Evidence for causality between BMI and IS

We performed a bidirectional MR analysis using the identified loci that demonstrated a significant association with the single phenotype GWAS of IS or BMI as IVs. All SNPs used in the MR analysis exhibited a high instrument strength, with an F-statistic greater than 10. We found evidence to support the causality of BMI on IS in four methods (MR Egger β = 0.21, se = 0.097, p = 0.034; weighted median β = 0.18, se = 0.06, p = 0.0027; IVW β = 0.16, se = 0.037, p = 1.63E−05), but those results with significant heterogeneity (IVW Q = 658.67, p = 4.29E−07). Egger intercept was not different from 0, which suggested that the IVs were not pleiotropic. The leave-one-out analysis showed that no SNP was driving the effect. We also found that BMI may increase the risk of LVS (MR Egger β = 0.477, se = 0.24, p = 0.046; weighted median β = 0.47, se = 0.14, p = 0.0008; IVW β = 0.29, se = 0.09, p = 0.001, Fig. 2) but with significant heterogeneity (IVW Q = 641.24, p =  5.45E−06, Figs. S1 and S2). In other words, the causality of IS and SVS with BMI was not stable. However, we did not observe the casual effect of BMI on SVS and CES. Furthermore, IS and its subtypes did not affect BMI supported by all methods in the reverse analyses (Fig. 2).

Fig. 2

Bi-directional Mendelian Randomization (MR) analyses between IS and BMI. A Causal effect of BMI on IS; B Causal effect of IS on BMI

Genetic correlations

LDSC was employed to calculate the liability-scale SNP heritability for BMI and IS and its subtypes. The liability-scale SNP heritability for BMI, IS, SVS, LVS, and CES were 21.2, 0.46, 44.46, 0.76, and 0.87%, respectively. We found that BMI was significantly genetic associated with IS (rg = 0.187, p<0.001) and SVS (rg = 0.281, p<0.001), but no genetic correlation observed between BMI and LVS, and CES, respectively (Table 1). Furthermore, HDL revealed that a significant genetic correlation between BMI and IS (rg = 0.243, p<0.001), and SVS (rg = 0.421, p = 0.0164), respectively.

Table 1 The global genetic correlation between BMI and ISLocal genetic correlations

The single GWAS was partitioned into 2495 loci to conduct the LAVA analysis, which aimed to explore the genetic correlation between IS and BMI. The threshold equals p = 0.05/2495, which is equivalent to 2.00E−5. Strong local correlations were found in two loci (chr22 27192924–27952441, p = 4.37E−06; chr18 6756497–7862479, p = 1.45E−05) for SVS with BMI (Table S1). However, we did not identify the local genetic association between BMI and IS because of a lower liability-scale SNP heritability of IS.

Identification of shared SNPs by cross trait meta-analysis

We then employed the CPASSOC and set a threshold of p < 5E−8 for meta-analysis to screen a pleiotropic SNP. A total of 36334 overlap SNPs were identified among IS and BMI (Table S2 and Fig. 3), 41441 were observed for SVS and BMI (Table S3). The SNP with the most significant statistic between BMI and IS, and BMI and SVS were rs7206790 (p = 1.165E−312) and rs7235626 (p = 3.26e−99). There were 9 novel SNPs shared between IS and BMI (pBMI and pIS<5E−6 and pCPASSOC <5E−8, Table S4) and 16 novel SNPs for SVS and BMI after excluding SNPs (Table S5) that were significant in the single-trait GWAS of IS or SVS, or BMI, or were in LD (LD r2 ≥ 0.02) with any of previously reported significant SNPs. We then employed FUMA platform to calculate the annotation information of the shared SNPs between two traits, respectively. A total of 784 genomic locus with 4529 independent significant SNPs (Table S6) and 1584 lead SNPs (Table S6) were identified between IS and BMI. There were 2797 genes were mapped according to the results of CPASSOC for IS and BMI (Table S6). However, there were 337 genomic locus with 1457 independent significant SNPs and 505 lead SNPs were obtained for SVS and BMI (Table S7). A total of 1337 genes were mapped based on the results of CPASSOC (Table S7).

Fig. 3

Manhattan plot showed the pleiotropy SNPs and genes revealed by CPASSOC (A) and MAGMA (B)

Finally, we also used MAGMA to map the common SNPs between IS and BMI. There were 18,483 and 18,689 genes were identified for BMI and IS, BMI and SVS, respectively (Table S8 and S9, Fig. 3). The intersect genes with Bonferroni correction mapped via MAGMT and FUMA were used for tissue and functional enrichment analysis (Table S10 and S11).

Colocalization analysis

We conducted colocalization analysis to confirm if the genomic region has pleiotropic effects on the loci annotated on FUMA. The results indicated that 4 loci (including 570, 42, 75 and 263) were associated with IS with a PPH4 value exceeding 70%. The significant SNPs included rs11066301, rs11066028, rs233721, rs2891403, rs16864515, rs12759907, rs235509, rs6893539, rs11241696, and rs7711753, which were used to map 17 genes including SH2B, ATXN2, BRAP, ACAD10, RP11-162P23.2, ALDH2, MAPKAPK5, TMEM116, ERP29, NAA25, TRAFD1, HECTD4, RPL6, PTPN11, RPH3A, PRRC2C and CEP120 (Table S12). However, no genomic region has pleiotropic effects on the loci for SVS (Table S13).

Tissue and functional level SNP heritability enrichment

The intersect genes with Bonferroni correction mapped via MAGMA and FUMA were used for tissue and functional enrichment analysis. MAGMA was employed to explore the tissue-level SNP heritability enrichment for IS and BMI, using GTEx with different tissues. We discovered that the brain tissues exhibited shared significant SNP-heritability enrichment for both IS and BMI (Table S14 and Fig. 4). In details, we found that SNPs associated with BMI and IS were enriched in 18 different brain regions, leading by frontal cortex, anterior cingulate cortex, cerebellum, and amygdala. However, the shared SNPs among BMI and SVS were only enriched in cerebellum (Table S14 and Fig. 4).

Fig. 4

MAGMA-based heritability enrichment estimates in tissues for BMI and IS. A IS, B SVS

Furthermore, we found that the common SNPs between BMI and IS were mainly enriched in the biological processes such as brain development and synaptic electrical activity revealed by GO analysis (Fig. 5 and Table S15). Results of KEEG suggested that those SNPs were enriched in biological pathway such as axon guidance, neurotrophin signaling pathway, and neurosynaptic potential regulation (Fig. 5 and Table S16). Moreover, the genes annotated by the shared SNPs between BMI and SVS were primarily enriched in immunoregulation such as T cell differentiation and mononuclear cell differentiation revealed by GO analysis (Table S15). Those genes were enriched in the signaling pathway involved with immunoregulation demonstrated by KEEG analysis (Table S16).

Fig. 5

Functional enrichment for pleiotropy genes between IS and BMI. A GO analysis, B KEEG analysis

Identification of shared functional genes for IS and BMI

In order to infer causality and identify the putative functional genes for BMI and IS, we analyzed GWAS summary data from GTEx using SMR based on the overlapped genes from results of FUMA and MAGMA. A total of 6, 2, 15, 4, 9, 25, 33, 24, 18, 11, 13, 15, 13 and 6 shared genes associated with BMI and IS in aorta artery, coronary artery, caudate basal ganglia, amygdala, anterior cingulate cortex BA24, cerebellar hemisphere, cerebellum, cortex, frontal cortex BA9, hippocampus, hypothalamus, nucleus accumbens basal ganglia, putamen basal ganglia, and substantia nigra, respectively. Among those genes, CHAF1A, CEP192, ULK4, and WARS2 were found in most of the 14 shared enriched tissues. Regarding the common genes between SVS and BMI, we identified 1(WNT3), 1(WNT3), 2(CYP2D6 and FUT2), 3(AS3MT, CYP2D6 and LRRFIP2), 2(AS3MT and CYP2D6), 3(ADAMTSL3, CISD2, and TYW5), 5(ADAMTSL3, CISD2, FAM212A, MANBA, and RNF123), 5(ADAMTSL3, AS3MT, CYP2D6, LRRFIP2 and NAGA), 2(AS3MT and CYP2D6), 2(AS3MT and CYP2D6), 4(CISD2, CYP2D6, MAMSTR and NAGA), 5(AS3MT, CYP2D6, FUT2, LRRFIP2 and MAMSTR), 4(CYP2D6, FUT2, LRRFIP2, and MAMSTR), and 1(CYP2D6) shared genes in aorta artery, coronary artery, caudate basal ganglia, amygdala, anterior cingulate cortex BA24, cerebellar hemisphere, cerebellum, cortex, frontal cortex BA9, hippocampus, hypothalamus, nucleus accumbens basal ganglia, putamen basal ganglia, and substantia nigra, respectively. However, only CYP2D6 and AS3MT were found in most of the 14 shared enriched tissues (Table S17).