Urinary Isoflavone Concentrations and Semen Parameters of Japanese Men Seeking Fertility Treatment

Men in the top daidzein quartile had 32–40% lower sperm counts than lowest-quartile men — but this is cross-sectional data from a fertility clinic.

Journal: Asian Journal of Andrology | Published: 2026-01-13 | Type: Cross-sectional | PMID: 41527991 Authors: Konishi S, Mizuno Y (University of Tokyo Graduate School of Medicine); Yamasaki K, Uchida M (Tsukuba Gakuen Hospital); Iwamoto T (Sanno Hospital, Tokyo) Funding/COI: Not listed for either

Summary

157 Japanese men presenting to fertility clinics provided same-day urine and semen samples. Higher urinary daidzein concentration correlated with lower sperm count across all three upper quartiles; genistein showed a significant association only in the third quartile. Equol showed no association. The authors conclude that high isoflavone intake "may be associated" with reduced sperm count — appropriately hedged language for what is, at bottom, a snapshot from a pre-selected population.

Claims

• Median sperm count: 112 × 10⁶ (IQR: 50–221 × 10⁶); median concentration: 39 × 10⁶/mL (IQR: 17–72 × 10⁶/mL)
• Daidzein Q2 vs Q1: −40% sperm count (95% CI: −59%, −13%)
• Daidzein Q3 vs Q1: −37% (95% CI: −56%, −9%)
• Daidzein Q4 vs Q1: −32% (95% CI: −53%, −3%)
• Genistein Q3 vs Q1: −50% (95% CI: −66%, −26%) — the only statistically robust genistein finding
• Genistein Q2 and Q4 CIs both cross zero (−47% to +12%; −51% to +3%) — not significant
• No association between any isoflavone quartile and total sperm motility
• No association between equol and sperm count (P > 0.05)

Study Quality

This is a cross-sectional study — it cannot establish whether high isoflavone exposure causes lower sperm counts, or whether men with certain metabolic profiles both eat more soy and produce fewer sperm. Urinary isoflavone levels were measured by LC-MS/MS and adjusted for urine specific gravity, which is methodologically sound. However, spot urine is a notoriously poor proxy for habitual dietary exposure; isoflavone excretion fluctuates substantially with recent meals. A 24-hour urine or repeated-sample protocol would be far more credible.

The daidzein dose-response pattern across Q2–Q4 looks consistent, which is at least suggestive. The genistein findings are weaker: only one of three upper quartiles is statistically significant, with the other two spanning zero. A monotonic dose-response for genistein this is not.

Red Flags

• Fatal selection bias: Every participant was already seeking fertility treatment. Men with normal sperm counts aren't in this cohort. Any association found here may not generalize to the broader population.
• Spot urine only: Single-day isoflavone measurement captures what a man ate recently, not his long-term dietary pattern.
• Missing confounder adjustment detail: The abstract does not specify what covariates (age, BMI, smoking, alcohol, abstinence interval) were controlled for — a significant gap.
• Genistein pattern is non-monotonic: Q3 is significant, Q2 and Q4 are not. This is the kind of noise pattern that emerges from multiple comparisons, not a clean dose-response.
• Funding and COI both unlisted: For a study touching dietary industry interests, this is a notable omission.
• n = 157 across two sites: Underpowered for the number of comparisons being run.

Strengths

• Objective biomarker measurement (LC-MS/MS) avoids dietary recall bias
• Specific gravity adjustment is appropriate
• Separating genistein, daidzein, and equol rather than lumping "total isoflavones" is the right approach
• Two-center recruitment adds modest geographic spread within Japan
• Authors are appropriately cautious in their language ("may be associated," "effect on fecundity remains unknown")

Verdict

The daidzein findings are intriguing and internally consistent enough to be worth noting, but this paper cannot carry the weight of a causal claim. It's a small cross-sectional snapshot of men who already have fertility problems, using single-day urine to estimate habitual soy intake, with genistein results that don't hold up across quartiles. File it as hypothesis-generating — a reasonable nudge toward a properly powered longitudinal study — not as evidence that soy harms fertility. The missing funding and COI disclosures are sloppy for a journal of this caliber.