Placental IGFBP1 levels during early pregnancy and the risk of insulin resistance and gestational diabetes

Gestational diabetes mellitus (GDM) affects roughly one in seven pregnancies globally and is linked to adverse maternal and offspring outcomes during pregnancy, at delivery, and later in life. A central contributor to GDM is reduced insulin sensitivity (increased insulin resistance), and among individuals with GDM, those with the lowest insulin sensitivity have the greatest risk of hyperglycemia-related complications. The placenta is thought to drive profound changes in maternal insulin physiology, including the normal late-pregnancy decline in insulin sensitivity, potentially via placental hormones; however, human studies have shown poor correlations between classic placental hormones and insulin sensitivity. This suggests that other placental factors may mediate changes in insulin sensitivity and help maintain euglycemia in pregnancy. The study’s purpose was to discover placental factors associated with insulin sensitivity during pregnancy and clarify their role in GDM pathophysiology using an unbiased placental transcriptomic approach and assessments of circulating protein levels across multiple pregnancy cohorts.

Prior research has established insulin resistance as a key contributor to GDM and its complications. Classic placental hormones (e.g., human chorionic gonadotropin, human placental lactogen, placental growth hormone) correlate poorly with insulin sensitivity in human pregnancy. IGFBP1, primarily a hepatic product outside pregnancy and highly expressed in placenta/decidua, modulates IGF-1 and IGF-2 action. In nonpregnant populations, lower IGFBP1 correlates with insulin resistance and predicts incident type 2 diabetes. IGF axis biology suggests complex context-dependent effects on glucose metabolism, with IGF-1 improving insulin sensitivity and IGF-2 regulating fetal growth. Previous reports on circulating IGFBP1 as a predictor of GDM have been inconsistent, with limited data from early pregnancy. These backgrounds motivated an agnostic search in placenta for transcripts linked to insulin sensitivity and subsequent evaluation of circulating IGFBP1 in relation to insulin sensitivity, GDM risk, and birth outcomes.

Design and cohorts: The study integrated (1) a genome-wide placental RNA-seq analysis in the Gen3G cohort and (2) circulating IGFBP1 measurements and metabolic phenotyping across three cohorts: Gen3G (Canada, population-based), SPRING (USA, high-risk longitudinal cohort), and MOMS (USA, nested case-control within a prenatal cohort). Participants with preexisting diabetes were excluded. Gen3G recruited 1,024 pregnant women (2010–2013), with placenta samples collected at delivery and phenotyping at ~9 weeks (V1) and ~26 weeks (V2, 75 g OGTT). For RNA-seq, 466 placentas were sequenced; after QC and phenotype availability, 434 were analyzed. Circulating IGFBP1 was measured in 837 Gen3G participants (V1 non-fasting, largely 1h post-50 g GCT; V2 fasting), 165 SPRING participants (OGTT at <15 and 24–32 weeks; 6–24 weeks postpartum), and a MOMS nested case-control set (55 GDM cases matched 2:1 to controls; fasting samples at 16–20 weeks). Placental RNA-seq: Placental tissue (~1 cm3) from the maternal-facing side including decidua was collected within 30 minutes of delivery, stabilized in RNAlater, and stored at −80°C. RNA was extracted and quality assessed; libraries were prepared (Illumina TruSeq Stranded mRNA) and sequenced on HiSeq 4000 (101-bp paired-end; mean ~113M reads/sample). Reads were processed following GTEx v8-like pipeline: alignment with STAR v2.5.3a to GRCh38; duplicates marked (Picard); quantification with RNA-SeQC v2.3.6 (GENCODE v26). QC excluded samples with >1% outlier genes, retaining 459; 434 had complete phenotype/covariates. Low-abundance genes were filtered (≥6 counts, TPM >0.5 in ≥20% samples, mappability ≥0.8), leaving 15,202 genes. Between-sample normalization used edgeR; voom (limma) produced log2 CPM values. Differential expression analysis: Multivariable linear models associated placental gene expression with insulin sensitivity (log2 Matsuda) at ~26 weeks, adjusting for maternal age, gravidity, first-trimester BMI, fetal sex, gestational age at delivery, and 37 surrogate variables (SmartSVA) to control technical/cell-type variability. Genes with P < 1×10−3 were reported. Circulating IGFBP1 assays: Plasma IGFBP1 (free) was measured in one laboratory using R&D Systems ELISA (DGB100). Assay precision: intra-assay CV 5.6%, inter-assay CV 9.5%. Measurements were blinded. Metabolic phenotyping: In Gen3G and SPRING, OGTT-derived indices were computed: Matsuda index for insulin sensitivity (validated in pregnancy). In MOMS, fasting samples enabled HOMA2-S calculation. Glucose and insulin were measured by standard clinical and research assays; values were transformed as needed for normality. Statistical analyses: IGFBP1 distributions were Box–Cox transformed in Gen3G; SPRING and MOMS used raw values (approximately normal). Pearson and partial correlations (adjusting for maternal age, gestational age at blood draw, and BMI) were used to relate IGFBP1 to insulin sensitivity indices across cohorts and to metabolic traits/birth anthropometry in Gen3G. Logistic regression estimated odds ratios for GDM per SD increase in early pregnancy IGFBP1 in Gen3G and SPRING; conditional logistic regression was used in MOMS (matched case–control). Models sequentially adjusted for maternal age, gravidity, gestational age at sampling, and BMI. ROC analyses in Gen3G compared prediction of GDM using clinical factors with and without first-trimester IGFBP1 (DeLong’s test). Longitudinal IGFBP1 trajectories across GDM physiologic subtypes (insulin-resistant, insulin-deficient, mixed) and normoglycemia were visualized using generalized additive models.

• Placental transcriptomics (Gen3G, n=434): 14 genes were associated with insulin sensitivity (log2 Matsuda) at P < 1×10−3; the strongest was IGFBP1 (β = 0.43; P = 2.5×10−7), with higher placental IGFBP1 expression linked to greater insulin sensitivity. No strong associations were observed for classic placental hormones or other IGF-pathway genes.
• Circulating IGFBP1 across pregnancy/postpartum: In SPRING normoglycemic participants (n=65), median plasma IGFBP1 increased from 66,610 pg/mL (first trimester) to 79,379 pg/mL (24–32 weeks) and declined postpartum to 16,588 pg/mL (P < 0.001 for pregnancy vs postpartum), supporting a placental/decidual source. In Gen3G (n=27) OGTT substudy, IGFBP1 was stable from fasting to 1 h (P=0.13) and decreased at 2 h (P=0.0007); the insulin rise (0–60 min) inversely related to IGFBP1 at 1 h (r = −0.39; P = 0.047).
• IGFBP1 and insulin sensitivity (cross-sectional): Strong positive correlations across cohorts and timepoints, robust to adjustment for age and gestational age, and attenuated but significant after BMI adjustment: • Gen3G: r ≈ 0.50 at 24–30 wks (n=816) and 7–15 wks (n=156); BMI-adjusted r ≈ 0.35 (all P < 0.001). • SPRING: r = 0.55 at 24–32 wks (n=119) and r = 0.57 postpartum (n=107); BMI-adjusted r = 0.34–0.48 (P < 0.001). • MOMS (16–20 wks): r = 0.60; BMI-adjusted r = 0.45 (P < 0.001) with HOMA-IS.
• Associations with metabolic traits and birth outcomes (Gen3G): Higher maternal BMI correlated with lower IGFBP1 (first trimester r = −0.27; late second trimester r = −0.54; both P < 0.001). Second-trimester IGFBP1 was inversely correlated with OGTT glucose (r ≈ −0.28 to −0.30) and insulin (r = −0.40), and with HOMA-IR (r = −0.43) and positively with HOMA2-S (r = 0.42) and Matsuda (r = 0.50) (all P < 0.001). Lower IGFBP1 at both time points was associated with higher birthweight z-score (r = −0.15 and −0.21; P < 0.001) and higher LGA risk (second trimester OR = 0.60; 95% CI 0.46–0.78; P = 0.0001), persisting after BMI adjustment (OR = 0.73; 95% CI 0.54–0.99; P = 0.045).
• Prediction of incident GDM (early pregnancy IGFBP1): In Gen3G (n=837; 8.4% GDM), first-trimester IGFBP1 alone yielded ROC AUC 0.64. Clinical factors alone (age, gravidity, family history, BMI, gestational age at sampling) AUC = 0.66. Adding IGFBP1 improved AUC to 0.72 (ΔAUC = 0.06; 95% CI −0.104 to −0.015; P = 0.008). Per 1 SD higher IGFBP1, adjusted OR for GDM = 0.44 (95% CI 0.30–0.64; P < 0.001). Replication: MOMS OR = 0.40 (95% CI 0.24–0.67; P < 0.001). SPRING: OR = 0.75 (95% CI 0.46–1.25; P = 0.28).
• GDM physiologic subtypes (Gen3G): All GDM subtypes had lower mean IGFBP1 in early pregnancy than normoglycemic participants. The insulin-resistant GDM group showed a blunted increase in IGFBP1 from first to second trimester; insulin-deficient GDM reached levels similar to normoglycemia by second trimester; mixed defect was intermediate. First-trimester IGFBP1 was inversely associated with both insulin-resistant and insulin-deficient GDM in adjusted models. Second-trimester IGFBP1 was associated with insulin-resistant GDM only (OR = 0.28; 95% CI 0.16–0.47; P < 0.001), not with insulin-deficient GDM.
• Placental vs circulating relationship: Circulating IGFBP1 levels correlated positively with placental IGFBP1 expression (e.g., Gen3G V1 r = 0.15, P = 0.002; V2 r = 0.14, P = 0.005, adjusted for gestational age), aligning with a placental/decidual contribution.

The study addresses the question of which placental factors relate to maternal insulin sensitivity and GDM risk. Unbiased placental RNA-seq identified IGFBP1 as the transcript most strongly associated with insulin sensitivity, supporting IGFBP1’s role in pregnancy glucose physiology. Convergent evidence across three independent cohorts shows that higher circulating IGFBP1 relates to greater insulin sensitivity during pregnancy and postpartum. The pregnancy rise and postpartum fall in IGFBP1, correlation with placental expression, and acute decrease after glucose-induced insulin rise suggest a placental/decidual source and insulin-mediated regulation akin to hepatocyte biology. Low early-pregnancy IGFBP1 predicted subsequent GDM beyond established risk factors, with stronger associations in population-based cohorts than in high-risk cohorts. Distinct IGFBP1 trajectories across GDM physiologic subtypes—especially attenuated increases in insulin-resistant GDM—implicate inadequate placental/decidual IGFBP1 upregulation as a potential contributor to excessive insulin resistance and hyperglycemia in this subtype. These findings provide a plausible compensatory mechanism in normal pregnancy and a precision-medicine angle for GDM classification and potential intervention targets.

Starting from unbiased placental transcriptomics, the study implicates IGFBP1 in regulating maternal insulin sensitivity during pregnancy. IGFBP1 is highly expressed in placenta/decidua, circulating levels increase through gestation and drop postpartum, and both placental and circulating levels correlate with insulin sensitivity. Low early-pregnancy IGFBP1 predicts later GDM independent of clinical risk factors, and insulin-resistant GDM shows an attenuated rise in IGFBP1 across pregnancy. These results suggest that insufficient placental/decidual IGFBP1 may contribute to GDM—particularly the insulin-resistant subtype—and that IGFBP1 could serve as part of a biomarker panel for early risk assessment. Future research should test whether IGFBP1 directly modulates maternal insulin-sensitive tissues in pregnancy, define mechanisms (IGF-dependent and -independent), identify cellular sources and regulators within placenta/decidua, evaluate urinary IGFBP1 as a practical biomarker, assess preconception IGFBP1 levels, and explore IGFBP1 augmentation as a precision prevention or therapeutic strategy for insulin-resistant GDM.

• Observational design precludes causal inference regarding IGFBP1’s role in insulin sensitivity and GDM pathogenesis.
• Placental RNA-seq used bulk tissue including placental and decidual cells; exact cellular sources of IGFBP1 cannot be resolved.
• Although overall sample sizes were large, the number of GDM cases and especially physiologic subtypes was modest, limiting precision for subtype analyses.
• Some classic placental hormones (e.g., CSH1, GH2) had gene mappability scores <0.8, warranting cautious interpretation of their null associations.
• Generalizability may be influenced by cohort composition (e.g., Gen3G largely White, population-based; SPRING high-risk).