Monogenic disease analysis establishes that fetal insulin accounts for half of human fetal growth

To the Editor: Extremes in birth weight are associated with adverse pregnancy outcomes and long-term risk of cardiometabolic disease. Fetal insulin has long been recognized as an important regulator of fetal growth, but the overall contribution of fetal insulin to birth weight in humans has not been quantified. Single-gene mutations resulting in absent fetal insulin secretion provide a unique opportunity to study the effects of fetal insulin on birth weight in humans. We sought to quantify the role of fetal insulin in fetal growth by studying birth weights in individuals without fetal insulin, either due to recessive loss-of-function mutations in the INS gene or pancreatic agenesis (Supplemental Methods and Supplemental Tables 1 and 2 for clinical details and genetics, respectively; supplemental material available online with this article; https://doi.org/10.1172/JCI165402DS1). We also investigated whether reduced insulin-mediated fetal growth affected postnatal growth once insulin was replaced. The study was approved by the Wales Research Ethics Committee (17/WA/0327). P values of less than 0.05 were considered statistically significant, and specific statistical tests are detailed in Figure 1. In the absence of fetal insulin birth weight is halved in humans. There was a substantial, global reduction in fetal growth in the absence of fetal insulin (Figure 1A and Supplemental Tables 1 and 3). Mean birth weight adjusted to 40 weeks’ gestation was 51% of normal […]


Identification of individuals with absent fetal insulin secretion
Individuals with recessive mutations in the INS gene (detected by targeted Sanger sequencing) or a mutation in a gene known to cause pancreatic agenesis (detected by targeted next generation sequencing as previously described (1)) were identified from an international, multi-ethnic cohort referred to the Exeter Genomics Laboratory for genetic diagnostic testing of neonatal diabetes.

Collection of clinical data
Clinical details were provided by referring clinicians. For the postnatal growth follow-up data, we collected serial measures of weight and length/height until at least 4 years of age (where possible). We also requested a most recent weight and height measurement, HbA1c and daily insulin dose. For two individuals, we used measurements at referral (age 11 and 22 years) as the most recently available measurement. All data was routinely collected in a clinical practice setting.

Standardization of anthropometric measurements
Birth weight (n=64) and length (n=17) were standardized for sex and gestational age at birth in weeks and days using the INTERGROWTH-21 st standards (3) and studied as standard deviation scores (SDS). Postnatal weight and length/height were standardized for sex and age at measurement (in months and weeks for the first 12 months, then months thereafter) using the WHO Child Growth Standards (4) with correction for gestational age at birth until the age of four years. There were 10 individuals with serial weight measurements and 9 individuals with serial length/height measurements (7 of whom had a corresponding birth length) that could be combined for analysis. We approximated the measurements within each individual to exact three-monthly windows for the first year of life (3, 6, 9 and 12 months) and six-monthly intervals for the next three years of life (18,24,30,36, 42 and 48 months) using a linear interpolation. Most recently available weight (n=16) and height (n=15) were standardized for sex and age of measurement using the UK-WHO/British 1990 Growth Reference (5) since it is integrated with the aforementioned WHO Child Growth Standards and provides standardization for age and sex up until 18 years of age. Most recently available weights and heights in adults were standardized to an age of 18 years.

Statistics
Data were summarized as n (%) for categorical data (sex, ethnic ancestry, parental consanguinity, congenital anomalies), medians and interquartile range (IQR) for non-normally distributed data (time to diagnosis of neonatal diabetes, gestational age at birth, birth length, weight SDS and length/height SDS) and mean and 95% confidence interval (CI) for normally distributed data (birth weight).
The relationship between birth weight and birth length without fetal insulin and gestational age were modelled using a univariable linear regression and prediction intervals and measures adjusted to 40 weeks' gestation were calculated from this model. P values <0.05 were considered statistically significant. All tests of statistical significance were two-tailed. All analyses were performed in R version 3.6.2 (R Foundation for Statistical Computing) or Stata version 16.0 (StataCorp, College Station, TX, U.S.A.). Figures were produced in R version 3.6.2 using the ggplot package (6).

Study approval
Written consent from participants (or their responsible guardians, where applicable) for use of their samples and clinical information for research was obtained. Samples and clinical information is stored securely in the Genetic Beta Cell Research Bank (https://www.diabetesgenes.org/current-research/genetic-beta-cell-research-bank/), approved by the Wales Research Ethics Committee (Reference 17/WA/0327).