New England Journal of Medicine MAY 08, 2008

Hyperglycemia and Adverse Pregnancy Outcome (HAPO) Study

The HAPO Study Cooperative Research Group (Boyd E. Metzger et al.)

Bottom Line

The HAPO study demonstrated that maternal glucose levels below those diagnostic of overt diabetes mellitus have strong, continuous, and independent associations with increased risks of adverse pregnancy and neonatal outcomes.

Key Findings

1. There was no identifiable threshold at which the risk of adverse outcomes significantly increased; rather, risks demonstrated a continuous and graded relationship across the spectrum of maternal glucose levels.

2. Adjusted odds ratios for birth weight above the 90th percentile were significantly increased for every 1-standard deviation increase in fasting, 1-hour, and 2-hour plasma glucose levels.

3. Fetal hyperinsulinemia, represented by cord serum C-peptide levels greater than the 90th percentile, showed strong positive associations with maternal glucose levels, supporting the Pedersen hypothesis.

4. Primary cesarean delivery and clinical neonatal hypoglycemia were also consistently associated with higher maternal glucose levels, although the association for hypoglycemia was less robust than for fetal anthropometric and insulin measures.

Study Design

Design

Prospective Observational Cohort Study

Double-Blind

Sample

23,316

Patients

Duration

6 wk

Median

Setting

Multicenter, 9 countries

Population Pregnant women who were not previously diagnosed with diabetes mellitus.

Intervention Not applicable (Observational study tracking glucose levels).

Comparator Not applicable (Observational study tracking glucose levels).

Outcome Composite of birth weight >90th percentile, primary cesarean delivery, clinical neonatal hypoglycemia, and cord serum C-peptide >90th percentile.

Study Limitations

• As an observational study, the findings establish association rather than causation, meaning the data cannot definitively prove that treating these specific lower levels of hyperglycemia reduces adverse outcomes.

• Clinical care was blinded to the study glucose results, which prevented any therapeutic intervention that might have modified the outcomes for women with higher glucose levels within the 'non-diabetic' range.

• The study focused on a 75-g OGTT at 24-32 weeks, which may not capture variations in glucose metabolism occurring at other stages of gestation.

• The generalizability of the results is influenced by the diverse international nature of the centers, though this also acts as a strength for global applicability.

Clinical Significance

This study served as the evidence base for the International Association of Diabetes in Pregnancy Study Groups (IADPSG) to revise diagnostic criteria for gestational diabetes mellitus (GDM). It established that lower glucose thresholds than previously used are clinically relevant, emphasizing the need for standardized screening to identify pregnancies at increased risk of adverse outcomes.

Historical Context

Before HAPO, the diagnostic criteria for GDM varied widely, often based on older standards intended to predict maternal risk of future Type 2 diabetes rather than perinatal outcomes. By confirming the continuous nature of risk between maternal glycemia and adverse outcomes, the study challenged the existing paradigm and addressed the long-standing debate over the diagnostic thresholds for gestational hyperglycemia.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

Based on the Pedersen hypothesis, what is the underlying physiological mechanism that explains the continuous relationship between maternal glucose levels and fetal macrosomia observed in the HAPO study?

Key Response

Maternal glucose crosses the placenta via facilitated diffusion, but maternal insulin does not. In response to maternal hyperglycemia, the fetal pancreas undergoes islet cell hyperplasia and secretes excess insulin. Since insulin is a potent growth factor in utero, this fetal hyperinsulinemia drives excessive fat deposition and somatic growth, leading to macrosomia even at glucose levels below the threshold for overt diabetes.

Resident

The HAPO study demonstrated a continuous risk of adverse outcomes with rising glucose levels. How did these findings challenge the traditional 'two-step' screening approach for Gestational Diabetes Mellitus (GDM)?

Key Response

Traditionally, GDM was diagnosed using thresholds (like Carpenter-Coustan) designed to predict the mother's future risk of type 2 diabetes. HAPO shifted the focus to immediate perinatal risks, showing that these risks (e.g., macrosomia and cord C-peptide levels) increase linearly with glucose. This led to the IADPSG 'one-step' 75g OGTT approach, which uses lower thresholds to identify and treat a larger population of women at risk for neonatal complications.

Fellow

HAPO utilized cord-blood serum C-peptide levels as a primary outcome. What is the clinical significance of this biomarker in reflecting the intrauterine environment compared to neonatal insulin levels?

Key Response

C-peptide is secreted in equimolar amounts with insulin but has a longer half-life and more stable clearance rates. Unlike insulin, C-peptide is not affected by maternal insulin antibodies or exogenous insulin administration. In HAPO, elevated cord C-peptide served as a direct surrogate for fetal hyperinsulinemia, confirming that even mild maternal hyperglycemia triggers a significant fetal metabolic response.

Attending

Considering the HAPO study's finding that there is no clear 'inflection point' for risk, how should we balance the increased medicalization of pregnancy against the absolute risk reduction when applying HAPO-derived diagnostic criteria in clinical practice?

Key Response

Adopting HAPO-based (IADPSG) criteria nearly triples the prevalence of GDM. While this identifies more 'at-risk' fetuses, many of these women have low absolute risk. Attending physicians must integrate this by emphasizing lifestyle modifications and avoiding 'iatrogenic' harm, such as unnecessary inductions or C-sections, which can occur simply due to the 'labeling' of a patient as having GDM.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

The HAPO study was strictly observational and excluded women with glucose levels exceeding predefined safety limits. How does this 'truncation' of the upper end of the glucose distribution affect the statistical power to define a definitive diagnostic threshold?

Key Response

By unblinding and treating women with the highest glucose levels for safety reasons, the study 'truncated' the distribution, potentially underestimating the strength of the association at higher levels. Because the relationship remained linear throughout the observable range without an 'elbow' or threshold, any diagnostic cutoff chosen (such as those by IADPSG) is a consensus-based policy decision rather than a mathematically derived biological breakpoint.

Journal Editor

As a peer reviewer, what is the most significant threat to the internal validity of the HAPO study's primary outcomes, specifically regarding the 'primary cesarean section' rate?

Key Response

The primary threat is 'unblinding' bias. Although glucose results were blinded to clinicians to prevent treatment bias, the clinicians were aware that the patients were part of a high-risk hyperglycemia study. If a clinician became unblinded (e.g., due to a high glucose result triggering a safety protocol) or if knowledge of the study's intent influenced the decision to perform a C-section for suspected macrosomia, it could artificially strengthen the association between glucose and surgical delivery.

Guideline Committee

ACOG and the NIH have historically resisted a full transition to HAPO-based IADPSG criteria, maintaining the two-step approach. Compare the evidence-based rationale for this 'high-threshold' approach versus the 'low-threshold' IADPSG approach.

Key Response

The IADPSG (based on HAPO) prioritizes sensitivity to identify all neonates at risk for metabolic over-stimulation. However, ACOG (referencing the NIH Consensus Development Conference) maintains the two-step Carpenter-Coustan or NDDG criteria because there is a lack of randomized controlled trial evidence showing that treating the 'mild' GDM cases identified only by IADPSG criteria actually improves long-term maternal or neonatal health outcomes enough to justify the increased healthcare costs and patient burden.

Clinical Landscape

Noteworthy Related Trials

2005

Australian Carbohydrate Intolerance Study in Pregnant Women (ACHOIS)

n = 1000 · NEJM

Tested

Dietary advice, blood glucose monitoring, and insulin if needed

Population

Women with mild gestational diabetes mellitus

Comparator

Routine antenatal care

Endpoint

Composite of serious perinatal complications

Key result: Treatment of mild gestational diabetes significantly reduced the risk of serious perinatal complications.

2009

Maternal-Fetal Medicine Units Network Trial

n = 958 · NEJM

Tested

Treatment for mild gestational diabetes (diet, exercise, and insulin)

Population

Pregnant women with mild gestational diabetes

Comparator

Routine obstetric care

Endpoint

Composite of perinatal death, small for gestational age, and other complications

Key result: Treatment reduced the risk of fetal overgrowth, shoulder dystocia, and cesarean delivery, though did not significantly reduce the primary composite outcome.

2020

Diabetes and Pregnancy Intervention Trial (DIPIT)

n = 4061 · Lancet Diabetes Endocrinol

Tested

Early screening and treatment for gestational diabetes

Population

Pregnant women

Comparator

Late screening

Endpoint

Neonatal composite outcome

Key result: Earlier identification and treatment of gestational diabetes did not lead to a significant difference in neonatal complications compared to standard screening.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis