Altered Glucose Metabolism in Mouse and Humans Conceived by IVF

Assisted reproduction technologies, mostly in vitro fertilization (IVF) and intracytoplasmic sperm injection, are increasingly used to treat infertility, with the number of children now in excess of 5 million worldwide (1). Of concern is emerging evidence to suggest that IVF children may be at an increased risk for the development of metabolic and cardiovascular diseases (2). In particular, studies (3–7) have reported that IVF children have increased fasting glucose levels, blood pressure, triglyceride levels, adiposity, and inflammatory biomarker levels, as well as systemic and pulmonary vascular dysfunction. To date, it has not been tested whether any differences in insulin sensitivity are apparent between these groups, which may underlie many of these differences. Moreover, it is unclear whether the increased numbers of risk factors observed in IVF children are related to dietary patterns, the underlying genetics of the parents, environmental factors, or the treatment procedures per se.

Animal models may provide evidence of biological plausibility and potential mechanisms. Impaired glucose tolerance, increased systolic blood pressure and body fat, altered fatty acid composition in liver and adipose tissue, endothelial dysfunction and increased stiffness, and shorter life span have all been reported in IVF-conceived mouse offspring (8–12). These studies suggest that the procedure of IVF itself may increase metabolic risk. However, these studies have rarely controlled for litter size or the maternal environment between groups, which may also impact outcomes. It is also unclear whether any differences in outcomes are due to the process of ovarian stimulation (OS) with high doses of gonadotropins prior to oocyte collection and/or the process of in vitro embryo culture.

The aim of this study was to examine insulin sensitivity and metabolic risk factors in young adults conceived by IVF or natural conception (NC) after consuming an energy-balanced diet (30% fat) and after 3 days of high-fat–overfeeding challenge (45% fat). In parallel, a study in C57BL/6J mice directly compared key metabolic factors in adult male offspring that were generated by IVF versus those that were naturally conceived. Importantly, mice conceived after OS alone were also examined, allowing us to separate out the effects of OS versus those of embryo culture.

Young adults conceived by IVF were recruited by advertising in local newspapers and on a university campus or from a database of IVF birth records in South Australia (Australia Company Number 008123466 Pty. Ltd.) and were matched by sex and BMI to NC individuals. Participants were excluded from the study if they were not of normal birth weight, reported any significant medical conditions, received any medications that may alter glucose or lipid metabolism (e.g., metformin), had first-degree relatives with type 2 diabetes or cardiovascular disease, or if they smoked or drank >140 g of alcohol per week. All parents were Caucasian, except for two sets of parents who were Chinese (one IVF male and one NC male participant).

Volunteers visited the clinical research facility for screening to determine eligibility, which included medical history and measurement of blood lipid and fasting glucose levels. Thirty-four individuals were recruited. Three IVF individuals were twins, and one NC individual was a twin. Body composition was measured by DEXA (Lunar DPX; Lunar Radiation, Madison, WI). Female participants were tested in the follicular phase of their menstrual cycle. Two female participants from each group were receiving oral contraception pills. Three female participants (two NC and one IVF participant) did not undergo repeat assessments after the overfeeding. The study protocol was approved by the Research Ethics Committee of Royal Adelaide Hospital. Written informed consent was obtained from all participants before commencement of the study.

Prior to metabolic testing at baseline, estimated energy requirements were calculated and individual menus were planned by a trained dietitian, as previously described (13). From day −3 to day 0, individuals were fed an energy-balanced diet (30% fat, 15% protein, and 55% carbohydrates). After baseline metabolic testing, individuals were switched to an overfeeding diet (+1,250 kcal/day) with a nutrient composition of 45% fat, 15% protein, and 40% carbohydrates for 3 days before metabolic assessments were repeated. Participants were provided with all foods and completed checklists reporting the foods consumed, which were reviewed at each metabolic testing visit. Energy intake was comparable between groups (Supplementary Table 1).

Participants attended the clinical research facility at 8:00 a.m. after a 12-h overnight fast. The procedures at the two visits were identical. Weight, height, and blood pressure were measured with the participant dressed in a hospital gown after voiding. The first phase of insulin secretion was assessed by an intravenous glucose tolerance test. After the fasting blood collection, a bolus dose of 25% glucose (0.3 g/kg with a maximum of 30 g) was injected into the antecubital vein within 1 min. Blood was collected at 1, 3, 4, 5, 6, 7, 8, and 10 min after glucose infusion. Insulin sensitivity was then measured using a 2-h hyperinsulinemic-euglycemic clamp (80 mU/m²/min), as described previously (14).

Glucose was analyzed using a glucose oxidase electrode (YSI Life Sciences). Serum insulin was assayed by radioimmunoassay (Millipore). Blood lipids were examined by photometric assays in the laboratory of SA Pathology (Adelaide, South Australia, Australia).

C57BL/6J mice were obtained at 6 weeks of age, and vasectomized (CBA × C57BL/6) F1 male mice were obtained at 8 weeks of age from the Animal Resource Centre (Perth, Western Australia, Australia). CBA × C57BL/6 F1 female mice were obtained at 6 weeks of age from the Laboratory Animal Services (Adelaide, South Australia, Australia). All mice were maintained on a 12-h light/dark cycle, with standard rodent chow diet (SF06–105; Specialty Feeds) and water available ad libitum. All mice were acclimatized for 2 weeks on chow before experimentation. All experiments were approved by the University of Adelaide Animal Ethics Committee and were conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes.

Mouse blastocysts were generated by NC (NC group) or by hormonal OS followed by either mating (OS group) or by IVF and embryo culture (IVF group). To generate pups for the NC group, female C57BL/6J mice were placed with male C57BL/6J mice overnight. For the OS group, female C57BL/6J mice were superovulated with consecutive injections of 7.5 IU of equine chorionic gonadotropin (Calbiochem) and 7.5 IU of human chorionic gonadotropin (hCG; Calbiochem), which were administered intraperitoneally 48 h apart. After injection with hCG, these female mice were placed with male C57BL/6J mice overnight. The following morning, female mice (NC and OS groups) with the presence of vaginal plugs were considered to be pregnant, and 3 days later were humanely killed by cervical dislocation. Blastocysts were collected by flushing dissected uteri with prewarmed HEPES-buffered minimal essential media (Invitrogen Australia Pty. Ltd.) supplemented with 5 mg/mL human serum albumin (ART-3001; SAGE Media). Blastocysts from the NC and OS groups were placed in Research Cleave media (Cook Medical, Brisbane, Queensland, Australia) for no more than 1 h prior to being transferred to the uteri of pseudopregnant mice (see below).

For the IVF group, the OS protocol was the same as that for the OS group. IVFs were performed using a modified version of the procedure described previously (15). At 13 h after injection with hCG, female C57BL/6J mice were humanely killed by cervical dislocation. Cumulus-oocyte complexes from the oviducts were placed in Research Fertilization media (Cook Medical) under paraffin oil (Merck Pty. Ltd.) and were incubated in a modular incubator chamber at 37°C in 6% CO₂, 5% O₂, 89% N₂ for 5–6 h with sperm collected from the cauda epididymis of male C57BL/6J mice that had been previously incubated for 1 h in Research Fertilization media for sperm capacitation. The putative zygotes were then placed in Research Cleave media and incubated in the modular incubator chamber a further 3 days to the blastocyst stage.

Unstimulated (CBA × C57BL/6) F1 female mice were mated with vasectomized males, and those with copulatory plugs the next morning were considered as being at day 0.5 of pseudopregnancy. At day 2.5 of pseudopregnancy, 7–10 blastocysts were transferred to the uteri of each pseudopregnant recipient mouse (seven to eight litters per group) anesthetized by intraperitoneal injection of Avertin (0.5 mg/g; Sigma-Aldrich). The analgesia agent carprofen (5 mg/kg; Rimadyl, Pfizer) was injected subcutaneously once after the surgery. All recipient females were fed chow diet (SF06-105; Specialty Feeds).

Pups were born on day 19.5 of pregnancy, were weighed weekly, and were weaned at 3 weeks of age onto a chow diet or a high-fat diet (HFD) for 8 weeks. Feed for the HFD was made in-house using the same recipe as that for the D12492 diet formula made by Research Diets (New Brunswick, NJ) (nutrient composition 60% fat, 20% protein, and 20% carbohydrate). Only male offspring were used for this study.

At 11 weeks of age, mice were fasted for 6 h and challenged with either an intraperitoneal injection of glucose (2 g/kg) or insulin (0.75 units/kg). Blood samples were obtained from the tail tip for assessment of glucose at 0, 15, 30, 60, and 120 min with a glucometer (Accu-Chek Performa; Roche Diagnostics) and insulin at 0, 30, 60, and 120 min by ultrasensitive ELISA (Merck Millipore). One week later, mice were either killed by cervical dislocation, and quadriceps, inguinal fat, epididymal fat, and liver were immediately excised, weighed, and snap frozen or an insulin stimulation test was performed (n = 4/group). For this test, mice were fasted for 6 h and, while under anesthesia (pentobarbital, 60 mg/kg i.p.; Sigma-Aldrich), samples of liver and one quadriceps muscle were collected and snap frozen. Insulin (1 unit/kg) was then injected into the inferior cava vein, and exactly 3 min later mice were killed by cervical dislocation and additional samples of liver and the contralateral quadriceps muscle were rapidly excised and snap frozen.

Liver and quadriceps tissues were lysed and protein concentration was determined using a Pierce BCA Kit (Thermo Scientific). Lysates (20 μg of protein) were resolved by SDS-PAGE and transferred onto polyvinylidene fluoride membranes. Membranes were probed for Akt (Cell Signaling Technology) and Phospho-Ser⁴⁷³ Akt (Cell Signaling Technology). All blots were applied with ECF substrate (Amersham Biosciences) and scanned for fluorescence by the Typhoon Trio+ (Amersham Biosciences) following the manufacturer instructions. The band intensity was measured using ImageJ software (National Institutes of Health, Bethesda, MD).

Total RNA was extracted from liver using Trizol (Invitrogen). cDNA was synthesized using the QuantiTect reverse transcription kit (Qiagen). Quantitative real-time PCR was performed as described previously (16) using the TaqMan primers and probes listed in Supplementary Table 1. The NormFinder program was used as described previously (17), and hypoxanthine phosphoribosyl transferase and cyclophilin A in seven potential reference genes were identified as the best combination. Data were analyzed using the 2^−ΔΔCt method.

Data are shown as the mean ± SEM, unless otherwise stated. Data were analyzed statistically with SPSS version 20 (SPSS, Chicago, IL). For the human data comparisons, baseline differences between groups were analyzed by Student t test, and response to diet intervention was assessed using repeated-measures ANOVA with Bonferroni post hoc analysis and an intention-to-treat approach without carrying forward data on the three dropouts with sex and group in the model. Sexes were combined for analysis since no differences in response were detected by sex. For animal studies, single comparisons were performed with Student t test or two-way ANOVA, whereas time courses were analyzed by repeated-measures ANOVA and Bonferroni post hoc analysis. When required, nonparametric tests (Mann-Whitney U test or Kruskal-Wallis test) were used as indicated. Differences were considered statistically significant at P < 0.05.

There was no significant difference in the z scores of birth weight adjusted for gestational age, maternal age, and sex (−0.11 ± 0.87) for the IVF group and the NC group (0.08 ± 1.1; P = 0.23) or between groups in terms of length of gestation or parental characteristics during pregnancy, including smoking, alcohol consumption, and health (data not shown), except that mothers undergoing IVF treatment were significantly older than those conceived naturally (34 ± 1 vs. 28 ± 1 years, P = 0.001).

There were no significant differences between groups at baseline with respect to age, weight, BMI, percent of fat mass, blood pressure, or cholesterol levels (Table 1). Fasting glucose levels, fasting insulin levels, and HOMA of insulin resistance (IR) were also not different among groups at baseline, but peripheral insulin sensitivity as measured by the hyperinsulinemic-euglycemic clamp was significantly lower in IVF versus NC individuals. First-phase secretion of insulin (shown by the insulin area under the curve [AUC] at 10 min) and glucose change (glucose AUC at 10 min) in response to the intravenous glucose tolerance test was not different between groups.

Body weight gain in response to 3 days of overfeeding was not significantly different between groups (NC 0.6 ± 0.2 vs. IVF 0.7 ± 0.2 kg). As we have previously observed, fasting glucose and insulin levels, and thus HOMA-IR, were increased significantly in response to 3 days of overfeeding in both groups (fasting glucose level: NC 0.06 ± 0.04 vs. IVF 0.1 ± 0.05 mmol/L; insulin level: NC 0.4 ± 0.6 vs. IVF 2 ± 0.6 μU/mL; and HOMA-IR: NC 0.1 ± 0.1 vs. IVF 0.4 ± 0.1 arbitrary units; diet effect P < 0.01). The increase in insulin or HOMA-IR did not reach statistical significance between IVF and NC individuals (both P = 0.1). However, there was a greater increase in systolic blood pressure in IVF versus NC individuals in response to overfeeding (NC 107 ± 2 vs. IVF 115 ± 3 mmHg; diet × group interaction P = 0.04; post hoc test P = 0.02). No other parameters described at baseline were altered after overfeeding, and no other group differences were detected (data not shown).

Litter size was not different among groups (NC 6.3 ± 0.6, OS 7.4 ± 0.7, IVF 6.9 ± 0.5, P = 0.5). The birth weights of IVF pups (1.51 ± 0.03, n = 26) and OS pups (1.51 ± 0.02, n = 28) were significantly lower than those of NC pups (1.68 ± 0.03, n = 21), and this difference was maintained until 3 weeks of age (Fig. 1). After weaning at 3 weeks of age, body weight remained significantly lower in OS and IVF mice on both diets. As expected, body weight and weight gain were increased by HFD. However, body weight gain, either before or after weaning, was not different between groups.

HFD decreased relative liver weight and increased relative adipose tissue weight, but there were no differences between IVF and control mice on either diet (Fig. 2). However, the relative weights of inguinal and epididymal fat pads were significantly lower in OS mice fed an HFD.

As expected, fasting glucose and insulin levels were increased by HFD (Fig. 3). Interestingly, IVF mice displayed higher fasting glucose levels compared with NC and OS mice, independent of diet. OS and IVF mice had lower fasting insulin levels than NC mice, only after HFD. In response to glucose challenge, IVF mice on both diets had impaired glucose tolerance, as evidenced by increased glucose AUC compared with NC mice, but the insulin response to glucose was not different among groups for chow-fed mice (Fig. 3). Insulin data were not available for mice on an HFD, but peripheral insulin sensitivity, as assessed by intraperitoneal insulin tolerance (data not shown) and by insulin-stimulated Akt-Ser⁴⁷³ phosphorylation in muscle, was not different among groups that were fed either chow or HFD. However, reduced Akt-Ser⁴⁷³ phosphorylation was observed in livers of IVF mice that were fed chow and HFD (Fig. 4).

Hepatic expression of gluconeogenesis gene G6pc and mitochondrial biogenesis markers Cpt1a, Pgc1α, and Tfam were not different between groups or diets (Fig. 5), and protein levels of Pgc1α and total OXPHOS were also not altered (data not shown). HFD increased the expression of glucokinase and decreased the expression of gluconeogenic gene Pck1 in all groups (Fig. 5). Compared with NC mice, the expression of lipogenesis gene Srebf1 was increased in IVF mice that were fed chow or HFD.

A suboptimal maternal environment during pregnancy predisposes offspring to chronic diseases later in life (18,19), with the preimplantation period also emerging as a critical stage for development and later adult health (20,21). Accumulating evidence suggests that children conceived by IVF may have increased risk of the development of metabolic syndrome, type 2 diabetes, and cardiovascular disease (3–7). In this study, we observed that adult humans conceived by IVF, the majority of whom were of normal weight, were more insulin resistant than NC individuals matched for BMI, sex, and age. In the carefully controlled mouse study, we showed that IVF-conceived mice displayed hyperglycemia, impaired glucose tolerance, and hepatic IR at both normal and high body weight. Thus, our study supports the hypothesis that IVF alters glucose metabolism and increases the risk of the development of metabolic diseases later in life.

This is the first study to test peripheral insulin sensitivity in IVF-conceived young adults by using gold standard assessment with the hyperinsulinemic-euglycemic clamp. We observed reduced peripheral insulin sensitivity in IVF adults, although we did not note any significant differences in fasting glucose or insulin levels, which is similar to observations made in cohorts of young nondiabetic individuals with a strong family history of type 2 diabetes (13,22). In contrast, some but not all studies have reported increased fasting glucose and triglycerides levels, blood pressure, or peripheral body fat in IVF-conceived children or adolescents (3–5). One study (23) has noted more favorable lipid profiles in prepubertal IVF children with higher HDL levels and lower triglyceride levels. Discrepancies between studies may be due to differences in the ages investigated, sample size, sampling of the comparison group, dietary intake, and/or parental characteristics, such as gestational weight gain, and maternal and paternal BMI. In this study, maternal age was higher in the IVF cohort, which may alter oocyte quality (24) and may have contributed to the observed effects. Further study is also needed to determine whether reduced insulin sensitivity is related to impaired suppression of hepatic glucose production, and whether these differences will be evident in cohorts of overweight individuals.

Given the young age of the cohort under investigation, we also examined the metabolic consequences of 3 days of high-fat overfeeding challenge. Typically, overfeeding diets rapidly increase glucose and insulin levels within 3 days (13). Impaired suppression of hepatic glucose production, as assessed by the hyperinsulinemic-euglycemic clamp, is also observed within 5–11 days in young healthy men (25,26). As noted in this study, longer lengths of time may be necessary to induce peripheral IR (14,27). We also observed that 3 days of high-fat overfeeding increased fasting insulin levels and HOMA-IR, which is reflective of hepatic IR (28,29). Interestingly, we have previously identified greater increases in HOMA-IR in response to a similar overfeeding protocol in nondiabetic individuals who have a family history of type 2 diabetes (13), who have a 1.7- to 6.1-fold greater risk of the development of type 2 diabetes (30). However, this did not reach statistical significance between IVF and NC individuals in this study. The elevated response of systolic blood pressure in IVF-conceived adults was unmasked by a 3-day overfeeding challenge. Consistently, increased blood pressure and vascular stiffness, and endothelial dysfunction have been reported previously in IVF mouse models and children (4,5,7,9,11). This may be related to insulin, since a large number of clinical studies have confirmed that higher systolic blood pressure is related to increased fasting insulin levels and IR, independent of age, BMI, sex, and race (31–33), although we did not observe a correlation between the changes in insulin levels and blood pressure in this study. IVF mice are also more susceptible to an HFD (50% fat), as evidenced by a nearly 25% shorter life span compared with NC mice (11). Together, these data suggest that individuals conceived by IVF may be more sensitive to the deleterious consequences of obesogenic environments.

In a previous mouse study (10), IVF male offspring displayed IR but normal glucose response to intraperitoneal glucose administration, and female offspring displayed impaired glucose tolerance after being fed a chow diet at 8 weeks of age. However, the genetic background, maternal environment, and litter sizes were not controlled in that study. To separate out these potential confounders and the effects of OS versus embryo culture, we developed an IVF mouse model using inbred C57BL/6J mice. In this study, IVF mice demonstrated impaired glucose metabolism, as evidenced by higher fasting glucose levels, impaired glucose tolerance, and impaired hepatic insulin signaling at 12 weeks of age while being fed both a chow diet and an HFD. Importantly, mice conceived by OS alone did not exhibit any of these differences while being fed either diet compared with controls, indicating that it is the process of embryo culture rather than hormonal stimulation that contributes to impaired glucose metabolism in male offspring. In support of this, others have reported (9,34,35) that embryo culture alters blastocyst formation, fetal development, and postnatal phenotype, including increased anxiety, poor spatial memory, and elevated systolic blood pressure. Recent studies (7,11) have demonstrated that IVF also led to vascular dysfunction in mouse and humans. Strikingly, this was observed even in IVF mice offspring conceived by transferring two cell embryos after just 30 h of in vitro culture, and these differences were transmitted to the next generation (11).

IVF singletons have a higher risk of low birth weight and preterm birth than their non-IVF siblings (36,37). Similarly, we observed that birth weight was lower in mice conceived either by OS alone or by IVF. This is the reverse of other mouse studies of IVF (10), but in those cases the maternal environment was not identical and IVF litters were much smaller than NC litters, issues that were carefully controlled in our study. Our study suggests that OS impairs fetal growth, and this is supported by human studies (38–41) and may be related to reductions in oocyte quality and embryo quality (42,43). In our mouse model, all blastocysts were transferred into unstimulated surrogate recipients with a natural uterine environment, suggesting that any differences in fetal growth were from influencing oocyte and/or embryo development and were not due to changes in endometrium receptivity. We also observed that mice conceived by OS had smaller fat mass gain while being fed an HFD, which may explain the lower fasting insulin levels observed after being fed an HFD in this group. In rodent models, low birth weight is associated with increased risk of type 2 diabetes (44,45). However, we did not observe any metabolic defects in OS mice, who were also born small.

In the current study, IVF mice displayed elevated fasting glucose levels and impaired glucose tolerance, but no difference in the insulin response to glucose, and this, combined with lower fasting insulin levels while being fed and HFD, may be indicative of impaired β-cell function. Alternatively, impaired suppression of basal hepatic glucose production may be responsible both for differences in fasting glycemia and impaired glucose tolerance. This is supported by our findings of reduced Akt-Ser⁴⁷³ phosphorylation in the liver after an insulin stimulation test. Increased gene expression of Srebf1, which is a master regulator of lipogenesis, was also observed in livers obtained from IVF mice, irrespective of diet. Of note, the phenomenon of selective IR via the Akt pathway, but continued sensitivity via the Srebf1 pathway, has been reported in other mouse models of type 2 diabetes (46,47).

In conclusion, IVF-conceived human individuals were more insulin resistant and tended to be more susceptible to the metabolic consequences of high-fat overfeeding. Our data in mice are partially supportive of these findings and further delineate the effects of embryo culture versus OS. These suggest that it is the process of embryo culture itself rather than genetic and/or environmental differences that contribute to impaired glucose metabolism and that OS impaired fetal growth, at least in mice. This study suggests there may be an increased risk of the development of metabolic and cardiovascular disease in IVF offspring later in life.

Acknowledgments. The authors thank Kylie Lange, Centre of Research Excellence in Translating Nutritional Science to Good Health, The University of Adelaide, for her expert statistical advice; Anita Peura, Discipline of Obstetrics and Gynaecology, The University of Adelaide, and George Hatzinikolas, Discipline of Medicine, The University of Adelaide, for their technical assistance in the mouse model; Julie Owens, School of Paediatrics and Reproductive Health, The University of Adelaide, for her critical comments in the mouse study; and Scott Standfield and Judith Wishart, Discipline of Medicine, The University of Adelaide, for their help in the insulin assays.

Funding. This study was supported by the Channel 7 Children’s Research Foundation.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.C. recruited human subjects, performed the mouse experiments, acquired and analyzed the human and mouse data, wrote and critically reviewed the manuscript, and approved the final manuscript. L.W. contributed to setting up the mouse model, critically reviewed the manuscript, and approved the final manuscript. J.Z. contributed to recruiting human subjects, acquired the data of the human study, critically reviewed the manuscript, and approved the final manuscript. F.W. contributed to acquiring the data in the mouse study, critically reviewed the manuscript, and approved the final manuscript. M.J.D. contributed to recruiting human subjects, critically reviewed the manuscript, and approved the final manuscript. G.A.W. contributed to the interpretation of the data, helped to draft the manuscript, critically reviewed the manuscript, and approved the final manuscript. R.J.N. contributed to recruiting human subjects and interpreting the data, critically reviewed the manuscript, and approved the final manuscript. R.L.R. designed and supervised the mouse study, interpreted the data, critically reviewed and revised the manuscript, and approved the final manuscript. L.K.H. conceived, designed, and supervised the human and mouse studies, interpreted the data, critically reviewed and revised the manuscript, and approved the final manuscript. L.K.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 69th Annual Meeting of the American Society for Reproductive Medicine held conjoint with the 21st World Congress of the International Federation of Fertility Societies, Boston, MA, 12–17 October 2013.