Serum Minerals and Male Age-related Hypogonadism: Multimodal Evidence Decoding Associations and Intervention Strategies

Serum iron associates with testosterone in middle-aged men — but the causal story is murkier than the headline suggests

Journal: The Journal of Clinical Endocrinology and Metabolism | Published: 2026-02-20 | Type: Journal Article (cross-sectional + Mendelian randomization) | PMID: 40972672 Authors: Xie Junfeng et al., Department of Urology, First Affiliated Hospital of China Medical University, Shenyang; Yao Jiaxi, Second Hospital of Chongqing Medical University Funding/COI: Not listed for either

Summary

Using NHANES 2013–2016 data from 687 age-stratified males, this paper finds that serum iron associates positively with total testosterone in middle-aged men, with a J-shaped curve and an inflection point at 12.18 µmol/L. Mendelian randomization tentatively supports a causal direction — iron influencing testosterone — but the signal fades when metabolic confounders are adjusted for. Other minerals tested (zinc, copper, phosphorus, calcium) didn't survive multivariable adjustment. The most novel claim is that a bilirubin degradation metabolite partially mediates the iron-testosterone relationship, acting as a negative mediator accounting for −68% of the effect.

Claims

• Serum iron positively associated with total testosterone in middle-aged males (β = 0.194–0.244, P = .016–.033; n = 687)
• The relationship is nonlinear (J-shaped, nonlinear P = .044), with an inflection point at 12.18 µmol/L serum iron
• Univariable Mendelian randomization supported iron → testosterone causation (P = 1.24 × 10⁻⁵)
• Multivariable MR: effect attenuated after adjustment for metabolic factors — causal relationship not cleanly established
• Zinc, copper, phosphorus, and calcium associations with sex hormones all lost significance in multivariable models
• Bilirubin metabolite C16H18N2O5 identified as a negative mediator (mediation proportion = −68.0%, P = .016)

Study Quality

The cross-sectional NHANES backbone is a well-characterized dataset but inherently cannot establish causation — the authors know this and appropriately add Mendelian randomization as a causal probe. That's a reasonable multi-method design. The MR uses SNP instruments for serum iron, and univariable results are statistically strong. However, the authors' own multivariable MR — which adjusts for metabolic factors — weakens the causal inference substantially, which they acknowledge but don't dwell on. That's a significant qualifier buried in the results.

The bilirubin metabolite mediation finding is the paper's most speculative element. Mediation analysis on observational data requires strong assumptions about temporal ordering and no unmeasured confounding, neither of which a cross-sectional NHANES analysis can guarantee. The metabolite identification itself comes from a secondary analysis layer, and with a mediation proportion of −68%, it's doing a lot of explanatory work for a pathway that has no independent replication.

Red Flags

• Funding and COI absent: No disclosure for either. In a paper making nutritional intervention claims from a Chinese university hospital, this is a transparency problem.
• n = 687 is modest for age-stratified subgroup analyses — middle-aged subgroup is smaller still; effect sizes should be interpreted cautiously
• Multivariable MR attenuated the iron signal: the headline causal result does not hold after metabolic adjustment; the paper's framing downplays this
• Cross-sectional design cannot establish whether iron levels precede testosterone changes or the reverse
• Bilirubin mediation is speculative: untested pathway, identified in the same dataset it was derived from, no external validation
• Conclusion language drifts toward advice: "personalized precision nutrition strategies" is a clinical recommendation, not a study finding
• Age stratification with a total n of 687 means some strata are too small to draw reliable estimates

Strengths

• Multi-method design (observational + MR + mediation) is appropriate for this research question
• Honest about the attenuation of MR results under multivariable adjustment
• Restricted cubic splines to detect nonlinearity is the right tool for this kind of dose-response question
• NHANES is a well-validated, nationally representative survey with careful mineral measurement protocols
• Age stratification is conceptually sound — mineral-hormone relationships plausibly differ across the lifespan

Verdict

This paper asks a legitimate question — does iron status causally influence testosterone? — and uses a reasonable toolkit to answer it. The honest answer from the data is: probably something, but it's complicated. The MR supports a directional effect, then undermines it. The bilirubin mediation pathway is genuinely novel but floated on thin statistical ice. The undisclosed funding and COI situation is a problem given the paper's drift toward nutritional intervention framing. Worth reading as hypothesis generation; not worth citing as evidence that iron supplementation does anything for testosterone.