Cohort studycited by the Tayebat archivefull text checked A-LEG-09 in-depth review
Soy formula and gene marks in baby girls
Soy Formula and Epigenetic Modifications: Analysis of Vaginal Epithelial Cells from Infant Girls in the IFED Study.
Overview
Harlid S et al.. Environ Health Perspect 2017 · PMID 27539829 · DOI 10.1289/EHP428
Small study found gene-methylation differences in soy-fed girls; health meaning unknown.
Does the study support the specific claim? Partly · Methodological quality: low
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Test epigenetic effects
- How was the study done?
Cohort + mouse
- How many people, animals or samples were studied?
50 girls
- What was actually found?
Methylation differences in one gene
- How large was the observed effect?
Not applicable
- What do the results mean in everyday life?
Biological effect possible.
- What are the study's limitations?
Small; no outcomes
- What does it say about Elawady's specific claim?
Weak support
- Do the results apply to Germany and Europe?
Moderate
- What can explicitly not be concluded from this study?
Cannot show disease.
Scientific in-depth review
1. Full reference
- Authors
- Harlid S, Adgent M, Jefferson WN, Panduri V, Umbach DM, Xu Z, Stallings VA, Williams CJ, Rogan WJ, Taylor JA
- Journal
- Environ Health Perspect
- Year
- 2017
- Identifier
- PMID 27539829 DOI 10.1289/EHP428 · PMC5332195
- Verification
- verified – PubMed E-utilities efetch, 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
NIH/NIEHS Intramural Research Program, NCRR (UL1RR024134) and NCATS (UL1TR000003); authors declare no competing financial interests.
2. Research question
Is soy formula associated with DNA methylation changes in vaginal cells?
3. Study design and population
- Study design
- Cohort study – Exploratory methylation substudy within a prospective observational cohort (IFED; feeding chosen by mothers, not randomised): small epigenome-wide scan, pyrosequencing follow-up in serial samples, plus neonatal mouse genistein-injection experiment
- Population
Healthy term singleton infant girls exclusively fed soy or cow-milk formula from birth (IFED cohort, Philadelphia); mothers mostly black, many with high-school education or less
- Country
USA (Children's Hospital of Philadelphia clinical site)
- Period
IFED enrolment Aug 2010 - Mar 2014; swabs collected after 17 Dec 2012; follow-up birth to 9 months
- Sample size
EWAS: 10 girls (4 soy, 6 cow) after 9 samples failed QC; pyrosequencing: 214 samples from 50 girls (28 soy, 22 cow), 205 samples from 49 girls in final analysis; mice: 60 genistein/vehicle pups + 10 untreated newborns
- Inclusion and exclusion criteria
Mothers >=18 y, English-speaking, no gestational diabetes/thyroid/endocrine disorders; infants term (37-42 wk), 2500-4500 g; excluded if feeding method changed; breastfed infants and boys not included (sample availability/DNA yield)
4. Methods and endpoints
- Exposure / intervention
Exclusive soy formula from birth (maternal intention); mouse: subcutaneous genistein 50 mg/kg/day on postnatal days 1-5
- Comparator
Exclusive cow-milk formula; mouse: corn-oil vehicle
- Endpoints
DNA methylation at PRR5L CpGs (cg00220721, cg22117805) in vaginal epithelial cells; mouse vaginal Prr5l mRNA
- Statistical methods
EWAS: robust linear regression at 427,097 CpGs without multiple-comparison correction; pyrosequencing: mixed models with natural cubic splines over age; pointwise differences without multiplicity correction; mouse: Mann-Whitney U
5. Actual results
- Results
EWAS: 5 CpGs with p<10^-4 and beta difference >0.3, three near the PRR5L transcription start site (higher methylation in soy-fed). Pyrosequencing: methylation ~0.8 at birth, declined with age (p<0.001); soy-fed girls had higher mean levels, but overall trajectories did not differ significantly (p=0.25); uncorrected pointwise differences significant (p<0.05) from about 126 days. TCGA tumour data: cg0022072 methylation inversely correlated with PRR5L mRNA (Spearman -0.53 to -0.55). Mice: genistein-exposed had significantly lower vaginal Prr5l mRNA at day 5 (p<0.01, ~50%); no difference by day 22.
- Absolute risks
not applicable (no clinical outcome)
- Authors' conclusion
Soy-formula-fed girls showed higher PRR5L methylation in vaginal cells, possibly associated with decreased expression of an oestrogen-responsive gene; persistence and health consequences, if any, are unknown; authors state the data do not constitute a clear contraindication for soy formula and are exploratory.
6. Strengths
- Prospective serial sampling from birth
- Independent samples used for pyrosequencing replication
- Complementary mouse experiment at serum-relevant genistein doses
7. Weaknesses
- Discovery scan of only 10 infants, without multiple-testing correction across 427,097 CpGs
- In the scan, soy samples (4-8 weeks) and cow samples (20-32 weeks) differed strongly in age, while methylation falls with age - major confounding of the discovery signal
- Overall trajectory difference not significant (p=0.25); significance only in uncorrected pointwise tests
- Feeding not randomised; no adjustment for maternal/social factors
- No exposure (isoflavone) measurement and no clinical outcome
- Suboptimal DNA input; 9 of 20 arrays failed QC
- Mouse exposure by injection, not oral
8. Risk of bias
- Orienting tool
- ROBINS-E
- Type of assessment
- structured assessment based on the full text by a single reviewer (no formal double review)
- Full formal assessment
- no
- Overall risk of bias
- high
- Rationale
Single-reviewer structured ROBINS-E judgement from full text; not formal (requires two independent reviewers and protocol). Raised from some concerns to high because of the age-imbalanced, uncorrected discovery scan and null global trajectory test.
| Domain | Judgement | Rationale |
|---|---|---|
| Confounding | high | Self-selected feeding groups; no covariate adjustment; discovery samples age-imbalanced between groups. |
| Exposure measurement | low | Exclusive feeding regimen documented prospectively; changers excluded. |
| Selection of participants | some concerns | Only 237 of 464 swabs yielded enough DNA; only samples after Dec 2012; selection of 20 highest-yield swabs for the array. |
| Post-exposure interventions | low | Feeding changers excluded. |
| Missing data | some concerns | Substantial sample loss for DNA yield and QC failures. |
| Outcome measurement | low | Pyrosequencing validated against array (R2 0.96); laboratory measures unlikely to be influenced by feeding knowledge. |
| Selection of reported result | high | Candidate loci selected from an uncorrected scan; only two CpGs followed up; significance claimed from uncorrected pointwise tests while the global test was null. |
Full-text review (25 October 2026): Full text shows uncorrected EWAS in 10 infants with large age imbalance between groups, null global trajectory test (p=0.25), injection-based mouse model, and authors' statement that results are no clear contraindication; bias raised from some_concerns to high.
9. Assessment of the statistical interpretation
Exploratory and hypothesis-generating. The global trajectory comparison was not significant (p=0.25); pointwise significance after 126 days is uncorrected for multiplicity. No effect size with CI for a clinical endpoint.
10. Relevance to the specific claim
Mixed/supportive for exposure effects in C-LEG-10.
11. Independent contrary findings
Strom 2001.
12. Transferability to Germany and Europe
Moderate.
13. Overall methodological judgement
Weak, exploratory biological signal of unknown clinical meaning; does not quantify phyto-oestrogen exposure.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | low |
| Risk of bias | high |
| Precision of results | low |
| Directness | indirect |
| Reproducibility | not replicated |
| Consistency with the overall evidence | unknown |
| Regional transferability | moderate |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Replication with multiplicity control and balanced ages; persistence of methylation changes; any clinical outcome.
15. Linked fact checks
Use by the Tayebat archive
Note: according to the archive, the studies were compiled by the archive operators. Whether Dr. Elawady himself cited this study is not established.
- What the archive says about the study
Bibliography only.
- A. What was actually studied?
Soy formula and infant vaginal-cell methylation.
- B. What did the authors find?
Higher PRR5L methylation in soy-fed girls in serial samples (global trajectory difference not significant, p=0.25; uncorrected pointwise significance after 126 days); genistein lowered Prr5l mRNA transiently in mice.
- C. What conclusions do the authors themselves draw?
May indicate an oestrogen-responsive epigenetic change; health consequences unknown; authors say it is not a clear contraindication to soy formula.
- D. What does the archive derive from it?
Supports soy-infant concern.
- E. Is this derivation scientifically justified?
Only partly: it shows a possible biological response to soy formula, but it does not measure isoflavone levels or any health outcome, and the authors explicitly do not regard it as a contraindication.
- Adequacy of the interpretation (by the archive)
- not applicable