Chandarana · Journal of the American Nutrition Association 2026 · narrative review · n=?

Why Serum Magnesium Fails: A Narrative Review of Magnesium Biochemistry, Compartmental Exchange, and Endpoint Selection for Supplementation Trials.

Level 5 - mechanism / opinion, no new human data

Narrative review and mechanism-based reasoning with no primary human data.

PubMed 42647115 · doi:10.1080/27697061.2026.2719468 · record verified 2026-08-27

What was done

This narrative review analyzed magnesium biochemistry, compartmental distribution kinetics, and biomarker endpoints used in oral supplementation trials. The authors evaluated serum, erythrocyte, ionized, urinary, and loading-test measurements across five clinical domains (essential hypertension, insulin resistance/type 2 diabetes, insomnia, muscle cramping, and migraine prophylaxis) to assess endpoint adequacy against physiological timescales.

What was found

No new primary data were collected; published biological variation metrics were synthesized: - Serum magnesium accounts for <1% of total body magnesium, with a reference change value of approximately 12%. - Erythrocyte magnesium integrates intracellular exposure over approximately 120 days, with a critical difference/reference change value of approximately 22%. - Because oral supplementation produces biochemical changes comparable to or smaller than these reference change values, trials relying on serum levels are methodologically underpowered to detect tissue repletion, explaining null biochemical findings alongside positive clinical outcomes.

Why it matters

This review explains why normal serum magnesium fails to rule out tissue depletion and proposes erythrocyte magnesium as a more physiologically sound group-level endpoint for supplementation trials lasting 3 months or longer.

Limits

As a narrative review, it lacks a systematic search protocol, formal risk of bias evaluation, and meta-analytic pooling. The proposed kinetic model and biomarker hierarchies rely on synthesis of secondary literature rather than new primary experimental data.

Cited by