Causal effects of metabolites on 5 stroke phenotype
The causal association between 486 metabolites and stroke of 5 phenotypes is shown in Supplementary Table 1.
For a better interpretation of metabolic changes, we excluded 177 unknown metabolites while including 309 with known structures and functions. Using these IVs, we estimated the causal association between these 309 metabolites and 5 stroke phenotypes and identified a total of 82 suggestive associations (p < 0.05, corresponding to 60 unique metabolites).
For ischemic stroke (cardioembolic), there were 8 causal relationships as follows: 2-methoxyacetaminophen sulfate*(p = 0.040,OR = 0.760, 95%CI:0.5854–0.987), arachidonate (20:4n6) (p = 0.0000768, OR = 1.904, 95%CI:1.384–2.620), cholate (p = 0.029, OR = 0.772, 95%CI:0.612–0.976), cholesterol (p = 1.99E-08, OR = 0.963, 95%CI:0.951–0.976),cortisol (p = 0.009, OR = 1.250, 95%CI:1.058–1.478), pelargonate (9:0) (p = 0.014,OR = 0.475, 95%CI:0.262–0.862), pentadecanoate (15:0) (p = 0.002,OR = 1.403, 95%CI: 1.138–1.729),tryptophan betaine (p = 0.049,OR = 2.049, 95%CI:1.003–4.186), which was shown in Fig. 2.
For ischemic stroke (large artery atherosclerosis), there were 14 causal relationships as follows: 1-linoleoylglycerophosphoethanolamine* (p = 2.89E-05, OR = 0.182, 95%CI: 0.082–0.405), 1-stearoylglycerophosphoinositol (p = 0.000791,OR = 3.166,95%CI: 1.615–6.206), 2-methoxyacetaminophen sulfate*(p = 3.87E-15,OR = 0.917,95%CI:0.899–0.938),5-oxoproline(p = 0.0179,OR = 0.287,95%CI = 0.1022–0.807),7-alpha-hydroxy-3-oxo-4-cholestenoate (7-Hoca)(p = 0.0205,OR = 3.852, 95%CI: 1.231–12.056), allantoin (p = 0.0274,OR = 0.587, 95%CI:0.366–0.942), arachidonate (20:4n6)(p = 0.0334, OR = 3.699,95%CI = 1.108–12.349),Cholesterol (p = 0.00195, OR = 4.199, 95%CI:1.694–10.409), dimethylarginine (SDMA + ADMA) (p = 0.0285,OR = 3.121, 95%CI: 1.127–8.643), fructose(p = 0.0379, OR = 5.341,95%CI = 1.0977–25.984),hexadecanedioate(p = 0.00945,OR = 0.613,95%CI:0.424–0.887), nonadecanoate (19:0)(p = 0.013,OR = 3.206, 95%CI:1.278–8.041),paraxanthine (p = 0.0249, OR = 0.672, 95%CI: 0.475–0.951),
Pyridoxate(p = 0.0499, OR = 0.550, 95%CI:0.303-1.000), which was shown in Fig. 3.
For ischemic stroke, there were 22 causal relationships as follows: 1-arachidonoylglycerophosphocholine* (p = 0.0160, OR = 1.354, 1.0582–1.734), 1-linoleoylglycerophosphoethanolamine* (p = 0.041, OR = 0.676, 95%CI:0.464–0.984), 1-methylxanthine (p = 0.0157, OR = 0.782, 0.640–0.955), 2-methoxyacetaminophen sulfate* (p = 0.035, OR = 0.991, 95%CI:0.983–0.999), 2-oleoylglycerophosphocholine* (p = 0.0463, OR = 0.679, 0.464–0.994), 5-oxoproline (p = 0.0148, OR = 0.563, 95%CI:0.355–0.893), ADpSGEGDFXAEGGGVR* (p = 0.00409, OR = 1.481, 95%CI:1.133–1.937) arachidonate (20:4n6) (p = 0.00463, OR = 1.721, 95%CI:1.182–2.506), bradykinin, des-arg(9) (p = 0.0212, OR = 1.092, 95%CI:1.013–1.177), cholesterol (p = 0.0165, OR = 1.633, 95%CI:1.0938–2.438), cortisol (p = 0.0358, OR = 0.668,95%CI:0.458–0.974), gamma-glutamylglutamine (p = 0.0173, OR = 0.603, 95%CI:0.398–0.915), gamma-glutamylmethionine* (p = 0.0425, OR = 0.741, 95%CI:0.555–0.990), indolelactate (p = 0.0217, OR = 0.652, 95%CI:0.453–0.939), linolenate [alpha or gamma; (18:3n3 or 6)] (p = 0.0461, OR = 0.648, 0.423–0.993), margarate (17:0) (p = 0.0396, OR = 0.628, 95%CI:0.403–0.978), octadecanedioate (p = 0.00655, OR = 0.677, 95%CI:0.510–0.897), piperine(p = 0.0166, OR = 1.210, 95%CI:1.0353–1.415), pyridoxate (p = 0.0398, OR = 0.774, 95%CI:0.607–0.988), pyroglutamylglycine (p = 0.0281, OR = 1.281, 95%CI:1.027–1.599), taurodeoxycholate (p = 0.0387, OR = 1.123, 95%CI:1.006–1.253), threonate (p = 0.0443, OR = 0.783, 95%CI:0.617–0.994), which was shown in Fig. 4.
For ischemic stroke (small-vessel), there were 14 causal relationships as follows: 1-eicosadienoylglycerophosphocholine*(p = 0.0222, OR = 0.507, 95%CI:0.283–0.907), 1-methylxanthine(p = 0.0112, OR = 0.549, 95%CI: 0.346–0.873,
1-palmitoleoylglycerophosphocholine*(p = 0.00508, OR = 0.289, 95%CI:0.121–0.688),2-methoxyacetaminophen sulfate* (p = 1.0348E-09, OR = 0.938, 95%CI:0.919–0.958), 4-acetaminophen sulfate (p = 0.0426, OR = 0.971, 95%CI: 0.944–0.999),4-androsten-3beta,17beta-diol disulfate 1* (p = 0.000714, OR = 0.633, 95%CI:0.486–0.825), ADpSGEGDFXAEGGGVR* (p = 0.000306, OR = 3.115, 95%CI:1.681–5.772), adrenate (22:4n6) (p = 0.00194, OR = 0.334, 95%CI:0.167–0.668),dihomo-linoleate (20:2n6) (p = 0.0315, OR = 2.539, 95%CI:1.086–5.933), dihomo-linolenate (20:3n3 or n6) (p = 0.000936,OR = 3.438, 95%CI: 1.654–7.145), Ibuprofen(p = 0.0142, OR = 0.913, 95%CI:0.850–0.982), indolepropionate (p = 0.0415, OR = 0.607, 95%CI:0.376–0.981), (p = 0.0172, OR = 1.862, 95%CI:1.117–3.106),kynurenine (p = 0.0350, OR = 2.973, 95%CI:1.080–8.186), linolenate [alpha or gamma; (18:3n3 or 6)](p = 0.0119,OR = 0.276,95%CI:0.101–0.753),N-(2-furoyl)glycine(p = 0.0495,OR = 1.099, 95%CI:1.000218-1.207), ornithine(p = 0.00799, OR = 0.203,95%CI:0.0627-0.660), palmitoleate (16:1n7) (p = 0.0368, OR = 0.457, 95%CI:0.219–0.953), piperin(p = 0.014,OR = 1.567, 95%CI:1.095–2.242),taurodeoxycholate(p = 0.0211,OR = 1.438, 95%CI:1.056–1.958), which was shown in Fig. 5.
For lacunar stroke, there were 18 causal relationships as follows: 1-linoleoylglycerol (1-monolinolein) (p = 0.0106, OR = 0.630, 95%CI:0.442–0.898),1-linoleoylglycerophosphoethanolamine* (p = 0.0239, OR = 2.062, 95%CI:1.1005–3.863),1-methylxanthine (p = 0.0381, OR = 0.645, 95%CI:0.426–0.976), 1-palmitoleoylglycerophosphocholine* (p = 0.0245, OR = 0.323, 95%CI:0.121–0.865), 2-methoxyacetaminophen sulfate* (p = 0.00117, OR = 0.969, 95%CI:0.951–0.988), 3-methyl-2-oxovalerate (p = 0.0301, OR = 2.299, 95%CI:1.083–4.880), 4-androsten-3beta,17beta-diol disulfate 1* (p = 0.0201, OR = 0.717, 95%CI:0.542–0.949), ADpSGEGDFXAEGGGVR* (p = 0.0101, OR = 1.984, 95%CI:1.1771–3.343), ADSGEGDFXAEGGGVR* (p = 0.0243, OR = 1.565, 95%CI:1.060–2.311), aspartate (p = 0.020, OR = 2.981, 95%CI:1.192–7.453), butyrylcarnitine (p = 0.0172, OR = 1.352, 95%CI:1.0549–1.733), glycine (p = 0.0272, OR = 0.594,95%CI:0.374–0.943),glycodeoxycholate (p = 0.0448, OR = 0.782, 95%CI: 0.615–0.994), homostachydrine* (p = 0.0389, OR = 0.533, 95%CI:0.294–0.968), mannose (p = 0.0251, OR = 0.490, 95% CI:0.263–0.915), methionine (p = 0.0470, OR = 0.205, 95% CI:0.0430–0.9790), phenylalanine (p = 0.0168, OR = 30.796, 95%CI:1.853-511.827), trans-4-hydroxyproline (p = 0.0302, OR = 2.194, 95% CI:1.078–4.466), which was shown in Fig. 6. The Venn diagram analysis of metabolites with causal associations with different phenotypes of stroke is shown in Fig. 7.
Sensitive analysis
The results of heterogeneity and pleiotropy testing are shown in Supplementary Table 2.
Horizontal pleiotropy was not observed in the intercept of MR Egger regression in causal relationship between metabolites and ischemic stroke (cardioembolic) (arachidonate (20:4n6), p = 0.255; Cholate, p = 0.902; Cholesterol, p = 0.625; Cortisol, p = 0.911; pelargonate (9:0), p = 0.796; pentadecanoate (15:0), p = 0.881; tryptophan betaine p = 0.880), ischemic stroke (large artery atherosclerosis) (1-linoleoylglycerophosphoethanolamine, p = 0.0692; 1-stearoylglycerophosphoinositol, p = 0.781; 2-methoxyacetaminophen sulfate,p = 0.931; 5-oxoproline ,p = 0.1178; 7-alpha-hydroxy-3-oxo-4-cholestenoate (7-Hoca),p = 0.906;allantoin, p = 0.0538; arachidonate (20:4n6), p = 0.907; cholesterolp = 0.329; dimethylarginine (SDMA + ADMA),p = 0.326;/fructose,p = 0.464; hexadecanedioate, p = 0.276; nonadecanoate (19:0), p = 0.493; paraxanthine, p = 0.837; pyridoxate, p = 0.400), ischemic stroke(1-arachidonoylglycerophosphocholine, p = 0.289; 1-linoleoylglycerophosphoethanolamine, p = 0.485;1-methylxanthine, p = 0.618; 2-methoxyacetaminophen sulfate, p = 0.867; 2-oleoylglycerophosphocholine, p = 0.633; 5-oxoproline, p = 0.376; ADpSGEGDFXAEGGGVR, p = 0.356; arachidonate (20:4n6), p = 0.823; bradykinin, des-arg(9),p = 0.219;cholesterol,p = 0.854;/cortisol,p = 0.769; gamma-glutamylglutamine, p = 0.186; gamma-glutamylmethionine, p = 0.986; indolelactate, p = 0.945; linolenate [alpha or gamma; (18:3n3 or 6)], p = 0.914; margarate (17:0) p = 0.461;octadecanedioate, p = 0.940; piperine, p = 0.962; pyridoxate, p = 0.140; pyroglutamylglycine, p = 0.409; taurodeoxycholate, p = 0.862; threonate, p = 0.224), ischemic stroke (small-vessel)(1-eicosadienoylglycerophosphocholine, p = 0.677; 1-methylxanthine, p = 0.726;1-palmitoleoylglycerophosphocholine, p = 0.187; 2-methoxyacetaminophen sulfate, p = 0.455; 4-acetaminophen sulfate, p = 0.572; 4-androsten-3beta,17beta-diol disulfate 1, p = 0.177;ADpSGEGDFXAEGGGVR,p = 0.516; adrenate (22:4n6), p = 0.427;dihomo-linoleate (20:2n6), p = 0.610;dihomo-linolenate (20:3n3 or n6), p = 0.657;ibuprofen, p = 0.870;indolepropionate, p = 0.951;isobutyrylcarnitine, p = 0.397;kynurenine, p = 0.716;linolenate [alpha or gamma; (18:3n3 or 6)], p = 0.798;N-(2-furoyl)glycine, p = 0.473;Ornithine, p = 0.779;palmitoleate (16:1n7), p = 0.190;piperine, p = 0.823;taurodeoxycholate, p = 0.055), lacunar stroke(1-linoleoylglycerol (1-monolinolein), p = 0.653; 1-linoleoylglycerophosphoethanolamine, p = 0.819; 1-methylxanthine, p = 0.946;1-palmitoleoylglycerophosphocholine, p = 0.197; 2-methoxyacetaminophen sulfate, p = 0.965; 3-methyl-2-oxovalerate, p = 0.182; 4-androsten-3beta,17beta-diol disulfate 1, p = 0.275; ADpSGEGDFXAEGGGVR, p = 0.641; ADSGEGDFXAEGGGVR, p = 0.547; Aspartate, p = 0.334; Butyrylcarnitine, p = 0.416; Glycine, p = 0.137; Glycodeoxycholate, p = 0.988; Homostachydrine, p = 0.811; Mannose, p = 0.075; Methionine, p = 0.254; Phenylalanine, p = 0.967; trans-4-hydroxyproline, p = 0.680).