Archival Report 档案报告

The Causal Relationships Between Gut Microbiota, Brain Volume, and Intelligence: A Two-Step Mendelian Randomization Analysis
肠道微生物群、脑容量与智力之间的因果关系：两步孟德尔随机分析法

Methods and Materials 方法与材料

Study Overview 研究概况

An overview of the study is shown in Figure 1. First, we used single nucleotide polymorphisms (SNPs) derived from the summary-level data as genetic instruments for the risk factor. Summary information on the data sources and sample sizes used in this study can be found in Table S1. Then, we performed a 2-sample bidirectional MR to assess the causal effect of each gut microbiome composition on intelligence and vice versa. Finally, we used a 2-step MR analysis to assess whether brain volume and neuroimaging phenotypes play a causal role in mediating the pathway linking the identified gut microbiome composition and intelligence. This study relied on summary-level data that have been made publicly available; ethical approval was obtained in all original studies.
研究概况见图 1。首先，我们使用从汇总级数据中得出的单核苷酸多态性（SNPs）作为风险因素的遗传工具。本研究中使用的数据来源和样本量的摘要信息见表 S1。然后，我们进行了 2 样本双向 MR，以评估每种肠道微生物组成分对智力的因果效应，反之亦然。最后，我们使用两步磁共振分析来评估脑容量和神经影像表型是否对已确定的肠道微生物组组成与智力之间的联系途径起着因果中介作用。本研究依赖于已公开的摘要级数据；所有原始研究均已获得伦理批准。

1. Download : Download high-res image (536KB)
   下载 ：下载高清图片 (536KB)
2. Download : Download full-size image
   下载 ：下载 : 下载全尺寸图片

Figure 1. Study workflow. GWAS, genome-wide association study.
图 1.研究工作流程。GWAS：全基因组关联研究。

Results 成果

Causal Effects of Gut Microbiota Composition on Intelligence
肠道微生物群组成对智力的因果效应

We conducted a 2-sample MR analysis to investigate the impact of gut microbiome abundance on intelligence. The IVW analyses revealed 29 taxa causally associated with intelligence at a nominally significant level (p < .05) (Table S2). More importantly, the genetic liability for 2 specific taxa, namely the genus Oxalobacter and the genus Fusicatenibacter, achieved statistical significance after false discovery rate correction. An abundance of genus Oxalobacter was negatively associated with intelligence (IVW odds ratio [OR] = 0.968; 95% CI, 0.952–0.985; p = 1.88 × 10⁻⁴) (Table 1). The estimates were similar in size in MR RAPS. We also found causal evidence that the abundance of genus Fusicatenibacter was positively associated with intelligence (IVW OR = 1.053; 95% CI, 1.024–1.082; p = 3.03 × 10⁻⁴) (Table 1). These results were further supported by MR RAPS and weighted median. The confidence intervals of the weighted mode and MR-Egger methods were wider than other methods, which could be attributed to their lower statistical power when compared with IVW (46). Scatterplots of SNP effects on these 2 taxa versus intelligence are presented in Figure 2, with colored lines representing the slopes of different MR analyses. Forest plots of individual and combined SNP MR-estimated effect sizes are also presented.
我们进行了双样本 MR 分析，以研究肠道微生物组丰度对智力的影响。IVW分析显示，29个类群与智力有因果关系，且具有名义上的显著性水平（p < .05）（表S2）。更重要的是，2 个特定类群（即牛杆菌属和 Fusicatenibacter 属）的遗传责任在经过错误发现率校正后达到了统计学意义。牛杆菌属的丰度与智力呈负相关（IVW 比值比 [OR] = 0.968; 95% CI, 0.952-0.985; p = 1.88 × 10 ⁻⁴ ）（表 1）。MR RAPS 的估计值大小相似。我们还发现了Fusicatenibacter属丰度与智力正相关的因果关系证据（IVW OR = 1.053; 95% CI, 1.024-1.082; p = 3.03 × 10 ⁻⁴ ）（表1）。MR RAPS 和加权中位数进一步证实了这些结果。加权模式和 MR-Egger 方法的置信区间比其他方法宽，这可能是由于与 IVW 相比，它们的统计能力较低（46）。SNP 对这两个类群的影响与智力的散点图见图 2，彩色线代表不同 MR 分析的斜率。图中还显示了单个和组合 SNP MR 估计效应大小的森林图。

Table 1. Significant MR Results for the Relationship Between Gut Microbiota and Intelligence
表 1.肠道微生物群与智力关系的显著 MR 结果

Method	Number of SNPs SNPs 数量	F Statistic F 统计	OR (95% CI)	p Value p 值	q Value q 价值
Effect of Genus Oxalobacter on Intelligence 牛杆菌属对智力的影响
IVW	10	99.9	0.968 (0.952–0.985) 0.968 (0.952-0.985)	1.88 × 10−4 1.88 × 10 −4	.040
MR RAPS 先生 RAPS			0.968 (0.950–0.986) 0.968 (0.950-0.986)	6.85 × 10−4 6.85 × 10 −4
Weighted Median 加权中位数			0.982 (0.959–1.005) 0.982 (0.959-1.005)	.115
Weighted Mode 加权模式			0.985 (0.949–1.022) 0.985 (0.949-1.022)	.377
MR Egger 埃格先生			0.999 (0.918–1.087) 0.999 (0.918-1.087)	.973
Effect of Genus Fusicatenibacter on Intelligence Fusicatenibacter 菌属对智力的影响
IVW	14	24.8	1.053 (1.024–1.082) 1.053 (1.024-1.082)	3.03 × 10−4 3.03 × 10 −4	.032
MR RAPS 先生 RAPS			1.053 (1.021–1.087) 1.053 (1.021-1.087)	1.01 × 10−3 1.01 × 10 −3
Weighted Median 加权中位数			1.050 (1.010–1.091) 1.050 (1.010-1.091)	.014
Weighted Mode 加权模式			1.015 (0.948–1.087) 1.015 (0.948-1.087)	.643
MR Egger 埃格先生			1.002 (0.896–1.121) 1.002 (0.896-1.121)	.964

p Values from the IVW MR test were adjusted using the Benjamini-Hochberg false discovery rate correction; for the resulting q value, the threshold was set to .05.
使用本杰明-霍奇伯格错误发现率校正法对 IVW MR 检验的 p 值进行调整；对于得出的 q 值，阈值设为 0.05。

IVW, inverse-variance weighted; MR, Mendelian randomization; OR, odds ratio; RAPS, robust adjusted profile score; SNP, single nucleotide polymorphism.
IVW，逆方差加权；MR，孟德尔随机化；OR，几率比；RAPS，稳健调整特征评分；SNP，单核苷酸多态性。

1. Download : Download high-res image (640KB)
   下载 ：下载高清图片（640KB）
2. Download : Download full-size image
   下载 ：下载 : 下载全尺寸图片

Figure 2. Mendelian randomization (MR) plots for the relationship of gut microbiome composition to intelligence. (A) Scatterplot of single nucleotide polymorphism (SNP) effects on genus Oxalobacter vs. intelligence, with the slope of each line corresponding to the estimated MR effect per method. Data are expressed as raw β values with 95% CI. (B) Forest plot of individual and combined SNP MR-estimated effect sizes for genus Oxalobacter on intelligence. (C) Scatterplot of SNP effects on genus Fusicatenibacter vs. intelligence. (D) Forest plot of individual and combined SNP MR-estimated effects sizes for the effect of genus Fusicatenibacter on intelligence. IVW, inverse-variance weighted; RAPS, robust adjusted profile score.
图 2.肠道微生物组组成与智力关系的孟德尔随机化（MR）图。(A) 单核苷酸多态性（SNP）对牛杆菌属的影响与智力的散点图，每条线的斜率对应于每种方法的估计 MR 影响。数据以原始 β 值和 95% CI 表示。(B) 单个和组合 SNP MR 对牛杆菌属智力影响大小的森林图。(C) SNP 对 Fusicatenibacter 属与智力的影响散点图。(D) Fusicatenibacter 属对智力影响的单个和合并 SNP MR 估计效应大小的森林图。IVW：逆方差加权；RAPS：稳健调整的特征得分。

All SNPs selected for inclusion and exclusion for these 2 taxa are presented in Tables S3 to S6 for replication. After selection, 10 and 14 SNPs significantly associated with Oxalobacter and Fusicatenibacter, respectively, were used in the MR analyses (see the Supplement). The F statistics for the genetic instruments indicated an absence of weak instrument bias (Tables S3 and S5). With the current sample size and given an estimated 5.2% (Oxalobacter) and 1.9% (Fusicatenibacter) of the phenotypic variance of gut microbiome composition, our study had sufficient power (>90%) when the expected causal effect for intelligence was ≥0.03 and ≥0.05, respectively (Table S7). We performed statistical fine-mapping and found that half of the selected genetic instruments were prioritized as potential causal variants (Table S8). Sensitivity analyses did not address any pleiotropy in the causal estimates (Table S9). Leave-one-out analysis revealed that none of the SNPs were responsible for driving the MR results (Figure S1), and funnel plots indicated that causal associations were less likely to be influenced by potential biases with SNPs symmetrically distributed (Figure S2). We found no evidence of reverse causality across the analyses in the MR Steiger test (Table S10).
表 S3 至表 S6 列出了这两个类群中所有被选入和排除的 SNPs，以供参考。经过筛选，分别有 10 个和 14 个与 Oxalobacter 和 Fusicatenibacter 显著相关的 SNPs 被用于 MR 分析（见附录）。遗传工具的 F 统计表明不存在弱工具偏差（表 S3 和 S5）。根据目前的样本量，并考虑到肠道微生物组组成的表型变异估计分别为 5.2%（牛杆菌）和 1.9%（镰刀菌），当智力的预期因果效应分别≥0.03 和≥0.05 时，我们的研究具有足够的功率（>90%）（表 S7）。我们进行了统计精细映射，发现所选遗传工具中有一半被优先列为潜在的因果变异（表 S8）。敏感性分析没有发现因果关系估计值中存在任何褶积现象（表 S9）。剔除分析表明，没有一个 SNPs 是导致 MR 结果的原因（图 S1），漏斗图显示，在 SNPs 对称分布的情况下，因果关联不太可能受到潜在偏差的影响（图 S2）。在 MR Steiger 检验中，我们在所有分析中都没有发现反向因果关系的证据（表 S10）。

Mediation Analyses 调解分析

Considering the close relationship between brain volume and human intelligence (19,20), we performed a 2-step MR analysis to investigate whether the effect of identified taxa on intelligence was mediated by regulating brain volume. First, we estimated the causal effect of identified taxa on brain volume using genetic instruments specific to Oxalobacter and Fusicatenibacter. Summary information on brain volume for the SNPs associated with these 2 taxa is listed in Tables S15 and S16. We identified that increased Fusicatenibacter was associated with increased brain volume (IVW beta = 0.086; 95% CI, 0.019–0.153; p = .012) (Figure 3A; Figure S6). We found no evidence of causal relationships for Oxalobacter on brain volume (IVW beta = 0.006; 95% CI, −0.033 to 0.046; p = .745). Then, we assessed the causal effect of brain volume on intelligence using genetic instruments that were associated with brain volume (Tables S17 and S18). We found strong evidence for causal effects of brain volume on intelligence (IVW beta = 0.199; 95% CI, 0.163–0.236; p = 4.32 × 10⁻²⁷) (Figure 3B; Figure S7). Sensitivity analyses did not address any pleiotropy in the causal estimates (Table S19; Figures S8–S10). Because both brain volume and intelligence included subjects from the UKB, sample overlap between the exposure and outcome GWASs may bias the estimate toward the confounded estimate. We estimated the bias due to participant overlap using the method described by Burgess et al. (52). We found that a maximum of 6.3% of brain volume participants overlapped with the intelligence GWAS (17,062/269,867), and the estimated bias was <0.82% in the MR study, indicating that our MR results were less likely to be affected by sample overlap between brain volume and intelligence. Finally, we revealed that the genus Fusicatenibacter indirectly affected intelligence by regulating brain volume. Specifically, the mediation effect was estimated to be 0.017 (95% CI, 0.003–0.031; p = .014) with a mediation proportion of 33.6% (95% CI, 6.8%–60.4%) (Table 3).
考虑到脑容量与人类智力之间的密切关系（19,20），我们分两步进行了磁共振分析，以研究确定的类群对智力的影响是否通过调节脑容量来介导。首先，我们利用牛杆菌和镰刀菌特有的遗传工具估算了已识别类群对脑体积的因果效应。表 S15 和表 S16 列出了与这两个类群相关的 SNPs 脑容量信息摘要。我们发现，Fusicatenibacter 的增加与脑容量的增加有关（IVW beta = 0.086; 95% CI, 0.019-0.153; p = .012）（图 3A；图 S6）。我们没有发现 Oxalobacter 对脑容量有因果关系的证据（IVW beta = 0.006；95% CI，-0.033 至 0.046；p = .745）。然后，我们利用与脑容量相关的遗传工具评估了脑容量对智力的因果效应（表 S17 和 S18）。我们发现了脑容量对智力的因果效应的有力证据（IVW beta = 0.199; 95% CI, 0.163-0.236; p = 4.32 × 10 ⁻²⁷ ）（图 3B；图 S7）。敏感性分析并没有解决因果关系估计中的任何多义性问题（表 S19；图 S8-S10）。由于脑容量和智力都包括了来自 UKB 的受试者，暴露和结果 GWAS 之间的样本重叠可能会使估计值偏向混杂估计值。我们使用 Burgess 等人（52）描述的方法估计了参与者重叠造成的偏差。我们发现，最多有 6.3% 的脑容量参与者与智力 GWAS（17,062/269,867）重叠，而 MR 研究中的估计偏差小于 0.82%，这表明我们的 MR 结果受脑容量和智力样本重叠影响的可能性较小。最后，我们发现镰刀菌属通过调节脑容量间接影响智力。具体来说，中介效应估计为 0.017（95% CI，0.003-0.031；p = .014），中介比例为 33.6%（95% CI，6.8%-60.4%）（表 3）。

1. Download : Download high-res image (410KB)
   下载 ：下载高清图片 (410KB)
2. Download : Download full-size image
   下载 ：下载 : 下载全尺寸图片

Figure 3. Mediation analysis of effects of gut microbiome composition on intelligence via brain volume. (A) Summary Mendelian randomization (MR) estimates derived from the inverse-variance weighted (IVW), weighted median, weighted mode, robust adjusted profile score (MR RAPS), and MR-Egger methods for effects of genus Oxalobacter on brain volume (left) and effects of genus Fusicatenibacter on brain volume (right). (B) Summary MR estimates for the effects of brain volume on intelligence.
图 3.肠道微生物组组成通过脑容量对智力影响的中介分析。(A) 用逆方差加权法（IVW）、加权中位数法、加权模式法、稳健调整剖面得分法（MR RAPS）和 MR-Egger 法得出的孟德尔随机化（MR）估计值摘要，用于分析 Oxalobacter 属对脑容量的影响（左）和 Fusicatenibacter 属对脑容量的影响（右）。(B) 脑容量对智力影响的 MR 估计摘要。

Table 3. The Mediation Effect of Gut Microbiota on Intelligence via Affecting Brain Volume
表 3.肠道微生物群通过影响脑容量对智力的中介效应

Exposure	Total Effect 总效果	Direct Effect A 直接影响 A	Direct Effect B 直接影响 B	Mediation Effect 调解效应		Mediated Proportion (95% CI) 调解比例（95% CI）
	β (95% CI)	β (95% CI)	β (95% CI)	β (95% CI)	p Value p 值
Genus Oxalobacter 牛杆菌属	−0.032 (−0.049 to −0.015) -0.032（-0.049 至 -0.015）	0.006 (−0.033 to 0.046) 0.006（-0.033 至 0.046）	0.199 (0.163 to 0.236) 0.199（0.163 至 0.236）	0.001 (−0.006 to 0.009) 0.001（-0.006 至 0.009）	.758	–
Genus Fusicatenibacter Fusicatenibacter 属	0.051 (0.023 to 0.079) 0.051（0.023 至 0.079）	0.086 (0.019 to 0.153) 0.086（0.019 至 0.153）		0.017 (0.003 to 0.031) 0.017（0.003 至 0.031）	.014	33.6% (6.8% to 60.4%) 33.6%（6.8% 至 60.4）

Total effect: indicates the effect of gut microbiota on intelligence; direct effect A: the effect of gut microbiota on brain volume; direct effect B: the effect of brain volume on intelligence; mediation effect: the effect of gut microbiota on intelligence through affecting brain volume. The total effect, direct effect A, and direct effect B were derived using the inverse-variance weighted method; the mediation effect was derived by using the delta method. All statistical tests were 2-sided. p < .05 was considered significant.
总效应：表示肠道微生物群对智力的影响；直接效应 A：肠道微生物群对脑量的影响；直接效应 B：脑量对智力的影响；中介效应：肠道微生物群通过影响脑量对智力的影响。总效应、直接效应 A 和直接效应 B 采用反方差加权法得出；中介效应采用 delta 法得出。所有统计检验均为双侧检验，P < .05 为显著。

Considering the bidirectional relationships between brain volume and intelligence (18,19), we further performed a 2-step MR analysis to evaluate whether the genus Fusicatenibacter could indirectly regulate brain volume by affecting intelligence (Table S20). We found that Fusicatenibacter could affect intelligence and then regulate brain volume, and the mediation effect was estimated to be 0.008 (95% CI, 0.003–0.014, p = .004) with a mediation proportion of 9.6% (95% CI, 3.1%–16.2%) (Tables S21 and S22; Figures 11 and 12). Our results indicate that brain volume and intelligence are mutually reinforcing in a virtuous cycle, with the genus Fusicatenibacter exerting a positive causal effect in both directions. There was no evidence for reverse causality of the mediators (brain volume or intelligence) on the exposure (genus Fusicatenibacter) (Tables S10 and S23); thus, the preconditions for causal mediation analyses were satisfied in our study.
考虑到脑容量与智力之间的双向关系（18,19），我们进一步进行了两步磁共振分析，以评估Fusicatenibacter属是否可以通过影响智力来间接调节脑容量（表S20）。我们发现，Fusicatenibacter 可影响智力，进而调节脑容量，其中介效应估计为 0.008 (95% CI, 0.003-0.014, p = .004)，中介比例为 9.6% (95% CI, 3.1%-16.2%) （表 S21 和 S22；图 11 和 12）。我们的研究结果表明，脑容量和智力在良性循环中相互促进，Fusicatenibacter 属在两个方向上都产生了积极的因果效应。没有证据表明中介因子（脑容量或智力）与暴露因子（镰刀菌属）之间存在反向因果关系（表 S10 和 S23）；因此，我们的研究满足了因果中介分析的先决条件。

In addition, given that intelligence is also related to a wide range of imaging phenotypes (4), we further investigated whether the subdivisions of the human cerebral cortex (25) could also act as potential mediators. Although there was no evidence for causal relationships after multiple testing correction, we found that the genetic liability for Oxalobacter and Fusicatenibacter made a causal contribution to 158 and 190 neuroimaging phenotypes, respectively, at a nominally significant level of p < .05 (Table S24). Then, we assessed the causal effect of these neuroimaging phenotypes on intelligence and found 50 mediating pathways from Oxalobacter or Fusicatenibacter to intelligence (Table S25). For example, the most significant mediator, the surface area of posterior insular area 2 [regions are based on the Human Connectome Project parcellation scheme (53)], was causally affected by Oxalobacter, which then influenced intelligence, and the mediation effect was estimated to be −0.009 (95% CI, −0.015 to −0.003; p = .003).
此外，考虑到智力也与多种影像表型相关（4），我们进一步研究了人类大脑皮层的分支（25）是否也可以作为潜在的中介。尽管多重检验校正后没有证据表明存在因果关系，但我们发现，Oxalobacter 和 Fusicatenibacter 的遗传责任分别对 158 种和 190 种神经影像表型有因果关系，且 p < .05 的水平具有名义显著性（表 S24）。然后，我们评估了这些神经影像表型对智力的因果效应，发现了 50 条从草酸杆菌或镰刀菌到智力的中介途径（表 S25）。例如，最显著的中介因子--后岛叶区2的表面积[区域基于人类连接组计划的划分方案（53）]受Oxalobacter的因果影响，进而影响智力，中介效应估计为-0.009 (95% CI, -0.015 to -0.003; p = .003)。

Discussion 讨论

The role that our microbiome plays in health and disease is one of the biggest scientific challenges (54). In this study, by using summary statistics obtained from the largest GWAS meta-analysis of gut microbiota and intelligence, we conducted a bidirectional 2-sample MR analysis to disentangle the causal relationship. We observed causal evidence indicating a risk effect of Oxalobacter and a protective effect of Fusicatenibacter on intelligence. In contrast, we found no evidence of a causal relationship between intelligence and the abundance of the gut microbiome. More interestingly, we conducted a mediation analysis and showed that the effect of genus Fusicatenibacter on intelligence was partially mediated by regulating brain volume. These findings may have implications for public health interventions that seek to enhance individual intelligence.
我们的微生物群在健康和疾病中扮演的角色是最大的科学挑战之一（54）。在本研究中，我们利用从最大规模的肠道微生物群与智力的 GWAS 元分析中获得的汇总统计数据，进行了双向 2 样本 MR 分析，以厘清其中的因果关系。我们观察到的因果关系证据表明，Oxalobacter 对智力有风险作用，而 Fusicatenibacter 对智力有保护作用。相比之下，我们没有发现智力与肠道微生物组丰度之间存在因果关系的证据。更有趣的是，我们进行了一项中介分析，结果表明镰刀菌属对智力的影响部分是通过调节脑容量来中介的。这些发现可能会对旨在提高个人智力的公共卫生干预措施产生影响。

The gut microbiota potentially affects the development and function of the immune, metabolic, and nervous systems through bidirectional communication along the gut-brain axis (55), which is believed to be involved in the intelligence/cognitive ability of the host (9). For example, compared with normal mice, germ-free mice showed impairments in tests of memory and reductions in hippocampal brain-derived neurotrophic factor (56), which is a neurotrophin crucial for neuronal development and survival, synaptic plasticity, and cognitive function. In contrast, the administration of probiotic strains can promote memory behavior through their production of lactate and the promotion of GABA (gamma-aminobutyric acid) accumulation in the hippocampus (57,58), which provides solid evidence for the role of the microbiome in intelligence. The association between the gut microbiome and cognitive ability in animals also occurs in humans, as mentioned previously. However, current conclusions rely predominantly on disparities in gut microbiome composition and the results of trials that involved the transplant of gut microbiota into gnotobiotic mice (59, 60, 61), which can be affected by various confounding factors. Because the gut microbiome is considered to be highly dynamic, causal association has been an unresolved issue in the field.
肠道微生物群可能会通过肠道-大脑轴的双向交流影响免疫、代谢和神经系统的发育和功能（55），这被认为与宿主的智力/认知能力有关（9）。例如，与正常小鼠相比，无菌小鼠在记忆测试中表现出障碍，海马脑源性神经营养因子减少（56），这种神经营养因子对神经元的发育和存活、突触可塑性和认知功能至关重要。相反，服用益生菌菌株可通过产生乳酸和促进 GABA（γ-氨基丁酸）在海马中的积累来促进记忆行为（57,58），这为微生物组在智力中的作用提供了确凿的证据。如前所述，肠道微生物组与动物认知能力之间的联系在人类中也同样存在。然而，目前的结论主要依赖于肠道微生物组组成的差异以及将肠道微生物群移植到非生物小鼠体内的试验结果（59、60、61），而这些结果可能会受到各种混杂因素的影响。由于肠道微生物群被认为是高度动态的，因此因果关系一直是该领域的一个悬而未决的问题。

To our knowledge, we have reported the first MR analysis to investigate the potential causal relationship between gut microbiota and intelligence. We found that the genetic liability for 2 taxa, Oxalobacter and Fusicatenibacter, reached statistical significance after false discovery rate correction. Oxalobacter is one of the key taxa involved in the gut microbiome diversity of individuals (62), and previous studies have shown that Oxalobacter formigenes plays an important role in oxalate absorption and secretion pathways in the gut (63). Observational studies have found controversial associations between Oxalobacter and cognitive ability. Some studies reported a negative association between the abundance of Oxalobacter and cognitive ability estimated by the Mini-Mental State Examination score (64). In contrast, another study reported that the reduced abundance of Oxalobacter was associated with mild cognitive impairment in older adults (65). Our MR analysis produced causal evidence indicating a risk effect of Oxalobacter on intelligence. We also provided evidence for a protective effect of Fusicatenibacter on intelligence. A recent cross-sectional analysis focusing on species-level features associated with cognition found that certain bacteria capable of producing short-chain fatty acids, such as Fusicatenibacter saccharivorans, were positively associated with better cognitive performance (15). Specifically, a higher abundance of Fusicatenibacter saccharivorans was linked to better scores on both the Mini-Mental State Examination and the Montreal Cognitive Assessment (15). Some taxa that only showed nominal statistical significance have also been reported to be associated with intelligence, such as Ruminococcaceae UCG003, Ruminococcus, and Phascolarctobacterium (see the Supplement) (66,67).
据我们所知，我们首次报告了研究肠道微生物群与智力之间潜在因果关系的磁共振分析。我们发现，经过误发现率校正后，2 个类群，即 Oxalobacter 和 Fusicatenibacter 的遗传责任达到了统计学显著性。牛杆菌是参与个体肠道微生物组多样性的关键类群之一（62），先前的研究表明，形牛杆菌在肠道草酸盐吸收和分泌途径中发挥着重要作用（63）。观察性研究发现，牛杆菌与认知能力之间的关系存在争议。一些研究报告称，草酸杆菌的丰度与以迷你精神状态检查（Mini-Mental State Examination）评分估算的认知能力之间存在负相关（64）。相反，另一项研究报告称，牛分枝杆菌丰度的降低与老年人的轻度认知障碍有关（65）。我们的磁共振分析提供的因果关系证据表明，牛分枝杆菌对智力有风险影响。我们还提供了证据，证明镰刀菌对智力有保护作用。最近一项侧重于与认知相关的物种水平特征的横断面分析发现，某些能够产生短链脂肪酸的细菌，如囊酵母菌（Fusicatenibacter saccharivorans），与更好的认知表现呈正相关（15）。具体来说，较高的 Fusicatenibacter saccharivorans 丰度与较高的迷你精神状态检查和蒙特利尔认知评估得分有关（15）。据报道，一些仅显示名义统计意义的类群也与智力有关，如反刍球菌科（Ruminococcaceae）UCG003、反刍球菌（Ruminococcus）和Phascolarctobacterium（见补编）（66,67）。

Another interesting conclusion arising from the current study is that the protective effect of Fusicatenibacter abundance on intelligence was partially mediated by increasing brain volume. In the first step, we identified that an increased abundance of the Fusicatenibacter was associated with increased intelligence. Differences in gut microbial composition have been reported to be associated with brain structure (68, 69, 70, 71). Specifically, multimodal neuroimaging fusion biomarkers have been reported to mediate the association between gut microbiota and cognition (72). However, little is known about the effect of the Fusicatenibacter on brain volume. The second step provided evidence that genetically determined higher brain volume was associated with higher intelligence. Jansen et al. (19) found causal evidence of effects of genetically predicted brain volume on intelligence (beta = 0.154, p = 1.88 × 10⁻²³) using the generalized summary-data-based MR package. This result supports our second-step estimate in terms of both direction and magnitude.
本研究得出的另一个有趣结论是，丰富的镰刀菌对智力的保护作用部分是通过增加脑容量来实现的。第一步，我们发现镰刀菌丰度的增加与智力的提高有关。据报道，肠道微生物组成的差异与大脑结构有关（68、69、70、71）。具体而言，多模态神经影像融合生物标志物据报道介导了肠道微生物群与认知之间的关联（72）。然而，人们对 Fusicatenibacter 对脑容量的影响知之甚少。第二步提供的证据表明，由基因决定的较高脑容量与较高智力有关。Jansen 等人（19）利用基于概括数据的 MR 软件包发现了遗传预测脑容量对智力影响的因果证据（β=0.154，p=1.88×10 ⁻²³ ）。这一结果在方向和幅度上都支持我们的第二步估计。

Admittedly, several limitations should be acknowledged when interpreting the results of this study. First, although we used the largest GWAS meta-analysis for gut microbiome composition reported to date, the number of subjects in the gut microbiome composition GWAS is relatively small, and genetic factors can only explain a small proportion of the variance in gut microbiome features; thus, the power to detect a causal relationship was limited. Second, consistent with previous microbiota MR studies, we used a relaxed p-value threshold to select genetic instruments due to the limited variant number associated with gut microbiome composition (17,28,29), which may introduce weak instrument bias. Nevertheless, we tested instrument strength and found that all instruments had F statistics exceeding 10, which is a conventional cutoff for strong instruments (52). Third, due to the lowest taxonomic level being genus in the exposure dataset, we were unable to explore the causal association between gut microbiota and intelligence at the species level. Fourth, although we identified several neuroimaging phenotypes that could serve as potential mediators of the relationship between the gut microbiome and intelligence, no significant results were identified after multiple testing corrections. This is probably due to the correlations among neuroimaging phenotypes; the multiple comparison adjustments may be excessive. Therefore, causality cannot be completely ruled out in the mediation results. Moreover, the study primarily involved individuals of European ancestry, suggesting that future research is needed to determine whether the results can be generalized to other populations.
诚然，在解释本研究的结果时，应该承认有几个局限性。首先，虽然我们使用了迄今为止报告的最大的肠道微生物组组成 GWAS meta 分析，但肠道微生物组组成 GWAS 的受试者数量相对较少，遗传因素只能解释肠道微生物组特征变异的一小部分；因此，检测因果关系的能力有限。其次，与之前的微生物群 MR 研究一致，由于与肠道微生物群组成相关的变异数量有限（17,28,29），我们在选择遗传工具时使用了宽松的 p 值阈值，这可能会带来微弱的工具偏差。尽管如此，我们还是测试了工具强度，发现所有工具的 F 统计量都超过了 10，这是强工具的常规临界值（52）。第三，由于暴露数据集中的最低分类级别为属，我们无法在物种级别探讨肠道微生物群与智力之间的因果关系。第四，尽管我们发现了几种神经影像表型可以作为肠道微生物组与智力之间关系的潜在中介，但经过多重检验校正后，没有发现显著的结果。这可能是由于神经影像表型之间存在相关性；多重比较调整可能过度。因此，不能完全排除中介结果中的因果关系。此外，该研究主要涉及欧洲血统的个体，这表明未来的研究需要确定这些结果是否可以推广到其他人群。