The Effects of Physical Activity on Functional Dyspepsia: A Mendelian Randomization and Randomized Controlled Study

Abstract

Background/Aims

The relationship between physical activity (PA) and functional dyspepsia (FD) remains unclear due to a lack of robust evidence. We conducted a Mendelian randomization (MR) analysis to assess potential genetic associations, followed by a randomized controlled trial (RCT) assessing the therapeutic effects of moderate-intensity exercise in FD patients.

Methods

In the MR analysis, genetic variants associated with PA traits from genome-wide association studies were analyzed. For the RCT, 260 FD patients were randomized to moderate-intensity exercise intervention or a control group. The primary endpoint was the adequate relief rate of dyspeptic symptoms.

Results

Using MR, a potential inverse association of strenuous sports activity against the development of FD was estimated (inverse variance weighte OR = 0.92, 95% CI = 0.88-0.98, P = 0.0047). Additionally, a suggestive inverse association was observed between other exercises (eg, swimming, cycling, keep fit, and fishing) (inverse variance weighte OR = 0.964; 95% CI, 0.937-0.993; P = 0.013) and FD risk. In the previous high-quality study, other exercises (eg, swimming, cycling, keep fit, and fishing) were considered moderate intensity exercise. Therefore, the MR study suggest that moderate intensity exercise may reduce the occurrence of FD. In the subsequent RCT, 114 participants in the exercise intervention group and 113 in the control group completed the study. The exercise group demonstrated a significantly higher adequate relief rate (46.9% vs 30.8%, P = 0.008) and greater reductions in symptom scores (all P < 0.05) versus controls.

Conclusions

PA, particularly moderate-intensity exercise, may reduce the risk of FD.

INTRODUCTION

Functional dyspepsia (FD) is a prevalent functional gastrointestinal disorder characterized by chronic or recurrent epigastric pain or burning, postprandial fullness, and early satiety.¹ With a global prevalence of approximately 16%,² FD imposes a substantial burden on healthcare systems and significantly impairs patients’ quality of life. The chronic and relapsing nature of the disease leads to persistent health deterioration, prompting over 40% of affected individuals to seek hospital-based care, resulting in considerable socioeconomic costs.³ Current therapeutic approaches for FD demonstrate limited efficacy, primarily offering only short-term symptomatic relief,⁴ underscoring the need for more effective and sustainable treatment strategies with minimal adverse effects.

Emerging evidence suggests that physical activity (PA) confers systemic health benefits, including positive effects on cardiovascular, neurological, and respiratory systems. Previous studies have demonstrated that regular exercise can enhance gastrointestinal motility and alleviate abdominal distension in healthy individuals, 5 while improving bowel function and colonic transit time in patients with chronic constipation.⁶ However, the potential therapeutic role of PA in FD remains underexplored. To date, only one small-scale randomized controlled trial (n = 72) conducted in India has evaluated exercise interventions for FD based on Rome IV criteria, demonstrating positive effects of aerobic exercise combined with conventional pharmacotherapy.⁷ Given the study’s limited sample size and short intervention period, further investigation is warranted to establish the efficacy of PA in FD management.

In this study, we employed a comprehensive approach to investigate the relationship between PA and FD risk. First, we performed Mendelian randomization (MR) analysis in a European population to examine potential genetic associations between PA levels and FD susceptibility. Subsequently, we conducted a randomized controlled trial to evaluate the therapeutic effects of moderate-intensity exercise on FD symptoms. We hypothesized that both genetically predicted PA levels and structured exercise interventions would demonstrate a protective association against FD development in the general population.

MATERIALS AND METHODS

Patients and Methods

Mendelian Randomization Study

Study design

The instrumental variables needed to meet the 3 assumptions: (1) genetic variants are significantly associated with physical activity, (2) they are not affected by potential confounding factors, and (3) they influence FD only through physical activity.

Genome-wide association study summary data for physical activity and functional dyspepsia

We drew summary data relative to PA from genome-wide association studies (GWAS) provided by Medical Research Council Integrative Epidemiology Unit (MRC-IEU) Consortium (n = 460 376). The genetic and phenotypic data used in this GWAS were obtained from the UK Biobank, a large-scale, population-based prospective study designed to investigate the genetic and environmental determinants of chronic diseases.⁸ This resource includes extensive health and lifestyle data from over 500 000 participants, along with genomic information, enabling comprehensive analyses of complex traits. We examined 6 self-reported physical activity phenotypes, assessed via UK Biobank questionnaires. These included: (1) Heavy do-it-yourself (DIY) activities (eg, weeding, lawn mowing, carpentry, and digging, ID: ukb-b-13184); (2) Light DIY activities (eg, pruning and watering the lawn, ID: ukb-b-11495); (3) None of the heavy DIY or light DIY activities (ID: ukb-b-15869); (4) Strenuous sports (ID: ukb-b-7663); (5) Walking for pleasure (ID: ukb-b-7337); and (6) Other exercises (eg: swimming, cycling, keep fit, and bowling, ID: ukb-b-8764). The physical activity assessment was based on participant self-reports regarding their engagement in these activities over the past 4 weeks. This approach aligns with validated physical activity measurement tools.⁹ The GWAS data for FD were obtained from the MRC-IEU Consortium (n = 463 010; ID: ukb-b-14814), with cases identified through ICD-10 code K30. All participants included in the MR analysis were of European ancestry. It should be noted that both the physical activity GWAS and the FD GWAS were derived from the UK Biobank cohort. This may lead to sample overlap, which can potentially introduce bias into MR estimates. The original GWAS was approved by relevant ethics committees, and informed consent was provided by all subjects (Supplementary Table).

Availability of data and materials

The data in the MR analysis could be attained from GWAS provided by MRC-IEU Consortium (Heavy DIY ID: ukb-b-13184; Light DIY ID: ukb-b-11495; None of the Heavy DIY or Light DIY ID: ukb-b-15869; Strenuous sports ID: ukb-b-7663; Walking for pleasure ID: ukb-b-7337; Other exercises eg: swimming, cycling, keep fit, bowling ID: ukb-b-8764; FD ID: ukb-b-14814). Data from the randomized controlled study section are available from the corresponding author on reasonable request.

Selection of genetic instruments

To construct the genetic instrument for physical activity, we selected single-nucleotide polymorphisms (SNPs) at the genome-wide significance level (P ≤ 5 × 10^-8). Independent SNPs were identified by applying a stringent linkage disequilibrium threshold (r² < 0.001) within a 10 000 kb sliding window. Palindromic SNPs with ambiguous strand orientation were excluded to ensure robustness. The genetic instruments need to meet the required threshold of F > 10 to ensure the powerful validity and the F statistic was calculated using the formula F = (β/SE).^2,10

Mendelian randomization analyses

The inverse variance weighted (IVW) method served as the primary analytical approach. Heterogeneity was assessed using Cochran’s Q test, with a significance threshold of P < 0.05. In the absence of heterogeneity, a fixed-effects model was applied; otherwise, a random-effects model was employed.¹¹ Complementary analyses included MR-Egger regression, weighted median, weighted mode, and simple mode methods. A significant association was defined as P < 0.008 (0.05/6 exposures), while associations with P-values between 0.008 and 0.05 were considered suggestive.¹² A positive causal association was inferred if the IVW results were statistically significant (P < 0.008), even if other methods did not reach significance, provided that no substantial pleiotropy or heterogeneity was detected and the effect direction OR was consistent across all methods.^11,13 If the OR value directions of these methods were inconsistent, a stricter threshold for the P-value ≤ 1×10^-8 of the instrumental variables was set.^13,14 Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) was used to detect the existence of outliers, and the intercept obtained by MR-Egger regression determines whether there was pleiotropy, P < 0.05 was considered significant. The leave-one-out analysis was used to evaluate the stability of MR Results.¹² All the analyses were performed by R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria).

Randomized Controlled Study

Study design

We conducted a single-center, randomized controlled trial involving outpatients diagnosed with FD. Eligible participants were enrolled between August 2020 and December 2021. Following screening based on predefined inclusion and exclusion criteria, patients were randomly assigned (1:1) to either the exercise intervention group or the control group using a block randomization method (block size = 4). Allocation concealment was ensured through sealed opaque envelopes. Written informed consent was obtained from all participants prior to enrollment.

Following randomization, participants in the exercise group received weekly telephone consultations from an exercise specialist to guide them in performing moderate-intensity exercise and to monitor adherence. In line with the methodology of Johannesson et al,¹⁵ the control group also received weekly phone calls to encourage maintenance of their usual lifestyle. Neither group received any pharmacological intervention for FD. To enhance study retention, control group participants were offered post-study exercise counseling and individualized exercise prescriptions. Outcome assessors and data analysts were blinded to group allocation. The intervention duration was 3 months. Since the MR study data we used were from public databases, no additional ethical approval from the institutional review board was required. The randomized controlled study was approved by the Ethics Committee of the Second Affiliated Hospital of Xi’an Jiaotong University (Approval No. 2020031) and registered at ClinicalTrials.gov (NCT04540549; registered 29 August 2020), and all participants provided written informed consent.

Inclusion criteria

Eligible participants met the following criteria: (1) age ≥ 18 years old; (2) diagnosed with FD according to the Rome IV criteria (Supplementary Table); (3) normal findings on gastroscopy, complete blood count, liver function tests, and Helicobacter pylori testing within the past year; (4) no use of prokinetic agents, proton pump inhibitors, or mucoprotective drugs within the preceding 2 weeks; (5) provided written informed consent and agreed to participate in the study; and (6) capable of engaging in moderate-intensity exercise.

Exclusion criteria

Participants were excluded if they met any of the following criteria: (1) There were organic diseases that may lead to dyspeptic symptoms, such as digestive tract ulcers and tumors, pre-existing hepatobiliary, pancreatic, malignant, or metabolic diseases¹⁶; (2) current pregnancy, lactation, or perioperative pregnancy planning; (3) history of abdominal surgery; (4) psychiatric or psychological disorders; (5) severe cardiac, hepatic, renal or respiratory dysfunction; (6) recent use of anxiolytics, antidepressants, corticosteroids, or nonsteroidal anti-inflammatory drugs (NSAIDs); and (7) predominant symptoms suggestive of gastroesophageal reflux disease (GERD).¹⁷

Data collection

(1) Demographic and anthropometric characteristics: age, height, weight, sex, marriage, education level; (2) lifestyle factors: smoking and drinking; (3) symptom assessment questionnaire: weekly overall dyspeptic symptoms assessment,¹⁸ 8 dyspeptic symptom score questionnaire, short-form leeds dyspepsia questionnaire (SFLDQ), short-form nepean dyspepsia index questionnaire (SFNDI), hospital anxiety and depression scale (HADS); and (4) safety and follow-up: adverse events and loss to follow-up.

Moderate-intensity exercise

According to the American College of Sports Medicine (ACSM), moderate-intensity exercise is recommended for maintaining cardiopulmonary health.¹⁹ Exercise intensity was assessed using the Borg Rating of Perceived Exertion (RPE) scale (6-20), with moderate intensity defined as fairly light (RPE 11-12) to somewhat hard (RPE 13-14).²⁰ The prescribed exercise regimen included: frequency: ≥ 5 sessions per week; duration: 30-60 minutes per session¹⁹; type: jogging or cycling; and intervention period: 3 months.

Questionnaires

Eight-item dyspepsia symptom score questionnaire. Eight-item dyspepsia symptom score questionnaire was used to assess the severity of dyspeptic symptoms, including epigastric pain, burning, postprandial fullness, early satiety, belching, bloating, nausea, and vomiting, with each symptom scoring range of 0-3 (0 = absent; 1 = mild; 2 = moderate; 3 = severe and interfering with daily activities).^21,22

Short-form leeds dyspepsia questionnaire. Comprising 8 items, the SF-LDQ is designed to assess dyspeptic symptoms. The tool measures the frequency and severity of four manifestations: indigestion, heartburn, regurgitation, and nausea. The SF-LDQ scores ranged from 0 to 32, with higher scores corresponding to more severe dyspepsia. A score of 0 was defined as no dyspepsia, a score of 1 to 4 was defined as very mild dyspepsia, a score of 5 to 8 was defined as mild dyspepsia, a score of 9 to 15 was defined as moderate dyspepsia, and a score of 15 or above was defined as severe dyspepsia.^23,24

Short-form nepean dyspepsia index questionnaire. SF-NDI was a 10-item questionnaire, including 5 subscales: tension/anxiety, interference with daily activities, eat/drink, knowledge/control, and work and study. Each subscale was composed of 2 items designed to evaluate the effect of dyspepsia on different aspects of patients’ quality of life. Every item was rated on a scale from 1 (indicating no impact or not applicable) to 5 (representing severe impact). The scores for all items within a subscale were summed, resulting in a total SF-NDI score that could vary from a minimum of 10 points to a maximum of 100 points. A larger total score indicated a worse living standard.²⁵

Hospital anxiety and depression scale. HADS included 2 subscales: hospital anxiety and hospital depression. Each subscale consisted of 7 items, totaling 14 items. Each item was scored from 0 to 3, with higher scores indicating higher levels of anxiety or depression. The total score of the 2 subscales of anxiety and depression was 21 points each, where 0-7 means no anxiety or depression, 8-10 means mild anxiety or depression, and 11-14 means moderate anxiety or depression, with a score of 15 to 21 indicating severe anxiety or depression.¹⁵

Study endpoints

The primary endpoint was the adequate relief rate of dyspeptic symptoms. This was assessed weekly by asking patients: “Over the past 7 days, have you had adequate relief of your dyspepsia symptoms?” with a binary (Yes/No) response. After a 3-month follow-up, adequate relief was defined as a “Yes” response in at least 50% of the weekly assessments between weeks 3 and 12 (total of 10 weeks, excluding the initial two weeks to ensure treatment stability).¹⁸ Secondary endpoints included changes in the 8-item dyspepsia symptom score, SF-LDQ score, SF-NDI score, and the HADS score.

Sample size calculation

The sample size was determined based on the anticipated adequate relief rate of dyspeptic symptoms, using a superiority/non-inferiority clinical trial design with reference to previous studies.¹⁸ Assuming an adequate relief rate of 40% in the control group (P0), we hypothesized a 20% improvement (absolute increase) in the intervention group (P1). With a two-sided significance level (α) of 0.05 and a power (1-β) of 90% (β = 0.1), the required sample size per group was calculated using the following formula: N = (Uα+Uβ)²2P (1-P)/(P1-P0)², P = (P0+P1)/2. The initial calculation yielded 108 participants per group. Accounting for an estimated 20% dropout rate, the total required sample size was 259 (rounded to 260 for balanced allocation). Thus, this study enrolled 260 participants, with 130 randomized to the moderate-intensity exercise group and 130 to the control group.

Statistical methods

All statistical analyses were performed using EmpowerStats and SPSS 20.0. Categorical variables were expressed as frequencies and percentages, and group comparisons were conducted using the chi-square test. When the expected cell count was less than 10, Fisher’s exact test was applied instead. Continuous variables were presented as mean ± SD, and differences between groups were assessed using Student’s t test or nonparametric tests, as appropriate. All analyses followed the intention-to-treat principle to minimize bias. The last observation carried forward method was applied to handle missing values. Graphical representations were generated using GraphPad Prism 8.0. A two-sided P-value < 0.05 was considered statistically significant.

RESULTS

Mendelian Randomization Study

Study design

In the MR analysis, we included 7662 FD cases and 455,348 normal controls.

After applying stringent SNP selection criteria, we identified the following instrument variables: Heavy DIY activity: 18 SNPs (F-statistic range: 30.6-51.7); Light DIY: 12 SNPs (F-statistic range: 29.8-84.9); No DIY (neither heavy nor light): 5 SNPs (F-statistic range: 30.9-39.3); Strenuous sports: 6 SNPs (F-statistic range: 31.9-49.2); Walking for pleasure: 20 SNPs (F-statistic range: 30.5-38.8); Other exercises (eg, swimming, cycling, keep fit, and bowling): 14 SNPs (F-statistic range: 30.3-61.1). All selected SNPs exhibited strong associations with their respective physical activity exposure traits (F-statistics > 10), indicating minimal risk of weak instrument bias. Detailed data sources and variant IDs are provided in Supplementary Table.

Causal effect of heavy do-it-yourself on functional dyspepsia

Q values of MR-Egger and IVW showed no significant heterogeneity between Heavy DIY and FD (P = 0.564 and 0.590, respectively) and fixed effect model was used. The IVW method indicated a suggestive inverse association (OR = 0.974; 95% CI, 0.954-0.994; P = 0.012) (Table 1). However, the effect direction of the IVW estimate was inconsistent with that of MR-Egger. To enhance robustness, we applied a stricter instrument selection threshold (P ≤ 1×10^-8), retaining 9 SNPs for analysis. Neither MR-Egger nor IVW detected significant heterogeneity (P > 0.05), and the fixed-effect model was again employed. In this study, no significant link was observed between Heavy DIY and FD using the IVW method (OR = 0.975; 95% CI: 0.948-1.003; P = 0.076) (Table 1 and Supplementary Fig. 1A). Horizontal pleiotropy was evaluated via MR-Egger intercept (P = 0.858) and MR-PRESSO (P = 0.560), neither of which indicated the presence of pleiotropic bias or outliers. Leave-one-out sensitivity analysis revealed that rs674094 disproportionately influenced the results; its exclusion resulted in a statistically significant association between Heavy DIY and FD (Supplementary Fig. 2A).

Table 1.

Mendelian Randomization Results for the Relationship Between Physical Activity and Functional Dyspepsia

Exposure	Method	OR (95% CI)	P-value	SNPs	Heterogeneity test (P-value)	Pleiotropy test (P-value)
Heavy DIY before tightening threshold	IVW (fixed)	0.974 (0.954-0.994)	0.012	18	0.590
	MR Egger	1.026 (0.899-1.170)	0.710	18	0.564	0.446
	Weighted median	0.966 (0.938-0.994)	0.019	18
	Simple mode	0.965 (0.915-1.018)	0.214	18
	Weighted mode	0.964 (0.917-1.014)	0.176	18
Heavy DIY after tightening threshold	IVW (fixed)	0.975 (0.948-1.003)	0.076	9	0.565
	MR Egger	1.001 (0.757-1.323)	0.995	9	0.460	0.858
	Weighted median	0.964 (0.929-1.000)	0.052	9
	Simple mode	0.958 (0.902-1.018)	0.202	9
	Weighted mode	0.959 (0.903-1.018)	0.206	9
Light DIY	IVW (fixed)	0.994 (0.972-1.017)	0.600	12	0.816
	MR Egger	0.984 (0.887-1.092)	0.769	12	0.749	0.851
	Weighted median	0.990 (0.961-1.020)	0.519	12
	Simple mode	0.975 (0.929-1.023)	0.318	12
	Weighted mode	0.987 (0.946-1.030)	0.567	12
None of the Heavy DIY or Light DIY	IVW (fixed)	1.059 (0.977-1.149)	0.161	5	0.140
	MR Egger	0.838 (0.194-3.626)	0.829	5	0.065	0.773
	Weighted median	1.037 (0.928-1.158)	0.524	5
	Simple mode	1.015 (0.866-1.191)	0.862	5
	Weighted mode	1.003 (0.855-1.176)	0.976	5
Strenuous sports	IVW (fixed)	0.924 (0.875-0.976)	0.005	6	0.057
	MR Egger	0.790 (0.544-1.146)	0.282	6	0.059	0.44
	Weighted median	0.929 (0.866-0.995)	0.037	6
	Simple mode	0.925 (0.823-1.040)	0.248	6
	Weighted mode	0.925 (0.833-1.026)	0.202	6
Walking for pleasure before tightening threshold	IVW (fixed)	0.975 (0.955-0.996)	0.022	20	0.421
	MR Egger	1.190 (0.896-1.580)	0.246	20	0.477	0.186
	Weighted median	0.979 (0.949-1.010)	0.184	20
	Simple mode	0.994 (0.938-1.052)	0.829	20
	Weighted mode	0.998 (0.943-1.055)	0.935	20
Walking for pleasure after tightening threshold	IVW (fixed)	0.988 (0.962-1.015)	0.377	12	0.208
	MR Egger	1.139 (0.730-1.776)	0.579	12	0.177	0.544
	Weighted median	1.004 (0.966-1.042)	0.854	12
	Simple mode	1.010 (0.948-1.076)	0.761	12
	Weighted mode	1.010 (0.949-1.073)	0.770	12
Other exercises (eg, swimming, cycling, keep fit, and bowling)	IVW (random)	0.964 (0.937-0.993)	0.013	14	0.038
	MR Egger	0.885 (0.709-1.104)	0.299	14	0.035	0.457
	Weighted median	0.966 (0.936-0.997)	0.031	14
	Simple mode	0.970 (0.920-1.023)	0.284	14
	Weighted mode	0.970 (0.926-1.017)	0.225	14

DIY, do-it-yourself; MR, Mendelian randomization; IVW, inverse variance weighted; SNPs, single-nucleotide polymorphisms.

If there was no heterogeneity, a fixed effects model was used, and if heterogeneity was found, a random effects model was used. The tighten threshold of the instrumental variables was P ≤ 1×10^-8.

Causal effect of light do-it-yourself on functional dyspepsia

According to Cochran Q-test, the P-value of MR-Egger was 0.749 and that of IVW was 0.816, indicating no significant heterogeneity, and fixed effect model was used. Our MR analysis revealed no significant causal effect of Light DIY on FD (IVW OR = 0.994; 95% CI: 0.972-1.017; P = 0.600) (Table 1 and Supplementary Fig. 1B). The MR-Egger regression indicated no substantial horizontal pleiotropy (P for intercept = 0.851), and the MR-PRESSO test detected no outlier SNPs (P = 0.812). Sensitivity analysis using the leaveone-out method confirmed that no individual SNP exerted a dominant effect on the overall association (Supplementary Fig. 2B).

Causal effect of none of the heavy do-it-yourself or light do-it-yourself on functional dyspepsia

According to Cochran Q-test, the P-value of MR-Egger was 0.065 and that of IVW was 0.14, indicating no significant heterogeneity, and fixed effect model was used. We found that none of the Heavy DIY or Light DIY had no causality on FD (IVW OR = 1.059; 95% CI, 0.977-1.149; P = 0.161) (Table 1 and Supplementary Fig. 1C). MR-Egger regression detected no evidence of directional pleiotropy (P intercept = 0.773), and the MR-PRESSO test confirmed the absence of outlier SNPs (P = 0.158). Sensitivity analysis using the leave-one-out method indicated that no single genetic variant disproportionately influenced the results (Supplementary Fig. 2C).

Causal effect of strenuous sports on functional dyspepsia

The MR analysis showed no evidence of significant heterogeneity in the causal estimate of strenuous sports on FD (Cochran’s Q-test: P = 0.059 for MR-Egger, P = 0.057 for IVW). Therefore, a fixed-effect model was applied. The IVW method revealed a potential inverse association of strenuous sports on FD (OR = 0.924; 95% CI: 0.875-0.976; P = 0.005) (Table 1 and Supplementary Fig. 1D), with consistent directionality across other MR methods. No significant horizontal pleiotropy was detected (MR-Egger intercept P = 0.44), and the MR-PRESSO global test (P = 0.094) confirmed the absence of outlier variants. However, leave-one-out sensitivity analysis identified rs12571549 as a highly influential variant; its exclusion nullified the association between strenuous sports and FD (Supplementary Fig. 2D), suggesting that the overall effect may be driven by this single genetic instrument.

Causal effect of walking for pleasure on functional dyspepsia

No significant heterogeneity was observed between walking for pleasure and FD, as indicated by the Q values of MR-Egger (P = 0.477) and IVW (P = 0.421), supporting the use of a fixed-effect model. The IVW method revealed a suggestive inverse association (OR = 0.975; 95% CI, 0.955-0.996; P = 0.022) (Table 1). However, the effect direction of the IVW estimate was inconsistent with that of MR-Egger. To enhance robustness, we applied a stricter instrument selection threshold (P ≤ 1×10^-8), retaining 12 SNPs for analysis. Again, no significant heterogeneity was detected (MR-Egger and IVW Q-test P > 0.05), and a fixed-effect model was employed. No significant association was observed for pleasure walking in relation to FD (IVW OR = 0.988; 95% CI, 0.962-1.015; P = 0.377) (Table 1 and Supplementary Figure 1E). Furthermore, the MR-Egger intercept test provided no evidence of horizontal pleiotropy (P = 0.544), and MR-PRESSO analysis confirmed the absence of outlier variants (P = 0.242). The robustness of the results was confirmed by a leave-one-out sensitivity analysis, which showed that they were not driven by any individual SNP (Supplementary Fig. 2E).

Causal effect of other exercises (eg, swimming, cycling, keep fit, and bowling) on functional dyspepsia

According to Cochran Q-test, the P-value of MR-Egger was 0.035 and that of IVW was 0.038, indicating the existence of heterogeneity, and random effect model was used. The IVW method demonstrated a suggestive protective effect of other exercises on FD (OR = 0.964; 95% CI, 0.937-0.993; P = 0.013) (Table 1 and Supplementary Fig. 1F). A suggestive association was also observed using the weighted median estimator (OR = 0.966; 95% CI, 0.936-0.997; P = 0.031). MR-Egger regression detected no evidence of directional pleiotropy (P intercept = 0.457), and the MR-PRESSO method identified no outlier SNPs. Leave-one-out sensitivity analysis revealed that no SNP was found to exert a disproportionate impact on the overall results (Supplementary Fig. 2F).

Randomized Controlled Study

Baseline of patient characteristics

The MR study analysis revealed that strong sports and other exercises (eg, swimming, cycling, keep fit, and fishing) were associated with a reduced risk of FD at the genetic level. Notably, the keep fit exercise recommended by the American College of Sports Medicine refers to moderate intensity exercise.¹⁹ Supporting this, prior high-quality research has classified other exercises (eg: swimming, cycling, keep fit, fishing) as moderate-intensity exercise.²⁶ This categorization aligns with previous literature that, based on Metabolic Equivalent of Task values, classified a similar composite variable including these activities as representing moderate-intensity physical activity. The evidence indicates that moderate-intensity exercise could mitigate the risk of developing FD. To further investigate this association, we conducted a RCT assessing the therapeutic effects of moderate-intensity exercise on FD symptoms.

From August 2020 to December 2021, a total of 260 patients who met the inclusion and exclusion criteria were collected. The block randomization method was used to randomly assign the patients to the moderate intensity exercise group and the control group in a 1:1 ratio with a block length of 4. During the 3-month intervention period, a total of 33 patients were lost to follow-up, including 16 patients in the exercise group and 17 patients in the control group. Finally, 114 patients (57.89% patients were female, average age was 44.36 ± 11.44 years) in the exercise group completed the study, and 113 patients (56.64% patients were female, average age was 47.35 ± 11.83 years) in the control group completed the study. The research flowchart was shown in Figure 1.

Figure.

Flow chart of the study.

Among FD subtypes, overlapping syndrome was the most prevalent in the exercise group (39.23%, 51/130), whereas epigastric pain syndrome was most common in the control group (40.00%, 52/130). Comparative analysis of baseline characteristics—including sex, age, body mass index, location, marriage, currently smoking, currently drinking, education level, daily exercise duration, and FD subtype—showed no statistically significant differences between groups (P > 0.05) (Table 2). These findings confirm that the 2 groups were well-balanced at baseline.

Table 2.

The Comparison of Baseline Characteristics Between Exercise Group and Control Group

	Exercise group (n = 130)	Control group (n = 130)	P-value
Sex			0.912
Male	56 (43.08%)	58 (44.62%)
Female	74 (56.92%)	72 (55.38%)
Age (yr)	45.14 ± 11.45	46.01 ± 12.26	0.542
BMI (kg/m2)	22.09 ± 3.05	22.12 ± 3.10	0.938
Location			0.224
Shaanxi	109 (83.85%)	101 (77.69%)
Other	21 (16.15%)	29 (22.31%)
Marriage			0.841
Unmarried	14 (10.77%)	13 (10.00%)
Married	116 (89.23%)	117 (90.00%)
Currently smoking			0.874
No	106 (81.54%)	107 (82.31%)
Yes	24 (18.46%)	23 (17.69%)
Currently drinking			0.620
No	108 (83.08%)	105 (80.77%)
Yes	22 (16.92%)	25 (19.23%)
Education level			0.785
Junior high school and below	53 (40.77%)	51 (39.23%)
High school	24 (18.46%)	26 (20.00%)
College and above	53 (40.77%)	53 (40.77%)
Daily exercise time			0.184
< 1/2 hr	68 (52.31%)	58 (44.62%)
1/2 hr < ≤ 1 hr	25 (19.23%)	28 (21.54%)
1 hr < ≤ 2 hr	16 (12.31%)	20 (15.38%)
> 2 hr	21 (16.15%)	24 (18.46%)
FD subtype			0.212
Postprandial distress syndrome	31 (23.85%)	40 (30.77%)
Epigastric pain syndrome	48 (36.92%)	52 (40.00%)
Overlapping syndromes	51 (39.23%)	38 (29.23%)

BMI, body mass index; FD, functional dyspepsia.

Values are expressed as the mean ± SD or n (%).

Adequate relief rate of dyspeptic symptoms

At the 12-week follow-up, the adequate relief rate of overall dyspeptic symptoms was significantly higher in the exercise group than in the control group (46.9% [61/130] vs 30.8% [40/130], P = 0.008).

Eight-item dyspepsia symptom score

At the 12-week follow-up, the exercise group demonstrated significantly greater reductions in the severity of epigastric pain, epigastric burning sensation, postprandial fullness, abdominal distension, and nausea compared to the control group (P < 0.05). In contrast, there was no significant difference in the decrease in severity scores of early satiety, belching, and vomiting in the exercise group compared with the control group (P > 0.05). Notably, the total dyspepsia symptom score decreased more markedly in the exercise group than in the control group (P < 0.05) (Table 3).

Table 3.

Comparison of 8-Item Dyspepsia Symptom Score, Short-form Leeds Dyspepsia Questionnaire Score, Short-form Nepean Dyspepsia Index Questionnaire Score, and Hospital Anxiety and Depression Scale Score Between Exercise Group and Control Group

	Exercise group (n = 130)			Control group (n = 130)			P for difference in change
	Baseline	12 wk	Change	Baseline	12 wk	Change
Eight-item dyspepsia symptom score
Epigastric pain	1.26 ± 1.03	0.65 ± 0.79	-0.61 ± 0.76	1.16 ± 0.95	1.05 ± 0.93	-0.11 ± 0.49	< 0.0001
Epigastric burning sensation	0.61 ± 0.95	0.28 ± 0.64	-0.33 ± 0.71	0.46 ± 0.84	0.35 ± 0.75	-0.11 ± 0.42	0.0023
Postprandial fullness	1.23 ± 1.05	0.69 ± 0.80	-0.54 ± 0.79	1.09 ± 0.96	1.01 ± 0.94	-0.08 ± 0.68	< 0.0001
Early satiety	0.34 ± 0.75	0.08 ± 0.34	-0.26 ± 0.69	0.33 ± 0.72	0.21 ± 0.59	-0.12 ± 0.48	0.0616
Belching	0.25 ± 0.66	0.12 ± 0.41	-0.13 ± 0.55	0.42 ± 0.79	0.31 ± 0.69	-0.12 ± 0.48	0.8093
Abdominal distension	1.07 ± 1.01	0.52 ± 0.77	-0.55 ± 0.85	1.03 ± 0.93	0.89 ± 0.89	-0.14 ± 0.73	< 0.0001
Nausea	0.36 ± 0.77	0.08 ± 0.37	-0.28 ± 0.70	0.28 ± 0.67	0.16 ± 0.53	-0.12 ± 0.46	0.0216
Vomiting	0.17 ± 0.54	0.05 ± 0.27	-0.12 ± 0.47	0.11 ± 0.42	0.06 ± 0.32	-0.05 ± 0.27	0.106
Total Eight-item dyspepsia symptom score	5.29 ± 2.79	2.47 ± 2.17	-2.82 ± 2.44	4.88 ± 2.43	4.05 ± 2.40	-0.84 ± 1.67	< 0.0001
SF-LDQ score
Indigestion frequency	3.11 ± 1.02	1.98 ± 1.42	-1.14 ± 1.24	3.19 ± 0.84	2.84 ± 0.90	-0.35 ± 0.63	< 0.0001
Indigestion interference with normal activities	3.08 ± 1.05	1.95 ± 1.40	-1.13 ± 1.29	3.13 ± 0.96	2.79 ± 0.93	-0.34 ± 0.74	< 0.0001
Heartburn frequency	0.94 ± 1.52	0.52 ± 1.15	-0.42 ± 1.04	0.57 ± 1.25	0.42 ± 1.06	-0.15 ± 0.83	0.0217
Heartburn interference with normal activities	0.88 ± 1.48	0.47 ± 1.06	-0.42 ± 1.15	0.52 ± 1.21	0.41 ± 1.03	-0.11 ± 0.77	0.0118
Regurgitation frequency	1.05 ± 1.50	0.36 ± 0.92	-0.68 ± 1.26	0.88 ± 1.44	0.70 ± 1.27	-0.18 ± 0.80	0.0002
Regurgitation interference with normal activities	0.92 ± 1.45	0.28 ± 0.79	-0.65 ± 1.23	0.73 ± 1.36	0.63 ± 1.23	-0.10 ± 1.04	0.0001
Nausea frequency	0.40 ± 0.95	0.07 ± 0.40	-0.33 ± 0.87	0.38 ± 0.97	0.25 ± 0.80	-0.14 ± 0.61	0.0393
Nausea interference with normal activities	0.35 ± 0.90	0.07 ± 0.40	-0.28 ± 0.80	0.38 ± 0.97	0.22 ± 0.73	-0.17 ± 0.69	0.215
Symptom frequency score	5.50 ± 2.63	2.93 ± 2.34	-2.57 ± 2.35	5.03 ± 2.26	4.21 ± 2.15	-0.82 ± 1.42	< 0.0001
Symptom interference with lifestyle score	5.25 ± 2.49	2.77 ± 2.16	-2.48 ± 2.28	4.76 ± 2.26	4.05 ± 2.08	-0.72 ± 1.61	< 0.0001
Total SF-LDQ score	10.75 ± 5.02	5.70 ± 4.44	-5.05 ± 4.47	9.79 ± 4.43	8.25 ± 4.17	-1.54 ± 2.89	< 0.0001
SF-NDI score
Tension/anxiety	4.32 ± 2.31	3.12 ± 1.46	-1.21 ± 1.56	4.62 ± 2.49	4.39 ± 2.38	-0.23 ± 0.70	< 0.0001
Interference	3.43 ± 2.16	2.51 ± 1.12	-0.92 ± 1.65	3.60 ± 2.19	3.45 ± 2.01	-0.15 ± 0.54	< 0.0001
Eat/drink	4.20 ± 2.09	2.99 ± 1.25	-1.21 ± 1.67	4.40 ± 2.28	4.08 ± 2.07	-0.32 ± 0.95	< 0.0001
Knowledge/control	4.62 ± 2.11	2.65 ± 0.91	-6.44 ± 4.00	4.44 ± 2.20	3.77 ± 1.76	-5.06 ± 3.42	0.0032
Work/study	3.54 ± 2.16	2.66 ± 1.21	-0.88 ± 1.54	3.59 ± 2.18	3.32 ± 1.99	-0.28 ± 0.81	0.0001
Total SF-NDI score	20.12 ± 6.00	13.93 ± 3.65	-6.18 ± 4.92	20.65 ± 7.30	19.00 ± 6.86	-1.65 ± 2.12	< 0.0001
HADS score
Anxiety score	9.01 ± 4.08	8.34 ± 3.49	-0.67 ± 1.44	9.05 ± 3.06	8.26 ± 2.65	-0.78 ± 1.15	0.476
Depression score	9.39 ± 4.12	8.92 ± 3.80	-0.48 ± 1.12	9.53 ± 3.25	9.12 ± 3.04	-0.42 ± 0.74	0.601
Total HADS score	18.40 ± 7.79	17.25 ± 6.93	-1.15 ± 2.32	18.58 ± 5.84	17.38 ± 5.31	-1.20 ± 1.43	0.822

SF-LDQ, Short-form leeds dyspepsia questionnaire; SF-NDI, Short-form Nepean dyspepsia index questionnaire; HADS, hospital anxiety and depression scale.

Values are expressed as the mean ± SD.

Short-form leeds dyspepsia questionnaire score

At the 12-week follow-up, the exercise group demonstrated significantly greater reductions in dyspeptic symptom frequency (-2.57 ± 2.35 vs -0.82 ± 1.42, P < 0.0001), symptom interference with lifestyle (-2.48 ± 2.28 vs -0.72 ± 1.61, P < 0.0001), and total SF-LDQ score (-5.05 ± 4.47 vs -1.54 ± 2.89, P < 0.0001) compared to the control group (Table 3). All differences were statistically significant.

Short-form nepean dyspepsia index questionnaire score

At the 12-week follow-up, the exercise group demonstrated significantly greater reductions in SF-NDI total scores compared to the control group (P < 0.05). Additionally, significant improvements were observed across multiple SF-NDI subdomains, including tension/anxiety, interference with daily activities, eat/drink behaviors, knowledge/control, and work/study performance (all P < 0.05) (Table 3).

Hospital anxiety and depression scale score

At the 12-week follow-up, intention-to-treat analysis revealed no statistically significant differences between the exercise and control groups in the reduction of anxiety scores (-0.67 ± 1.44 vs -0.78 ± 1.15, P = 0.476), depression scores (-0.48 ± 1.12 vs -0.42 ± 0.74, P = 0.601), or total HADS scores (-1.15 ± 2.32 vs -1.20 ± 1.43, P = 0.822) (Table 3).

Adverse reactions

Throughout the 12-week intervention period, no serious adverse events were reported in either the exercise or the control group. These findings suggest that moderate-intensity exercise over 12 weeks was well-tolerated and safe in the studied population.

DISCUSSION

Our MR analysis revealed that strong sports and other exercises (eg, swimming, cycling, keep fit, and fishing) suggested a potential inverse association with FD at the genetic level. Notably, the keep fit exercise recommended by the American College of Sports Medicine refers to moderate intensity exercise.¹⁹ Supporting this, prior high-quality research has classified other exercises (eg, swimming, cycling, keep fit, and fishing) as moderate-intensity exercise.²⁶ The evidence indicates that moderate-intensity exercise could mitigate the risk of developing FD. Subsequently, our prospective RCT demonstrated that moderate-intensity exercise not only increased the remission rate of FD but also alleviated symptom severity and improved patients’ quality of life. This study provides novel evidence, integrating genetic and clinical data, to support physical activity as a viable therapeutic intervention for FD. Furthermore, given its cost-effectiveness, accessibility, and lack of reliance on expensive resources, exercise represents a practical and economical treatment option for FD management.

The association between physical activity and FD remains underexplored. While some experts recommended physical activity as a primary therapy for FD,²⁷ robust evidence supporting its efficacy in alleviating FD symptoms is lacking. Several observational studies have reported an inverse relationship between exercise and FD. According to a study by Ford et al²⁸ involving 3160 participants via mailed survey, individuals with FD tended to engage in less physical exercise, and after adjusting for age, sex, body mass index, smoking, and excluding overlap with IBS, FD patients had lower frequency of moderate intensity exercise compared with control group. An internet questionnaire survey by Miwa et al²⁹ showed that the frequency of exercise of subjects with FD, IBS or both was significantly lower than that of the control subjects and suggested that exercise might be effective in improving the symptoms of FGID. An online survey of 2725 people in India found that uninvestigated dyspepsia may be related to lack of exercise.³⁰ Furukawa et al³¹ conducted a cross-sectional study in a population of Japanese college students. Their research demonstrated that exercise frequency and intensity may be inversely related to the occurrence of FD. However, these studies were limited by their reliance on the Rome III criteria and cross-sectional designs. A small RCT by Rane et al. (n = 36 per group) suggested that aerobic exercise as adjuvant therapy may improve FD symptoms,⁷ but the short intervention duration and limited sample size constrain its generalizability. In contrast, our study included 130 patients per group and demonstrated through 12-week follow-up that moderate-intensity exercise exerts a measurable therapeutic effect on FD.

The role of physical activity in the pathogenesis of FD remains unclear. In patients with FD, duodenal microbiota diversity is reduced, and regular physical exercise may alter the intestinal microbiome, increasing its diversity and possibly increasing the number of beneficial bacteria.^32-34 Physical activity has been shown to increase the flow of endogenous endorphins in the blood, which can reduce the severity of symptoms.³⁵ Meanwhile, endorphins also interact with receptors in the brain to reduce pain perception. 36 Physical exercise triggers the release of endorphins in the brain, which aids in stress relief and improves mood.³⁷ Research shows that endogenous cannabinoid can cause the release of hypothalamic endorphins,³⁸ and physical exercise can reduce depression symptoms by increasing the activity of endogenous cannabinoid to regulate the amygdala. Previous studies have shown that the benefits of exercise are systemic. Exercise can induce the release of bioactive factors into the blood circulation, resulting in systemic beneficial effects.³⁹ Interestingly, while gastrointestinal symptoms and quality of life improved significantly, we did not observe a corresponding reduction in HADS. This may suggest that the therapeutic benefit of moderate-intensity exercise in FD operates primarily through mechanisms directly affecting gastrointestinal physiology (eg, motility, visceral sensitivity, and microbiota) rather than, or in addition to, a secondary effect mediated by improved mood. Alternatively, the intervention duration or intensity may have been insufficient to induce measurable changes in generalized anxiety and depression levels in this patient cohort.

There were several limitations in the study. First, the MR analysis relied on physical activity phenotypes derived from self-reported questionnaires in the UK Biobank. Measurement error and misclassification inherent to self-report may attenuate the estimated causal associations, which is a common limitation when utilizing publicly available GWAS summary statistics for exposure definition. Also, the MR analysis utilized GWAS summary statistics for both exposure and outcome derived from the UK Biobank, leading to potential sample overlap. While we employed stringent genetic instrument selection criteria, this overlap may still introduce some bias and should be considered when interpreting the MR results. Second, the MR analysis included a limited number of SNPs assessing the causal relationship between strenuous sports and FD, which may have led to an overrepresentation of rs12571549 in the leave-one-out sensitivity analysis. Third, as with most FD treatment studies, a placebo effect could not be entirely excluded. Although physical activity interventions cannot be blinded, we minimized bias by having the control group receive weekly phone consultations to reinforce habitual lifestyle adherence. Forth, the MR analysis was conducted using genetic data from individuals of European ancestry, whereas the RCT was performed in a Chinese population. Differences in genetic background, lifestyle, and environmental factors between these populations may limit the direct comparability and generalizability of the integrated findings. Future multi-ethnic studies are needed to confirm the trans-ethnic relevance of these associations. Finally, consistent with previous high-quality studies,¹⁵ the exercise intervention was not directly supervised but instead monitored via telephone follow-up, and concomitant medication use was not restricted—reflecting real-world clinical practice. While this design enhances external validity, it may introduce variability in treatment adherence.

In conclusion, our study supports the beneficial role of physical activity in reducing the risk of FD in the general population. Strong sports and other exercises (eg, swimming, cycling, keep fit, and fishing), may be suggestively associated with a reduced risk of FD. Additionally, moderate-intensity exercise appears to alleviate gastrointestinal symptoms, enhance the rate of adequate symptom relief, and improve overall quality of life in FD patients. Further well-designed clinical trials are needed to validate these findings and elucidate the potential therapeutic benefits of physical activity for FD management.

SUPPLEMENTARY MATERIAL

Note: To access the supplementary table and figures mentioned in this article, visit the online version of Journal of Neurogastroenterology and Motility at http://www.jnmjournal.org/, and at http://doi.org/10.5056/jnm25159.