Introduction

Gastroesophageal reflux disease (GERD), according to the Montreal classification, is defined as a condition in which reflux of the stomach contents causes troublesome, annoying symptoms and/or complications1. Heartburn, chest pain and regurgitation are typical symptoms of GERD, a broad range of manifestations also have been reported to be associated with GERD, including chronic cough, wheezing, globus sensation, dental erosions, posterior laryngitis and idiopathic pulmonary fibrosis2,3. The proposed pathogenesis of GERD involves multiple factors, such as transient lower oesophageal sphincter relaxations, lower oesophageal sphincter pressure irregularities, hiatus hernias, delayed gastric emptying, and visceral hypersensitivity4. GERD remains widespread in the western world, with its prevalence continuing to rise5. Around 13.3% of people suffer from GERD globally, while the estimated prevalence of GERD in North America is 15.4%, and annual cost related to GERD are estimated at $10 billion in the United States6,7. Increasing age was recognized as a risk factor of GERD6,7,8, however, the genetic mechanisms by which increasing age impacts GERD have not yet been investigated due to various constraints.

Genome-wide association studies (GWAS) can identify genetic associations between genotypes and phenotypes by examining allele frequency variations among ancestrally similar individuals who exhibit phenotypic differences9. Various genetic methods were developed in recent years to explore the genetic connections between different phenotypes. Genetic correlation analysis of different complex traits has been developed through methods like linkage disequilibrium (LD) score regression (LDSC)10and high-definition likelihood inference for genetic correlations (HDL)11. These two methods help elucidate shared genetic architectures and provide insight into the genetic underpinnings of complex phenotypes. Another method, called “pleiotropic analysis under composite null hypothesis (PLACO)”12, which aims to detect pleiotropy at the SNP level using a level-α intersection-union test, was also developed recently. This method is especially valuable for identifying pleiotropic loci that influence multiple traits simultaneously. FUMA is an integrated web-based platform that leverages information from various biological resources to support the functional annotation of GWAS results, prioritize genes, and enable interactive visualization13. Bayesian colocalization analysis14is another important method that can identify shared causal variants of different phenotypes. MAGMA (Multi-marker Analysis of Genomic Annotation)15, employs a multiple regression method to incorporate linkage disequilibrium (LD) between markers and accurately detect multi-marker effects, offering a fast and adaptable tool for gene and gene-set analysis of GWAS genotypes, which is essential for understanding complex gene interactions. All these methods described above make it possible to explore the genetic relationship between increasing age and GERD. Therefore, we performed a genome-wide cross-traits analysis through various statistical genetic approaches sequentially, to investigate the underlying shared genetic etiology between GERD and several age-related phenotypes.

Methods

GWAS summary statistics acquisition and quality control

All of the GWAS statistics of GERD and age-related phenotypes were obtained from publicly available datasets with European ancestry with consent and ethical approval of relevant authorities. No formal ethical approval was obtained for this publicly available GWAS datasets study.

We employed five age-related phenotypes in our study, including frailty index (FI), facial aging (FA), telomere length (TL), longevity, and parental lifespan (PL). GWAS for GERD was extracted from the FinnGen biobank (https://www.finngen.fi/en). The detailed information of these GWAS summary datasets were shown in Table 1.

Table 1 The characteristics of GWAS studies.
Full size table

To ensure the accuracy and reliability of the GWAS summary statistics used in our study, we excluded SNPs in the MHC region (specifically the 25 Mb to 35 Mb interval on chromosome 6). This region is recognized for its high levels of linkage disequilibrium and highly complex gene structure, which could result in false-positive results.

Genetic correlation analysis

We used the LDSC (Linkage disequilibrium score regression) and HDL (high-definition likelihood) analysis to investigate genome-wide genetic correlations between GERD and five age-related phenotypes. LDSC estimates genetic correlations between traits by examining the genetic overlap, adjusting for the linkage disequilibrium structure in the genome. HDL complements LDSC by examining the heritability of traits and their genetic correlations. HDL focuses on estimating the heritable portion of genetic variation in traits. Compared to LDSC, it is reported that HDL reduces the variance of genetic correlation estimates by about 60%, more effectively uncovering the genetic links between complex traits. In the results of LDSC and HDL, the genetic correlation (rg) value quantifies the genetic relationship between two different traits, where a positive rg value suggests that the traits share common genetic factors, while a negative rg value indicates an inverse genetic relationship of these traits. A P-value less than 0.05 indicates statistical significance, and FDR correction was employed to control for multiple testing and further ensure the reliability of our results.

Identification of pleiotropic, genomic risk loci – PLACO, FUMA and Bayesian colocalization

We then employed the pleiotropic analysis under composite null hypothesis (PLACO) to identify potential pleiotropy SNPs of GERD and age-related phenotypes. This method identifies loci that may influence multiple diseases or traits simultaneously, helping to uncover potential common genetic variants of GERD and age-related phenotypes. SNPs recognized by PLACO with P-value less than 5E-08 (the genome-wide significant level) were considered as significant pleiotropic variants. And then, we utilized the Functional Mapping and Annotation tool (FUMA) to explore the genomic regions of these risk variants, and provides functional annotation for the genetic variants, helping to interpret the biological relevance of the identified loci and their connection to GERD and age-related traits. Additional Bayesian colocalization analysis, which can identify shared causal variants of different phenotypes, was performed to confirm shared pleiotropic loci between GERD and age-related phenotypes, a colocalized locus was identified when the posterior probability of H4 (PP.H4) was greater than 0.7.

Identification of pleiotropic gene and additional tissue specific analysis -MAMGA

To explore the shared genetic architecture of the identified genomic risk loci, we mapped the nearest genes based on the results of FUMA. In addition, an adaptable gene and gene-set analysis of GWAS data (the Multi-marker Analysis of GenoMic Annotation, MAGMA) method was employed to determine the biological functions of these identified pleiotropic loci, which helped to reveal biological pathways that may be involved in the genetic overlap between GERD and age-related traits. Specifically, we utilized MAGMA gene analysis to find pleiotropic genes by incorporating LD between markers properly and to detect multi-marker effects (FDR correction). MAGMA gene-set analysis was conducted to explore biological functions of identified lead SNPs, a total of 10,678 gene sets, including curated (c2.all) and GO terms (c5.bp, c5.cc, c5.mf) from MSigDB were ultimately tested16. To avoid false positives, FDR correction was conducted for all tested gene sets. Finally, we conducted a genome-wide tissue-specific enrichment analysis using 54 GTEx tissues17, focusing on the pleiotropic results generated by PLACO. We calculated the average log2-transformed expression of all identified pleiotropic genes across these tissues and examined tissue specificity through differentially expressed genes (DEGs) in each tissue, with both upregulated and downregulated DEGs determined based on the signs of the t-statistics.

Software and packages

The major statistical analysis in our study was performed in R (version 4.2.2). LDSC analysis was conducted with “LDSC” software (v1.0.1)10. PLACO analysis between GERD and age-related phenotypes were performed with “PLACO” package12. Bayesian colocalization analysis14in this study was performed with the “coloc” package (version 5.2.1). Function analysis and MAGMA was performed by FUMA web tool and MAGMA software13,15.

Results

Genetic correlation analysis – LDSC and HDL

We evaluated the genetic correlation between GERD and five age-related phenotypes by the LDSC and HDL methods, and results from both of these two genetic methods were highly consistent. FI was identified to be positively genetical correlated with GERD, while TL, PL and longevity were identified to be negatively genetical correlated with GERD. However, both of the LDSC and HDL analysis found no significant genetic correlation between FA and GERD. It was worth noting that FI, TL, PL and longevity remained significantly genetical correlated with GERD after applying the FDR correction in both of the LDSC and HDL analysis. The results of genetic correlation analysis by LDSC and HDL methods were shown in Table 2 and Supplement table S1-S2. Considering that no genetic correlation was found between FA and GERD, this trait pair was excluded in our further analysis.

Table 2 Genetic correlation between GERD and age-related phenotypes.
Full size table

Identification of pleiotropic and genomic risk loci between GERD and age-related phenotypes

Based on the shared genetic associations identified by LDSC and HDL, we then performed the novel pleiotropy analyses (PLACO) to find potential pleiotropic loci between GERD and four age-related phenotypes (FI, TL, PL and longevity). Due to the identified positively genetic correlation between GERD and FI, 73 SNPs (T.placo > 0, p.placo < 5E-8) were identified as pleiotropic variants between GERD and FI by PLACO (supplement Table S3). A total of 148 pleiotropic SNPs (T.placo < 0,p.placo < 5E-8) were identified between GERD and TL (supplement Table S4), considering the negatively genetic correlation identified by LDSC and HDL. Only 1 pleiotropic SNPs were found between GERD and PL (supplement Table S5), while no pleiotropic SNPs were found between GERD and longevity.

Based on the PLACO results described above, we then performed FUMA annotation to identify genomic risk loci between GERD and age-related phenotypes. A total of.

23 genomic risk loci were identified by FUMA (supplement Table S6). Additional Colocalization analysis based on these 23 pleiotropic genomic risk loci found 4 colocalized loci with PP.H4 greater than 0.7 (Table 3 and supplement Table S7).

Table 3 Four colocalized loci in colocalization analysis based on 23 pleiotropic genomic risk loci identified by FUMA (PP.H4 > 0.7).
Full size table

Pleiotropic gene and additional tissue specific analysis –MAGMA

Using the identify pleiotropic loci of PLACO and FUMA, we mapped a total of 220 pleiotropic gene by MAGMA (FDR correlation, Figs. 1, 2 and 3 and supplement table S8-S10), specifically 66 pleiotropic genes for GERD-FI trait pair, 3 pleiotropic genes for GERD-PL and 151 pleiotropic genes for GERD-TL. Further functional enrichment analysis identified that these pleiotropic genes were involved in several typical senescence-related biological pathways, such as g1 to s cell cycle control (supplement Fig. 1 and supplement table S11). Additional tissue specific analysis based on pleiotropic gene identified by MAGMA between GERD and FI found that these genes were enriched into some structures of the brain, like Brain Cortex, Brain Anterior cingulate cortex and Brain Frontal Cortex (FDR correction, Fig. 4 and supplement table S12). However, no specific tissues were enriched by MAGMA analysis between GERD and TL (as well as PL) after applying the FDR correction (supplement table S12).

Fig. 1
figure 1

Pleiotropic genes identified by MAGMA between GERD and FI (geneManhattan plot). Note: The blue line indicates a significance threshold of 0.05, while the red line represents the adjusted P-value significance level after FDR correction.

Full size image
Fig. 2
figure 2

Pleiotropic genes identified by MAGMA between GERD and PL (geneManhattan plot). Note: The blue line indicates a significance threshold of 0.05, while the red line represents the adjusted P-value significance level after FDR correction.

Full size image
Fig. 3
figure 3

Pleiotropic genes identified by MAGMA between GERD and TL (geneManhattan plot). Note: The blue line indicates a significance threshold of 0.05, while the red line represents the adjusted P-value significance level after FDR correction.

Full size image
Fig. 4
figure 4

MAGMA tissue specific analysis between GERD and FI, PL, TL. Note: The blue line indicates a significance threshold of 0.05, while the red line represents the adjusted P-value significance level after FDR correction.

Full size image

Discussion

Aging is an intricate biological process, accompanied by the development of age-related diseases and events inevitably. A HUNT study found that increasing age was positively associated with the development of new-onset GERD, with an odds ratio of 1.01 per year (95% CI 1.00–1.02)7. In addition, a meta-analysis also identified that the prevalence of GERD was significantly higher in individuals aged 50 years and older (OR 1.32; 95% CI 1.12–1.54)6. These findings suggest that increasing age is a notable risk factor for the development of GERD. To the best of our knowledge, no studies have investigated the genetic association between age and GERD yet. Hence, we employed comprehensive genetic methodologies sequentially to explore the shared genetic mechanisms between GERD and several age-related phenotypes. Strong genetic correlations were observed between GERD and FI, TL, as well as PL. No significant genetic correlation was identified between FA and GERD, which may be attributed to the complex interplay of biological, environmental, and individual factors (such as, gender differences in wrinkle formation, lifestyle habits, and genetic variability in skin properties) influencing facial aging, these factors could render facial aging an unreliable indicator of actual biological age18. Furthermore, a set of pleiotropic loci and genes between age-related phenotypes and GERD were identified by PLACO, FUMA, Bayesian colocalization and additional MAGMA analysis, providing novel insights into the genetic interplay between GERD and aging-related traits, and paving the way for future studies to develop targeted interventions or therapeutic strategies for GERD in the context of aging.

Frailty is a clinical phenotype that associated with aging closely, characterized by gradual multisystem deterioration, heightened susceptibility to stressors, and increased unfavorable health outcomes19. Previous studies indicate that frailty increases with age dramatically, which subsequently elevates the risk of developing mental disorders, cardiovascular diseases, and cancer20,21,22. However, the genetic association between frailty and GERD remain unclear. Our results provide new evidence that frailty was significantly genetic association with GERD by LDSC and HDL analysis. Previous hospital-based study found that a higher Frequency Scale for the Symptoms of Gastroesophageal Reflux Disease (FSSG) score was significantly associated with frailty in elderly patients23. Another retrospective study pointed out that frailty in elderly patients is associated with decreased lower esophageal sphincter (LES) pressure23, which one of the proposed pathogenesis mechanisms of GERD. In addition, tissue specific analysis based on pleiotropic gene identified by MAGMA between GERD and FI found that these pleiotropic genes were enriched into some structures of the brain, including Brain Cortex, Brain Anterior cingulate cortex and Brain Frontal Cortex, even if the FDR correction was performed. The central nervous system is the most vulnerable component affected by aging and frailty, leading to cognitive decline and other associated symptoms24. GERD can also influence the cerebral cortex architecture via the brain-gut axis, potentially increasing the risks of cognitive impairment25. This study brings a new insight that the formation of GERD may also be interrelated with the aging of the central nervous system. Further analysis (PLACO, FUMA and Bayesian colocalization) identified rs589292 (whose nearest gene was SCAI) and rs4884400 (whose nearest gene was OLFM4) were pleiotropic and genomic risk loci of these two diseases. In our study, SCAI and OLFM4 may act as key pleiotropic variants between GERD and frailty, SCAI was reported to be associated with neuropsychiatric disorders, such as migraines and depression in several studies26,27,28. Similarly, OLFM4 was reported to be associated with major depressive disorder and insomnia29,30. Hence, these two genes may also be involved in the shared genetic architecture of GERD and frailty via the brain-gut axis, however, the specific mechanism of SCAI (or OLFM4) that influence the development of frailty and GERD remains elucidate.

Telomeres are complexes composed of DNA and protein that can protect the ends of chromosomes, however, processes that degrade telomeric DNA can impair telomere function and cause genomic instability31. Telomeres would shorten with each round of DNA replication during the organism’s aging process32. Therefore, telomere length (TL) has been identified as a pivotal indicator of biological aging, cellular senescence, and various disease progression33,34,35. Recent Mendelian randomization study suggests that leukocyte telomere length (LTL) may have a potential causal relationship with GERD36, but more comprehensive study about genetic association between GERD and TL have not reported yet. In this study, we identified the genetic association between GERD and TL, and further pleiotropic loci analysis found that rs13012094 (whose nearest gene was FAM49A) was a pleiotropic genomic risk loci of GERD and TL. Few studies about the function of FAM49A were reported, further research is needed to elucidate the mechanisms by which the FAM49A gene influences GERD, as well as TL.

Our gene-wide association study also has several limitations. Firstly, similar to other GWAS analysis, this research depended on summary-level data, yet individual-level datasets were unavailable. Further population stratification by age or gender could not be performed in this study. Future studies aim to access and analyze individual-level datasets to enable more granular analyses, including age- and gender-specific stratifications. Secondly, the accuracy of gene-set analysis by MAGMA may be affected by the insufficient number of pleiotropic genes. To further improve the reliability of gene-set analyses using MAGMA, future research focusing on leveraging advanced data imputation techniques or integrating cross-population datasets, may increase the effective sample size of pleiotropic genes and provide a more comprehensive genetic landscape, enhancing the precision of the analyses. Thirdly, our study was limited to individuals of the European ancestry, which may have an impact on its applicability to other populations (e.g., the Asian and African populations). Including the Asian and African cohorts in future studies could broaden the scope of our findings, improving their applicability across different ethnic groups.

Conclusion

Our research provided strong evidence of genetic correlation between GERD and age-related phenotypes, especially frailty index and telomere length. In addition, we identified a set of pleiotropic genetic loci and genes of GERD and age-related phenotypes, as well as their potential shared biological mechanisms.