V an Haarlem and colleagues reported that higher dietary niacin and cobalamin intake is associated with a lower incidence of open-angle glaucoma (OAG). While this prospective study has notable strengths, I wish to raise several methodological and interpretive considerations.
First, dietary intake was assessed only at baseline using food‑frequency questionnaires. With a mean follow‑up of approximately 11 years, dietary habits may have changed due to aging, health status, or disease awareness. Although the authors argue that misclassification would be non‑differential, changes in diet could have been influenced by prodromal symptoms, potentially biasing results away from the null. A sensitivity analysis excluding early incident cases (e.g., first 2-4 years) would help address reverse causality.
Second, the dose‑response curves show that the protective association for cobalamin plateaus at relatively low intakes (≈2-4 µg/d). The authors suggest that meeting the Dutch recommended intake (2.8 µg/d) suffices for risk reduction. However, serum homocysteine, a metabolite linked to cobalamin, was not significantly associated with iOAG in this study. This suggests that homocysteine reduction may not explain the observed benefit. Alternative mechanisms (e.g., direct neuroprotection or myelin synthesis) require further investigation.
Third, the finding that niacin’s protective effect was stronger among individuals with high genetic risk for OAG is intriguing. The genetic risk score (GRS) was derived from published variants, but it is unclear whether these variants are primarily IOP-related or IOP-independent. Clarifying this would help determine whether niacin modifies IOP-dependent pathways. Additionally, presenting multiplicative interaction tests (rather than stratified estimates alone) would strengthen the evidence for effect modification.
Fourth, in Model 4 the authors adjusted for IOP measured at the incident visit. For participants receiving IOP‑lowering medication, a fixed correction factor (division by 0.7) was applied. This assumes uniform drug response across individuals, which may not be valid. A sensitivity analysis limited to untreated participants or using baseline IOP would provide more robust support for IOP‑independent effects.
Finally, multiple comparisons were not accounted for. The study examined five B vitamins across several outcomes (iOAG, IOP, RNFL, GCL) and multiple models. While the main findings for niacin and cobalamin are statistically significant, a correction such as Bonferroni would be informative to rule out type I error in this exploratory nutritional analysis.
In summary, this study provides valuable prospective evidence linking niacin and cobalamin intake with lower OAG incidence. Addressing the above points would further strengthen the interpretation and clinical translation of these findings.
CRediT authorship contribution statement
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
