Breastfeeding Leaves a Metabolic Fingerprint That Predicts Heart Disease and Diabetes Risk
A 5-metabolite plasma score linked to breastfeeding duration predicts lower T2D and CVD risk decades later, revealing biological pathways behind the benefit.
Summary
Researchers identified a plasma metabolomic signature associated with lifetime breastfeeding duration in over 6,000 women from multiple cohorts. Using elastic net regression on 181 untargeted metabolites, they derived a 5-metabolite score—comprising three triglycerides (C54:2, C56:2, C56:3), cotinine, and indole-3-propionate—measured in mid-life blood samples. This score was significantly inversely associated with type 2 diabetes incidence (HR=0.76) and cardiovascular disease incidence (HR=0.88), independent of traditional risk factors. Findings were replicated in the Women's Health Initiative and PREDIMED trial, suggesting breastfeeding imprints lasting metabolic changes that may explain its long-term cardiometabolic protection.
Detailed Summary
Breastfeeding is known to reduce a mother's long-term risk of type 2 diabetes (T2D) and cardiovascular disease (CVD), but the biological mechanisms linking lactation history to cardiometabolic health decades later have remained poorly understood. This study takes a metabolomics-first approach to uncover those pathways, identifying a plasma metabolite signature of breastfeeding duration that also predicts future disease risk.
The discovery cohort comprised 4,349 women from the Nurses' Health Studies (NHS and NHSII), with replication in 2,088 postmenopausal women from the Women's Health Initiative (WHI). Researchers profiled 181 untargeted plasma metabolites via liquid chromatography-mass spectrometry from mid-life blood samples and used elastic net regularized regression to derive a metabolite-based breastfeeding score based on self-reported lifetime total breastfeeding duration across all pregnancies.
The resulting score incorporated just 5 metabolites: three large triglycerides (C54:2, C56:2, and C56:3 TAG), cotinine (a nicotine metabolite used as a smoking biomarker), and indole-3-propionate (IPA), a gut microbiome-derived tryptophan metabolite with anti-inflammatory and insulin-sensitizing properties. The score showed a modest but statistically significant correlation with breastfeeding duration and was successfully replicated in the WHI cohort.
Prospective analyses using multivariable Cox regression showed that higher metabolite-based breastfeeding scores were significantly associated with lower incidence of T2D (HR=0.76, 95% CI=0.71–0.82) and CVD (HR=0.88, 95% CI=0.84–0.93), independent of established cardiometabolic risk factors. These associations were externally replicated in both the WHI and the PREDIMED trial, strengthening confidence in the findings. The inclusion of cotinine in the score likely reflects residual confounding by smoking behavior, while IPA's presence points to gut microbiome-mediated metabolic reprogramming as a plausible mechanism.
The study is notable for its large sample size, multi-cohort replication strategy, and the use of blood samples collected years to decades after the last breastfeeding episode, demonstrating that breastfeeding leaves durable metabolic imprints detectable in mid-life. These findings open new avenues for understanding how reproductive history shapes long-term metabolic health and may inform future biomarker-based risk stratification in women.
Key Findings
- A 5-metabolite plasma score (3 triglycerides, cotinine, indole-3-propionate) significantly correlates with lifetime breastfeeding duration.
- Higher metabolite-based breastfeeding score linked to 24% lower T2D incidence (HR=0.76) in mid-life women.
- Score also associated with 12% lower CVD incidence (HR=0.88), independent of traditional cardiovascular risk factors.
- Findings replicated in two independent cohorts: Women's Health Initiative and PREDIMED trial.
- Indole-3-propionate, a gut microbiome metabolite, may mediate breastfeeding's long-term metabolic benefits.
Methodology
Elastic net regularized regression was applied to 181 untargeted plasma metabolites (LC-MS) from 4,349 NHS/NHSII women to derive a metabolite-based breastfeeding score, replicated in 2,088 WHI participants. Prospective T2D and CVD associations were assessed via multivariable Cox regression with external replication in WHI and PREDIMED cohorts.
Study Limitations
Breastfeeding duration was self-reported and recalled retrospectively, introducing potential recall bias. Blood samples were collected in mid-life, not immediately postpartum, so causal directionality of metabolite changes cannot be fully established. The study population was predominantly female nurses, limiting generalizability to more diverse populations.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.