No Genetic Evidence for an Effect of Erythrocyte Zinc on Major Depression: Multivariable MR with GBD-2021 Triangulation

Objectives: To systematically evaluate the causal effect of erythrocyte zinc levels on major depressive disorder (MDD) using multiple Mendelian randomization (MR) approaches.
Methods: Two-sample MR was conducted using GWAS data for erythrocyte zinc (QIMR; n=2,603; ieu-a-1079) and MDD (PGC + UK Biobank; n=500,199; ieu-b-102). Four independent SNPs passed quality control and showed strong instrument strength. The primary analysis was inverse-variance weighted (IVW), complemented by MR-Egger, weighted median, MR-PRESSO, and robust adjusted profile score (RAPS). Multivariable MR (MVMR) adjusted for copper and iron. Sensitivity analyses included Steiger directionality and instrument diagnostics. We also performed a post hoc power analysis to evaluate detection thresholds for modest causal effects. To contextualize genetic findings, we conducted a brief ecological triangulation using GBD 2021 (1990–2021; DALY ASR): global trends, 2021 cross-sectional associations with SDI, and cross-correlations (raw vs first-differenced).
Results: No significant causal effect of zinc on MDD was observed (IVW OR=0.997, 95% CI 0.959, 1.037, P=0.99), with consistent null findings across all MR methods. No evidence of pleiotropy or outlier variants was detected. MVMR estimates remained non-significant after adjusting for copper and iron (β=0.053, 95% CI –0.037, 0.143, P=0.23). Power analysis indicated >80% ability to detect ORs<0.97 or >1.03, reinforcing the robustness of null findings. Cross-platform validation using an independent zinc GWAS was attempted but infeasible due to non-overlapping SNPs. A brief GBD 2021 check indicated asynchronous population trends and weak, SDI-dependent associations, with no stable lead–lag after detrending. Findings were consistent with Steiger testing and detrended CCF, showing no stable lead–lag structure.
Conclusions: Genetic evidence does not support a causal effect of zinc on MDD; estimates were consistently null across MR and MVMR with adequate power. A brief ecological triangulation using GBD 2021 showed asynchronous population trends and weak, SDI-dependent associations, aligning with the genetic null and suggesting that improving zinc status alone is unlikely to reduce depression burden at scale. Observational links may reflect residual confounding, reverse causation, or context-dependent biology. Future work should expand and diversify zinc GWAS, develop tissue- and time-specific instruments, and test nonlinearity, subgroup susceptibility, and gene–environment interactions.
Keywords: Erythrocyte zinc；Major depressive disorder；Mendelian randomization；Causal inference；Multivariable MR；Genetic epidemiology；GBD

研究目的：采用多种孟德尔随机化（Mendelian randomization, MR）方法，系统评估红细胞锌水平对重性抑郁障碍（Major Depressive Disorder, MDD）的因果效应。

研究方法：本研究采用两样本孟德尔随机化设计，使用红细胞锌水平的全基因组关联研究（Genome-Wide Association Study, GWAS）数据（QIMR队列；样本量n=2603；数据集编号ieu-a-1079）以及重性抑郁障碍的GWAS数据（精神疾病基因组学联合会[PGC]+英国生物银行[UK Biobank]；样本量n=500199；数据集编号ieu-b-102）。共有4个独立的单核苷酸多态性（Single Nucleotide Polymorphism, SNPs）通过质量控制，且具备较强的工具变量效力。主要分析采用逆方差加权（inverse-variance weighted, IVW）法，并辅以MR-Egger、加权中位数法、MR-PRESSO以及稳健校正轮廓得分（robust adjusted profile score, RAPS）。多变量孟德尔随机化（Multivariable MR, MVMR）分析校正了铜和铁的混杂效应。敏感性分析涵盖Steiger方向性检验与工具变量诊断分析。本研究还开展了事后功效分析，以评估中等程度因果效应的检测阈值。为了对遗传学研究结果进行情境化解读，本研究采用2021年全球疾病负担研究（Global Burden of Disease Study 2021, GBD 2021，覆盖1990-2021年；伤残调整寿命年年龄标化率[DALY ASR]）开展了简要的生态学三角验证分析，包括全球流行趋势、2021年与社会人口发展指数（Socio-demographic Index, SDI）的横断面关联，以及交叉相关性分析（原始数据与一阶差分数据对比）。

研究结果：未观察到锌水平对重性抑郁障碍的显著因果效应（逆方差加权分析比值比[OR]=0.997，95%置信区间[CI]：0.959~1.037，P=0.99），所有孟德尔随机化方法均得到一致的零结果。未检测到多效性或异常变异位点的证据。校正铜和铁的混杂效应后，多变量孟德尔随机化分析结果仍无统计学意义（β=0.053，95%置信区间：-0.037~0.143，P=0.23）。功效分析结果显示，本研究有超过80%的效力可检测到比值比小于0.97或大于1.03的因果效应，进一步验证了零结果的稳健性。本研究尝试采用独立的锌水平GWAS数据进行跨平台验证，但由于单核苷酸多态性位点无重叠而未能实施。基于2021年全球疾病负担研究的简要分析结果显示，人群趋势存在异步性，且关联强度较弱且依赖于社会人口发展指数，去趋势后未发现稳定的先导-滞后关系。本研究结果与Steiger检验及去趋势交叉相关函数（Cross-Correlation Function, CCF）分析结果一致，均未发现稳定的先导-滞后结构。

结论：遗传学证据不支持红细胞锌水平对重性抑郁障碍存在因果效应；孟德尔随机化与多变量孟德尔随机化分析均得到一致的零结果，且研究具备充足的统计功效。基于2021年全球疾病负担研究的简要生态学三角验证分析结果显示人群趋势异步、关联弱且依赖社会人口发展指数，与遗传学零结果一致，提示仅通过改善锌营养状态不太可能大规模降低抑郁负担。观察性研究中发现的锌与抑郁的关联，可能源于残余混杂、反向因果或情境依赖的生物学机制。未来研究应扩大锌水平全基因组关联研究的样本量并丰富其多样性，开发组织特异性与时间特异性的工具变量，并检验非线性关联、亚组易感性及基因-环境交互作用。

关键词：红细胞锌；重性抑郁障碍；孟德尔随机化；因果推断；多变量孟德尔随机化；遗传流行病学；全球疾病负担研究（GBD）