson regression, while non-standardized birthweight and cord C-peptide (indicator of foetal insulin response to transplacental glucose transfer) linear regressions were additionally adjusted for gestational age and sex.

F I G U R E 1
Associations between plasma myo-inositol in (A) early (7 weeks) or (B,C) late (28 weeks) pregnancy with gestational glycaemia at 28 weeks.Gestational glycaemia (fasting, and 1 h and 2 h) was assessed by a three time point 75 g oral glucose tolerance test at 28 weeks.Plasma myo-inositol (μmol/L) and glycaemia (mmol/L) were log e transformed and adjusted for study site, Asian ethnicity, preconception body mass index, parity, maternal age, household income level, family history of diabetes, and smoking during pregnancy in linear regression models.(C) In addition, adjusted for the predicted effect of early pregnancy plasma myo-inositol on glycaemia.β coefficient represents the percentage change in log e -transformed glycaemia relative to each percentage change in log e -transformed early pregnancy plasma myo-inositol.Pearson correlation between plasma myo-inositol in early pregnancy (7 weeks, log e transformed) and adjusted (for above covariates) (D) 1 h and (E) 2 h gestational glycaemia at 28 weeks.The overlap in plasma myo-inositol concentrations in the control and intervention (taking supplement containing myo-inositol) groups are indicated.R 2 provides an estimate of how much of the variance in glycaemia is explained by plasma myoinositol after accounting for the listed covariates.N (number) indicates those with available data.CI, confidence interval.*p < .05,***p = .001.F I G U R E 2 Associations between early pregnancy plasma myo-inositol and gestational glycaemia at 28 weeks in subgroups or with additional adjustments.Stratified by preconception risk factors: (A) ethnicity (non-Asian, Asian); (B) FHx 37 (no FHx, with FHx); (C) metabolic risk defined by the International Diabetes Federation (IDF) (low, no risk factor; moderate, at least one risk factor but did not fulfil criteria for MetS; high, fulfils criteria for MetS (central obesity with two other risk factors).The five IDF factors are: central obesity defined as a waist circumference ≥88 cm for non-Asian, ≥80 cm for South and East Asian women, hyperglycaemia defined as fasting plasma glucose ≥100 mg/dl (5.6 mmol/L), hypertriglyceridemia defined as triglycerides ≥150 mg/dl, high-density lipoprotein cholesterol <50 mg/dl; hypertension with systolic blood pressure >130 mmHg or diastolic blood pressure >85 mmHg, or on anti-hypertensive treatment; see TableS1.Additional adjustment for potential confounders: (D) other components of the NiPPeR intervention known to influence gestational glycaemia (folate, and vitamins B12 and D); (E) inherent variations in inositol processing (pre-intervention plasma scyllo-inositol/myo-inositol ratio and urine/plasma myo-inositol ratio).Plasma myo-inositol and gestational glycaemia (fasting, and 1 h and 2 h in a 75 g oral glucose tolerance test) were log e -transformed and analysed by linear regression adjusted for study site, Asian ethnicity, preconception body mass index, parity, maternal age, household income level, FHx and smoking during pregnancy.β coefficient represents the percentage change in log e -transformed glycaemia (mmol/L) relative to each percentage change in log e -transformed early pregnancy plasma myo-inositol (μmol/L).*p < .05,**p < .01,***p = .001.N indicates number of women with available data.CI, confidence interval; FHx, family history of type 2 diabetes mellitus; MetS, metabolic syndrome.
our present study indicates that periconception myo-inositol supplementation, which increases early pregnancy plasma myo-inositol concentration, may slightly increase later postprandial glycaemia.Further research is required to replicate these findings, identify mechanisms and investigate the potential long-term implications for the offspring.Future studies should investigate the optimal timing for starting myo-inositol supplementation aimed at regulating gestational glycaemia, and evaluate the benefit-risk ratio of prenatal myo-inositol supplementation.AUTHOR CONTRIBUTIONS SC, WC and KG conceptualized and designed the study.SC, JW, SB, HN, SE, TK, LL, JMRN, JG, ISZ, KG and WC contributed to data collection and assimilation.SC, HZ, JW, HFC, LWC, SB, KG and WC contributed to statistical analysis.SC and HZ led the writing of the article.All authors contributed to interpretation of the data, critical revision of F I G U R E 3 Schematic diagram of the postulated underlying mechanism for the role of early and late pregnancy plasma myo-inositol in gestational glycaemia regulation.

1
Characteristics of participants who provided inositol and gestational glycaemia data.
Plasma myo-inositol and gestational glycaemia (fasting, 1 h, 2 h PG) were log e -transformed and analysed by linear regression adjusted for site, preconception body mass index, parity, family history of type 2 diabetes mellitus, maternal age, household income level, smoking during pregnancy and the predicted effect of early pregnancy plasma myo-inositol on glycaemia.Monte Carlo simulation with bootstrapping (100 iterations; i.e. resampling with replacement for 100 times with each time ran with a selection of 90% of the population).If the confidence interval does not cross 0, it is regarded as statistical significance of p < .05. β coefficients represent the percentage change in log e -transformed glycaemia (mmol/L) relative to each percentage change in log e -transformed late pregnancy plasma myo-inositol (μmol/L).