Genetic overlap between AD and GIT disorders
We first tested for SNP-level genetic overlap between AD and GIT disorders using the SNP effect concordance analysis (SECA) method . Briefly, SECA performs a bi-directional analysis, assessing the effects of AD-associated SNPs (dataset 1) on each of the GIT disorders (dataset 2) and vice versa. We found a significant concordance of SNP risk effects between the AD GWAS and each of the GERD, PUD, PGM, gastritis-duodenitis, IBS and diverticular disease GWAS, indicating that there is a strong genetic overlap between AD and each of these phenotypes. Table 2 summarises the results of the primary test for the concordance of effects in which 144 SNP subsets were tested, with AD as P1 (dataset 1) and GERD as P2 (dataset 2). All these SNP subsets showed significant concordance of effects (Odds ratio [OR] > 1 and P < 0.05) with a permuted P-value (Ppermuted) = 9.99 × 10−4. A total of 26,963 linkage disequilibrium (LD)-independent SNPs (r2 < 0.1) were common to both the AD and GERD GWAS (at P1GWAS-data = P2GWAS-data ≤ 1), 13,955 (52%) of which exhibit significant effect concordance across the two GWAS (OR = 1.18, PFisher’s-exact = 4.65 × 10−11) [Table 2].
As expected, a pattern of increasing strength of association between AD and GERD (measured using the OR values) was observed as the P-values for the SNP subsets (P1 and P2) decrease. For example, at AD (P1GWAS-data < 0.05) and GERD (P2GWAS-data < 0.05), the proportion of SNP effect concordance is 58% (OR = 1.84, PFisher’s-exact = 1.74 × 10−6), increasing to 61% at P1 = P2 < 0.01. In a reverse analysis (GERD as dataset 1 (P1) and AD as dataset 2 (P2)) using SECA, we also found all the 144 subsets of SNPs (OR > 1 and P < 0.05, Ppermuted = 9.99 × 10−4) showing significant concordance of SNP risk effects across the two disorders [Additional file 2: Supplementary Note 1]. These results indicate that AD-associated SNPs are also associated with GERD, and vice versa—supporting evidence of highly significant genetic overlap between the two disorders.
SECA analyses also revealed a similar significant genetic overlap between AD and each of PGM, gastritis-duodenitis, diverticulosis and PUD (Table 3). While there was significant genetic overlap between AD and IBS, the strength of association was comparatively less than for the other GIT disorders (Additional file 2: Supplementary Note 1). For AD and IBD, SECA revealed significant concordance of SNP risk effects when AD was assessed (as dataset 1) against IBD (as dataset 2) [OR > 1 and PGWAS-data < 0.05, Ppermuted = 0.025] but not the other way around (Additional file 2: Supplementary Note 1). The observed significant overlap between AD (dataset 1) and IBD (dataset 2) was much weaker than for the rest of the GIT disorders assessed. Table 3 summarises the results of our SECA-based genetic overlap assessment between AD and GIT traits.
Genetic correlation between AD and GIT disorders
We used the linkage disequilibrium score regression (LDSC) method to further assess and quantify the SNP-level genetic correlation between AD and GIT disorders. The apolipoprotein E (APOE) region has a large effect on the risk of AD; hence, we excluded APOE and the 500 kilobase (kb) flanking region (hg19, 19:44,909,039 – 45,912,650) from our AD GWAS for this analysis. Given the complex LD structure in the human major histocompatibility complex (MHC), we also excluded SNPs in the 26 to 36 megabase region of chromosome six from the data. In analyses both with and without the APOE and MHC regions, LDSC reveals a significant genetic correlation between AD and several of the GIT traits (Fig. 2 and Additional file 1: Table S2).
LDSC reveals a positive and significant genetic correlation (rg) of AD (without APOE and MHC regions) with GERD (rg = 0.19, P = 8.78 × 10-7), PUD (rg = 0.26, P = 2.92 × 10-4), PGM (rg = 0.15, P = 1.43 × 10-4), gastritis-duodenitis (rg = 0.19, P = 5.40 × 10-3), IBS (rg = 0.16, P = 2.36 × 10-2), and diverticular disease (rg = 0.18, P = 1.59 × 10-3). These results (Fig. 2) are all consistent with findings in SECA. Moreover, our results were based on the unconstrained genetic covariance intercept, hence the significance of these estimates may be conservative given the negligible, or complete absence of, sample overlap between the pairs of traits assessed.
Using LDSC, we did not find a significant genetic correlation between AD and IBD (rg = -0.05, P = 3.80 × 10-1) [Fig. 2 and Additional file 1: Table S2], a result that is partially consistent with our SECA findings—highlighting how SECA differs (a bidirectional assessment of the relationships) as well as complements LDSC. Additional file 1: Table S2, comprehensively describes the findings of these analyses. We also performed cross-trait LDSC analysis assessing the relationship between each of the GWAS included in this study (Fig. 3 and Additional file 1: Table S2). Notably, there was no evidence for a significant relationship of IBD with any of the other GIT disorders, except IBS (rg = 0.14, P = 4.41 × 10-3) [Fig. 3 and Additional file 1: Table S2]. Conversely, we found a significant genetic correlation between all the other pairs of GIT disorders (Fig. 3 and Additional file 1: Table S2). It is noteworthy that the GWAS for medication use in PUD and GERD (PGM) was strongly correlated with disorders of the gastric mucosa (PUD [rg = 0.76, P = 4.41 × 10-101], gastritis-duodenitis [rg = 0.76, P = 4.41 × 10-20] and GERD [rg = 0.99, P = 0.000]), supporting its inclusion in the present study (Fig. 3 and Additional file 1: Table S2).
SNPs and loci shared by AD and GIT disorders
Leveraging the significant genetic overlap and correlation as well as the substantial sample sizes of GERD and PUD, we performed cross-disorder meta-analyses of AD with each of these disorders. PGM has a very large number of cases and overall sample size (Table 1) and is strongly correlated with GERD (rg = 0.99, P = 0.000) and PUD (rg = 0.76, P = 4.41 × 10-101) [Fig. 3 and Additional file 1: Table S2], hence, we utilised it in meta-analysis with AD. Our analyses identified shared SNPs and susceptibility loci, some of which are novel for both AD and GIT disorders. The primary objective of this analysis was to identify SNPs and loci which were not genome-wide significant in the individual AD or GIT disorder GWAS (i.e., 5 × 10−8 < PGWAS-data < 0.05) but reached genome-wide significance (Pmeta-analysis < 5 × 10−8) following a meta-analysis (Table 4). We additionally identified SNPs and loci which were already established (genome-wide significant, PGWAS-data < 5 × 10−8) in AD (Sentinel AD SNPs/loci), which were also significantly associated with a GIT disorder, and vice versa, following the GWAS meta-analysis.
AD and PGM GWAS meta-analysis
A total of 42 SNPs reached genome-wide significance (Pmeta-analysis < 5 × 10−8) in the cross-disorder meta-analysis of AD and PGM GWAS (Additional file 1: Table S3). None of these 42 SNPs was genome-wide significant in the individual AD or PGM GWAS (before meta-analysis) [PGWAS-data > 5 × 10−8] but they were at least nominally significant (PGWAS-data < 0.05) in each of the traits (5 × 10−8 < PGWAS-data < 0.05). Of the 42 genome-wide significant SNPs, 11 were independent (at r2 < 0.6), from which we characterised seven lead SNPs at seven genomic loci (r2 < 0.1) [Table 4]. That is, seven independent loci reached genome-wide significance for the AD and PGM. A search in the PhenoScanner  (accessed on 04/05/2021), revealed that one of the 11 independent SNPs, rs11083749 (on chromosome 19q13.32, NECTIN2), has been reported for association with AD at a genome-wide significant level. Our study provides evidence that this SNP and locus are also associated with PGM given the substantial reduction in the GWAS meta-analysis P-value.
Of the remaining nine independent SNPs, at six genomic loci, none was previously found to be associated with AD, GERD, or PUD at a genome-wide level of significance, suggesting them to be novel SNPs and loci for the analysed traits (Table 4). Moreover, the results for m-value posterior probability and the BE P-value indicate that all the identified SNPs and loci, except the 3p21.31 locus (SNPs rs709210 [ HYAL2] and rs7642934 [SEMA3F]), are associated with both AD and PGM. The m-value for each of the remaining SNPs (excluding the 3p21.31 locus) was > 0.90, predicting that they have effects in both GWAS (Table 4). Notably, the 3p21.31 locus (HYAL2 and SEMA3F), was subsequently identified to have effects both in AD and GIT-trait (based on the binary effect P-value results) in the meta-analysis of AD and PUD GWAS (Table 4).
We identified an additional 23 SNPs, at three independent loci (r2 < 0.1), that reached genome-wide suggestive association (Pmeta-analysis < 1 × 10−5) in the meta-analysis of AD and PGM (Additional file 1: Table S4). Of these, the rs33998678 SNP (at 16q22.1, IL34) is in high LD (r2 = 0.91) with one of the genome-wide significant SNP loci (rs34644948, at 16q22.1, MTSS2, Table 4) identified in the meta-analysis of AD and PGM. The finding, thus, supports the involvement of the locus (16q22.1, MTSS2) in both AD and GERD or PUD (traits represented by PGM GWAS). Similarly, the rs663576 (at 17q21.32, PHOSPHO1) is moderately correlated (r2 = 0.41) with another genome-wide significant SNP (rs2584662 at 17q21.33, PHB, Table 4), identified in the meta-analysis. This locus (17q21.33) was reproduced in the meta-analysis of AD and GERD (SNP rs2584662 near PHB), lending support for its involvement in AD and PUD or GERD. Notably, all the three loci reaching suggestive associations were predicted, using m-value and the BE P-value methods, to have effects in both AD and PGM.
AD and GERD GWAS meta-analysis
A meta-analysis of AD and GERD identified a total of 119 SNPs reaching genome-wide significant association (Pmeta-analysis < 5 × 10−8, Additional file 1: Table S5), none of which was previously genome-wide significant in the individual AD or GERD GWAS (5 × 10−8 < PGWAS-data < 0.05). From these, we characterised 11 independent SNPs (r2 < 0.6) and seven lead SNPs (r2 < 0.1) at seven genomic loci (Table 4). The identified loci included those implicated in the meta-analysis of AD and PGM at a genome-wide level of significance (Pmeta-analysis < 5 × 10−8)—1p31.3, 3p21.31, 6p21.32, 17q21.33 and 19q13.32—and at a genome-wide suggestive association level (Pmeta-analysis < 1 × 10-5)—16q22.1 and 1q32.2.
Also, we found one (nearby) locus, 17q21.32 (SNP rs8067459, ZNF652, Table 4) reaching genome-wide significance in the AD and GERD meta-analysis. This locus was genome-wide suggestive in the AD vs PGM GWAS meta-analysis (SNP rs663576, 17q21.32, LD between rs663576 and rs8067459 = 0.86), providing additional evidence for the locus being shared by AD and the GIT disorders (GERD and PUD). An additional 175 independent SNPs at 121 loci reached a genome-wide suggestive association (Pmeta-analysis < 1 × 10-5), reproducing some of the genome-wide significant loci in the AD and PGM or the AD and GERD meta-analysis. The loci include 1p31.3 (rs2840677, PDE4B), 1q31.1 (rs10753964, BRINP3) and 1q32.2 (rs4147104, CD46) [Additional file 1: Table S6). Thus, the results support the loci being shared by AD and GERD. Other SNPs and loci reproduced at the suggestive level of association (or in high LD with identified loci) are highlighted in Additional file 1: Table S6.
AD and PUD GWAS meta-analysis
We identified 22 SNPs reaching genome-wide significance in the meta-analysis of AD and PUD GWAS (Pmeta-analysis < 5 × 10−8, Additional file 1: Table S7). From these, we characterised seven independent SNPs at six genomic loci (Table 4) associated with both AD and PUD. Both the m-value (> 0.90) and the BE methods predict that the identified SNPs and loci have effects in AD and PUD (Table 4). Of the loci identified in the AD and PGM meta-analysis, four were replicated in the AD and PUD meta-analysis. Two of the four loci, the 19q13.32 (rs28363848 near BCL3), and the 6p21.32 (rs9270599, HLA-DRA), were replicated at a genome-wide level of significance, while the remaining two—rs709210, 3p21.31, P(FE) = 5.24 × 10-3, HYAL2; and rs6695557, 1p31.3, P(FE) = 2.94 × 10-4, PDE4B—were replicated at a nominal level (significant reduction in P-value after AD and PUD meta-analysis, Additional file 1: Table S8). The SNP rs530324, at 8p21.1 (SCARA3, Table 4), identified in the AD and PUD meta-analysis, is in strong LD (r2 = 0.91) with another SNP (rs4732732, CLU) which reached a suggestive association for AD and PUD (Additional file 1: Table S9). The finding, thus, provides additional evidence for the involvement of the locus (8p21.1) in both AD and PUD. Additional file 1: Table S9, presents 24 independent SNPs, at 21 genomic loci, reaching genome-wide suggestive association (Pmeta-analysis < 1 × 10-5) for AD and PUD.
Shared genomic regions
Using a colocalization analysis in GWAS-PW , we assessed shared genomic regions between AD and each of PGM and GERD (Additional file 1: Table S10). The results of this analysis confirm that all the loci identified in the meta-analyses (except in chromosome 3) are shared by AD and the respective GIT traits (model 4 posterior probability [PPA 4] > 0.9, Additional file 1: Table S10). While the findings also suggest that the causal variants might be different (in some of the loci—PPA 3 < 0.5), we note that when variants in a locus are in strong LD, which is likely the case here, GWAS-PW is limited in its ability to correctly distinguish model 3 (PPA 3) from model 4 (PPA 4) . Additional shared genomic regions, in chromosomes 1, 6, 16, 17 and 19 having PPA 4 > 0.90 were identified for AD and the GIT traits (Additional file 1: Table S10). Also, we identified another locus on chromosome 17, having PPA 3 > 0.80, and implicating the SNP rs2526380 (17q22, TSPOAP1) in both AD and GERD. The posterior probability that this SNP is a causal variant under model 3  is high at 0.99 (Additional file 1: Table S10).
Results of causal association analysis between AD and GIT disorders
We assessed the causal relationship between AD (as the outcome variable) and GERD (as the exposure variable) using the two-sample Mendelian randomisation (MR) method. We found no evidence of a causal relationship between AD and GERD, irrespective of the direction of the analysis (AD or GERD as the outcome or exposure variable) [Table 5]. For sensitivity testing, we implemented five additional models of MR analysis—MR-Egger, weighted median, simple mode, weighted mode and the MR-PRESSO (Mendelian Randomization Pleiotropy RESidual Sum and Outlier)—since a true finding will be consistent across the multiple methods. Results of all these methods agree with those of the Inverse Variance Weighted (IVW) model supporting a lack of evidence for a causal association between AD and GERD (Table 5 and Additional file 1: Table S11). We carried out further MR analysis assessing AD against each of PUD, PGM, IBS, diverticular disease, and IBD. Findings similarly reveal no evidence for a causal relationship between AD and each of the GIT-disorders assessed (Additional file 1: Table S11, and Additional file 2: Supplementary Note 2).
We also used the Latent Causal Variable (LCV) approach  to test for a causal relationship between AD and each of the GIT disorders. The results of LCV suggest a partial causal influence of gastritis-duodenitis (genetic causal proportion [GCP] = -0.69, P = 0.0026), and lansoprazole (GCP = -0.38, P = 0.001129) on AD, Table 6. The result was in the reverse direction for diverticular disease (GCP = 0.23, P = 0.000272). Using another set of GWAS (Table 6), we applied LCV methods to test the reproducibility of the partial causal association found. None of the partial causal association results was reproduced.
Gene-based association analysis
Using a gene-based analysis of the SNPs that overlapped between the AD and PGM GWAS, we identified a total of 18,763 protein-coding genes for each of the traits. Applying a threshold P-value of 2.66 × 10-6 (0.05/18763—Bonferroni correction for testing 18,763 genes), we identified 64 genome-wide significant (Pgene < 2.66 × 10-6) genes for AD (Additional file 1: Table S12), 75 for PGM (Additional file 1: Table S13), and 44 for GERD (Additional file 1: Table S14). Using the Fisher’s Combined P-value (FCP) method, a total of 44 genome-wide significant (PFCP < 2.66 × 10-6) genes shared by AD and PGM were identified, 11 of which were not previously significant in the individual AD or PGM GWAS (Additional file 1: Table S15). It is noteworthy that some of the identified AD and PGM shared genes are in chromosomal locations found in our meta-analysis, including 1p31.3 (PDE4B), 3p21.31, (SEMA3F, HYAL2), 6p21.32 (HLA-DRA) and 19q13.32 (several apolipoprotein genes). We replicated a similar pattern of findings using the AD and the GERD GWAS (Additional file 1: Table S16).
Biological pathways and mechanisms shared by AD and GIT disorders
To identify significantly enriched biological pathways, mechanisms, and processes for AD, GIT disorders (GERD and PGM having the largest sample size), or their comorbidity, we performed pathway-based functional enrichment analyses in the g: Profiler platform . These analyses enable us to functionally interpret genes overlapping between AD and GIT disorders and can provide biological insight from their commonalities. First, we investigated genes overlapping AD and GERD (at Pgene < 0.05, FCP < 0.02) and identified several biological pathways that were overrepresented (Fig. 4 and Additional file 1: Table S17), implying they have a role in the mechanisms underlying both AD and GERD.
Pathways related to membrane trafficking and metabolism, alteration, lowering or inhibition of lipids were significantly enriched for both traits (Additional file 1: Table S17). These included plasma lipoprotein assembly, remodelling, and clearance (Padjusted = 2.01 × 10-3), cholesterol metabolism (Padjusted = 4.99 × 10-2), plasma lipoprotein assembly (Padjusted = 3.45 × 10-5), and triglyceride-rich plasma lipoprotein particle (Padjusted = 5.23 × 10-9), among others. Also, lipase inhibitors (Padjusted = 6.08 × 10-3) and the statin (3-Hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors) pathway (Padjusted = 3.99 × 10-2) were significantly enriched for AD and GERD (Additional file 1: Table S17).
Pathways related to the immune system were also overrepresented for both AD and GERD as evidenced by the identification of immune or autoimmune-related disorders such as asthma (Padjusted = 3.53 × 10-3), systemic lupus erythematosus (Padjusted = 7.88 × 10-3), and type I diabetes mellitus (Padjusted = 2.47 × 10-2). Other immune-related pathways identified include intestinal immune network for IgA production (Padjusted = 4.07 × 10-2), programmed cell death protein 1 (PD-1) signalling (Padjusted = 5.24 × 10-3), translocation of ZAP-70 to immunological synapse (Padjusted = 2.44 × 10-3), and interferon-gamma signalling pathways (Padjusted = 2.45 × 10-2) [Additional file 1: Table S17].
Following enrichment mapping and auto-annotation, the identified biological pathways were clustered into six themes of biological mechanisms, namely: ‘lipoprotein particle clearance,’ ‘receptor signalling pathway,’ ‘side membrane vesicle and cell adhesion,’ ‘peptide antigen binding,’ ‘intestinal immune network,’ and ‘interferon-gamma signalling’ (Fig. 4). Moreover, a pathway-based analysis using genes that were overlapping between the AD and PGM GWAS (at Pgene < 0.05) replicated some of the pathways identified for AD and GERD, including ‘plasma lipoprotein assembly, remodelling, and clearance’ (Padjusted = 3.01 × 10-4), ‘peptide antigen binding’ (Padjusted = 2.28 × 10-3), and ‘triglyceride-rich plasma lipoprotein particle’ (Padjusted = 6.60 × 10-8) [Additional file 1: Table S18, and Additional file 2: Supplementary Note 3].