Soy after breast cancer: recurrence and mortality diverge
A pooled survivor analysis extends the evidence beyond one cohort without turning observation into treatment.

Illustrative food image, not a photograph of the study intervention.
A later pooled analysis
Nechuta and colleagues combined information from two US cohorts and one Chinese cohort in a 2012 analysis of 9,514 women with invasive breast cancer. The exposure was soy food intake after diagnosis, estimated with food-frequency questionnaires. This is a distinct publication from the earlier Shanghai survivor report, but pooling cohorts does not necessarily create wholly independent evidence from every previous cohort publication. Its contribution is a larger, cross-country analysis with common statistical modeling. The women were not randomly assigned to eat soy, and the study did not test an isoflavone treatment against placebo.
Nechuta SJ, Caan BJ, Chen WY, Lu W et al.. The American journal of clinical nutrition, 2012. PMID 22648714 · DOI
Separate the three outcomes
Over a mean 7.4 years of follow-up, investigators identified 1,171 deaths, including 881 attributed to breast cancer, and 1,348 recurrences. In the pooled analysis, the category consuming at least 10 milligrams of isoflavones daily had a recurrence hazard ratio of 0.75, with a 95% confidence interval of 0.61 to 0.92. The corresponding estimates for all-cause mortality and breast cancer mortality were 0.87 and 0.83, respectively, with both confidence intervals crossing one. An association with recurrence should therefore not be retold as a demonstrated mortality reduction.
Nechuta SJ, Caan BJ, Chen WY, Lu W et al.. The American journal of clinical nutrition, 2012. PMID 22648714 · DOI
What pooling does not randomize
The models adjusted for sociodemographic, clinical, and lifestyle factors. That improves the comparison but does not make reported dietary choices equivalent to randomized assignments. Other differences between women with different soy intakes may still matter. The investigators observed inverse recurrence associations in both countries despite large differences in intake, which adds useful context without removing this design limitation. The numerical threshold describes an exposure category in the analysis, not a dose prescribed by an experiment. It should not become a daily supplement target or a personalized estimate of recurrence reduction.
Nechuta SJ, Caan BJ, Chen WY, Lu W et al.. The American journal of clinical nutrition, 2012. PMID 22648714 · DOI
Food evidence stays food evidence
A careful summary is that higher postdiagnosis soy food intake was associated with less recurrence in this pooled survivor analysis, while the mortality estimates remained statistically inconclusive. The paper does not establish that starting soy after diagnosis causes that outcome, that a particular soy dish reproduces it, or that concentrated extracts are interchangeable with the reported foods. Use the study to ask an informed question about food within cancer care, rather than to replace treatment with a dietary calculation. Its value lies in distinguishing outcomes and study designs, not in attaching a survival claim to dinner.
Nechuta SJ, Caan BJ, Chen WY, Lu W et al.. The American journal of clinical nutrition, 2012. PMID 22648714 · DOI